Bayesian-Maximum-Entropy Reweighting of IDP Ensembles Based on NMR Chemical Shifts

Advancing Molecular Dynamics: Toward Standardization, Integration, and Data Accessibility in Structural Biology

Molecular dynamics (MD) simulations have become a valuable tool in structural biology, offering insights into complex biological systems that are difficult to obtain through experimental techniques alone. The lack of available data sets and structures in most published computational work has limited other researchers' use of these models. In recent years, the emergence of online sharing platforms and MD database initiatives favor the deposition of ensembles and structures to accompany publications, favoring reuse of the data sets. However, the lack of uniform metadata collection, formats, and what data are deposited limits the impact and its use by different communities that are not necessarily experts in MD. This Perspective highlights the need for standardization and better resource sharing for processing and interpreting MD simulation results, akin to efforts in other areas of structural biology. As the field moves forward, we will see an increase in popularity and benefits of MD-based integrative approaches combining experimental data and simulations through probabilistic reasoning, but these too are limited by uniformity in experimental data availability and choices on how the data are modeled that are not trivial to decipher from papers. Other fields have addressed similar challenges comprehensively by establishing task forces with different degrees of success. The large scope and number of communities to represent the breadth of types of MD simulations complicates a parallel approach that would fit all. Thus, each group typically decides what data and which format to upload on servers like Zenodo. Uploading data with FAIR (findable, accessible, interoperable, reusable) principles in mind including optimal metadata collection will make the data more accessible and actionable by the community. Such a wealth of simulation data will foster method development and infrastructure advancements, thus propelling the field forward.

Atomistic molecular simulations of Aβ-Zn conformational ensembles

The amyloid-forming Aβ peptide is able to interact with metal cations to form very stable complexes that influence fibril formation and contribute to the onset of Alzheimer's disease. Multiple structures of peptides derived from Aβ in complex with different metals have been resolved experimentally to provide an atomic-level description of the metal-protein interactions. However, Aβ is intrinsically disordered, and hence more amenable to an ensemble description. Molecular dynamics simulations can now reach the timescales needed to generate ensembles for these type of complexes. However, this requires accurate force fields both for the protein and the protein-metal interactions. Here we use state-of-the-art methods to generate force field parameters for the Zn(II) cations in a set of complexes formed by different Aβ variants and combine them with the Amber99SB*-ILDN optimized force field. Upon comparison of NMR experiments with the simulation results, further optimized with a Bayesian/Maximum entropy approach, we provide an accurate description of the molecular ensembles for most Aβ-metal complexes. We find that the resulting conformational ensembles are more heterogeneous than the NMR models deposited in the Protein Data Bank.

Delineating Structural Propensities of the 4E-BP2 Protein via Integrative Modeling and Clustering

The intrinsically disordered 4E-BP2 protein regulates mRNA cap-dependent translation through interaction with the predominantly folded eukaryotic initiation factor 4E (eIF4E). Phosphorylation of 4E-BP2 dramatically reduces the level of eIF4E binding, in part by stabilizing a binding-incompatible folded domain. Here, we used a Rosetta-based sampling algorithm optimized for IDRs to generate initial ensembles for two phospho forms of 4E-BP2, non- and 5-fold phosphorylated (NP and 5P, respectively), with the 5P folded domain flanked by N- and C-terminal IDRs (N-IDR and C-IDR, respectively). We then applied an integrative Bayesian approach to obtain NP and 5P conformational ensembles that agree with experimental data from nuclear magnetic resonance, small-angle X-ray scattering, and single-molecule Förster resonance energy transfer (smFRET). For the NP state, inter-residue distance scaling and 2D maps revealed the role of charge segregation and pi interactions in driving contacts between distal regions of the chain (∼70 residues apart). The 5P ensemble shows prominent contacts of the N-IDR region with the two phosphosites in the folded domain, pT37 and pT46, and, to a lesser extent, delocalized interactions with the C-IDR region. Agglomerative hierarchical clustering led to partitioning of each of the two ensembles into four clusters with different global dimensions and contact maps. This helped delineate an NP cluster that, based on our smFRET data, is compatible with the eIF4E-bound state. 5P clusters were differentiated by interactions of C-IDR with the folded domain and of the N-IDR with the two phosphosites in the folded domain. Our study provides both a better visualization of fundamental structural poses of 4E-BP2 and a set of falsifiable insights on intrachain interactions that bias folding and binding of this protein.

Hybrid computational methods combining experimental information with molecular dynamics

A goal of structural biology is to understand how macromolecules carry out their biological roles by identifying their metastable states, mechanisms of action, pathways leading to conformational changes, and the thermodynamic and kinetic relationships between those states. Integrative modeling brings structural insights into systems where traditional structure determination approaches cannot help. We focus on the synergies and challenges of integrative modeling combining experimental data with molecular dynamics simulations.

Heterogeneous interactions and polymer entropy decide organization and dynamics of chromatin domains

Chromatin is known to be organized into multiple domains of varying sizes and compaction. While these domains are often imagined as static structures, they are highly dynamic and show cell-to-cell variability. Since processes such as gene regulation and DNA replication occur in the context of these domains, it is important to understand their organization, fluctuation, and dynamics. To simulate chromatin domains, one requires knowledge of interaction strengths among chromatin segments. Here, we derive interaction-strength parameters from experimentally known contact maps and use them to predict chromatin organization and dynamics. Taking two domains on the human chromosome as examples, we investigate its three-dimensional organization, size/shape fluctuations, and dynamics of different segments within a domain, accounting for hydrodynamic effects. Considering different cell types, we quantify changes in interaction strengths and chromatin shape fluctuations in different epigenetic states. Perturbing the interaction strengths systematically, we further investigate how epigenetic-like changes can alter the spatio-temporal nature of the domains. Our results show that heterogeneous weak interactions are crucial in determining the organization of the domains. Computing effective stiffness and relaxation times, we investigate how perturbations in interactions affect the solid- and liquid-like nature of chromatin domains. Quantifying dynamics of chromatin segments within a domain, we show how the competition between polymer entropy and interaction energy influence the timescales of loop formation and maintenance of stable loops.

Integrative Conformational Ensembles of Sic1 Using Different Initial Pools and Optimization Methods

Intrinsically disordered proteins play key roles in regulatory protein interactions, but their detailed structural characterization remains challenging. Here we calculate and compare conformational ensembles for the disordered protein Sic1 from yeast, starting from initial ensembles that were generated either by statistical sampling of the conformational landscape, or by molecular dynamics simulations. Two popular, yet contrasting optimization methods were used, ENSEMBLE and Bayesian Maximum Entropy, to achieve agreement with experimental data from nuclear magnetic resonance, small-angle X-ray scattering and single-molecule Förster resonance energy transfer. The comparative analysis of the optimized ensembles, including secondary structure propensity, inter-residue contact maps, and the distributions of hydrogen bond and pi interactions, revealed the importance of the physics-based generation of initial ensembles. The analysis also provides insights into designing new experiments that report on the least restrained features among the optimized ensembles. Overall, differences between ensembles optimized from different priors were greater than when using the same prior with different optimization methods. Generating increasingly accurate, reliable and experimentally validated ensembles for disordered proteins is an important step towards a mechanistic understanding of their biological function and involvement in various diseases.

Using Unassigned NMR Chemical Shifts to Model RNA Secondary Structure

NMR-derived chemical shifts are sensitive probes of RNA structure. However, the need to assign NMR spectra hampers their utility as a direct source of structural information. In this report, we describe a simple method that uses unassigned 2D NMR spectra to model the secondary structure of RNAs. As in the case of assigned chemical shifts, we could use unassigned chemical shift data to reweight conformational libraries such that the highest weighted structure closely resembles their reference NMR structure. Furthermore, the application of our approach to the 3'- and 5'-UTR of the SARS-CoV-2 genome yields structures that are, for the most part, consistent with the secondary structure models derived from chemical probing data. Therefore, we expect the framework we describe here will be useful as a general strategy for rapidly generating preliminary structural RNA models directly from unassigned 2D NMR spectra. As we demonstrated for the 337-nt and 472-nt UTRs of SARS-CoV-2, our approach could be especially valuable for modeling the secondary structures of large RNA.

Integrative structure determination of histones H3 and H4 using genetic interactions

Integrative structure modeling is increasingly used for determining the architectures of biological assemblies, especially those that are structurally heterogeneous. Recently, we reported on how to convert in vivo genetic interaction measurements into spatial restraints for structural modeling: first, phenotypic profiles are generated for each point mutation and thousands of gene deletions or environmental perturbations. Following, the phenotypic profile similarities are converted into distance restraints on the pairs of mutated residues. We illustrate the approach by determining the structure of the histone H3-H4 complex. The method is implemented in our open-source IMP program, expanding the structural biology toolbox by allowing structural characterization based on in vivo data without the need to purify the target system. We compare genetic interaction measurements to other sources of structural information, such as residue coevolution and deep-learning structure prediction of complex subunits. We also suggest that determining genetic interactions could benefit from new technologies, such as CRISPR-Cas9 approaches to gene editing, especially for mammalian cells. Finally, we highlight the opportunity for using genetic interactions to determine recalcitrant biomolecular structures, such as those of disordered proteins, transient protein assemblies, and host-pathogen protein complexes.

Unifying coarse-grained force fields for folded and disordered proteins

Liquid-liquid phase separation drives the formation of biological condensates that play essential roles in transcriptional regulation and signal sensing. Computational modeling could provide high-resolution structural characterizations of these condensates and help uncover physicochemical interactions that dictate their stability. However, many protein molecules involved in phase separation often contain multiple ordered domains connected with flexible, structureless linkers. Simulating such proteins necessitates force fields with consistent accuracy for both folded and disordered proteins. We provide a critical review of existing coarse-grained force fields for disordered proteins and highlight the challenges in their application to folded proteins. After discussing existing algorithms for force field parameterization, we propose an optimization strategy that should lead to computer models with improved transferability across protein types.

Simultaneous Assignment and Structure Determination of Proteins From Sparsely Labeled NMR Datasets

Sparsely labeled NMR samples provide opportunities to study larger biomolecular assemblies than is traditionally done by NMR. This requires new computational tools that can handle the sparsity and ambiguity in the NMR datasets. The MELD (modeling employing limited data) Bayesian approach was assessed to be the best performing in predicting structures from sparsely labeled NMR data in the 13th edition of the Critical Assessment of Structure Prediction (CASP) event—and limitations of the methodology were also noted. In this report, we evaluate the nature and difficulty in modeling unassigned sparsely labeled NMR datasets and report on an improved methodological pipeline leading to higher-accuracy predictions. We benchmark our methodology against the NMR datasets provided by CASP 13.

Resolving Dynamics in the Ensemble: Finding Paths through Intermediate States and Disordered Protein Structures

Proteins have been found to inhabit a diverse set of three-dimensional structures. The dynamics that govern protein interconversion between structures happen over a wide range of time scales─picoseconds to seconds. Our understanding of protein functions and dynamics is largely reliant upon our ability to elucidate physically populated structures. From an experimental structural characterization perspective, we are often limited to measuring the ensemble-averaged structure both in the steady-state and time-resolved regimes. Generating kinetic models and understanding protein structure-function relationships require atomistic knowledge of the populated states in the ensemble. In this Perspective, we present ensemble refinement methodologies that integrate time-resolved experimental signals with molecular dynamics models. We first discuss integration of experimental structural restraints to molecular models in disordered protein systems that adhere to the principle of maximum entropy for creating a complete set of ensemble structures. We then propose strategies to find kinetic pathways between the refined structures, using time-resolved inputs to guide molecular dynamics trajectories and the use of inference to generate tailored stimuli to prepare a desired ensemble of protein states.

Integrating single-molecule spectroscopy and simulations for the study of intrinsically disordered proteins

Over the last two decades, intrinsically disordered proteins and protein regions (IDRs) have emerged from a niche corner of biophysics to be recognized as essential drivers of cellular function. Various techniques have provided fundamental insight into the function and dysfunction of IDRs. Among these techniques, single-molecule fluorescence spectroscopy and molecular simulations have played a major role in shaping our modern understanding of the sequence-encoded conformational behavior of disordered proteins. While both techniques are frequently used in isolation, when combined they offer synergistic and complementary information that can help uncover complex molecular details. Here we offer an overview of single-molecule fluorescence spectroscopy and molecular simulations in the context of studying disordered proteins. We discuss the various means in which simulations and single-molecule spectroscopy can be integrated, and consider a number of studies in which this integration has uncovered biological and biophysical mechanisms.

On the potential of machine learning to examine the relationship between sequence, structure, dynamics and function of intrinsically disordered proteins

Intrinsically disordered proteins (IDPs) constitute a broad set of proteins with few uniting and many diverging properties. IDPs-and intrinsically disordered regions (IDRs) interspersed between folded domains-are generally characterized as having no persistent tertiary structure; instead they interconvert between a large number of different and often expanded structures. IDPs and IDRs are involved in an enormously wide range of biological functions and reveal novel mechanisms of interactions, and while they defy the common structure-function paradigm of folded proteins, their structural preferences and dynamics are important for their function. We here discuss open questions in the field of IDPs and IDRs, focusing on areas where machine learning and other computational methods play a role. We discuss computational methods aimed to predict transiently formed local and long-range structure, including methods for integrative structural biology. We discuss the many different ways in which IDPs and IDRs can bind to other molecules, both via short linear motifs, as well as in the formation of larger dynamic complexes such as biomolecular condensates. We discuss how experiments are providing insight into such complexes and may enable more accurate predictions. Finally, we discuss the role of IDPs in disease and how new methods are needed to interpret the mechanistic effects of genomic variants in IDPs.

Fitting Side-Chain NMR Relaxation Data Using Molecular Simulations

Proteins display a wealth of dynamical motions that can be probed using both experiments and simulations. We present an approach to integrate side-chain NMR relaxation measurements with molecular dynamics simulations to study the structure and dynamics of these motions. The approach, which we term ABSURDer (average block selection using relaxation data with entropy restraints), can be used to find a set of trajectories that are in agreement with relaxation measurements. We apply the method to deuterium relaxation measurements in T4 lysozyme and show how it can be used to integrate the accuracy of the NMR measurements with the molecular models of protein dynamics afforded by the simulations. We show how fitting of dynamic quantities leads to improved agreement with static properties and highlight areas needed for further improvements of the approach.

Multiscale modeling of genome organization with maximum entropy optimization

Three-dimensional (3D) organization of the human genome plays an essential role in all DNA-templated processes, including gene transcription, gene regulation, and DNA replication. Computational modeling can be an effective way of building high-resolution genome structures and improving our understanding of these molecular processes. However, it faces significant challenges as the human genome consists of over 6 × 109 base pairs, a system size that exceeds the capacity of traditional modeling approaches. In this perspective, we review the progress that has been made in modeling the human genome. Coarse-grained models parameterized to reproduce experimental data via the maximum entropy optimization algorithm serve as effective means to study genome organization at various length scales. They have provided insight into the principles of whole-genome organization and enabled de novo predictions of chromosome structures from epigenetic modifications. Applications of these models at a near-atomistic resolution further revealed physicochemical interactions that drive the phase separation of disordered proteins and dictate chromatin stability in situ. We conclude with an outlook on the opportunities and challenges in studying chromosome dynamics.

Consistent Force Field Captures Homologue-Resolved HP1 Phase Separation

Many proteins have been shown to function via liquid-liquid phase separation. Computational modeling could offer much needed structural details of protein condensates and reveal the set of molecular interactions that dictate their stability. However, the presence of both ordered and disordered domains in these proteins places a high demand on the model accuracy. Here, we present an algorithm to derive a coarse-grained force field, MOFF, which can model both ordered and disordered proteins with consistent accuracy. It combines maximum entropy biasing, least-squares fitting, and basic principles of energy landscape theory to ensure that MOFF recreates experimental radii of gyration while predicting the folded structures for globular proteins with lower energy. The theta temperature determined from MOFF separates ordered and disordered proteins at 300 K and exhibits a strikingly linear relationship with amino acid sequence composition. We further applied MOFF to study the phase behavior of HP1, an essential protein for post-translational modification and spatial organization of chromatin. The force field successfully resolved the structural difference of two HP1 homologues despite their high sequence similarity. We carried out large-scale simulations with hundreds of proteins to determine the critical temperature of phase separation and uncover multivalent interactions that stabilize higher-order assemblies. In all, our work makes significant methodological strides to connect theories of ordered and disordered proteins and provides a powerful tool for studying liquid-liquid phase separation with near-atomistic details.

Refinement of α-Synuclein Ensembles Against SAXS Data: Comparison of Force Fields and Methods

The inherent flexibility of intrinsically disordered proteins (IDPs) makes it difficult to interpret experimental data using structural models. On the other hand, molecular dynamics simulations of IDPs often suffer from force-field inaccuracies, and long simulation times or enhanced sampling methods are needed to obtain converged ensembles. Here, we apply metainference and Bayesian/Maximum Entropy reweighting approaches to integrate prior knowledge of the system with experimental data, while also dealing with various sources of errors and the inherent conformational heterogeneity of IDPs. We have measured new SAXS data on the protein α-synuclein, and integrate this with simulations performed using different force fields. We find that if the force field gives rise to ensembles that are much more compact than what is implied by the SAXS data it is difficult to recover a reasonable ensemble. On the other hand, we show that when the simulated ensemble is reasonable, we can obtain an ensemble that is consistent with the SAXS data, but also with NMR diffusion and paramagnetic relaxation enhancement data.

Full structural ensembles of intrinsically disordered proteins from unbiased molecular dynamics simulations

Molecular dynamics (MD) simulation is widely used to complement ensemble-averaged experiments of intrinsically disordered proteins (IDPs). However, MD often suffers from limitations of inaccuracy. Here, we show that enhancing the sampling using Hamiltonian replica-exchange MD (HREMD) led to unbiased and accurate ensembles, reproducing small-angle scattering and NMR chemical shift experiments, for three IDPs of varying sequence properties using two recently optimized force fields, indicating the general applicability of HREMD for IDPs. We further demonstrate that, unlike HREMD, standard MD can reproduce experimental NMR chemical shifts, but not small-angle scattering data, suggesting chemical shifts are insufficient for testing the validity of IDP ensembles. Surprisingly, we reveal that despite differences in their sequence, the inter-chain statistics of all three IDPs are similar for short contour lengths (< 10 residues). The results suggest that the major hurdle of generating an accurate unbiased ensemble for IDPs has now been largely overcome.

Conformational Ensembles of an Intrinsically Disordered Protein Consistent with NMR, SAXS, and Single-Molecule FRET

Intrinsically disordered proteins (IDPs) have fluctuating heterogeneous conformations, which makes their structural characterization challenging. Although challenging, characterization of the conformational ensembles of IDPs is of great interest, since their conformational ensembles are the link between their sequences and functions. An accurate description of IDP conformational ensembles depends crucially on the amount and quality of the experimental data, how it is integrated, and if it supports a consistent structural picture. We used integrative modeling and validation to apply conformational restraints and assess agreement with the most common structural techniques for IDPs: Nuclear Magnetic Resonance (NMR) spectroscopy, Small-angle X-ray Scattering (SAXS), and single-molecule Förster Resonance Energy Transfer (smFRET). Agreement with such a diverse set of experimental data suggests that details of the generated ensembles can now be examined with a high degree of confidence. Using the disordered N-terminal region of the Sic1 protein as a test case, we examined relationships between average global polymeric descriptions and higher-moments of their distributions. To resolve apparent discrepancies between smFRET and SAXS inferences, we integrated SAXS data with NMR data and reserved the smFRET data for independent validation. Consistency with smFRET, which was not guaranteed a priori, indicates that, globally, the perturbative effects of NMR or smFRET labels on the Sic1 ensemble are minimal. Analysis of the ensembles revealed distinguishing features of Sic1, such as overall compactness and large end-to-end distance fluctuations, which are consistent with biophysical models of Sic1's ultrasensitive binding to its partner Cdc4. Our results underscore the importance of integrative modeling and validation in generating and drawing conclusions from IDP conformational ensembles.

Integrating NMR and simulations reveals motions in the UUCG tetraloop

We provide an atomic-level description of the structure and dynamics of the UUCG RNA stem-loop by combining molecular dynamics simulations with experimental data. The integration of simulations with exact nuclear Overhauser enhancements data allowed us to characterize two distinct states of this molecule. The most stable conformation corresponds to the consensus three-dimensional structure. The second state is characterized by the absence of the peculiar non-Watson-Crick interactions in the loop region. By using machine learning techniques we identify a set of experimental measurements that are most sensitive to the presence of non-native states. We find that although our MD ensemble, as well as the consensus UUCG tetraloop structures, are in good agreement with experiments, there are remaining discrepancies. Together, our results show that (i) the MD simulation overstabilize a non-native loop conformation, (ii) eNOE data support its presence with a population of ≈10% and (iii) the structural interpretation of experimental data for dynamic RNAs is highly complex, even for a simple model system such as the UUCG tetraloop.

How to learn from inconsistencies: Integrating molecular simulations with experimental data

Molecular simulations and biophysical experiments can be used to provide independent and complementary insights into the molecular origin of biological processes. A particularly useful strategy is to use molecular simulations as a modeling tool to interpret experimental measurements, and to use experimental data to refine our biophysical models. Thus, explicit integration and synergy between molecular simulations and experiments is fundamental for furthering our understanding of biological processes. This is especially true in the case where discrepancies between measured and simulated observables emerge. In this chapter, we provide an overview of some of the core ideas behind methods that were developed to improve the consistency between experimental information and numerical predictions. We distinguish between situations where experiments are used to refine our understanding and models of specific systems, and situations where experiments are used more generally to refine transferable models. We discuss different philosophies and attempt to unify them in a single framework. Until now, such integration between experiments and simulations have mostly been applied to equilibrium data, and we discuss more recent developments aimed to analyze time-dependent or time-resolved data.

Computing, Analyzing, and Comparing the Radius of Gyration and Hydrodynamic Radius in Conformational Ensembles of Intrinsically Disordered Proteins

The level of compaction of an intrinsically disordered protein may affect both its physical and biological properties, and can be probed via different types of biophysical experiments. Small-angle X-ray scattering (SAXS) probe the radius of gyration (R_g) whereas pulsed-field-gradient nuclear magnetic resonance (NMR) diffusion, fluorescence correlation spectroscopy, and dynamic light scattering experiments can be used to determine the hydrodynamic radius (R_h). Here we show how to calculate R_g and R_h from a computationally generated conformational ensemble of an intrinsically disordered protein. We further describe how to use a Bayesian/Maximum Entropy procedure to integrate data from SAXS and NMR diffusion experiments, so as to derive conformational ensembles in agreement with those experiments.

Universal Sample Size Invariant Measures for Uncertainty Quantification in Density Estimation

Previously, we developed a high throughput non-parametric maximum entropy method (PLOS ONE, 13(5): e0196937, 2018) that employs a log-likelihood scoring function to characterize uncertainty in trial probability density estimates through a scaled quantile residual (SQR). The SQR for the true probability density has universal sample size invariant properties equivalent to sampled uniform random data (SURD). Alternative scoring functions are considered that include the Anderson-Darling test. Scoring function effectiveness is evaluated using receiver operator characteristics to quantify efficacy in discriminating SURD from decoy-SURD, and by comparing overall performance characteristics during density estimation across a diverse test set of known probability distributions.