Incorporating physics to overcome data scarcity in predictive modeling of protein function: A case study of BK channels.

Machine learning has played transformative roles in numerous chemical and biophysical problems such as protein folding where large amount of data exists. Nonetheless, many important problems remain challenging for data-driven machine learning approaches due to the limitation of data scarcity. One approach to overcome data scarcity is to incorporate physical principles such as through molecular modeling and simulation. Here, we focus on the big potassium (BK) channels that play important roles in cardiovascular and neural systems. Many mutants of BK channel are associated with various neurological and cardiovascular diseases, but the molecular effects are unknown. The voltage gating properties of BK channels have been characterized for 473 site-specific mutations experimentally over the last three decades; yet, these functional data by themselves remain far too sparse to derive a predictive model of BK channel voltage gating. Using physics-based modeling, we quantify the energetic effects of all single mutations on both open and closed states of the channel. Together with dynamic properties derived from atomistic simulations, these physical descriptors allow the training of random forest models that could reproduce unseen experimentally measured shifts in gating voltage, ∆V1/2, with a RMSE ~ 32 mV and correlation coefficient of R ~ 0.7. Importantly, the model appears capable of uncovering nontrivial physical principles underlying the gating of the channel, including a central role of hydrophobic gating. The model was further evaluated using four novel mutations of L235 and V236 on the S5 helix, mutations of which are predicted to have opposing effects on V1/2 and suggest a key role of S5 in mediating voltage sensor-pore coupling. The measured ∆V1/2 agree quantitatively with prediction for all four mutations, with a high correlation of R = 0.92 and RMSE = 18 mV. Therefore, the model can capture nontrivial voltage gating properties in regions where few mutations are known. The success of predictive modeling of BK voltage gating demonstrates the potential of combining physics and statistical learning for overcoming data scarcity in nontrivial protein function prediction.

Inner pore hydration free energy controls the activation of big potassium channels.

Hydrophobic gating is an emerging mechanism in regulation of protein ion channels where the pore remains physically open but becomes dewetted to block ion permeation. Atomistic molecular dynamics simulations have played a crucial role in understanding hydrophobic gating by providing the molecular details to complement mutagenesis and structural studies. However, existing studies rely on direct simulations and do not quantitatively describe how the sequence and structural changes may control the delicate liquid-vapor equilibrium of confined water in the pore of the channel protein. To address this limitation, we explore two enhanced sampling methods, namely metadynamics and umbrella sampling, to derive free-energy profiles of pore hydration in both the closed and open states of big potassium (BK) channels, which are important in cardiovascular and neural systems. It was found that metadynamics required substantially longer sampling times and struggled to generate stably converged free-energy profiles due to the slow dynamics of cooperative pore water diffusion even in the barrierless limit. Using umbrella sampling, well-converged free-energy profiles can be readily generated for the wild-type BK channels as well as three mutants with pore-lining mutations experimentally known to dramatically perturb the channel gating voltage. The results show that the free energy of pore hydration faithfully reports the gating voltage of the channel, providing further support for hydrophobic gating in BK channels. Free-energy analysis of pore hydration should provide a powerful approach for quantitative studies of how protein sequence, structure, solution conditions, and/or drug binding may modulate hydrophobic gating in ion channels.

Machine Learning Generation of Dynamic Protein Conformational Ensembles.

Machine learning has achieved remarkable success across a broad range of scientific and engineering disciplines, particularly its use for predicting native protein structures from sequence information alone. However, biomolecules are inherently dynamic, and there is a pressing need for accurate predictions of dynamic structural ensembles across multiple functional levels. These problems range from the relatively well-defined task of predicting conformational dynamics around the native state of a protein, which traditional molecular dynamics (MD) simulations are particularly adept at handling, to generating large-scale conformational transitions connecting distinct functional states of structured proteins or numerous marginally stable states within the dynamic ensembles of intrinsically disordered proteins. Machine learning has been increasingly applied to learn low-dimensional representations of protein conformational spaces, which can then be used to drive additional MD sampling or directly generate novel conformations. These methods promise to greatly reduce the computational cost of generating dynamic protein ensembles, compared to traditional MD simulations. In this review, we examine recent progress in machine learning approaches towards generative modeling of dynamic protein ensembles and emphasize the crucial importance of integrating advances in machine learning, structural data, and physical principles to achieve these ambitious goals.

Physics-based modeling provides predictive understanding of selectively promiscuous substrate binding by Hsp70 chaperones.

To help cells cope with protein misfolding and aggregation, Hsp70 molecular chaperones selectively bind a variety of sequences ("selective promiscuity"). Statistical analyses from substrate-derived peptide arrays reveal that DnaK, the E. coli Hsp70, binds to sequences containing three to five branched hydrophobic residues, although otherwise the specific amino acids can vary considerably. Several high-resolution structures of the substrate -binding domain (SBD) of DnaK bound to peptides reveal a highly conserved configuration of the bound substrate and further suggest that the substrate-binding cleft consists of five largely independent sites for interaction with five consecutive substrate residues. Importantly, both substrate backbone orientations (N- to C- and C- to N-) allow essentially the same backbone hydrogen-bonding and side-chain interactions with the chaperone. In order to rationalize these observations, we performed atomistic molecular dynamics simulations to sample the interactions of all 20 amino acid side chains in each of the five sites of the chaperone in the context of the conserved substrate backbone configurations. The resulting interaction energetics provide the basis set for deriving a predictive model that we call Paladin (Physics-based model of DnaK-Substrate Binding). Trained using available peptide array data, Paladin can distinguish binders and nonbinders of DnaK with accuracy comparable to existing predictors and further predicts the detailed configuration of the bound sequence. Tested using existing DnaK-peptide structures, Paladin correctly predicted the binding register in 10 out of 13 substrate sequences that bind in the N- to C- orientation, and the binding orientation in 16 out of 22 sequences. The physical basis of the Paladin model provides insight into the origins of how Hsp70s bind substrates with a balance of selectivity and promiscuity. The approach described here can be extended to other Hsp70s where extensive peptide array data is not available.

Accelerating the Generalized Born with Molecular Volume and Solvent Accessible Surface Area Implicit Solvent Model Using Graphics Processing Units.

The generalized Born with molecular volume and solvent accessible surface area (GBMV2/SA) implicit solvent model provides an accurate description of molecular volume and has the potential to accurately describe the conformational equilibria of structured and disordered proteins. However, its broader application has been limited by the computational cost and poor scaling in parallel computing. Here, we report an efficient implementation of both the electrostatic and nonpolar components of GBMV2/SA on graphics processing unit (GPU) within the CHARMM/OpenMM module. The GPU-GBMV2/SA is numerically equivalent to the original CPU-GBMV2/SA. The GPU acceleration offers ~60- to 70-fold speedup on a single NVIDIA TITAN X (Pascal) graphics card for molecular dynamic simulations of both folded and unstructured proteins of various sizes. The current implementation can be further optimized to achieve even greater acceleration with minimal reduction on the numerical accuracy. The successful development of GPU-GBMV2/SA greatly facilitates its application to biomolecular simulations and paves the way for further development of the implicit solvent methodology. © 2019 Wiley Periodicals, Inc.

Kikuchi disease presenting with aseptic meningitis.

Kikuchi disease is a self-limited disease characterized primarily by regional lymphadenopathy. Kikuchi disease was first described in 1972 as a lymphadenitis with specific histopathologic findings. Extranodal manifestations have been reported, including rare neurologic complications such as aseptic meningitis. This case report discusses a patient who presented to the ED with signs and symptoms suggestive of aseptic meningitis and was ultimately diagnosed with Kikuchi disease. We also review the epidemiology, clinical presentation, and laboratory findings typically found in patients with Kikuchi disease. Inclusion of Kikuchi disease in the differential diagnosis for meningitis may help establish a diagnosis in patients also presenting with regional lymphadenopathy.

Incorporating physics to overcome data scarcity in predictive modeling of protein function: a case study of BK channels.

Machine learning has played transformative roles in numerous chemical and biophysical problems such as protein folding where large amount of data exists. Nonetheless, many important problems remain challenging for data-driven machine learning approaches due to the limitation of data scarcity. One approach to overcome data scarcity is to incorporate physical principles such as through molecular modeling and simulation. Here, we focus on the big potassium (BK) channels that play important roles in cardiovascular and neural systems. Many mutants of BK channel are associated with various neurological and cardiovascular diseases, but the molecular effects are unknown. The voltage gating properties of BK channels have been characterized for 473 site-specific mutations experimentally over the last three decades; yet, these functional data by themselves remain far too sparse to derive a predictive model of BK channel voltage gating. Using physics-based modeling, we quantify the energetic effects of all single mutations on both open and closed states of the channel. Together with dynamic properties derived from atomistic simulations, these physical descriptors allow the training of random forest models that could reproduce unseen experimentally measured shifts in gating voltage, ΔV 1/2 , with a RMSE ∼ 32 mV and correlation coefficient of R ∼ 0.7. Importantly, the model appears capable of uncovering nontrivial physical principles underlying the gating of the channel, including a central role of hydrophobic gating. The model was further evaluated using four novel mutations of L235 and V236 on the S5 helix, mutations of which are predicted to have opposing effects on V 1/2 and suggest a key role of S5 in mediating voltage sensor-pore coupling. The measured ΔV 1/2 agree quantitatively with prediction for all four mutations, with a high correlation of R = 0.92 and RMSE = 18 mV. Therefore, the model can capture nontrivial voltage gating properties in regions where few mutations are known. The success of predictive modeling of BK voltage gating demonstrates the potential of combining physics and statistical learning for overcoming data scarcity in nontrivial protein function prediction. Author Summary: Deep machine learning has brought many exciting breakthroughs in chemistry, physics and biology. These models require large amount of training data and struggle when the data is scarce. The latter is true for predictive modeling of the function of complex proteins such as ion channels, where only hundreds of mutational data may be available. Using the big potassium (BK) channel as a biologically important model system, we demonstrate that a reliable predictive model of its voltage gating property could be derived from only 473 mutational data by incorporating physics-derived features, which include dynamic properties from molecular dynamics simulations and energetic quantities from Rosetta mutation calculations. We show that the final random forest model captures key trends and hotspots in mutational effects of BK voltage gating, such as the important role of pore hydrophobicity. A particularly curious prediction is that mutations of two adjacent residues on the S5 helix would always have opposite effects on the gating voltage, which was confirmed by experimental characterization of four novel mutations. The current work demonstrates the importance and effectiveness of incorporating physics in predictive modeling of protein function with scarce data.

Using Metadynamics To Explore the Free Energy of Dewetting in Biologically Relevant Nanopores.

Water confined within hydrophobic spaces can undergo cooperative dewetting transitions due to slight changes in water density and pressure that push water toward the vapor phase. Many transmembrane protein ion channels contain nanoscale hydrophobic pores that could undergo dewetting transitions, sometimes blocking the flow of ions without physical blockages. Standard molecular dynamics simulations have been extensively applied to study the behavior of water in nanoscale pores, but the large free energy barriers of dewetting often prevent direct sampling of both wet and dry states and quantitative studies of the hydration thermodynamics of biologically relevant pores. Here, we describe a metadynamics protocol that uses the number of waters within the pore as the collective variable to drive many reversible transitions between relevant hydration states and calculate well-converged free energy profiles of pore hydration. By creating model nanopore systems and changing their radius and morphology and including various cosolvents, we quantify how these pore properties and cosolvents affect the dewetting transition. The results reveal that the dewetting free energy of nanoscale pores is determined by two key thermodynamic parameters, namely, the effective surface tension coefficients of water-air and water-pore interfaces. Importantly, while the effect of salt can be fully captured in the water activity dependence, amphipathic cosolvents such as alcohols modify both dry and wet states of the pore and dramatically shift the wet-dry equilibrium. The metadynamics approach could be applied to studies of dewetting transitions within nanoscale pores of proteins and provide new insights into why different pore properties evolved in biological systems.

Computationally-Aided Modeling of Hsp70-Client Interactions: Past, Present, and Future.

Hsp70 molecular chaperones play central roles in maintaining a healthy cellular proteome. Hsp70s function by binding to short peptide sequences in incompletely folded client proteins, thus preventing them from misfolding and/or aggregating, and in many cases holding them in a state that is competent for subsequent processes like translocation across membranes. There is considerable interest in predicting the sites where Hsp70s may bind their clients, as the ability to do so sheds light on the cellular functions of the chaperone. In addition, the capacity of the Hsp70 chaperone family to bind to a broad array of clients and to identify accessible sequences that enable discrimination of those that are folded from those that are not fully folded, which is essential to their cellular roles, is a fascinating puzzle in molecular recognition. In this article we discuss efforts to harness computational modeling with input from experimental data to develop a predictive understanding of the promiscuous yet selective binding of Hsp70 molecular chaperones to accessible sequences within their client proteins. We trace how an increasing understanding of the complexities of Hsp70-client interactions has led computational modeling to new underlying assumptions and design features. We describe the trend from purely data-driven analysis toward increased reliance on physics-based modeling that deeply integrates structural information and sequence-based functional data with physics-based binding energies. Notably, new experimental insights are adding to our understanding of the molecular origins of "selective promiscuity" in substrate binding by Hsp70 chaperones and challenging the underlying assumptions and design used in earlier predictive models. Taking the new experimental findings together with exciting progress in computational modeling of protein structures leads us to foresee a bright future for a predictive understanding of selective-yet-promiscuous binding exploited by Hsp70 molecular chaperones; the resulting new insights will also apply to substrate binding by other chaperones and by signaling proteins.

Asynchronous learning: student utilization out of sync with their preference.

BACKGROUND: Asynchronous learning is gaining popularity. Data are limited regarding this learning method in medical students rotating in emergency medicine (EM). In EM, faculty time is limited to give in-person lectures. The authors sought to create an online curriculum that students could utilize as an additional learning modality. OBJECTIVE: The goal was to evaluate effectiveness, participation, and preference for this mode of learning. METHODS: We developed five online, narrated PowerPoint presentations. After orientation, access to the online curriculum was provided to the students, which they could review at their leisure. RESULTS: One hundred and seven fourth-year medical students participated. They reported the curriculum to be of high quality. Pretest scores were similar for those that viewed all lectures - compliant group (CG) (9.5 [CI 4.8-14.1]) and those that did not view any - non-compliant group (NCG) (9.6 [CI 5.9-13.4]). There was no statistical significant difference in posttest scores between the groups although there was improvement overall: CG 14.6 (CI 6.9-22.1); NCG 11.4 (CI 5.7-17.1). A majority (69.2%) favored inclusion of asynchronous learning, but less than a quarter (22.4%) reported viewing all five modules and more than a third (36.4%) viewed none. CONCLUSION: Despite student-expressed preference for an online curriculum, they used the online resource less than expected. This should give pause to educators looking to convert core EM topics to an online format. However, when high-quality online lectures are utilized as a learning tool, this study demonstrates that they had neither a positive nor a negative impact on test scores.

Drosophila Bys is nuclear and shows dynamic tissue-specific expression during development.

Although the bys-like family of genes has been conserved from yeast to humans, it is not apparent to what extent the function of Bys-like proteins has been conserved across phylogenetic groups. Human Bystin is thought to function in a novel cell adhesion complex involved in embryo implantation. The product of the yeast bys-like gene, Enp1, is nuclear and has a role in pre-ribosomal RNA (pre-rRNA) splicing and ribosome biogenesis. To gain insight into the function of the Drosophila melanogaster bys-like family member, termed bys, we examined bys mRNA expression and the localization of Bys protein. In embryos, bys mRNA is expressed in a tissue-specific pattern during gastrulation. In the larval wing imaginal disc, bys mRNA is expressed in the ventral and dorsal regions of the wing pouch, regions that give rise to epithelia that adhere to one another after the wing disc everts. The bys mRNA expression patterns could be interpreted as being consistent with a role for Bys in events requiring cell-cell interactions. However, embryonic bys mRNA expression patterns mirror those of genes that are potential targets of the growth regulator Myc and encode nucleolar proteins implicated in cell growth. Additionally, in Schneider line 2 (S2) cells, an epitope-tagged Bys protein is localized to the nucleus, suggesting that Drosophila Bys function may be conserved with that of yeast Enp1.