Understanding the Reversible Binding of a Multichain Protein-Protein Complex through Free-Energy Calculations.

We demonstrate that the binding affinity of a multichain protein-protein complex, insulin dimer, can be accurately predicted using a streamlined route of standard binding free-energy calculations. We find that chains A and C, which do not interact directly during binding, stabilize the insulin monomer structures and reduce the binding affinity of the two monomers, therefore enabling their reversible association. Notably, we confirm that although classical methods can estimate the binding affinity of the insulin dimer, conventional molecular dynamics, enhanced sampling algorithms, and classical geometrical routes of binding free-energy calculations may not fully capture certain aspects of the role played by the noninteracting chains in the binding dynamics. Therefore, this study not only elucidates the role of noninteracting chains in the reversible binding of the insulin dimer but also offers a methodological guide for investigating the reversible binding of multichain protein-protein complexes utilizing streamlined free-energy calculations.

Binding mechanism of full-length Aß40 peptide to a mixed lipid bilayer.

The destructive effect of Aß peptides on membranes is an important source of its cytotoxicity in the pathogenesis of Alzheimer's disease. We have investigated the binding mechanism between the Aß42 peptide and bilayer in our former work. However, as another abundant form of Aß peptides in the physiological environment, the binding mechanism between Aß40 peptide and the lipid bilayer still remains ambiguous. Hence, we performed all-atom simulations on the Aß40 peptides with the lipid bilayer herein using replica exchange with the solute tempering 2 method. We obtained four major binding models with the hydrophobic C-terminus as the most preferable binding region. Hydrophobic residues and positively charged residues are the principal residues involved in the peptide-bilayer interactions. Aß40 peptides in our simulation mainly adopt a ß-rich conformation in both bound and unbound states. Besides, we determined peptide-water interactions and found that bound peptides prefer forming hydrogen bonds with water molecules than unbound peptides. Our findings herein may provide new insights for the in-depth understanding of the membrane-destructive mechanism of Aß peptides.

In-situ synthesis of quantum dots in the nucleus of live cells.

The cell nucleus is the main site for the storage and replication of genetic material, and the synthesis of substances in the nucleus is rhythmic, regular and strictly regulated by physiological processes. However, whether exogenous substances, such as nanoparticles, can be synthesized in situ in the nucleus of live cells has not been reported. Here, we have achieved in-situ synthesis of CdSxSe1-x quantum dots (QDs) in the nucleus by regulation of the glutathione (GSH) metabolic pathway. High enrichment of GSH in the nucleus can be accomplished by the addition of GSH with the help of the Bcl-2 protein. Then, high-valence Se is reduced to low-valence Se by glutathione-reductase-catalyzed GSH, and interacts with the Cd precursor formed through Cd and thiol-rich proteins, eventually generating QDs in the nucleus. Our work contributes to a new understanding of the syntheses of substances in the cell nucleus and will pave the way for the development of advanced 'supercells'.

Collective Variable-Based Enhanced Sampling: From Human Learning to Machine Learning.

Enhanced-sampling algorithms relying on collective variables (CVs) are extensively employed to study complex (bio)chemical processes that are not amenable to brute-force molecular simulations. The selection of appropriate CVs characterizing the slow movement modes is of paramount importance for reliable and efficient enhanced-sampling simulations. In this Perspective, we first review the application and limitations of CVs obtained from chemical and geometrical intuition. We also introduce path-sampling algorithms, which can identify path-like CVs in a high-dimensional free-energy space. Machine-learning algorithms offer a viable approach to finding suitable CVs by analyzing trajectories from preliminary simulations. We discuss both the performance of machine-learning-derived CVs in enhanced-sampling simulations of experimental models and the challenges involved in applying these CVs to realistic, complex molecular assemblies. Moreover, we provide a prospective view of the potential advancements of machine-learning algorithms for the development of CVs in the field of enhanced-sampling simulations.

Analyzing Molecular Dynamics Trajectories Thermodynamically through Artificial Intelligence.

Molecular dynamics simulations produce trajectories that correspond to vast amounts of structure when exploring biochemical processes. Extracting valuable information, e.g., important intermediate states and collective variables (CVs) that describe the major movement modes, from molecular trajectories to understand the underlying mechanisms of biological processes presents a significant challenge. To achieve this goal, we introduce a deep learning approach, coined DIKI (deep identification of key intermediates), to determine low-dimensional CVs distinguishing key intermediate conformations without a-priori assumptions. DIKI dynamically plans the distribution of latent space and groups together similar conformations within the same cluster. Moreover, by incorporating two user-defined parameters, namely, coarse focus knob and fine focus knob, to help identify conformations with low free energy and differentiate the subtle distinctions among these conformations, resolution-tunable clustering was achieved. Furthermore, the integration of DIKI with a path-finding algorithm contributes to the identification of crucial intermediates along the lowest free-energy pathway. We postulate that DIKI is a robust and flexible tool that can find widespread applications in the analysis of complex biochemical processes.

Aggregation, structure and water permeability of membrane-embedded helical Aß oligomers.

It is widely recognized that membranes can facilitate the aggregation of amyloid-ß (Aß) peptides, while Aß can in turn cause membrane damage. Many studies focus on the peptide-membrane interactions of Aß oligomers with ß-rich structures. However, the exact aggregation and toxicity mechanism of the membrane-embedded helical Aß oligomers remain ambiguous. Herein, the molecular dynamics simulations were performed on membrane-embedded helical Aß42 peptides. Initiated by eight Aß42 monomers embedded in a lipid bilayer, the monomers aggregate into oligomers with stable transmembrane helix structures. With the aggregation of peptides, the membrane perturbations caused by Aß aggregates decrease. The molecular architectures of oligomers were characterized and a helix-rich octamer stabilized by an annular network of hydrogen bonds was observed. The oligomers demonstrate the capability to assist transmembrane water transport. Our study may provide new insights for the investigation of transmembrane Aß oligomers.

Explainable Graph Neural Networks with Data Augmentation for Predicting pKa of C-H Acids.

The pKa of C-H acids is an important parameter in the fields of organic synthesis, drug discovery, and materials science. However, the prediction of pKa is still a great challenge due to the limit of experimental data and the lack of chemical insight. Here, a new model for predicting the pKa values of C-H acids is proposed on the basis of graph neural networks (GNNs) and data augmentation. A message passing unit (MPU) was used to extract the topological and target-related information from the molecular graph data, and a readout layer was utilized to retrieve the information on the ionization site C atom. The retrieved information then was adopted to predict pKa by a fully connected network. Furthermore, to increase the diversity of the training data, a knowledge-infused data augmentation technique was established by replacing the H atoms in a molecule with substituents exhibiting different electronic effects. The MPU was pretrained with the augmented data. The efficacy of data augmentation was confirmed by visualizing the distribution of compounds with different substituents and by classifying compounds. The explainability of the model was studied by examining the change of pKa values when a specific atom was masked. This explainability was used to identify the key substituents for pKa. The model was evaluated on two data sets from the iBonD database. Dataset1 includes the experimental pKa values of C-H acids measured in DMSO, while dataset2 comprises the pKa values measured in water. The results show that the knowledge-infused data augmentation technique greatly improves the predictive accuracy of the model, especially when the number of samples is small.

Interpretable Perturbator for Variable Selection in near-Infrared Spectral Analysis.

A perturbator was developed for variable selection in near-infrared (NIR) spectral analysis based on the perturbation strategy in deep learning for developing interpretation methods. A deep learning predictor was first constructed to predict the targets from the spectra in the training set. Then, taking the output of the predictor as a reference, the perturbator was trained to derive the perturbation-positive (P+) and perturbation-negative (P-) features from the spectra. Therefore, the weight (σ) of the perturbator layer can be a criterion to evaluate the importance of the variables in the spectra. Ranking the spectral variables by the criterion, the number of the variables used in the quantitative model can be obtained through cross-validation. Three NIR data sets were used to evaluate the proposed method. The root mean squared error was found to be comparable with or superior to that obtained by the commonly used methods. Moreover, the selected spectral variables are interpretable in identifying the key spectral features related to the prediction target. Therefore, the proposed method provides not only an effective tool for optimizing quantitative model, but also an efficient way for explaining spectra of multicomponent samples.

Binding Thermodynamics of Fourth-Generation EGFR Inhibitors Revealed by Absolute Binding Free Energy Calculations.

The overexpression or mutation of the kinase domain of the epidermal growth factor receptor (EGFR) is strongly associated with non-small-cell lung cancer (NSCLC). EGFR tyrosine kinase inhibitors (TKIs) have proven to be effective in treating NSCLC patients. However, EGFR mutations can result in drug resistance. To elucidate the mechanisms underlying this resistance and inform future drug development, we examined the binding affinities of BLU-945, a recently reported fourth-generation TKI, to wild-type EGFR (EGFRWT) and its double-mutant (L858R/T790M; EGFRDM) and triple-mutant (L858R/T790M/C797S; EGFRTM) forms. We compared the binding affinities of BLU-945, BLU-945 analogues, CH7233163 (another fourth-generation TKI), and erlotinib (a first-generation TKI) using absolute binding free energy calculations. Our findings reveal that BLU-945 and CH7233163 exhibit binding affinities to both EGFRDM and EGFRTM stronger than those of erlotinib, corroborating experimental data. We identified K745 and T854 as the key residues in the binding of fourth-generation EGFR TKIs. Electrostatic forces were the predominant driving force for the binding of fourth-generation TKIs to EGFR mutants. Furthermore, we discovered that the incorporation of piperidinol and sulfone groups in BLU-945 substantially enhanced its binding capacity to EGFR mutants. Our study offers valuable theoretical insights for optimizing fourth-generation EGFR TKIs.

Deep-Learning-Assisted Enhanced Sampling for Exploring Molecular Conformational Changes.

We present a novel strategy to explore conformational changes and identify stable states of molecular objects, eliminating the need for a priori knowledge. The approach applies a deep learning method to extract information about the movement modes of the molecular object from a short, high-dimensional, and parameter-free preliminary enhanced-sampling simulation. The gathered information is described by a small set of deep-learning-based collective variables (dCVs), which steer the production-enhanced-sampling simulation. Considering the challenge of adequately exploring the configurational space using the low-dimensional, suboptimal dCVs, we incorporate a method designed for ergodic sampling, namely, Gaussian-accelerated molecular dynamics (MD), into the framework of CV-based enhanced sampling. MD simulations on both toy models and nontrivial examples demonstrate the remarkable computational efficiency of the strategy in capturing the conformational changes of molecular objects without a priori knowledge. Specifically, we achieved the blind folding of two fast folders, chignolin and villin, within a time scale of hundreds of nanoseconds and successfully reconstructed the free-energy landscapes that characterize their reversible folding. All in all, the presented strategy holds significant promise for investigating conformational changes in macromolecules, and it is anticipated to find extensive applications in the fields of chemistry and biology.

Standard Binding Free-Energy Calculations: How Far Are We from Automation?

Recent success stories suggest that in silico protein-ligand binding free-energy calculations are approaching chemical accuracy. However, their widespread application remains limited by the extensive human intervention required, posing challenges for the neophyte. As such, it is critical to develop automated workflows for estimating protein-ligand binding affinities with minimum personal involvement. Key human efforts include setting up and tuning enhanced-sampling or alchemical-transformation algorithms as a preamble to computational binding free-energy estimations. Additionally, preparing input files, bookkeeping, and postprocessing represent nontrivial tasks. In this Perspective, we discuss recent progress in automating standard binding free-energy calculations, featuring the development of adaptive or parameter-free algorithms, standardization of binding free-energy calculation workflows, and the implementation of user-friendly software. We also assess the current state of automated standard binding free-energy calculations and evaluate the limitations of existing methods. Last, we outline the requirements for future algorithms and workflows to facilitate automated free-energy calculations for diverse protein-ligand complexes.

Mechanism of Binding of Polyproline to Ice via Interfacial Water: An Experimental and Theoretical Study.

The molecular mechanism underlying inhibition of ice growth by polyproline (PPro), a minimal antifreeze glycoprotein mimic, remains unclear. In this work, the change in the structure of water during the growth of ice in PPro solutions was investigated using a combination of near-infrared spectroscopy and molecular dynamics (MD) simulations. The results show that only high concentrations of PPro solutions can effectively inhibit ice growth, as indicated by the variation in the spectral intensity of ice with time. When PPro exhibits an antifreeze effect, the spectral intensity of hydrated water associated with PPro in a solution is weakened. The experiments and MD simulations reveal that the quantity of the interfacial water between the ice crystal and the hydrophobic groups of PPro progressively reaches a plateau. Most significantly, we present clear evidence that the stable existence of this interfacial water is critical for the antifreeze activity of PPro.

Achieving Accurate Standard Protein-Protein Binding Free Energy Calculations through the Geometrical Route and Ergodic Sampling.

A new strategy for the prediction of binding free energies of protein-protein complexes is reported in the present article. By combining an ergodic-sampling algorithm with the so-called "geometrical route", which introduces a series of geometrical restraints as a preamble to the physical separation of the two partners, we achieve accurate binding free energy calculations for medium-sized protein-protein complexes within the microsecond timescale. The ergodic-sampling algorithm, namely, Gaussian-accelerated molecular dynamics (GaMD), implicitly helps explore the conformational change of the two binding partners as they associate reversibly by raising the energy wells. Therefore, independent simulations capturing the isomerization of proteins are no longer needed, reducing both the computational cost and human effort. Numerical applications indicate errors on the order of 0.1 kcal/mol for the Abl-SH3 domain binding a decapeptide, of 2.6 kcal/mol for the barnase-barstar complex, and of 0.2 kcal/mol for human leukocyte elastase binding the third domain of the turkey ovomucoid inhibitor. Compared with the classical geometrical route, which resorts to collective variables to describe the isomerization of proteins, our new strategy possesses remarkable convergence properties and robustness for protein-protein complexes owing to improved ergodic sampling. We are confident that the strategy presented in this study will have a broad range of applications, helping us understand recognition-association phenomena in the areas of physical, biological, and medicinal chemistry.

Structures of water on the surface of anatase TiO2 studied by diffuse reflectance near-infrared spectroscopy.

Investigating the structures of water on metal oxides is helpful for understanding the mechanism of the adsorptions in the presence of water. In this work, the structures of adsorbed water molecules on anatase TiO2 (101) were studied by diffuse reflectance near-infrared spectroscopy (DR-NIRS). With resolution enhanced spectrum by continuous wavelet transform (CWT), the spectral features of adsorbed water at different sites were found. In the spectrum of dried TiO2 powder, there is only the spectral feature of the water adsorbed at 5-coordinated titanium atoms (Ti5c). With the increase of the adsorbed water, the spectral feature of the water at 2-coordinated oxygen atoms (O2c) emerges first, and then that of the water interacting with the adsorbed water can be observed. When adenosine triphosphate (ATP) was adsorbed on TiO2, the intensity of the peaks related to the adsorbed water decreases, indicating that the adsorbed water is replaced by ATP due to the strong affinity to Ti5c. Therefore, there is a clear correlation between the peak intensity of the adsorbed water and the adsorbed quantity of ATP. Water can be a NIR spectroscopic probe to detect the quantity of the adsorbed ATP. A partial least squares (PLS) model was established to predict the content of adsorbed ATP by the spectral peaks of water. The recoveries of validation samples are in the range of 92.00-114.96% with the relative standard deviations (RSDs) in a range of 2.13-5.82%.

Ultra-high resolution near-infrared spectrum by wavelet packet transform revealing the hydrogen bond interactions.

Resolution is always an obstacle to analyzing the fine structure of a spectrum. The problem is particularly serious in the analysis of the near-infrared (NIR) spectra of aqueous solutions, because the spectrum is generally composed of overlapping broad peaks making the understanding of the structures and the interactions notoriously difficult. In this work, wavelet packet transform (WPT) was adopted to enhance the resolution of the NIR spectra of aqueous mixtures. Due to the microscopic ability of WPT in both position and frequency, the fine details of a spectrum can be observed in the spectral components of different frequencies obtained by WPT decomposition. Ultra-high resolution spectrum can be obtained from the high-frequency component representing the spectral features. Spectral features of different hydrogen-bonded OH, as well as the OH in HOH and HOD, were identified from the high-resolution NIR spectra of water and heavy water mixtures and validated by the variation of the spectral intensity with the mole ratio of H2O and D2O. The high-resolution spectrum was further applied in analyzing the interaction of amine and water. The spectral features of the hydrogen bonding between CH/NH in tert-butylamine (TBA) and OH in water were observed. The structures of CH bonded to one water molecule, and the structures of NH connecting with one and two water molecules were identified.

Revealing the Mechanism of Irreversible Binding of Antifreeze Glycoproteins to Ice.

Antifreeze glycoproteins (AFGPs) are a special kind of antifreeze proteins with strong flexibility. Whether their antifreeze activity is achieved by reversibly or irreversibly binding to ice is widely debated, and the molecular mechanism of irreversible binding remains unclear. In this work, the antifreeze mechanism of the smallest AFGP isoform, AFGP8, is investigated at the atomic level. The results indicate that AFGP8 can bind to ice both reversibly through its hydrophobic methyl groups (peptide binding) and irreversibly through its hydrophilic disaccharide moieties (saccharide binding). Although peptide binding occurs faster than saccharide binding, free-energy calculations indicate that the latter is energetically more favorable. In saccharide binding, at least one disaccharide moiety is frozen in the grown ice, resulting in irreversible binding, while the other moieties significantly perturb the water hydrogen-bonding network, thus inhibiting ice growth more effectively. The present study reveals the coexistence of reversible and irreversible bindings of AFGP8, both contributing to the inhibition of ice growth and further provides molecular mechanism of irreversible binding.

Meta-Analysis Reveals That Absolute Binding Free-Energy Calculations Approach Chemical Accuracy.

Systematic and quantitative analysis of the reliability of formally exact methods that in silico calculate absolute protein-ligand binding free energies remains lacking. Here, we provide, for the first time, evidence-based information on the reliability of these methods by statistically studying 853 cases from 34 different research groups through meta-analysis. The results show that formally exact methods approach chemical accuracy (error = 1.58 kcal/mol), even if people are challenging difficult tasks such as blind drug screening in recent years. The geometrical-pathway-based methods prove to possess a better convergence ability than the alchemical ones, while the latter have a larger application range. We also reveal the importance of always using the latest force fields to guarantee reliability and discuss the pros and cons of turning to an implicit solvent model in absolute binding free-energy calculations. Moreover, based on the meta-analysis, an evidence-based guideline for in silico binding free-energy calculations is provided.

Metabolic Engineering of the Isopentenol Utilization Pathway Enhanced the Production of Terpenoids in Chlamydomonas reinhardtii.

Eukaryotic green microalgae show considerable promise for the sustainable light-driven biosynthesis of high-value fine chemicals, especially terpenoids because of their fast and inexpensive phototrophic growth. Here, the novel isopentenol utilization pathway (IUP) was introduced into Chlamydomonas reinhardtii to enhance the hemiterpene (isopentenyl pyrophosphate, IPP) titers. Then, diphosphate isomerase (IDI) and limonene synthase (MsLS) were further inserted for limonene production. Transgenic algae showed 8.6-fold increase in IPP compared with the wild type, and 23-fold increase in limonene production compared with a single MsLS expressing strain. Following the culture optimization, the highest limonene production reached 117 µg/L, when the strain was cultured in a opt2 medium supplemented with 10 mM isoprenol under a light: dark regimen. This demonstrates that transgenic algae expressing the IUP represent an ideal chassis for the high-value terpenoid production. The IUP will facilitate further the metabolic and enzyme engineering to enhance the terpenoid titers by significantly reducing the number of enzyme steps required for an optimal biosynthesis.

Accurate Description of Solvent-Exposed Salt Bridges with a Non-polarizable Force Field Incorporating Solvent Effects.

The strength of salt bridges resulting from the interaction of cations and anions is modulated by their environment. However, polarization of the solvent molecules by the charged moieties makes the accurate description of cation-anion interactions in an aqueous solution by means of a pairwise additive potential energy function and classical combination rules particularly challenging. In this contribution, aiming at improving the representation of solvent-exposed salt-bridge interactions with an all-atom non-polarizable force field, we put forth here a parametrization strategy. First, the interaction of a cation and an anion is characterized by hybrid quantum mechanical/molecular mechanics (QM/MM) potential of mean force (PMF) calculations, whereby constantly exchanging solvent molecules around the ions are treated at the quantum mechanical level. The Lennard-Jones (LJ) parameters describing the salt-bridge ion pairs are then optimized to match the reference QM/MM PMFs through the so-called nonbonded FIX, or NBFIX, feature of the CHARMM force field. We apply the new set of parameters, coined CHARMM36m-SBFIX, to the calculation of association constants for the ammonium-acetate and guanidinium-acetate complexes, the osmotic pressures for glycine zwitterions, guanidinium, and acetate ions, and to the simulation of both folded and intrinsically disordered proteins. Our findings indicate that CHARMM36m-SBFIX improves the description of solvent-exposed salt-bridge interactions, both structurally and thermodynamically. However, application of this force field to the standard binding free-energy calculation of a protein-ligand complex featuring solvent-excluded salt-bridge interactions leads to a poor reproduction of the experimental value, suggesting that the parameters optimized in an aqueous solution cannot be readily transferred to describe solvent-excluded salt-bridge interactions. Put together, owing to their sensitivity to the environment, modeling salt-bridge interactions by means of a single, universal set of LJ parameters remains a daunting theoretical challenge.

Spectral Encoder to Extract the Features of Near-Infrared Spectra for Multivariate Calibration.

An autoencoder architecture was adopted for near-infrared (NIR) spectral analysis by extracting the common features in the spectra. Three autoencoder-based networks with different purposes were constructed. First, a spectral encoder was established by training the network with a set of spectra as the input. The features of the spectra can be encoded by the nodes in the bottleneck layer, which in turn can be used to build a sparse and robust model. Second, taking the spectra of one instrument as the input and that of another instrument as the reference output, the common features in both spectra can be obtained in the bottleneck layer. Therefore, in the prediction step, the spectral features of the second can be predicted by taking the reverse of the decoder as the encoder. Furthermore, transfer learning was used to build the model for the spectra of more instruments by fine-tuning the trained network. NIR datasets of plant, wheat, and pharmaceutical tablets measured on multiple instruments were used to test the method. The multi-linear regression (MLR) model with the encoded features was found to have a similar or slightly better performance in prediction compared with the partial least-squares (PLS) model.