Prioritizing Virtual Screening with Interpretable Interaction Fingerprints.

Machine learning-based drug discovery success depends on molecular representation. Yet traditional molecular fingerprints omit both the protein and pointers back to structural information that would enable better model interpretability. Therefore, we propose LUNA, a Python 3 toolkit that calculates and encodes protein-ligand interactions into new hashed fingerprints inspired by Extended Connectivity FingerPrint (ECFP): EIFP (Extended Interaction FingerPrint), FIFP (Functional Interaction FingerPrint), and Hybrid Interaction FingerPrint (HIFP). LUNA also provides visual strategies to make the fingerprints interpretable. We performed three major experiments exploring the fingerprints' use. First, we trained machine learning models to reproduce DOCK3.7 scores using 1 million docked Dopamine D4 complexes. We found that EIFP-4,096 performed (R2 = 0.61) superior to related molecular and interaction fingerprints. Second, we used LUNA to support interpretable machine learning models. Finally, we demonstrate that interaction fingerprints can accurately identify similarities across molecular complexes that other fingerprints overlook. Hence, we envision LUNA and its interface fingerprints as promising methods for machine learning-based virtual screening campaigns. LUNA is freely available at https://github.com/keiserlab/LUNA.

Rhapsody: predicting the pathogenicity of human missense variants.

MOTIVATION: The biological effects of human missense variants have been studied experimentally for decades but predicting their effects in clinical molecular diagnostics remains challenging. Available computational tools are usually based on the analysis of sequence conservation and structural properties of the mutant protein. We recently introduced a new machine learning method that demonstrated for the first time the significance of protein dynamics in determining the pathogenicity of missense variants. RESULTS: Here, we present a new interface (Rhapsody) that enables fully automated assessment of pathogenicity, incorporating both sequence coevolution data and structure- and dynamics-based features. Benchmarked against a dataset of about 20 000 annotated variants, the methodology is shown to outperform well-established and/or advanced prediction tools. We illustrate the utility of Rhapsody by in silico saturation mutagenesis studies of human H-Ras, phosphatase and tensin homolog and thiopurine S-methyltransferase. AVAILABILITY AND IMPLEMENTATION: The new tool is available both as an online webserver at http://rhapsody.csb.pitt.edu and as an open-source Python package (GitHub repository: https://github.com/prody/rhapsody; PyPI package installation: pip install prody-rhapsody). Links to additional resources, tutorials and package documentation are provided in the 'Python package' section of the website. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

Complementary computational and experimental evaluation of missense variants in the ROMK potassium channel.

The renal outer medullary potassium (ROMK) channel is essential for potassium transport in the kidney, and its dysfunction is associated with a salt-wasting disorder known as Bartter syndrome. Despite its physiological significance, we lack a mechanistic understanding of the molecular defects in ROMK underlying most Bartter syndrome-associated mutations. To this end, we employed a ROMK-dependent yeast growth assay and tested single amino acid variants selected by a series of computational tools representative of different approaches to predict each variants' pathogenicity. In one approach, we used in silico saturation mutagenesis, i.e. the scanning of all possible single amino acid substitutions at all sequence positions to estimate their impact on function, and then employed a new machine learning classifier known as Rhapsody. We also used two additional tools, EVmutation and Polyphen-2, which permitted us to make consensus predictions on the pathogenicity of single amino acid variants in ROMK. Experimental tests performed for selected mutants in different classes validated the vast majority of our predictions and provided insights into variants implicated in ROMK dysfunction. On a broader scope, our analysis suggests that consolidation of data from complementary computational approaches provides an improved and facile method to predict the severity of an amino acid substitution and may help accelerate the identification of disease-causing mutations in any protein.

Structural dynamics is a determinant of the functional significance of missense variants.

Accurate evaluation of the effect of point mutations on protein function is essential to assessing the genesis and prognosis of many inherited diseases and cancer types. Currently, a wealth of computational tools has been developed for pathogenicity prediction. Two major types of data are used to this aim: sequence conservation/evolution and structural properties. Here, we demonstrate in a systematic way that another determinant of the functional impact of missense variants is the protein's structural dynamics. Measurable improvement is shown in pathogenicity prediction by taking into consideration the dynamical context and implications of the mutation. Our study suggests that the class of dynamics descriptors introduced here may be used in conjunction with existing features to not only increase the prediction accuracy of the impact of variants on biological function, but also gain insight into the physical basis of the effect of missense variants.

Patterns of coevolving amino acids unveil structural and dynamical domains.

Patterns of interacting amino acids are so preserved within protein families that the sole analysis of evolutionary comutations can identify pairs of contacting residues. It is also known that evolution conserves functional dynamics, i.e., the concerted motion or displacement of large protein regions or domains. Is it, therefore, possible to use a pure sequence-based analysis to identify these dynamical domains? To address this question, we introduce here a general coevolutionary coupling analysis strategy and apply it to a curated sequence database of hundreds of protein families. For most families, the sequence-based method partitions amino acids into a few clusters. When viewed in the context of the native structure, these clusters have the signature characteristics of viable protein domains: They are spatially separated but individually compact. They have a direct functional bearing too, as shown for various reference cases. We conclude that even large-scale structural and functionally related properties can be recovered from inference methods applied to evolutionary-related sequences. The method introduced here is available as a software package and web server (spectrus.sissa.it/spectrus-evo_webserver).

Unifying view of mechanical and functional hotspots across class A GPCRs.

G protein-coupled receptors (GPCRs) are the largest superfamily of signaling proteins. Their activation process is accompanied by conformational changes that have not yet been fully uncovered. Here, we carry out a novel comparative analysis of internal structural fluctuations across a variety of receptors from class A GPCRs, which currently has the richest structural coverage. We infer the local mechanical couplings underpinning the receptors' functional dynamics and finally identify those amino acids whose virtual deletion causes a significant softening of the mechanical network. The relevance of these amino acids is demonstrated by their overlap with those known to be crucial for GPCR function, based on static structural criteria. The differences with the latter set allow us to identify those sites whose functional role is more clearly detected by considering dynamical and mechanical properties. Of these sites with a genuine mechanical/dynamical character, the top ranking is amino acid 7x52, a previously unexplored, and experimentally verifiable key site for GPCR conformational response to ligand binding.

EIF3H Orchestrates Hippo Pathway-Mediated Oncogenesis via Catalytic Control of YAP Stability.

EIF3H is presumed to be a critical translational initiation factor. Here, our unbiased screening for tumor invasion factors has identified an unexpected role for EIF3H as a deubiquitylating enzyme that dictates breast tumor invasion and metastasis by modulating the Hippo-YAP pathway. EIF3H catalyzed YAP for deubiquitylation, resulting in its stabilization. Structure-based molecular modeling and simulations coupled with biochemical characterization unveiled a unique catalytic mechanism for EIF3H in dissociating polyubiquitin chains from YAP through a catalytic triad consisting of Asp90, Asp91, and Gln121. Trp119 and Tyr 140 on EIF3H directly interacted with the N-terminal region of YAP1, facilitating complex formation of EIF3H and YAP1 for YAP1 deubiquitylation. Stabilization of YAP via elevated EIF3H promoted tumor invasion and metastasis. Interference of EIF3H-mediated YAP deubiquitylation blocked YAP-induced tumor progression and metastasis in breast cancer models. These findings point to a critical role for YAP regulation by EIF3H in tumor invasion and metastasis. SIGNIFICANCE: This work demonstrates that EIF3H is a novel bona fide deubiquitinase that counteracts YAP ubiquitylation and proteolysis, and stabilization of YAP by EIF3H promotes tumor invasion and metastasis.

Trimerization of dopamine transporter triggered by AIM-100 binding: Molecular mechanism and effect of mutations.

Recent work demonstrated the propensity of dopamine transporters (DATs) to form trimers or higher oligomers, enhanced upon binding a furopyrimidine, AIM-100. AIM-100 binding promotes DAT endocytosis and thereby moderates dopaminergic transmission. Despite the neurobiological significance of these events, the molecular mechanisms that underlie the stabilization of DAT trimer and the key interactions that modulate the trimerization of DAT, and not serotonin transporter SERT, remain unclear. In the present study, we determined three structural models, termed trimer-W238, -C306 and -Y303, for possible trimerization of DATs . To this aim, we used structural data resolved for DAT and its structural homologs that share the LeuT fold, advanced computational modeling and simulations, site-directed mutagenesis experiments and live-cell imaging assays. The models are in accord with the versatility of LeuT fold to stabilize dimeric or higher order constructs. Selected residues show a high propensity to occupy interfacial regions. Among them, D231-W238 in the extracellular loop EL2, including the intersubunit salt-bridge forming pair D231/D232-R237 (not present in SERT) (in trimer-W238), the loop EL3 (trimers-C306 and -Y303), and W497 on the intracellularly exposed IL5 loop (trimer-C306) and its spatial neighbors (e.g. K525) near the C-terminus are computationally predicted and experimentally confirmed to play important roles in enabling the correct folding and/or oligomerization of DATs in the presence of AIM-100. The study suggests the possibility of controlling the effective transport of dopamine by altering the oligomerization state of DAT upon small molecule binding, as a possible intervention strategy to modulate dopaminergic signaling. This article is part of the issue entitled 'Special Issue on Neurotransmitter Transporters'.

The Interplay between Structural Stability and Plasticity Determines Mutation Profiles and Chaperone Dependence in Protein Kinases.

We present a novel comparative analysis of representative protein kinases to characterize the main dynamic and energetic determinants of functional regulation shared among different families. The relationships between stability and plasticity are also used to rationalize kinase tendencies to interact with the molecular chaperone Hsp90. These questions are tackled through newly developed molecular-dynamics-based methods of analysis of internal energy and dynamics applied to a total of 37 different systems, which represent wild-type and mutated proteins, including active and inactive states. Energetic decomposition analysis is coupled to multiple structural alignments and dynamic decomposition methods and identifies, across different families, common elements that underlie fold stabilization and conformational regulation. This analysis also exposes which substructures play a key role in determining chaperone dependence. Overall, the results highlight common interaction networks that underpin kinase stabilization, are modulated by mutations (even if located at a distance), and underlie their tendencies to act as clients or nonclients of Hsp90.

Shared dynamics of LeuT superfamily members and allosteric differentiation by structural irregularities and multimerization.

The LeuT-fold superfamily includes secondary active transporters from different functional families, which share a common tertiary structure, despite having a remarkably low sequence similarity. By identifying the common structural and dynamical features upon principal component analysis of a comprehensive ensemble of 90 experimentally resolved structures and anisotropic network model evaluation of collective motions, we provide a unified point of view for understanding the reasons why this particular fold has been selected by evolution to accomplish such a broad spectrum of functions. The parallel identification of conserved sequence features, localized at specific sites of transmembrane helices, sheds light on the role of broken helices (TM1 and TM6 in LeuT) in promoting ion/substrate binding and allosteric interconversion between the outward- and inward-facing conformations of transporters. Finally, the determination of the dynamics landscape for the structural ensemble provides a promising framework for the classification of transporters based on their dynamics, and the characterization of the collective movements that favour multimerization.This article is part of a discussion meeting issue 'Allostery and molecular machines'.

SPECTRUS: A Dimensionality Reduction Approach for Identifying Dynamical Domains in Protein Complexes from Limited Structural Datasets.

Identifying dynamical, quasi-rigid domains in proteins provides a powerful means for characterizing functionally oriented structural changes via a parsimonious set of degrees of freedom. In fact, the relative displacements of few dynamical domains usually suffice to rationalize the mechanics underpinning biological functionality in proteins and can even be exploited for structure determination or refinement purposes. Here we present SPECTRUS, a general scheme that, by solely using amino acid distance fluctuations, can pinpoint the innate quasi-rigid domains of single proteins or large complexes in a robust way. Consistent domains are usually obtained by using either a pair of representative structures or thousands of conformers. The functional insights offered by the approach are illustrated for biomolecular systems of very different size and complexity such as kinases, ion channels, and viral capsids. The decomposition tool is available as a software package and web server at spectrus.sissa.it.