Leveraging machine learning models for peptide-protein interaction prediction.

Peptides play a pivotal role in a wide range of biological activities through participating in up to 40% protein-protein interactions in cellular processes. They also demonstrate remarkable specificity and efficacy, making them promising candidates for drug development. However, predicting peptide-protein complexes by traditional computational approaches, such as docking and molecular dynamics simulations, still remains a challenge due to high computational cost, flexible nature of peptides, and limited structural information of peptide-protein complexes. In recent years, the surge of available biological data has given rise to the development of an increasing number of machine learning models for predicting peptide-protein interactions. These models offer efficient solutions to address the challenges associated with traditional computational approaches. Furthermore, they offer enhanced accuracy, robustness, and interpretability in their predictive outcomes. This review presents a comprehensive overview of machine learning and deep learning models that have emerged in recent years for the prediction of peptide-protein interactions.

Energetics of substrate transport in proton-dependent oligopeptide transporters.

The PepT So transporter mediates the transport of peptides across biological membranes. Despite advancements in structural biology, including cryogenic electron microscopy structures resolving PepT So in different states, the molecular basis of peptide recognition and transport by PepT So is not fully elucidated. In this study, we employed molecular dynamics simulations, Markov State Models (MSMs), and Transition Path Theory (TPT) to investigate the transport mechanism of an alanine-alanine peptide (Ala-Ala) through the PepT So transporter. Our simulations revealed conformational changes and key intermediate states involved in peptide translocation. We observed that the presence of the Ala-Ala peptide substrate lowers the free energy barriers associated with transition to the inward-facing state. Furthermore, we elucidated the proton transport model and analyzed the pharmacophore features of intermediate states, providing insights for rational drug design. These findings highlight the significance of substrate binding in modulating the conformational dynamics of PepT So and identify critical residues that facilitate transport.

Structure-function analysis of plant G-protein regulatory mechanisms identifies key Gα-RGS protein interactions.

Heterotrimeric GTP-binding protein alpha subunit (Gα) and its cognate regulator of G-protein signaling (RGS) protein transduce signals in eukaryotes spanning protists, amoeba, animals, fungi, and plants. The core catalytic mechanisms of the GTPase activity of Gα and the interaction interface with RGS for the acceleration of GTP hydrolysis seem to be conserved across these groups; however, the RGS gene is under low selective pressure in plants, resulting in its frequent loss. Our current understanding of the structural basis of Gα:RGS regulation in plants has been shaped by Arabidopsis Gα, (AtGPA1), which has a cognate RGS protein. To gain a comprehensive understanding of this regulation beyond Arabidopsis, we obtained the x-ray crystal structures of Oryza sativa Gα, which has no RGS, and Selaginella moellendorffi (a lycophyte) Gα that has low sequence similarity with AtGPA1 but has an RGS. We show that the three-dimensional structure, protein-protein interaction with RGS, and the dynamic features of these Gα are similar to AtGPA1 and metazoan Gα. Molecular dynamic simulation of the Gα-RGS interaction identifies the contacts established by specific residues of the switch regions of GTP-bound Gα, crucial for this interaction, but finds no significant difference due to specific amino acid substitutions. Together, our data provide valuable insights into the regulatory mechanisms of plant G-proteins but do not support the hypothesis of adaptive co-evolution of Gα:RGS proteins in plants.

Substrate Prediction for RiPP Biosynthetic Enzymes via Masked Language Modeling and Transfer Learning.

Ribosomally synthesized and post-translationally modified peptide (RiPP) biosynthetic enzymes often exhibit promiscuous substrate preferences that cannot be reduced to simple rules. Large language models are promising tools for predicting such peptide fitness landscapes. However, state-of-the-art protein language models are trained on relatively few peptide sequences. A previous study comprehensively profiled the peptide substrate preferences of LazBF (a two-component serine dehydratase) and LazDEF (a three-component azole synthetase) from the lactazole biosynthetic pathway. We demonstrated that masked language modeling of LazBF substrate preferences produced language model embeddings that improved downstream classification models of both LazBF and LazDEF substrates. Similarly, masked language modelling of LazDEF substrate preferences produced embeddings that improved the performance of classification models of both LazBF and LazDEF substrates. Our results suggest that the models learned functional forms that are transferable between distinct enzymatic transformations that act within the same biosynthetic pathway. Our transfer learning method improved performance and data efficiency in data-scarce scenarios. We then fine-tuned models on each data set and showed that the fine-tuned models provided interpretable insight that we anticipate will facilitate the design of substrate libraries that are compatible with desired RiPP biosynthetic pathways.

Functional regulation of aquaporin dynamics by lipid bilayer composition.

With the diversity of lipid-protein interactions, any observed membrane protein dynamics or functions directly depend on the lipid bilayer selection. However, the implications of lipid bilayer choice are seldom considered unless characteristic lipid-protein interactions have been previously reported. Using molecular dynamics simulation, we characterize the effects of membrane embedding on plant aquaporin SoPIP2;1, which has no reported high-affinity lipid interactions. The regulatory impacts of a realistic lipid bilayer, and nine different homogeneous bilayers, on varying SoPIP2;1 dynamics are examined. We demonstrate that SoPIP2;1's structure, thermodynamics, kinetics, and water transport are altered as a function of each membrane construct's ensemble properties. Notably, the realistic bilayer provides stabilization of non-functional SoPIP2;1 metastable states. Hydrophobic mismatch and lipid order parameter calculations further explain how lipid ensemble properties manipulate SoPIP2;1 behavior. Our results illustrate the importance of careful bilayer selection when studying membrane proteins. To this end, we advise cautionary measures when performing membrane protein molecular dynamics simulations.

Binding Position Dependent Modulation of Smoothened Activity by Cyclopamine.

Cyclopamine is a natural alkaloid that is known to act as an agonist when it binds to the Cysteine Rich Domain (CRD) of the Smoothened receptor and as an antagonist when it binds to the Transmembrane Domain (TMD). To study the effect of cyclopamine binding to each binding site experimentally, mutations in the other site are required. Hence, simulations are critical for understanding the WT activity due to binding at different sites. Additionally, there is a possibility that cyclopamine could bind to both sites simultaneously especially at high concentration, the implications of which remain unknown. We performed three independent sets of simulations to observe the receptor activation with cyclopamine bound to each site independently (CRD, TMD) and bound to both sites simultaneously. Using multi-milliseconds long aggregate MD simulations combined with Markov state models and machine learning, we explored the dynamic behavior of cyclopamine's interactions with different domains of WT SMO. A higher population of the active state at equilibrium, a lower activation free energy barrier of ~ 2 kcal/mol, and expansion of the hydrophobic tunnel to facilitate cholesterol transport agrees with the cyclopamine's agonistic behavior when bound to the CRD of SMO. A higher population of the inactive state at equilibrium, a higher free energy barrier of ~ 4 kcal/mol and restricted the hydrophobic tunnel to impede cholesterol transport showed cyclopamine's antagonistic behavior when bound to TMD. With cyclopamine bound to both sites, there was a slightly larger inactive population at equilibrium and an increased free energy barrier (~ 3.5 kcal/mol). The tunnel was slightly larger than when solely bound to TMD, and showed a balance between agonism and antagonism with respect to residue movements exhibiting an overall weak antagonistic effect.

Contribution of Signaling Partner Association to Strigolactone Receptor Selectivity.

The parasitic plant witchweed, Striga hermonthica, results in agricultural losses of billions of dollars per year. It perceives its host via plant hormones called strigolactones, which act as germination stimulants for witchweed. Strigolactone signaling involves substrate binding to the strigolactone receptor, followed by substrate hydrolysis and a conformational change from an inactive, or open state, to an active, or closed state. In the active state, the receptor associates with a signaling partner, MAX2. Recently, it was shown that this MAX2 association process acts as a strong contributor to the uniquely high signaling activity observed in ShHTL7; however, it is unknown why ShHTL7 has enhanced MAX2 association affinity. Using an umbrella sampling molecular dynamics approach, we characterized the association processes of AtD14, ShHTL7, a mutant of ShHTL7, and ShHTL6 with MAX2 homologue OsD3. From these results, we show that ShHTL7 has an enhanced standard binding free energy of OsD3 compared to those of the other receptors. Additionally, our results suggest that the overall topology of the T2/T3 helix region is likely an important modulator of MAX2 binding. Thus, differences in MAX2 association, modulated by differences in the T2/T3 helix region, are a contributor to differences in signaling activity between different strigolactone receptors.

The Experimentalist's Guide to Machine Learning for Small Molecule Design.

Initially part of the field of artificial intelligence, machine learning (ML) has become a booming research area since branching out into its own field in the 1990s. After three decades of refinement, ML algorithms have accelerated scientific developments across a variety of research topics. The field of small molecule design is no exception, and an increasing number of researchers are applying ML techniques in their pursuit of discovering, generating, and optimizing small molecule compounds. The goal of this review is to provide simple, yet descriptive, explanations of some of the most commonly utilized ML algorithms in the field of small molecule design along with those that are highly applicable to an experimentally focused audience. The algorithms discussed here span across three ML paradigms: supervised learning, unsupervised learning, and ensemble methods. Examples from the published literature will be provided for each algorithm. Some common pitfalls of applying ML to biological and chemical data sets will also be explained, alongside a brief summary of a few more advanced paradigms, including reinforcement learning and semi-supervised learning.

Leveraging Machine Learning Models for Peptide-Protein Interaction Prediction.

Peptides play a pivotal role in a wide range of biological activities through participating in up to 40% protein-protein interactions in cellular processes. They also demonstrate remarkable specificity and efficacy, making them promising candidates for drug development. However, predicting peptide-protein complexes by traditional computational approaches, such as Docking and Molecular Dynamics simulations, still remains a challenge due to high computational cost, flexible nature of peptides, and limited structural information of peptide-protein complexes. In recent years, the surge of available biological data has given rise to the development of an increasing number of machine learning models for predicting peptide-protein interactions. These models offer efficient solutions to address the challenges associated with traditional computational approaches. Furthermore, they offer enhanced accuracy, robustness, and interpretability in their predictive outcomes. This review presents a comprehensive overview of machine learning and deep learning models that have emerged in recent years for the prediction of peptide-protein interactions.

Characterization of binding kinetics and intracellular signaling of new psychoactive substances targeting cannabinoid receptor using transition-based reweighting method.

New psychoactive substances (NPS) targeting cannabinoid receptor 1 pose a significant threat to society as recreational abusive drugs that have pronounced physiological side effects. These greater adverse effects compared to classical cannabinoids have been linked to the higher downstream ß-arrestin signaling. Thus, understanding the mechanism of differential signaling will reveal important structure-activity relationship essential for identifying and potentially regulating NPS molecules. In this study, we simulate the slow (un)binding process of NPS MDMB-Fubinaca and classical cannabinoid HU-210 from CB1 using multi-ensemble simulation to decipher the effects of ligand binding dynamics on downstream signaling. The transition-based reweighing method is used for the estimation of transition rates and underlying thermodynamics of (un)binding processes of ligands with nanomolar affinities. Our analyses reveal major interaction differences with transmembrane TM7 between NPS and classical cannabinoids. A variational autoencoder-based approach, neural relational inference (NRI), is applied to assess the allosteric effects on intracellular regions attributable to variations in binding pocket interactions. NRI analysis indicate a heightened level of allosteric control of NPxxY motif for NPS-bound receptors, which contributes to the higher probability of formation of a crucial triad interaction (Y7.53-Y5.58-T3.46) necessary for stronger ß-arrestin signaling. Hence, in this work, MD simulation, data-driven statistical methods, and deep learning point out the structural basis for the heightened physiological side effects associated with NPS, contributing to efforts aimed at mitigating their public health impact.

Adaptive Sampling Methods for Molecular Dynamics in the Era of Machine Learning.

Molecular dynamics (MD) simulations are fundamental computational tools for the study of proteins and their free energy landscapes. However, sampling protein conformational changes through MD simulations is challenging due to the relatively long time scales of these processes. Many enhanced sampling approaches have emerged to tackle this problem, including biased sampling and path-sampling methods. In this Perspective, we focus on adaptive sampling algorithms. These techniques differ from other approaches because the thermodynamic ensemble is preserved and the sampling is enhanced solely by restarting MD trajectories at particularly chosen seeds rather than introducing biasing forces. We begin our treatment with an overview of theoretically transparent methods, where we discuss principles and guidelines for adaptive sampling. Then, we present a brief summary of select methods that have been applied to realistic systems in the past. Finally, we discuss recent advances in adaptive sampling methodology powered by deep learning techniques, as well as their shortcomings.

Resolving binding pathways and solvation thermodynamics of plant hormone receptors.

Plant hormones are small molecules that regulate plant growth, development, and responses to biotic and abiotic stresses. They are specifically recognized by the binding site of their receptors. In this work, we resolved the binding pathways for eight classes of phytohormones (auxin, jasmonate, gibberellin, strigolactone, brassinosteroid, cytokinin, salicylic acid, and abscisic acid) to their canonical receptors using extensive molecular dynamics simulations. Furthermore, we investigated the role of water displacement and reorganization at the binding site of the plant receptors through inhomogeneous solvation theory. Our findings predict that displacement of water molecules by phytohormones contributes to free energy of binding via entropy gain and is associated with significant free energy barriers for most systems analyzed. Also, our results indicate that displacement of unfavorable water molecules in the binding site can be exploited in rational agrochemical design. Overall, this study uncovers the mechanism of ligand binding and the role of water molecules in plant hormone perception, which creates new avenues for agrochemical design to target plant growth and development.

Sequence basis for selectivity of ephrin-B2 ligand for Eph receptors and pathogenic henipavirus G glycoproteins.

IMPORTANCE: Ephrin-B2 (EFNB2) is a ligand for six Eph receptors in humans and regulates multiple cell developmental and signaling processes. It also functions as the cell entry receptor for Nipah virus and Hendra virus, zoonotic viruses that can cause respiratory and/or neurological symptoms in humans with high mortality. Here, we investigate the sequence basis of EFNB2 specificity for binding the Nipah virus attachment G glycoprotein over Eph receptors. We then use this information to engineer EFNB2 as a soluble decoy receptor that specifically binds the attachment glycoproteins of the Nipah virus and other related henipaviruses to neutralize infection. These findings further mechanistic understanding of protein selectivity and may facilitate the development of diagnostics or therapeutics against henipavirus infection.

Functional Regulation of Aquaporin Dynamics by Lipid Bilayer Composition.

With the diversity of lipid-protein interactions, any observed membrane protein dynamics or functions directly depend on the lipid bilayer selection. However, the implications of lipid bilayer choice are seldom considered unless characteristic lipid-protein interactions have been previously reported. Using molecular dynamics simulation, we characterize the effects of membrane embedding on plant aquaporin SoPIP2;1, which has no reported high-affinity lipid interactions. The regulatory impacts of a realistic lipid bilayer, and nine different homogeneous bilayers, on varying SoPIP2;1 dynamics were examined. We demonstrate that SoPIP2;1s structure, thermodynamics, kinetics, and water transport are altered as a function of each membrane construct's ensemble properties. Notably, the realistic bilayer provides stabilization of non-functional SoPIP2;1 metastable states. Hydrophobic mismatch and lipid order parameter calculations further explain how lipid ensemble properties manipulate SoPIP2;1 behavior. Our results illustrate the importance of careful bilayer selection when studying membrane proteins. To this end, we advise cautionary measures when performing membrane protein molecular dynamics simulations.

Sequence controlled secondary structure is important for the site-selectivity of lanthipeptide cyclization.

Lanthipeptides are ribosomally synthesized and post-translationally modified peptides that are generated from precursor peptides through a dehydration and cyclization process. ProcM, a class II lanthipeptide synthetase, demonstrates high substrate tolerance. It is enigmatic that a single enzyme can catalyze the cyclization process of many substrates with high fidelity. Previous studies suggested that the site-selectivity of lanthionine formation is determined by substrate sequence rather than by the enzyme. However, exactly how substrate sequence contributes to site-selective lanthipeptide biosynthesis is not clear. In this study, we performed molecular dynamic simulations for ProcA3.3 variants to explore how the predicted solution structure of the substrate without enzyme correlates to the final product formation. Our simulation results support a model in which the secondary structure of the core peptide is important for the final product's ring pattern for the substrates investigated. We also demonstrate that the dehydration step in the biosynthesis pathway does not influence the site-selectivity of ring formation. In addition, we performed simulation for ProcA1.1 and 2.8, which are well-suited candidates to investigate the connection between order of ring formation and solution structure. Simulation results indicate that in both cases, C-terminal ring formation is more likely which was supported by experimental results. Our findings indicate that the substrate sequence and its solution structure can be used to predict the site-selectivity and order of ring formation, and that secondary structure is a crucial factor influencing the site-selectivity. Taken together, these findings will facilitate our understanding of the lanthipeptide biosynthetic mechanism and accelerate bioengineering efforts for lanthipeptide-derived products.

Distinct activation mechanisms regulate subtype selectivity of Cannabinoid receptors.

Design of cannabinergic subtype selective ligands is challenging because of high sequence and structural similarities of cannabinoid receptors (CB1 and CB2). We hypothesize that the subtype selectivity of designed selective ligands can be explained by the ligand binding to the conformationally distinct states between cannabinoid receptors. Analysis of ~ 700 µs of unbiased simulations using Markov state models and VAMPnets identifies the similarities and distinctions between the activation mechanism of both receptors. Structural and dynamic comparisons of metastable intermediate states allow us to observe the distinction in the binding pocket volume change during CB1 and CB2 activation. Docking analysis reveals that only a few of the intermediate metastable states of CB1 show high affinity towards CB2 selective agonists. In contrast, all the CB2 metastable states show a similar affinity for these agonists. These results mechanistically explain the subtype selectivity of these agonists by deciphering the activation mechanism of cannabinoid receptors.

The Sequence Basis for Selectivity of Ephrin-B2 Ligand for Eph Receptors and Pathogenic Henipavirus G Glycoproteins: Selective Ephrin-B2 Decoys for Nipah and Hendra Virus.

Ephrin-B2 (EFNB2) is a ligand for six Eph receptors in humans and functions as a cell entry receptor for several henipaviruses including Nipah virus (NiV), a pathogenic zoonotic virus with pandemic potential. To understand the sequence basis of promiscuity for EFNB2 binding to the attachment glycoprotein of NiV (NiV-G) and Eph receptors, we performed deep mutagenesis on EFNB2 to identify mutations that enhance binding to NiV-G over EphB2, one of the highest affinity Eph receptors. The mutations highlight how different EFNB2 conformations are selected by NiV-G versus EphB2. Specificity mutations are enriched at the base of the G-H binding loop of EFNB2, especially surrounding a phenylalanine hinge upon which the G-H loop pivots, and at a phenylalanine hook that rotates away from the EFNB2 core to engage Eph receptors. One EFNB2 mutant, D62Q, possesses pan-specificity to the attachment glycoproteins of closely related henipaviruses and has markedly diminished binding to the six Eph receptors. However, EFNB2-D62Q has high residual binding to EphB3 and EphB4. A second deep mutational scan of EFNB2 identified combinatorial mutations to further enhance specificity to NiV-G. A triple mutant of soluble EFNB2, D62Q-Q130L-V167L, has minimal binding to Eph receptors but maintains binding, albeit reduced, to NiV-G. Soluble EFNB2 decoy receptors carrying the specificity mutations were potent neutralizers of chimeric henipaviruses. These findings demonstrate how specific residue changes at the shared binding interface of a promiscuous ligand (EFNB2) can influence selectivity for multiple receptors, and may also offer insight towards the development of henipavirus therapeutics and diagnostics.

The Effects of N-Linked Glycosylation on SLC6 Transporters.

Membrane transporters of the solute carrier 6 (SLC6) family mediate various physiological processes by facilitating the translocation of amino acids, neurotransmitters, and other metabolites. In the body, the activity of these transporters is tightly controlled through various post-translational modifications with implications on protein expression, stability, membrane trafficking, and dynamics. While N-linked glycosylation is a universal regulatory mechanism among eukaryotes, a consistent mechanism of how glycosylation affects the SLC6 transporter family remains elusive. It is generally believed that glycans influence transporter stability and membrane trafficking; however, the role of glycosylation on transporter dynamics remains disputable, with differing conclusions among individual transporters across the SLC6 family. In this study, we collected over 1 ms of aggregated all-atom molecular dynamics (MD) simulation data to systematically identify the impact of N-glycans on SLC6 transporter dynamics. We modeled four human SLC6 transporters, the serotonin, dopamine, glycine, and B0AT1 transporters, by first simulating all possible combinations of a glycan attached to each glycosylation site followed by investigating the effect of larger, oligo-N-linked glycans to each transporter. The simulations reveal that glycosylation does not significantly affect the transporter structure but alters the dynamics of the glycosylated extracellular loop and surrounding regions. The structural consequences of glycosylation on the loop dynamics are further emphasized with larger glycan molecules attached. However, no apparent differences in ligand stability or movement of the gating helices were observed, and as such, the simulations suggest that glycosylation does not have a profound effect on conformational dynamics associated with substrate transport.

Active Learning of the Conformational Ensemble of Proteins Using Maximum Entropy VAMPNets.

Rapid computational exploration of the free energy landscape of biological molecules remains an active area of research due to the difficulty of sampling rare state transitions in molecular dynamics (MD) simulations. In recent years, an increasing number of studies have exploited machine learning (ML) models to enhance and analyze MD simulations. Notably, unsupervised models that extract kinetic information from a set of parallel trajectories have been proposed including the variational approach for Markov processes (VAMP), VAMPNets, and time-lagged variational autoencoders (TVAE). In this work, we propose a combination of adaptive sampling with active learning of kinetic models to accelerate the discovery of the conformational landscape of biomolecules. In particular, we introduce and compare several techniques that combine kinetic models with two adaptive sampling regimes (least counts and multiagent reinforcement learning-based adaptive sampling) to enhance the exploration of conformational ensembles without introducing biasing forces. Moreover, inspired by the active learning approach of uncertainty-based sampling, we also present MaxEnt VAMPNet. This technique consists of restarting simulations from the microstates that maximize the Shannon entropy of a VAMPNet trained to perform the soft discretization of metastable states. By running simulations on two test systems, the WLALL pentapeptide and the villin headpiece subdomain, we empirically demonstrate that MaxEnt VAMPNet results in faster exploration of conformational landscapes compared with the baseline and other proposed methods.

Activation mechanism of the human Smoothened receptor.

Smoothened (SMO) is a membrane protein of the class F subfamily of G protein-coupled receptors (GPCRs) and maintains homeostasis of cellular differentiation. SMO undergoes conformational change during activation, transmitting the signal across the membrane, making it amenable to bind to its intracellular signaling partner. Receptor activation has been studied at length for class A receptors, but the mechanism of class F receptor activation remains unknown. Agonists and antagonists bound to SMO at sites in the transmembrane domain (TMD) and the cysteine-rich domain have been characterized, giving a static view of the various conformations SMO adopts. Although the structures of the inactive and active SMO outline the residue-level transitions, a kinetic view of the overall activation process remains unexplored for class F receptors. We describe SMO's activation process in atomistic detail by performing 300 µs of molecular dynamics simulations and combining it with Markov state model theory. A molecular switch, conserved across class F and analogous to the activation-mediating D-R-Y motif in class A receptors, is observed to break during activation. We also show that this transition occurs in a stage-wise movement of the transmembrane helices: TM6 first, followed by TM5. To see how modulators affect SMO activity, we simulated agonist and antagonist-bound SMO. We observed that agonist-bound SMO has an expanded hydrophobic tunnel in SMO's core TMD, whereas antagonist-bound SMO shrinks this tunnel, further supporting the hypothesis that cholesterol travels through a tunnel inside Smoothened to activate it. In summary, this study elucidates the distinct activation mechanism of class F GPCRs and shows that SMO's activation process rearranges the core TMD to open a hydrophobic conduit for cholesterol transport.