Understanding protein sequence-to-function relationship is crucial to assist studies of genetic diseases, protein evolution, and protein engineering. The sequence-to-function relationship of proteins is inherently complex due to multi-site high-dimensional correlation and structural dynamics. Deep learning algorithms such as (graph) convolutional neural networks and recently transformers have become very popular for learning the protein sequence-to-function mapping from deep mutational scanning data and available structures. However, it remains very challenging for these models to achieve accurate extrapolation when predicting functional effect of variants with positions or mutation types not seen in the training data. We propose that incorporating the physics of protein interactions and dynamics can be an effective approach to overcome the extrapolation limitations. Specifically, we demonstrate that biophysics-based modeling can be used to quantify the energetic effects of mutations and that incorporating these physical energetics directly within the convolution and graph convolution neural networks can significantly improve the performance of positional and mutational extrapolation compared to models without biophysics-inspired features. Our results support the effectiveness of leveraging physical knowledge in overcoming the limitation of data scarcity.

Abstract The spatial distribution of small molecules within cells shapes their biological activity, yet these distributions are generally assumed to be governed passively by reaction-driven electrochemical gradients. Here we show that aminoglycoside antibiotics actively control their own subcellular organization by undergoing phase separation with RNAs. Combining in vitro reconstitution, bacterial assays, and molecular dynamics simulations, we discovered that aminoglycosides coacervate with RNA through multivalent electrostatic interactions, displacing and releasing RNA-bound Mg 2+ . This condensate-dependent Mg 2+ release remodels the cytosolic labile Mg 2+ pool and activates magnesium signaling. This effect dampens the magnesium-starvation regulation, sustains ribosome activity, and shifts the cellular electrochemical state, promoting bacterial fitness. Because condensation occurs only above a defined concentration threshold, it generates a non-monotonic dose-response in which higher antibiotic concentrations paradoxically enhance bacterial survival. This antibiotic condensate-dependent Mg 2+ signaling confers tolerance to multiple ribosome-targeting antibiotics simultaneously even in cells lacking resistance gene, while condensate dissolution restores antibiotic efficacy. Our findings establish antibiotic-driven phase separation as a previously unrecognized mechanism to encode cellular signaling and identify antibiotic condensates as a distinct functional unit underlying drug tolerance.

Orthoflavivirus, such as West Nile Virus (WNV), dengue virus (DENV), and ZIKA virus (ZIKV), are globally distributed pathogens that pose substantial threats to human health. Currently, there are still no effective antiviral drugs for WNV or ZIKV. Despite the availability of two licensed DENV vaccines, their use remains limited due to potential risks, highlighting an urgent need for antiviral drug development. The highly conserved orthoflavivirus protease NS2B/NS3 is required for viral replication, making it a promising antiflavivirus target. A major challenge, however, is the highly charged active site of this enzyme, which requires charged chemical matters with low bioavailability. An alternative and more attractive strategy is to target potential allosteric sites or folding intermediate states of the protease. In this work, we employ the topology-based coarse-grained Go̅ modeling to explore the coupled binding and folding pathways of WNV NS2B/NS3 protease and study the effects of the widely used experimental construct with a G4SG4 linker between NS2B and NS3 on stability and folding. Our results provide a holistic conformational landscape of protease binding and folding, including several key intermediate states. We find that the presence of the G4SG4 linker alters the folding pathways and destabilizes the NS2B C-terminus. The latter is consistent with experimental observations that the G4SG4-linked protease has lower activity and adopts an open state without the substrate in crystal structures. Together, these findings provide, for the first time, a complete picture of the binding and folding of the NS2B/NS3 protease and identify important folding intermediate states that could be targeted for allosteric antiviral drug development.

Activation of the chemotaxis kinase CheA, controlled by ligand binding or covalent modification of its associated chemoreceptor, modulates the chemotaxis behavior of bacteria such as Escherichia coli. The activation of CheA, which results in large changes to its autophosphorylation rate, is proposed to involve control of interdomain interactions between the catalytic P4 domain and the substrate P1 domain. While both productive and non-productive interdomain P1/P4 interactions have been identified and modeled, observation of how these interdomain interactions impact activity has not been determined in functional signaling complexes. Here, we report evidence for changes to CheA P1-P4 interactions upon activation based on changes in chemical shift perturbations of NMR resonances of isolated CheA-P1 domains upon addition of signaling complexes of CheA-P3P4P5, chemoreceptor cytoplasmic fragments, and CheW. We found that CheA-P1 interacts non-productively with CheA-P4 in kinase-off complexes, and forms P1/P1’ dimers during this interaction. In contrast, P1 domains in free CheA do not form an observable population of P1/P1’ dimers during their non-productive P1/P4 interaction, so incorporation of CheA into signaling complexes must induce conformational changes to the P4 domain that allow for P1/P1’ dimerization and could further stabilize the non-productive interaction. Activation of CheA in signaling complexes leads to higher affinity interactions between P1 and P4 domains in a productive mode that could lead to higher autophosphorylation activity. The presence of AMPPCP, an ATP-analog, focuses the productive interaction into a more specific binding mode that could be necessary for catalysis. Thus, the activation of CheA in signaling complexes is controlled by changes to the P4 domain that control the transient interactions of P1 and P4 domains. Loss of the non-productive interdomain interaction happens simultaneously with the gain of the high affinity productive interdomain interaction to activate the kinase CheA.

The big potassium (BK) channels remain open with a small limiting probability of P o ~ 10-7 at minimal Ca2+ and negative voltages < -100 mV. The molecular origin of such "intrinsic opening" are not understood. Here, free energy analysis of K+ permeation shows that intrinsic opening of BK channels is likely an inherent property of the vapor barrier, generated by hydrophobic dewetting of the inner pore in the Ca2+-free state. The vapor barrier only gives rise to a finite free energy barrier, of ~ 8 kcal/mol, and cannot completely shut down K+ flow even when the voltage sensor domains are fully deactivated. The resulting "leaking" currents can be measured at negative voltages as intrinsic opening. The shallow limiting slope of P o at negative voltages arises primarily from the electric field effects on the permeating ion through the vapor barrier. We further demonstrate that the vapor barrier can be perturbed by truncation of the cytosolic domains, inner pore mutations, and ligand binding to the pore, leading to predicable changes in limiting slope measurements highly consistent with existing experimental data. Therefore, the intrinsic opening opens up an opportunity for direct experimental study of hydrophobic gating in BK and many other ion channels.

Dysregulated lipid metabolism is a key driver of Alzheimer's disease (AD), yet how membrane lipid composition influences tau-membrane interaction remains poorly understood. Here, we combine single-molecule total internal reflection fluorescence microscopy with atomistic molecular dynamics (MD) simulations to elucidate the molecular basis of tau association with the supported lipid bilayer. NMR titration suggests that tau associates with negatively charged lipid headgroups via electrostatic interactions involving residues ∼120-400, which encompass the positively charged proline-rich region (PRR) and microtubule-binding repeat domains. Importantly, whereas prior studies have generally suggested that cholesterol uniformly enhances protein binding, our work reveals a much more complex and lipid-dependent mechanism: cholesterol suppresses tau binding to phosphatidylcholine-phosphatidylglycerol (PC/PG) bilayers but enhances tau binding to phosphatidylcholine-phosphatidylserine (PC/PS) bilayers. Large-scale all-atom MD simulations with a polybasic model peptide, KR8, accurately recapitulate this dichotomy at the molecular level and further reveal that the contrasting regulatory effects of cholesterol arise from lipid-dependent shifts in the preferred insertion depth of KR8, together with local conformational rearrangements of its membrane-interacting basic residues at the bilayer interface. Given that tau-membrane association contributes to aggregation and prion-like propagation, these results identify a previously unrecognized lipid-specific regulatory mechanism by which cholesterol modulates tau-membrane interactions and provide mechanistic insight into how cholesterol dysregulation contributes to AD pathogenesis.

Cells use post-translational modifications (PTMs) to reconfigure biomolecular condensates across length scales, space, and time.1,2 While charged PTMs are well-known electrostatic switches,3,4 how ubiquitous neutral PTMs shape condensate plasticity and hierarchy remains unclear. Here, we establish a set of design principles for using site-specific lipidation, a class of neutral hydrophobic PTMs, to rationally control properties and interactions of engineered biomolecular condensates. Through systematic analysis of over 80 lipidated synthetic intrinsically disordered proteins (IDPs), we uncovered two distinct axes of control. First, the interplay between the lipid and the local three-residue sequence of its attachment site acts as a programmable switch for cohesion-the homotypic interactions that define the material state of the condensed phase-directing assemblies toward dynamic liquids, arrested gels, or ordered fibrillar solids. Second, the lipid, together with the global properties of the IDP scaffold, tunes adhesion-the heterotypic interactions that govern condensate miscibility and hierarchical organization. We harnessed these principles to rationally engineer complex, multi-phase architectures and create hybrid hydrogels with programmed microstructure and material properties that guide the morphogenesis of functional intestinal organoids. These findings establish a new framework for lipoengineering advanced biomaterials and provide a blueprint for dissecting structure-property relationships across diverse classes of PTMs.

Abstract Investigating the conformational dynamics of intrinsically disordered proteins (IDPs) is essential to understanding how their structural heterogeneity underlies function and how their dysregulation contributes to diseases. Here, we utilized an MspA nanopore-based approach for studying the conformational dynamics and interactions of IDPs at the single-molecule level. The platform was demonstrated using the intrinsically disordered transactivation domain of tumor suppressor p53 (p53-TAD), one of the important proteins in cancer biology. We showed that MspA can stably capture p53-TAD and resolve up to six distinct current states with frequent interconversions, revealing a rich conformational landscape. The nanopore also detected the effect of a cancer-associated double mutational variant, N29K/N30D. Combining experiments with steered molecular dynamics simulations, we showed that the mutant sampled compact conformational states more frequently than wild type, consistent with previous NMR studies. Importantly, the MspA platform enabled direct monitoring of E3 ligase MDM2 binding to p53-TAD and resolved how this interaction is inhibited by anti-cancer compound epigallocatechin gallate (EGCG). Notably, EGCG stabilizes one of the six states sampled by p53-TAD, providing a mechanistic explanation for its inhibitory effect. Together, these findings demonstrate the promise of the nanopore platform for label-free monitoring of IDP conformational dynamics, modulation, binding and inhibition at single-molecule resolution.

Intrinsically disordered proteins and regions (IDPs) are ubiquitous cellular regulators. Uncovering how their transient, multivalent interactions organize and fine-tune cellular processes requires transferable methods capable of deriving dynamic conformational ensembles across diverse environments at scale. Here, we present HyRes, a physics-based, hybrid-resolution protein model with atomistic backbones and intermediate-resolution sidechains that bridges the gap between atomistic accuracy and computational efficiency. Evaluated across ~100 IDPs, HyRes generates atomistic ensembles that match or outperform state-of-the-art all-atom force fields in reproducing experimental chain dimensions, transient tertiary contacts, and local secondary structures. Demonstrating exceptional transferability, HyRes accurately captures dynamic IDP interactions in dilute phases, condensed phases, and amyloid fibril fuzzy coats. Finally, we leverage HyRes' scalability to generate disordered ensembles for ~30,000 IDPs from the human proteome and DisProt, revealing strong correlation between residual structures and cellular function and localization. HyRes and this open-access database provide unprecedented resources for IDP biology and deep learning.

Autophosphorylation of CheA is key to initiation of the phosphorylation cascade that eventually controls the direction of downstream flagellar motors for chemotaxis signaling in motile bacteria. The phospho-transfer reaction, from ATP bound in the P4 catalytic domain to a specific His residue in the P1 substrate domain in CheA, can be significantly accelerated within core signaling unit complexes containing chemoreceptors, CheA and CheW. Previous studies have proposed that CheA autophosphorylation activity is regulated by changing the dynamics of P4 and/or altering its interactions with P1 in response to signals transmitted from chemoreceptors. However, the positions of CheA P1 and P2 domains in the core signaling unit are not well-characterized because they form only transient interactions with other domains. Though previous studies have identified possible domain-domain interaction surfaces, especially for P1 and P4, a bottom-up analysis of these interactions has not been performed. Here, we employed extensive molecular simulations to analyze interactions among CheA domains using a hybrid resolution (HyRes) protein model designed for dynamic protein structures and interactions. The results revealed multiple major modes of dynamic P1/P4 interactions. In particular, P1 was found to bind dynamically near the preferential binding surfaces on P4 even in the ATP free state. ATP binding to P4 reduces the motion of P1 binding and promotes a trans- productive-like mode that brings His48 in close contact with the bound ATP for possible autophosphorylation. The predicted non-productive and productive P1/P4 interaction modes appear highly consistent with existing NMR, mutagenesis, and chemical modification data. Together, these findings provide a more complete picture of the dynamics of domain-domain interactions of CheA and new insights into the possible regulation mechanism of its autophosphorylation.

Amyloid fibrils formed by the protein α-synuclein are implicated in the pathogenesis of synucleinopathies. In addition to their rigid cross-β core, these fibrils have intrinsically disordered regions on their surface, which are important for interactions with other cellular components, such as chaperones. Chaperones play a vital role in preventing and reversing amyloid formation in neurodegenerative diseases. How they recognize misfolded proteins is an active field of research. DNAJB1 is a cochaperone that recognizes fibrils and recruits other chaperones such as Hsp70 and Apg2, which collectively disaggregate fibrils formed by α-synuclein, tau, and huntingtin. Because DNAJB1 was reported to bind the C-terminus of α-synuclein and S129 in this C-terminus is predominantly phosphorylated in patient-derived fibrils, we wanted to determine the effect of this post-translational modification on DNAJB1 binding. Using electron micrographs, NMR spectroscopy, and binding assays, we show that phosphorylation at S129 reduces the dynamics of the intrinsically disordered C-terminus of α-synuclein fibrils and increases the binding of DNAJB1 to this very C-terminus. MD simulations further suggest that the reduced dynamics is due to increased interaction of the phosphorylated C-terminus with the fibril core. DNAJB1 binds the exact same region at the C-terminus, indicating the phosphorylation at S129 might have a dual effect of reducing fibril surface dynamics and increasing chaperone recognition.

Voltage-dependence gating of ion channels underlies numerous physiological and pathophysiological processes, and disruption of normal voltage gating is the cause of many channelopathies. Here, long timescale atomistic simulations were performed to directly probe voltage-induced gating transitions of the big potassium (BK) channels, where the voltage sensor domain (VSD) movement has been suggested to be distinct from that of canonical Kv channels but remains poorly understood. Using a Core-MT construct without the gating ring, multiple voltage activation transitions were observed at 750 mV, allowing detailed analysis of the activated state of BK VSD and key mechanistic features. Even though the S4 helix remains the principal voltage sensor in BK, its vertical displacement is only ~3 Å and accompanied by significant lateral movements. The nature of the predicted VSD movement is in strong agreement with recent Cryo-EM structural studies of mutant BK channels with constitutively activated VSD. Free energy analysis based on the predicted activation transition yielded a total gating charge of 0.44 e per VSD, consistent with the experimental range of 0.48–0.65 e. We further show that the ability of modest physical movements with a small total gating charge to drive effective voltage gating of BK can be attributed to large gradients in the local electric field as reshaped by the protein. Furthermore, the S4 movement is coupled to the pore opening through a non-canonical pathway that involves the tightly packed S4-S5-S6 interface. These distinct mechanistic features may be relevant to voltage gating of other ion channels where VSDs are not domain-swapped with respect to the pore-gate domain.