Atomic Detail Research Articles

Masking the transmembrane region of the amyloid β precursor protein as a safe means to lower amyloid β production.

Reducing brain levels of both soluble and insoluble forms of amyloid beta (Aβ) remains the primary goal of most therapies that target Alzheimer's disease (AD). However, no treatment has so far resulted in patient benefit, and clinical trials of the most promising drug candidates have generally failed due to significant adverse effects. This highlights the need for safer and more selective ways to target and modulate Aβ biogenesis. Peptide technology has advanced to allow reliable synthesis, purification, and delivery of once-challenging hydrophobic sequences. This is opening up new routes to target membrane processes associated with disease. Here we deploy a combination of atomic detail molecular dynamics (MD) simulations, living-cell Förster resonance energy transfer (FRET), and in vitro assays to elucidate the atomic-detail dynamics, molecular mechanisms, and cellular activity and selectivity of a membrane-active peptide that targets the Aβ precursor protein (APP). We demonstrate that Aβ biogenesis can be downregulated selectively using an APP occlusion peptide (APPOP). APPOP inhibits Aβ production in a dose-dependent manner, with a mean inhibitory concentration (IC50) of 450nM toward exogenous APP and 50nM toward endogenous APP in primary rat cortical neuronal cultures. APPOP does not impact the γ-secretase cleavage of Notch-1, or exhibit toxicity toward cultured primary rat neurons, suggesting that it selectively shields APP from proteolysis. Drugs targeting AD need to be given early and for very long periods to prevent the onset of clinical symptoms. This necessitates being able to target Aβ production precisely and without affecting the activity of key cellular enzymes such as γ-secretase for other substrates. Peptides offer a powerful way for targeting key pathways precisely, thereby reducing the risk of adverse effects. Here we show that protecting APP from proteolytic processing offers a promising route to safely and specifically lower Aβ burden. In particular, we show that the amyloid pathway can be targeted directly and specificically. This reduces the risk of off-target effects and paves the way for a safe prophylactic treatment.

Energy transfer during hydrogen atom collisions with surfaces

Hydrogen adsorption is an important cornerstone of many natural and technological processes, yet its atomistic details are not fully understood. Here, we review recent advances to understand the interactions between atomic hydrogen and substrates that lead to adsorption and discuss open questions.

Molecular Simulations of Liquid Jet Explosions and Shock Waves Induced by X-Ray Free-Electron Lasers.

X-ray free-electron lasers (XFELs) produce x-ray pulses with high brilliance and short pulse duration. These properties enable structural investigations of biomolecular nanocrystals, and they allow one to resolve the dynamics of biomolecules down to the femtosecond timescale. Liquid jets are widely used to deliver samples into the XFEL beam. The impact of the x-ray pulse leads to vaporization and explosion of the liquid jet, while the expanding gas triggers the formation of shock wave trains traveling along the jet, which may affect biomolecular samples before they have been probed. Here, we used molecular dynamics simulations to reveal the structural dynamics of shock waves after an x-ray impact. Analysis of the density and temperature in the jet revealed shock waves that form close to the explosion center, travel along the jet with supersonic velocities, and decay exponentially with an attenuation length proportional to the jet diameter. A trailing shock wave formed after the first shock wave, similar to the shock wave trains in experiments. High shock wave velocities in our simulations are compatible with the phenomenon of "fast sound," as emerging at large sound frequencies. Although using purely classical models in the simulations, the resulting explosion geometry and shock wave dynamics closely resemble experimental findings, and they highlight the importance of atomistic details for modeling shock wave attenuation.

Mending cracks atom-by-atom in rutile TiO2 with electron beam radiolysis

Rich electron-matter interactions fundamentally enable electron probe studies of materials such as scanning transmission electron microscopy (STEM). Inelastic interactions often result in structural modifications of the material, ultimately limiting the quality of electron probe measurements. However, atomistic mechanisms of inelastic-scattering-driven transformations are difficult to characterize. Here, we report direct visualization of radiolysis-driven restructuring of rutile TiO2 under electron beam irradiation. Using annular dark field imaging and electron energy-loss spectroscopy signals, STEM probes revealed the progressive filling of atomically sharp nanometer-wide cracks with striking atomic resolution detail. STEM probes of varying beam energy and precisely controlled electron dose were found to constructively restructure rutile TiO2 according to a quantified radiolytic mechanism. Based on direct experimental observation, a “two-step rolling” model of mobile octahedral building blocks enabling radiolysis-driven atomic migration is introduced. Such controlled electron beam-induced radiolytic restructuring can be used to engineer novel nanostructures atom-by-atom.

Biphasic NMR of Hyperpolarized Suspensions-Real-Time Monitoring of Solute-to-Solid Conversion to Watch Materials Grow.

Nuclear magnetic resonance (NMR) spectroscopy is a key method for the determination of molecular structures. Due to its intrinsically high (i.e., atomistic) resolution and versatility, it has found numerous applications for investigating gases, liquids, and solids. However, liquid-state NMR has found little application for suspensions of solid particles as the resonances of such systems are excessively broadened, typically beyond the detection threshold. Herein, we propose a route to overcoming this critical limitation by enhancing the signals of particle suspensions by >3.000-fold using dissolution dynamic nuclear polarization (d-DNP) coupled with rapid solid precipitation. For the proof-of-concept series of experiments, we employed calcium phosphate (CaP) as a model system. By d-DNP, we boosted the signals of phosphate 31P spins before rapid CaP precipitation inside the NMR spectrometer, leading to the inclusion of the hyperpolarized phosphate into CaP-nucleated solid particles within milliseconds. With our approach, within only 1 s of acquisition time, we obtained spectra of biphasic systems, i.e., micrometer-sized dilute solid CaP particles coexisting with their solution-state precursors. Thus, this work is a step toward real-time characterization of the solid-solution equilibrium. Finally, integrating the hyperpolarized data with molecular dynamics simulations and electron microscopy enabled us to shed light on the CaP formation mechanism in atomistic detail.

Calcium binding and permeation in TRPV channels: Insights from molecular dynamics simulations.

Some calcium channels selectively permeate Ca2+, despite the high concentration of monovalent ions in the surrounding environment, which is essential for many physiological processes. Without atomistic and dynamical ion permeation details, the underlying mechanism of Ca2+ selectivity has long been an intensively studied, yet controversial, topic. This study takes advantage of the homologous Ca2+-selective TRPV6 and non-selective TRPV1 and utilizes the recently solved open-state structures and a newly developed multisite calcium model to investigate the ion binding and permeation features in TRPV channels by molecular dynamics simulations. Our results revealed that the open-state TRPV6 and TRPV1 show distinct ion binding patterns in the selectivity filter, which lead to different ion permeation features. Two Ca2+ ions simultaneously bind to the selectivity filter of TRPV6 compared with only one Ca2+ in the case of TRPV1. Multiple Ca2+ binding at the selectivity filter of TRPV6 permeated in a concerted manner, which could efficiently block the permeation of Na+. Cations of various valences differentiate between the binding sites at the entrance of the selectivity filter in TRPV6. Ca2+ preferentially binds to the central site with a higher probability of permeation, repelling Na+ to a peripheral site. Therefore, we believe that ion binding competition at the selectivity filter of calcium channels, including the binding strength and number of binding sites, determines Ca2+ selectivity under physiological conditions.

Elucidation of binding mechanism, affinity, and complex structure between mWT1 tumor-associated antigen peptide and HLA-A*24:02.

We have applied our advanced computational and experimental methodologies to investigate the complex structure and binding mechanism of a modified Wilms' Tumor 1 (mWT1) protein epitope to the understudied Asian-dominant allele HLA-A*24:02 (HLA-A24) in aqueous solution. We have applied our developed multicanonical molecular dynamics (McMD)-based dynamic docking method to analyze the binding pathway and mechanism, which we verified by comparing the highest probability structures from simulation with our experimentally solved x-ray crystal structure. Subsequent path sampling MD simulations elucidated the atomic details of the binding process and indicated that first an encounter complex is formed between the N-terminal's positive charge of the 9-residue mWT1 fragment peptide and a cluster of negative residues on the surface of HLA-A24, with the major histocompatibility complex (MHC) molecule preferring a predominantly closed conformation. The peptide first binds to this closed MHC conformation, forming an encounter complex, after which the binding site opens due to increased entropy of the binding site, allowing the peptide to bind to form the native complex structure. Further sequence and structure analyses also suggest that although the peptide loading complex would help with stabilizing the MHC molecule, the binding depends in a large part on the intrinsic affinity between the MHC molecule and the antigen peptide. Finally, our computational tools and analyses can be of great benefit to study the binding mechanism of different MHC types to their antigens, where it could also be useful in the development of higher affinity variant peptides and for personalized medicine.

Plant derived active compounds of ayurvedic neurological formulation, Saraswatharishta as a potential dual leucine zipper kinase inhibitor: an in-silico study

Recent findings have highlighted the essential role of dual leucine zipper kinase (DLK) in neuronal degeneration. Saraswatharishta (SWRT), an ayurvedic formulation utilized in traditional Indian medicine, has demonstrated effectiveness in addressing neurodegenerative diseases. Herein, we aim to delve into the atomistic details of the mode of action of phytochemicals present in SWRT against DLK. Our screening process encompassed over 500 distinct phytochemicals derived from the main ingredients of the SWRT formulation. Through a comparative analysis of docking scores and relative poses, we successfully identified four novel compounds, which underwent further investigation via 2 × 500 ns long molecular dynamics (MD) simulations. Among the top four compounds, CID16066851 sourced from the Acorus calamus displayed the most stable complex with DLK. The molecular mechanics Poisson − Boltzmann surface area (MM-PBSA) calculations highlighted the significance of electrostatic and van der Waals interactions in the binding recognition process. Additionally, we identified key residues, namely Phe192, Leu243, Val139, and Leu141, as hotspots that predominantly govern the DLK-inhibitor interaction. Notably, the leading compounds are sourced from the Acorus calamus, Syzygium aromaticum, Zingiber officinale, and Anethum sowa plants present in the SWRT formulation. Overall, the findings of our study hold promise for future drug development endeavors combating neurodegenerative conditions. Communicated by Ramaswamy H. Sarma

Unraveling the Atomic Shuffles of Twinning Nucleation in Hexagonal Close-Packed Rhenium Nanocrystals.

Reining in deformation twinning is crucial for the mechanical properties of hexagonal close-packed (HCP) metals and hinges on an explicit understanding of the twinning nucleation mechanism. Unfortunately, it is often suggested rather than conclusively demonstrated that twinning nucleation can be mediated by pure atomic shuffles. Herein, by utilizing in situ high-resolution transmission electron microscopy, we have dissected the atomic shuffling mechanism during the {101̅2} twinning nucleation in rhenium nanocrystals, which revealed the emergence of an intermediate body-centered tetragonal (BCT) structure. Specifically, the double-layered prismatic planes initially shuffle into single-layered {11̅0}BCT planes; subsequently, adjacent {22̅0}BCT planes shuffle in opposite directions to form the basal planes of the twin embryo. This shuffling mechanism is further corroborated by molecular dynamic simulations. The finding provides direct evidence of shuffle-dominated twinning nucleation with atomic details that may lead to better control of this critical twinning mode in HCP metals.

Heterogeneity in M. tuberculosis β-lactamase inhibition by Sulbactam

For decades, researchers have elucidated essential enzymatic functions on the atomic length scale by tracing atomic positions in real-time. Our work builds on possibilities unleashed by mix-and-inject serial crystallography (MISC) at X-ray free electron laser facilities. In this approach, enzymatic reactions are triggered by mixing substrate or ligand solutions with enzyme microcrystals. Here, we report in atomic detail (between 2.2 and 2.7 Å resolution) by room-temperature, time-resolved crystallography with millisecond time-resolution (with timepoints between 3 ms and 700 ms) how the Mycobacterium tuberculosis enzyme BlaC is inhibited by sulbactam (SUB). Our results reveal ligand binding heterogeneity, ligand gating, cooperativity, induced fit, and conformational selection all from the same set of MISC data, detailing how SUB approaches the catalytic clefts and binds to the enzyme noncovalently before reacting to a trans-enamine. This was made possible in part by the application of singular value decomposition to the MISC data using a program that remains functional even if unit cell parameters change up to 3 Å during the reaction.

Cryo-EM structure determination of small therapeutic protein targets at 3 Å-resolution using a rigid imaging scaffold

Cryoelectron microscopy (Cryo-EM) has enabled structural determination of proteins larger than about 50 kDa, including many intractable by any other method, but it has largely failed for smaller proteins. Here, we obtain structures of small proteins by binding them to a rigid molecular scaffold based on a designed protein cage, revealing atomic details at resolutions reaching 2.9 Å. We apply this system to the key cancer signaling protein KRAS (19 kDa in size), obtaining four structures of oncogenic mutational variants by cryo-EM. Importantly, a structure for the key G12C mutant bound to an inhibitor drug (AMG510) reveals significant conformational differences compared to prior data in the crystalline state. The findings highlight the promise of cryo-EM scaffolds for advancing the design of drug molecules against small therapeutic protein targets in cancer and other human diseases.

On the Role of Temperature in the Depolymerization of PET by FAST-PETase: an Atomistic Point of View on Possible Active-Site Pre-Organization and Substrate-Destabilization Effect.

Enzyme FAST-PETase, recently obtained by machine learning approach, can depolymerize polyethylene terephthalate (PET), a synthetic resin employed in plastics and in clothing fibers. Therefore it represents a promising solution for the recycling of PET-based materials. In the present study, a model of PET was adopted to describe the substrate, and all-atoms classical molecular dynamics (MD) simulations on apo- and substrate bound- FAST-PETase were carried out at 30°C and 50°C, to provide atomistic details on the binding step of the catalytic cycle. Comparative analysis sheds light on the interactions occurring between the FAST-PETase and 4PET at 50 °C, the optimal working condition of the enzyme. Pre-organization of the enzyme active and binding sites has been highlighted while MD of FAST-PETase:4PET pointed out on the occurrence of solvent-inaccessible conformations of the substrate promoted by the enzyme. Indeed, neither of these conformations were observed during MD of substrate alone in solution performed at 30 °C, 50 °C and 150 °C. The analysis led us to propose that, at 50 °C, the FAST-PETase is pre-organized to bind the PET and that the interactions occurring in the binding site can promote more reactive conformation of PET substrate, thus enhancing the catalytic activity of the enzyme.

Multiscale Modeling of Heterogeneous Interfaces in All Solid-State Batteries

Ion transport and degradation at the solid-electorlyte/cathode interfaces dictate the performance of all solid-state batteries. For example, it has been routinely observed that in many superionic oxide conductors, Li-ion transport kinetics are orders of magnitude slower at the interfaces, such as grain boundaries or interfaces with common electrode materials, compared to the bulk phases. Upon electrochemical cycling, interdiffusion and/or electrolyte/cathode decomposition at these interfaces can further deteriorate Li-ion transport kinetics. Therefore, the ability to understand, predict and precisely control ion transport kinetics and the degree of interfacial degradation are critical and would have a transformative impact on the development of solid-state battery technologies for practical applications. Here I will address the need of a multiscale and multiphysics modeling framework to understand Li-ion transport mechanisms at complex interfaces and predict the kinetics of Li-ion transport in solid-electrolyte/cathode composite architectures that are relevant for all solid-state Li batteries. I will also emphasize the importance of using a closely coupled theory-experiment approach to reveal the atomic details of the interfaces and unravel the degradation mechanisms of interfaces and its impact on Li-ion transport. I will use the garnet-type solid-state electrotype and lithium cobalt oxide cathode as an example to demonstrate the predictive power of such a multiscale, multiphysics modeling framework towards establishing the correlations between ion transport and local atomic structure, chemical environment, and mesoscopic microstructure features in the cathode/electrolyte composites.This work was sponsored by the Office of Energy Efficiency and Renewable Energy, Vehicle Technologies Office and was performed under the auspices of the U.S. Department of Energy by Lawrence Livermore National Laboratory under Contract DE-AC52-07NA27344. An award of computer time was provided by the Innovative and Novel Computational Impact on Theory and Experiment (INCITE) program. This research used resources of the Argonne Leadership Computing Facility, which is a DOE Office of Science User Facility supported under Contract DE-AC02-06CH11357.

Exploring rigid-backbone protein docking in biologics discovery: a test using the DARPin scaffold.

Accurate protein-protein docking remains challenging, especially for artificial biologics not coevolved naturally against their protein targets, like antibodies and other engineered scaffolds. We previously developed ProPOSE, an exhaustive docker with full atomistic details, which delivers cutting-edge performance by allowing side-chain rearrangements upon docking. However, extensive protein backbone flexibility limits its practical applicability as indicated by unbound docking tests. To explore the usefulness of ProPOSE on systems with limited backbone flexibility, here we tested the engineered scaffold DARPin, which is characterized by its relatively rigid protein backbone. A prospective screening campaign was undertaken, in which sequence-diversified DARPins were docked and ranked against a directed epitope on the target protein BCL-W. In this proof-of-concept study, only a relatively small set of 2,213 diverse DARPin interfaces were selected for docking from the huge theoretical library from mutating 18 amino-acid positions. A computational selection protocol was then applied for enrichment of binders based on normalized computed binding scores and frequency of binding modes against the predefined epitope. The top-ranked 18 designed DARPin interfaces were selected for experimental validation. Three designs exhibited binding affinities to BCL-W in the nanomolar range comparable to control interfaces adopted from known DARPin binders. This result is encouraging for future screening and engineering campaigns of DARPins and possibly other similarly rigid scaffolds against targeted protein epitopes. Method limitations are discussed and directions for future refinements are proposed.

Characterizing the Folding Transition-State Ensembles in the Energy Landscape of an RNA Tetraloop.

Molecular dynamics (MD) simulations have become increasingly powerful and can now describe the folding/unfolding of small biomolecules in atomic detail. However, a major challenge in MD simulations is to represent the complex energy landscape of biomolecules using a small number of reaction coordinates. In this study, we investigate the folding pathways of an RNA tetraloop, gcGCAAgc, using five classical MD simulations with a combined simulation time of approximately 120 μs. Our approach involves analyzing the tetraloop dynamics, including the folding transition state ensembles, using the energy landscape visualization method (ELViM). The ELViM is an approach that uses internal distances to compare any two conformations, allowing for a detailed description of the folding process without requiring root mean square alignment of structures. This method has previously been applied to describe the energy landscape of disordered β-amyloid peptides and other proteins. The ELViM results in a non-linear projection of the multidimensional space, providing a comprehensive representation of the tetraloop's energy landscape. Our results reveal four distinct transition-state regions and establish the paths that lead to the folded tetraloop structure. This detailed analysis of the tetraloop's folding process has important implications for understanding RNA folding, and the ELViM approach can be used to study other biomolecules.

Unfaulting mechanisms of interstitial Frank loops in fluorite-structured ThO2

Unfaulting of Frank loops in irradiated fluorite-structured oxides is of significance for understanding microstructural evolution. However, the mechanisms have not yet been directly observed. To this end, we utilize molecular dynamics to reveal the atomistic details related to the unfaulting process of interstitial Frank loops in ThO2, which involve a single pair or multiple pairs of co-propagating Shockley partials to annihilate the stacking faults. We find that the unfaulting is achieved via a synchronous shear of the partial pairs to remove the extrinsic stacking fault in the cation sublattice and the intrinsic stacking fault in the anion sublattice. The high mobility of oxygen at the dislocation core may reduce the activation barriers of dislocation nucleation and migration. These findings provide a fundamental understanding of the transformation of faulted loops in irradiated ThO2, and could be transferable to other fluorite-structured oxides.

Tunable properties of poly(vinylidene fluoride)-derived polymers for advancing battery performance and enabling diverse applications

This study uses molecular dynamics (MD) simulations to examine the thermal, structural, mechanical, and dynamic characteristics of poly(vinylidene fluoride) (PVDF) and its copolymers. We constructed and simulated three copolymers by replacing 0 to 50 mol% of PVDF monomers with trifluoroethylene (TrFE), hexafluoropropylene (HFP), and chlorotrifluoroethylene (CTFE) in order to investigate their properties. Additionally, we conducted a simulation of a polyelectrolyte model using the copolymers we constructed, along with LiFSI salt and FEC solvent, in order to evaluate the ionic conductivity. The calculated properties, such as melting temperature, thermal conductivity, Young’s modulus, and ionic conductivity, were found to be dependent on the molar fraction of substituted monomers. We observed specific peaks in the ionic conductivity calculated at room temperature, with maximum values of 3.32 × 10−5 S/cm for P(VDF-3 mol% HFP) and 2.56 × 10−5 S/cm for P(VDF-10 mol% TrFE). The molecular dynamics simulation results were consistent with experimental findings and provided insights into the adjustable atomistic details of fluorinated polymers derived from PVDF.

Investigation of molecular details of a bacterial cationic amino acid transporter (GkApcT) during arginine transportation using molecular dynamics simulation and umbrella sampling techniques.

Cationic amino acid transporters (CATs) facilitate arginine transport across membranes and maintain its levels in various tissues and organs, but their overexpression has been associated with severe cancers. A recent study identified the alternating access mechanism and critical residues involved in arginine transportation in a cationic amino acid transporter from Geobacillus kaustophilus (GkApcT). Here, we used molecular dynamics (MD) simulation methods to investigate the transportation mechanism of arginine (Arg) through GkApcT. The results revealed that arginine strongly interacts with specific binding site residues (Thr43, Asp111, Glu115, Lys191, Phe231, Ile234, and Asp237). Based on the umbrella sampling, the main driving force for arginine transport is the polar interactions of the arginine with channel-lining residues. An in-depth description of the dissociation mechanism and binding energy analysis brings valuable insight into the interactions between arginine and transporter residues, facilitating the design of effective CAT inhibitors in cancer cells. The membrane-protein system was constructed by uploading the prokaryotic CAT (PDB ID: 6F34) to the CHARMM-GUI web server. Molecular dynamics simulations were done using the GROMACS package, version 5.1.4, with the CHARMM36 force field and TIP3P water model. The MM-PBSA approach was performed for determining the arginine binding free energy. Furthermore, the hotspot residues were identified through per-residue decomposition analysis. The characteristics of the channel such as bottleneck radius and channel length were analyzed using the CaverWeb 1.1 web server. The proton wire inside the transporter was investigated based on the classic Grotthuss mechanism. We also investigated the atomistic details of arginine transportation using the path-based free energy umbrella sampling technique (US).

Understanding Molecular Aggregation of Ligand-Protected Atomically-Precise Metal Nanoclusters.

Monolayer-protected atomically precise nanoclusters (MPCs) are an important class of molecules due to their unique structural features and diverse applications, including bioimaging, sensors, and drug carriers. Understanding the atomistic and dynamical details of their self-assembly process is crucial for designing system-specific applications. Here, we applied molecular dynamics and on-the-fly probability-based enhanced sampling simulations to study the aggregation of Au25(pMBA)18 MPCs in aqueous and methanol solutions. The MPCs interact via both hydrogen bonds and π-stacks between the aromatic ligands to form stable dimers, oligomers, and crystals. The dimerization free energy profiles reveal a pivotal role of the ligand charged state and solvent mediating the molecular aggregation. Furthermore, MPCs' ligands exhibit suppressed conformational flexibility in the solid phase due to facile intercluster hydrogen bonds and π-stacks. Our work provides unprecedented molecular-level dynamical details of the aggregation process and conformational dynamics of MPCs ligands in solution and crystalline phases.

Toward a general neural network force field for protein simulations: Refining the intramolecular interaction in protein.

Molecular dynamics (MD) is an extremely powerful, highly effective, and widely used approach to understanding the nature of chemical processes in atomic details for proteins. The accuracy of results from MD simulations is highly dependent on force fields. Currently, molecular mechanical (MM) force fields are mainly utilized in MD simulations because of their low computational cost. Quantum mechanical (QM) calculation has high accuracy, but it is exceedingly time consuming for protein simulations. Machine learning (ML) provides the capability for generating accurate potential at the QM level without increasing much computational effort for specific systems that can be studied at the QM level. However, the construction of general machine learned force fields, needed for broad applications and large and complex systems, is still challenging. Here, general and transferable neural network (NN) force fields based on CHARMM force fields, named CHARMM-NN, are constructed for proteins by training NN models on 27 fragments partitioned from the residue-based systematic molecular fragmentation (rSMF) method. The NN for each fragment is based on atom types and uses new input features that are similar to MM inputs, including bonds, angles, dihedrals, and non-bonded terms, which enhance the compatibility of CHARMM-NN to MM MD and enable the implementation of CHARMM-NN force fields in different MD programs. While the main part of the energy of the protein is based on rSMF and NN, the nonbonded interactions between the fragments and with water are taken from the CHARMM force field through mechanical embedding. The validations of the method for dipeptides on geometric data, relative potential energies, and structural reorganization energies demonstrate that the CHARMM-NN local minima on the potential energy surface are very accurate approximations to QM, showing the success of CHARMM-NN for bonded interactions. However, the MD simulations on peptides and proteins indicate that more accurate methods to represent protein-water interactions in fragments and non-bonded interactions between fragments should be considered in the future improvement of CHARMM-NN, which can increase the accuracy of approximation beyond the current mechanical embedding QM/MM level.