Structure prediction of cyclic peptides by molecular dynamics + machine learning

Generative β-hairpin design using a residue-based physicochemical property landscape

De novo peptide design is a new frontier that has broad application potential in the biological and biomedical fields. Most existing models for de novo peptide design are largely based on sequence homology that can be restricted based on evolutionarily derived protein sequences and lack the physicochemical context essential in protein folding. Generative machine learning for de novo peptide design is a promising way to synthesize theoretical data that are based on, but unique from, the observable universe. In this study, we created and tested a custom peptide generative adversarial network intended to design peptide sequences that can fold into the β-hairpin secondary structure. This deep neural network model is designed to establish a preliminary foundation of the generative approach based on physicochemical and conformational properties of 20 canonical amino acids, for example, hydrophobicity and residue volume, using extant structure-specific sequence data from the PDB. The beta generative adversarial network model robustly distinguishes secondary structures of β hairpin from α helix and intrinsically disordered peptides with an accuracy of up to 96% and generates artificial β-hairpin peptide sequences with minimum sequence identities around 31% and 50% when compared against the current NCBI PDB and nonredundant databases, respectively. These results highlight the potential of generative models specifically anchored by physicochemical and conformational property features of amino acids to expand the sequence-to-structure landscape of proteins beyond evolutionary limits.

Medium-sized peptides from microbial sources with potential for antibacterial drug development

Covering: 1993 to the end of 2022As the rapid development of antibiotic resistance shrinks the number of clinically available antibiotics, there is an urgent need for novel options to fill the existing antibiotic pipeline. In recent years, antimicrobial peptides have attracted increased interest due to their impressive broad-spectrum antimicrobial activity and low probability of antibiotic resistance. However, macromolecular antimicrobial peptides of plant and animal origin face obstacles in antibiotic development because of their extremely short elimination half-life and poor chemical stability. Herein, we focus on medium-sized antibacterial peptides (MAPs) of microbial origin with molecular weights below 2000 Da. The low molecular weight is not sufficient to form complex protein conformations and is also associated to a better chemical stability and easier modifications. Microbially-produced peptides are often composed of a variety of non-protein amino acids and terminal modifications, which contribute to improving the elimination half-life of compounds. Therefore, MAPs have great potential for drug discovery and are likely to become key players in the development of next-generation antibiotics. In this review, we provide a detailed exploration of the modes of action demonstrated by 45 MAPs and offer a concise summary of the structure-activity relationships observed in these MAPs.

PepBinding: A Workflow for Predicting Peptide Binding Structures by Combining Peptide Docking and Peptide Gaussian Accelerated Molecular Dynamics Simulations

Predicting protein-peptide interactions is crucial for understanding peptide binding processes and designing peptide drugs. However, traditional computational modeling approaches face challenges in accurately predicting peptide-protein binding structures due to the slow dynamics and high flexibility of the peptides. Here, we introduce a new workflow termed "PepBinding" for predicting peptide binding structures, which combines peptide docking, all-atom enhanced sampling simulations using the Peptide Gaussian accelerated Molecular Dynamics (Pep-GaMD) method, and structural clustering. PepBinding has been demonstrated on seven distinct model peptides. In peptide docking using HPEPDOCK, the peptide backbone root-mean-square deviations (RMSDs) of their bound conformations relative to X-ray structures ranged from 3.8 to 16.0 Å, corresponding to the medium to inaccurate quality models according to the Critical Assessment of PRediction of Interactions (CAPRI) criteria. The Pep-GaMD simulations performed for only 200 ns significantly improved the docking models, resulting in five medium and two acceptable quality models. Therefore, PepBinding is an efficient workflow for predicting peptide binding structures and is publicly available at https://github.com/MiaoLab20/PepBinding.

Accurate Structure Prediction for Cyclic Peptides Containing Proline Residues with High-Temperature Molecular Dynamics

Cyclic peptides (CPs) are emerging as promising drug candidates. Numerous natural CPs and their analogs are effective therapeutics against various diseases. Notably, many of them contain peptidyl cis-prolyl bonds. Due to the high rotational barrier of peptide bonds, conventional molecular dynamics simulations struggle to effectively sample the cis/trans-isomerization of peptide bonds. Previous studies have highlighted the high accuracy of the residue-specific force field (RSFF) and the high sampling efficiency of high-temperature molecular dynamics (high-T MD). Herein, we propose a protocol that combines high-T MD with RSFF2C and a recently developed reweighting method based on probability densities for accurate structure prediction of proline-containing CPs. Our method successfully predicted 19 out of 23 CPs with the backbone rmsd < 1.0 Å compared to X-ray structures. Furthermore, we performed high-T MD and density reweighting on the sunflower trypsin inhibitor (SFTI-1)/trypsin complex to demonstrate its applicability in studying CP-complexes containing cis-prolines. Our results show that the conformation of SFTI-1 in aqueous solution is consistent with its bound conformation, potentially facilitating its binding.

Assessing the Performance of Peptide Force Fields for Modeling the Solution Structural Ensembles of Cyclic Peptides

Molecular dynamics simulation is a powerful tool for characterizing the solution structural ensembles of cyclic peptides. However, the ability of simulation to recapitulate experimental results and make accurate predictions largely depends on the force fields used. In our work here, we evaluate the performance of seven state-of-the-art force fields in recapitulating the experimental NMR results in water of 12 benchmark cyclic peptides, consisting of 6 cyclic pentapeptides, 4 cyclic hexapeptides, and 2 cyclic heptapeptides. The results show that RSFF2+TIP3P, RSFF2C+TIP3P, and Amber14SB+TIP3P exhibit similar and the best performance, all recapitulating the NMR-derived structure information on 10 cyclic peptides. Amber19SB+OPC successfully recapitulates the NMR-derived structure information on 8 cyclic peptides. In contrast, OPLS-AA/M+TIP4P, Amber03+TIP3P, and Amber14SBonlysc+GB-neck2 could only recapitulate the NMR-derived structure information on 5 cyclic peptides, the majority of which are not well-structured.

HighFold: accurately predicting structures of cyclic peptides and complexes with head-to-tail and disulfide bridge constraints

In recent years, cyclic peptides have emerged as a promising therapeutic modality due to their diverse biological activities. Understanding the structures of these cyclic peptides and their complexes is crucial for unlocking invaluable insights about protein target-cyclic peptide interaction, which can facilitate the development of novel-related drugs. However, conducting experimental observations is time-consuming and expensive. Computer-aided drug design methods are not practical enough in real-world applications. To tackles this challenge, we introduce HighFold, an AlphaFold-derived model in this study. By integrating specific details about the head-to-tail circle and disulfide bridge structures, the HighFold model can accurately predict the structures of cyclic peptides and their complexes. Our model demonstrates superior predictive performance compared to other existing approaches, representing a significant advancement in structure-activity research. The HighFold model is openly accessible at https://github.com/hongliangduan/HighFold.

Deep learning for advancing peptide drug development: Tools and methods in structure prediction and design

Peptides can bind challenging disease targets with high affinity and specificity, offering enormous opportunities for addressing unmet medical needs. However, peptides' unique features, including smaller size, increased structural flexibility, and limited data availability, pose additional challenges to the design process compared to proteins. This review explores the dynamic field of peptide therapeutics, leveraging deep learning to enhance structure prediction and design. Our exploration encompasses various facets of peptide research, ranging from dataset curation handling to model development. As deep learning technologies become more refined, we channel our efforts into peptide structure prediction and design, aligning with the fundamental principles of structure-activity relationships in drug development. To guide researchers in harnessing the potential of deep learning to advance peptide drug development, our insights comprehensively explore current challenges and future directions of peptide therapeutics.

Unlocking novel therapies: cyclic peptide design for amyloidogenic targets through synergies of experiments, simulations, and machine learning

Existing therapies for neurodegenerative diseases like Parkinson's and Alzheimer's address only their symptoms and do not prevent disease onset. Common therapeutic agents, such as small molecules and antibodies struggle with insufficient selectivity, stability and bioavailability, leading to poor performance in clinical trials. Peptide-based therapeutics are emerging as promising candidates, with successful applications for cardiovascular diseases and cancers due to their high bioavailability, good efficacy and specificity. In particular, cyclic peptides have a long in vivo stability, while maintaining a robust antibody-like binding affinity. However, the de novo design of cyclic peptides is challenging due to the lack of long-lived druggable pockets of the target polypeptide, absence of exhaustive conformational distributions of the target and/or the binder, unknown binding site, methodological limitations, associated constraints (failed trials, time, money) and the vast combinatorial sequence space. Hence, efficient alignment and cooperation between disciplines, and synergies between experiments and simulations complemented by popular techniques like machine-learning can significantly speed up the therapeutic cyclic-peptide development for neurodegenerative diseases. We review the latest advancements in cyclic peptide design against amyloidogenic targets from a computational perspective in light of recent advancements and potential of machine learning to optimize the design process. We discuss the difficulties encountered when designing novel peptide-based inhibitors and we propose new strategies incorporating experiments, simulations and machine learning to design cyclic peptides to inhibit the toxic propagation of amyloidogenic polypeptides. Importantly, these strategies extend beyond the mere design of cyclic peptides and serve as template for the de novo generation of (bio)materials with programmable properties.

Molecular chameleons in drug discovery

Molecular chameleons possess a flexibility that allows them to dynamically shield or expose polar functionalities in response to the properties of the environment. Although the concept of molecular chameleons was introduced already in 1970, interest in them has grown considerably since the 2010s, when drug discovery has focused to an increased extent on new chemical modalities. Such modalities include cyclic peptides, macrocycles and proteolysis-targeting chimeras, all of which reside in a chemical space far from that of traditional small-molecule drugs. Both cell permeability and aqueous solubility are required for the oral absorption of drugs. Engineering these properties, and potent target binding, into the larger new modalities is a more daunting task than for traditional small-molecule drugs. The ability of chameleons to adapt to different environments may be essential for success. In this Review, we provide both general and theoretical insights into the realm of molecular chameleons. We discuss why chameleons have come into fashion and provide a do-it-yourself toolbox for their design; we then provide a glimpse of how advanced in silico methods can support molecular chameleon design.

The ionic liquid-based electrolytes during their charging process: Movable endpoints of overscreening effect near the electrode interface

HYPOTHESIS

Adding solvents to ionic liquids (ILs) can lead to the suppression of the overscreening effect near an electrode interface. Also, this suppression can be observed in neat ILs by elongating the length of the nonpolar chains on their ions. Most neat ILs, unlike the ideal model, do not exhibit a crowding effect in experiments. Through molecular dynamics (MD) simulations, researchers can model and analyze these systems in order to understand them.

SIMULATIONS

In this study, the dynamic change near the electrode interface of ILs-based electrolytes was investigated using MD simulations. The phenomena observed in MD simulations are generally understandable because factors can attenuate charge densities calculated from these simulations.

FINDINGS

The study findings reveal that both the solvents or nonpolar chains contributed to the formation of nonpolar domains. Also, the microscopic mechanisms and influences of these nonpolar domains were clearly identified. The results are important for real life applications. Some ions form a "point to surface" layer near the electrode of neat ILs. When ILs contain long nonpolar chains, they can suppress the crowding effect through self-assembly behavior. However, when they do not have any chains or short nonpolar chains, it can be difficult to stop the overscreening effect. This means it can become challenging to begin the next stage of the crowding effect.

Computational Prediction of Cyclic Peptide Structural Ensembles and Application to the Design of Keap1 Binders

The Nrf2 transcription factor is a master regulator of the cellular response to oxidative stress, and Keap1 is its primary negative regulator. Activating Nrf2 by inhibiting the Nrf2-Keap1 protein-protein interaction has shown promise for treating cancer and inflammatory diseases. A loop derived from Nrf2 has been shown to inhibit Keap1 selectively, especially when cyclized, but there are no reliable design methods for predicting an optimal macrocyclization strategy. In this work, we employed all-atom, explicit-solvent molecular dynamics simulations with enhanced sampling methods to predict the relative degree of preorganization for a series of peptides cyclized with a set of bis-thioether "staples". We then correlated these predictions to experimentally measured binding affinities for Keap1 and crystal structures of the cyclic peptides bound to Keap1. This work showcases a computational method for designing cyclic peptides by simulating and comparing their entire solution-phase ensembles, providing key insights into designing cyclic peptides as selective inhibitors of protein-protein interactions.

Exploring a Structural Data Mining Approach to Design Linkers for Head-to-Tail Peptide Cyclization

Peptides have recently regained interest as therapeutic candidates, but their development remains confronted with several limitations including low bioavailability. Backbone head-to-tail cyclization, i.e., setting a covalent peptide bond linking the last amino acid with the first one, is one effective strategy of peptide-based drug design to stabilize the conformation of bioactive peptides while preserving peptide properties in terms of low toxicity, binding affinity, target selectivity, and preventing enzymatic degradation. Starting from an active peptide, it usually requires the design of a linker of a few amino acids to make it possible to cyclize the peptide, possibly preserving the conformation of the initial peptide and not affecting its activity. However, very little is known about the sequence-structure relationship requirements of designing linkers for peptide cyclization in a rational manner. Recently, we have shown that large-scale data-mining of available protein structures can lead to the precise identification of protein loop conformations, even from remote structural classes. Here, we transpose this approach to linkers, allowing head-to-tail peptide cyclization. First we show that given a linker sequence and the conformation of the linear peptide, it is possible to accurately predict the cyclized peptide conformation. Second, and more importantly, we show that it seems possible to elaborate on the information inferred from protein structures to propose effective candidate linker sequences constrained by length and amino acid composition, providing the first framework for the rational design of head-to-tail cyclization linkers. Finally, we illustrate this for two peptides using a limited set of amino-acids likely not to interfere with peptide function. For a linear peptide derived from Nrf2, the peptide cyclized starting from the experimental structure showed a 26-fold increase in the binding affinity. For urotensin II, a peptide already cyclized by a disulfide bond that exerts a broad array of biological activities, we were able, starting from models of the structure, to design a head-to-tail cyclized peptide, the first synthesized bicyclic 14-residue long urotensin II analogue, showing a retention of in vitro activity. Although preliminary, our results strongly suggest that such an approach has strong potential for cyclic peptide-based drug design.

Cell-permeable chameleonic peptides: Exploiting conformational dynamics in de novo cyclic peptide design

Membrane-traversing peptides offer opportunities for targeting intracellular proteins and oral delivery. Despite progress in understanding the mechanisms underlying membrane traversal in natural cell-permeable peptides, there are still several challenges to designing membrane-traversing peptides with diverse shapes and sizes. Conformational flexibility appears to be a key determinant of membrane permeability of large macrocycles. We review recent developments in the design and validation of chameleonic cyclic peptides, which can switch between alternative conformations to enable improved permeability through cell membranes, while still maintaining reasonable solubility and exposed polar functional groups for target protein binding. Finally, we discuss the principles, strategies, and practical considerations for rational design, discovery, and validation of permeable chameleonic peptides.

Training Neural Network Models Using Molecular Dynamics Simulation Results to Efficiently Predict Cyclic Hexapeptide Structural Ensembles

Cyclic peptides have emerged as a promising class of therapeutics. However, their de novo design remains challenging, and many cyclic peptide drugs are simply natural products or their derivatives. Most cyclic peptides, including the current cyclic peptide drugs, adopt multiple conformations in water. The ability to characterize cyclic peptide structural ensembles would greatly aid their rational design. In a previous pioneering study, our group demonstrated that using molecular dynamics results to train machine learning models can efficiently predict structural ensembles of cyclic pentapeptides. Using this method, which was termed StrEAMM (Structural Ensembles Achieved by Molecular Dynamics and Machine Learning), linear regression models were able to predict the structural ensembles for an independent test set with R² = 0.94 between the predicted populations for specific structures and the observed populations in molecular dynamics simulations for cyclic pentapeptides. An underlying assumption in these StrEAMM models is that cyclic peptide structural preferences are predominantly influenced by neighboring interactions, namely, interactions between (1,2) and (1,3) residues. Here we demonstrate that for larger cyclic peptides such as cyclic hexapeptides, linear regression models including only (1,2) and (1,3) interactions fail to produce satisfactory predictions (R² = 0.47); further inclusion of (1,4) interactions leads to moderate improvements (R² = 0.75). We show that when using convolutional neural networks and graph neural networks to incorporate complex nonlinear interaction patterns, we can achieve R² = 0.97 and R² = 0.91 for cyclic pentapeptides and hexapeptides, respectively.

Cyclic peptide structure prediction and design using AlphaFold

Deep learning networks offer considerable opportunities for accurate structure prediction and design of biomolecules. While cyclic peptides have gained significant traction as a therapeutic modality, developing deep learning methods for designing such peptides has been slow, mostly due to the small number of available structures for molecules in this size range. Here, we report approaches to modify the AlphaFold network for accurate structure prediction and design of cyclic peptides. Our results show this approach can accurately predict the structures of native cyclic peptides from a single sequence, with 36 out of 49 cases predicted with high confidence (pLDDT > 0.85) matching the native structure with root mean squared deviation (RMSD) less than 1.5 Å. Further extending our approach, we describe computational methods for designing sequences of peptide backbones generated by other backbone sampling methods and for de novo design of new macrocyclic peptides. We extensively sampled the structural diversity of cyclic peptides between 7-13 amino acids, and identified around 10,000 unique design candidates predicted to fold into the designed structures with high confidence. X-ray crystal structures for seven sequences with diverse sizes and structures designed by our approach match very closely with the design models (root mean squared deviation < 1.0 Å), highlighting the atomic level accuracy in our approach. The computational methods and scaffolds developed here provide the basis for custom-designing peptides for targeted therapeutic applications.

LassoHTP: A High-Throughput Computational Tool for Lasso Peptide Structure Construction and Modeling

Lasso peptides are a subclass of ribosomally synthesized and post-translationally modified peptides with a slipknot conformation. With superior thermal stability, protease resistance, and antimicrobial activity, lasso peptides are promising candidates for bioengineering and pharmaceutical applications. To enable high-throughput computational prediction and design of lasso peptides, we developed a software, LassoHTP, for automatic lasso peptide structure construction and modeling. LassoHTP consists of three modules, including the scaffold constructor, mutant generator, and molecular dynamics (MD) simulator. With a user-provided sequence and conformational annotation, LassoHTP can either generate the structure and conformational ensemble as is or conduct random mutagenesis. We used LassoHTP to construct eight known lasso peptide structures de novo and to simulate their conformational ensembles for 100 ns MD simulations. For benchmarking, we calculated the root mean square deviation (RMSD) of these ensembles with reference to their experimental crystal or NMR PDB structures; we also compared these RMSD values against those of the MD ensembles that are initiated from the PDB structures. Dihedral principal component analysis was also conducted. The results show that the LassoHTP-initiated ensembles are similar to those of the PDB-initiated ensembles. LassoHTP offers a computational platform to develop strategies for lasso peptide prediction and design.

Ensemble Generation for Linear and Cyclic Peptides Using a Reservoir Replica Exchange Molecular Dynamics Implementation in GROMACS

The profile of shapes presented by a cyclic peptide modulates its therapeutic efficacy and is represented by the ensemble of its sampled conformations. Although some algorithms excel at creating a diverse ensemble of cyclic peptide conformations, they seldom address the entropic contribution of flexible conformations and often have significant practical difficulty producing an ensemble with converged and reliable thermodynamic properties. In this study, an accelerated molecular dynamics (MD) method, namely, reservoir replica exchange MD (R-REMD or Res-REMD), was implemented in GROMACS ver. 4.6.7 and benchmarked on two small cyclic peptide model systems: a cyclized furin cleavage site of the severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) spike (cyclo-(CGPRRARSG)) and oxytocin (disulfide-bonded CYIQNCPLG). Additionally, we also benchmarked Res-REMD on alanine dipeptide and Trpzip2 to demonstrate its validity and efficiency over REMD. For Trpzip2, Res-REMD coupled with an umbrella-sampling-derived reservoir generated similar folded fractions as regular REMD but on a much faster time scale. For cyclic peptides, Res-REMD appeared to be marginally faster than REMD in ensemble generation. Finally, Res-REMD was more effective in sampling rare events such as trans to cis peptide bond isomerization. We provide a GitHub page with the modified GROMACS source code for running Res-REMD at https://github.com/PlotkinLab/Reservoir-REMD.

Design of Protein Segments and Peptides for Binding to Protein Targets

Recent years have witnessed a rise in methods for accurate prediction of structure and design of novel functional proteins. Design of functional protein fragments and peptides occupy a small, albeit unique, space within the general field of protein design. While the smaller size of these peptides allows for more exhaustive computational methods, flexibility in their structure and sparsity of data compared to proteins, as well as presence of noncanonical building blocks, add additional challenges to their design. This review summarizes the current advances in the design of protein fragments and peptides for binding to targets and discusses the challenges in the field, with an eye toward future directions.

Biological Membrane-Penetrating Peptides: Computational Prediction and Applications

Peptides comprise a versatile class of biomolecules that present a unique chemical space with diverse physicochemical and structural properties. Some classes of peptides are able to naturally cross the biological membranes, such as cell membrane and blood-brain barrier (BBB). Cell-penetrating peptides (CPPs) and blood-brain barrier-penetrating peptides (B3PPs) have been explored by the biotechnological and pharmaceutical industries to develop new therapeutic molecules and carrier systems. The computational prediction of peptides’ penetration into biological membranes has been emerged as an interesting strategy due to their high throughput and low-cost screening of large chemical libraries. Structure- and sequence-based information of peptides, as well as atomistic biophysical models, have been explored in computer-assisted discovery strategies to classify and identify new structures with pharmacokinetic properties related to the translocation through biomembranes. Computational strategies to predict the permeability into biomembranes include cheminformatic filters, molecular dynamics simulations, artificial intelligence algorithms, and statistical models, and the choice of the most adequate method depends on the purposes of the computational investigation. Here, we exhibit and discuss some principles and applications of these computational methods widely used to predict the permeability of peptides into biomembranes, exhibiting some of their pharmaceutical and biotechnological applications.

Improved Recovery and Selectivity of Lanthanide-Ion-Binding Cyclic Peptide Hosts by Changing the Position of Acidic Amino Acids

The development of an effective host molecule to separate lanthanide (Ln) ions and a method for predicting its guest recognition/self-assembly behavior based on primary chemical structures are highly sought after in both academia and industry. Herein, we report the improvement of one-pot Ln ion recovery and a performance prediction method for four new cyclic peptide hosts that differ in the position of acidic amino acids. These cyclic peptide hosts could recognize Ln3+ directly through a 1:1 complexation–precipitation process and exhibited high Lu3+ selectivity in spite of similar ion size and electronegativity when the positions of the acidic amino acids were changed. This unpredictable selectivity was explained by considering the dipole moment, lowest unoccupied molecular orbital, and cohesion energy. In addition, a semi-empirical function using these parameters was proposed for screening the sequence and estimating the isolated yields without long-time molecular dynamics calculations. The insights obtained from this study can be employed for the development of high-performance peptides for the selective recovery of Ln and other metal ions, as well as for the construction of diverse supramolecular recognition systems.