GraphVAMPnets for uncovering slow collective variables of self-assembly dynamics.

Uncovering slow collective variables (CVs) of self-assembly dynamics is important to elucidate its numerous kinetic assembly pathways and drive the design of novel structures for advanced materials through the bottom-up approach. However, identifying the CVs for self-assembly presents several challenges. First, self-assembly systems often consist of identical monomers, and the feature representations should be invariant to permutations and rotational symmetries. Physical coordinates, such as aggregate size, lack high-resolution detail, while common geometric coordinates like pairwise distances are hindered by the permutation and rotational symmetry challenges. Second, self-assembly is usually a downhill process, and the trajectories often suffer from insufficient sampling of backward transitions that correspond to the dissociation of self-assembled structures. Popular dimensionality reduction methods, such as time-structure independent component analysis, impose detailed balance constraints, potentially obscuring the true dynamics of self-assembly. In this work, we employ GraphVAMPnets, which combines graph neural networks with a variational approach for Markovian process (VAMP) theory to identify the slow CVs of the self-assembly processes. First, GraphVAMPnets bears the advantages of graph neural networks, in which the graph embeddings can represent self-assembly structures in high-resolution while being invariant to permutations and rotational symmetries. Second, it is built upon VAMP theory, which studies Markov processes without forcing detailed balance constraints, which addresses the out-of-equilibrium challenge in the self-assembly process. We demonstrate GraphVAMPnets for identifying slow CVs of self-assembly kinetics in two systems: the aggregation of two hydrophobic molecules and the self-assembly of patchy particles. We expect that our GraphVAMPnets can be widely applied to molecular self-assembly.

Disease mutations and phosphorylation alter the allosteric pathways involved in autoinhibition of protein phosphatase 2A.

Mutations in protein phosphatase 2A (PP2A) are connected to intellectual disability and cancer. It has been hypothesized that these mutations might disrupt the autoinhibition and phosphorylation-induced activation of PP2A. Since they are located far from both the active and substrate binding sites, it is unclear how they exert their effect. We performed allosteric pathway analysis based on molecular dynamics simulations and combined it with biochemical experiments to investigate the autoinhibition of PP2A. In the wild type (WT), the C-arm of the regulatory subunit B56δ obstructs the active and substrate binding sites exerting a dual autoinhibition effect. We find that the disease mutant, E198K, severely weakens the allosteric pathways that stabilize the C-arm in the WT. Instead, the strongest allosteric pathways in E198K take a different route that promotes exposure of the substrate binding site. To facilitate the allosteric pathway analysis, we introduce a path clustering algorithm for lumping pathways into channels. We reveal remarkable similarities between the allosteric channels of E198K and those in phosphorylation-activated WT, suggesting that the autoinhibition can be alleviated through a conserved mechanism. In contrast, we find that another disease mutant, E200K, which is in spatial proximity of E198, does not repartition the allosteric pathways leading to the substrate binding site; however, it may still induce exposure of the active site. This finding agrees with our biochemical data, allowing us to predict the activity of PP2A with the phosphorylated B56δ and provide insight into how disease mutations in spatial proximity alter the enzymatic activity in surprisingly different mechanisms.

Multiscale computation delivers organophosphorus reactivity and stereoselectivity to immunoglobulin scavengers.

Quantum mechanics/molecular mechanics (QM/MM) maturation of an immunoglobulin (Ig) powered by supercomputation delivers novel functionality to this catalytic template and facilitates artificial evolution of biocatalysts. We here employ density functional theory-based (DFT-b) tight binding and funnel metadynamics to advance our earlier QM/MM maturation of A17 Ig-paraoxonase (WTIgP) as a reactibody for organophosphorus toxins. It enables regulation of biocatalytic activity for tyrosine nucleophilic attack on phosphorus. The single amino acid substitution l-Leu47Lys results in 340-fold enhanced reactivity for paraoxon. The computed ground-state complex shows substrate-induced ionization of the nucleophilic l-Tyr37, now H-bonded to l-Lys47, resulting from repositioning of l-Lys47. Multiple antibody structural homologs, selected by phenylphosphonate covalent capture, show contrasting enantioselectivities for a P-chiral phenylphosphonate toxin. That is defined by crystallographic analysis of phenylphosphonylated reaction products for antibodies A5 and WTIgP. DFT-b analysis using QM regions based on these structures identifies transition states for the favored and disfavored reactions with surprising results. This stereoselection analysis is extended by funnel metadynamics to a range of WTIgP variants whose predicted stereoselectivity is endorsed by experimental analysis. The algorithms used here offer prospects for tailored design of highly evolved, genetically encoded organophosphorus scavengers and for broader functionalities of members of the Ig superfamily, including cell surface-exposed receptors.

A comprehensive mechanism for 5-carboxylcytosine-induced transcriptional pausing revealed by Markov state models.

RNA polymerase II (Pol II) surveils the genome, pausing as it encounters DNA lesions and base modifications and initiating signals for DNA repair among other important regulatory events. Recent work suggests that Pol II pauses at 5-carboxycytosine (5caC), an epigenetic modification of cytosine, because of a specific hydrogen bond between the carboxyl group of 5caC and a specific residue in fork loop 3 of Pol II. This hydrogen bond compromises productive NTP binding and slows down elongation. Apart from this specific interaction, the carboxyl group of 5caC can potentially interact with numerous charged residues in the cleft of Pol II. However, it is not clear how other interactions between Pol II and 5caC contribute to pausing. In this study, we use Markov state models (a type of kinetic network models) built from extensive molecular dynamics simulations to comprehensively study the impact of 5caC on Pol II translocation. We describe two translocation intermediates with specific interactions that prevent the template base from loading into the Pol II active site. In addition to the previously observed state with 5caC constrained by fork loop 3, we discovered a new intermediate state with a hydrogen bond between 5caC and fork loop 2. Surprisingly, we find that 5caC may curb translocation by suppressing kinking of the helix bordering the active site (the bridge helix) because its high flexibility is critical to translocation. Our work provides new insights into how epigenetic modifications of genomic DNA can modulate Pol II translocation, inducing pauses in transcription.

8-Oxo-guanine DNA damage induces transcription errors by escaping two distinct fidelity control checkpoints of RNA polymerase II.

RNA polymerase II (Pol II) has an intrinsic fidelity control mechanism to maintain faithful genetic information transfer during transcription. 8-Oxo-guanine (8OG), a commonly occurring damaged guanine base, promotes misincorporation of adenine into the RNA strand. Recent structural work has shown that adenine can pair with the syn conformation of 8OG directly upstream of the Pol II active site. However, it remains unknown how 8OG is accommodated in the active site as a template base for the incoming ATP. Here, we used molecular dynamics (MD) simulations to investigate two consecutive steps that may contribute to the adenine misincorporation by Pol II. First, the mismatch is located in the active site, contributing to initial incorporation of adenine. Second, the mismatch is in the adjacent upstream position, contributing to extension from the mismatched bp. These results are supported by an in vitro transcription assay, confirming that 8OG can induce adenine misincorporation. Our simulations further suggest that 8OG forms a stable bp with the mismatched adenine in both the active site and the adjacent upstream position. This stability predominantly originates from hydrogen bonding between the mismatched adenine and 8OG in a noncanonical syn conformation. Interestingly, we also found that an unstable bp present directly upstream of the active site, such as adenine paired with 8OG in the canonical anti conformation, largely disrupts the stability of the active site. Our findings have uncovered two main factors contributing to how 8OG induces transcriptional errors and escapes Pol II transcriptional fidelity control checkpoints.

Conformational selection turns on phenylalanine hydroxylase.

Phenylalanine hydroxylase catalyzes a critical step in the phenylalanine catabolic pathway, and impairment of the human enzyme is linked to phenylketonuria. Phenylalanine is also a positive allosteric regulator of the enzyme, and the allosteric binding site has been determined by crystallography. However, the allosteric activation mechanism remains unclear. Using large-scale simulations to explore how phenylalanine binds to the regulatory site, Ge et al. discovered gating motions of the protein that suggest a conformational selection mechanism.

Searching for signatures of positive selection in cytochrome b gene associated with subterranean lifestyle in fast-evolving arvicolines (Arvicolinae, Cricetidae, Rodentia).

BACKGROUND: Mitochondrial genes encode proteins involved in oxidative phosphorylation. Variations in lifestyle and ecological niche can be directly reflected in metabolic performance. Subterranean rodents represent a good model for testing hypotheses on adaptive evolution driven by important ecological shifts. Voles and lemmings of the subfamily Arvicolinae (Rodentia: Cricetidae) provide a good example for studies of adaptive radiation. This is the youngest group within the order Rodentia showing the fastest rates of diversification, including the transition to the subterranean lifestyle in several phylogenetically independent lineages. RESULTS: We evaluated the signatures of selection in the mitochondrial cytochrome b (cytB) gene in 62 Arvicolinae species characterized by either subterranean or surface-dwelling lifestyle by assessing amino acid sequence variation, exploring the functional consequences of the observed variation in the tertiary protein structure, and estimating selection pressure. Our analysis revealed that: (1) three of the convergent amino acid substitutions were found among phylogenetically distant subterranean species and (2) these substitutions may have an influence on the protein complex structure, (3) cytB showed an increased ω and evidence of relaxed selection in subterranean lineages, relative to non-subterranean, and (4) eight protein domains possess increased nonsynonymous substitutions ratio in subterranean species. CONCLUSIONS: Our study provides insights into the adaptive evolution of the cytochrome b gene in the Arvicolinae subfamily and its potential implications in the molecular mechanism of adaptation. We present a framework for future characterizations of the impact of specific mutations on the function, physiology, and interactions of the mtDNA-encoded proteins involved in oxidative phosphorylation.

Markov State Models to Study the Functional Dynamics of Proteins in the Wake of Machine Learning.

Markov state models (MSMs) based on molecular dynamics (MD) simulations are routinely employed to study protein folding, however, their application to functional conformational changes of biomolecules is still limited. In the past few years, the field of computational chemistry has experienced a surge of advancements stemming from machine learning algorithms, and MSMs have not been left out. Unlike global processes, such as protein folding, the application of MSMs to functional conformational changes is challenging because they mostly consist of localized structural transitions. Therefore, it is critical to properly select a subset of structural features that can describe the slowest dynamics of these functional conformational changes. To address this challenge, we recommend several automatic feature selection methods such as Spectral-OASIS. To identify states in MSMs, the chosen features can be subject to dimensionality reduction methods such as TICA or deep learning based VAMPNets to project MD conformations onto a few collective variables for subsequent clustering. Another challenge for the application of MSMs to the study of functional conformational changes is the ability to comprehend their biophysical mechanisms, as MSMs built for these processes often require a large number of states. We recommend the recently developed quasi-MSMs (qMSMs) to address this issue. Compared to MSMs, qMSMs encode the non-Markovian dynamics via the generalized master equation and can significantly reduce the number of states. As a result, qMSMs can be built with a handful of states to facilitate the interpretation of functional conformational changes. In the wake of machine learning, we believe that the rapid advancement in the MSM methodology will lead to their wider application in studying functional conformational changes of biomolecules.

COGNAT: a web server for comparative analysis of genomic neighborhoods.

BACKGROUND: In prokaryotic genomes, functionally coupled genes can be organized in conserved gene clusters enabling their coordinated regulation. Such clusters could contain one or several operons, which are groups of co-transcribed genes. Those genes that evolved from a common ancestral gene by speciation (i.e. orthologs) are expected to have similar genomic neighborhoods in different organisms, whereas those copies of the gene that are responsible for dissimilar functions (i.e. paralogs) could be found in dissimilar genomic contexts. Comparative analysis of genomic neighborhoods facilitates the prediction of co-regulated genes and helps to discern different functions in large protein families. AIM: We intended, building on the attribution of gene sequences to the clusters of orthologous groups of proteins (COGs), to provide a method for visualization and comparative analysis of genomic neighborhoods of evolutionary related genes, as well as a respective web server. RESULTS: Here we introduce the COmparative Gene Neighborhoods Analysis Tool (COGNAT), a web server for comparative analysis of genomic neighborhoods. The tool is based on the COG database, as well as the Pfam protein families database. As an example, we show the utility of COGNAT in identifying a new type of membrane protein complex that is formed by paralog(s) of one of the membrane subunits of the NADH:quinone oxidoreductase of type 1 (COG1009) and a cytoplasmic protein of unknown function (COG3002). REVIEWERS: This article was reviewed by Drs. Igor Zhulin, Uri Gophna and Igor Rogozin.

COGcollator: a web server for analysis of distant relationships between homologous protein families.

BACKGROUND: The Clusters of Orthologous Groups (COGs) of proteins systematize evolutionary related proteins into specific groups with similar functions. However, the available databases do not provide means to assess the extent of similarity between the COGs. AIM: We intended to provide a method for identification and visualization of evolutionary relationships between the COGs, as well as a respective web server. RESULTS: Here we introduce the COGcollator, a web tool for identification of evolutionarily related COGs and their further analysis. We demonstrate the utility of this tool by identifying the COGs that contain distant homologs of (i) the catalytic subunit of bacterial rotary membrane ATP synthases and (ii) the DNA/RNA helicases of the superfamily 1. REVIEWERS: This article was reviewed by Drs. Igor N. Berezovsky, Igor Zhulin and Yuri Wolf.