The study of lipid bilayers has lagged behind that of other complex molecular assemblies due to the dynamic nature of lipid bilayers and the paucity of experimental techniques to study them in atomic detail. This is particularly true when analyzing interactions between lipid components of membranes. We now report a multi-resolution in silico analysis of a lipid bilayer based on the asymmetric lipid composition of the human erythrocyte plasma membrane. Our approach combines a coarse-grained bilayer simulation with resolution transformation to multiple smaller all-atom simulations. The coarse-grained simulation was consistent with previously published studies, as demonstrated by analysis of cholesterol translocation and neighbor enrichment. Subsequently, small patches from the coarse-grained simulations were transformed to all-atom resolution; quality control analysis of this data was again consistent with area per lipid and chain orientational order measurements of model bilayers. All-atom simulations permitted us to examine intermolecular and intramolecular hydrogen-bonding with a particular focus on sphingolipids, which unlike glycerophospholipids, donate hydrogen bonds in the bilayer interface region. We determined the relative contribution of each donor and acceptor hydrogen bond in lipid-lipid interactions and took advantage of atomic level resolution to study hydrogen-bonding for d-erythro-, 3-keto-, phyto- and dihydro-sphingolipids. Removing one of the hydrogen bond donors in sphingomyelin (SM) (by modeling 3-keto-SM rather than d-erythro-SM) resulted in an unexpected compensatory increase in hydrogen bond donation by the amide proton to neighboring SM. Phyto-SM increased hydrogen-bonding by the 3-position hydroxyl, while dihydro-SM reduced donation to cholesterol. Our results not only illustrate the usefulness of all-atom simulations to reveal membrane interactions less accessible by other methods, but also strengthen our recent suggestions that lipid bilayers are finely-tuned molecular assemblies in which atomic interactions play critical roles in maintaining and regulating membrane structure and function.
The voltage-gated potassium channel Kv2.1, encoded by the epileptic encephalopathy-associated gene KCNB1, is a primary driver of delayed-rectifier K+ currents in neurons. These currents contribute to high-frequency firing by preventing depolarization block due to Na+ channel inactivation. Wild-type (WT) channels are localized at the soma, proximal dendrites, and the initial segment of the axon, forming aggregates (clusters) via their C-terminal proximal restriction and clustering domain (PRC). This study investigated the biophysical and functional consequences of two C-terminal truncation mutations (Y529∗ and R579∗), identified in patients with epileptic encephalopathy, which disrupt this critical clustering domain. The mutations induced a change of Kv2.1 subcellular distribution towards distal dendrites and the axon. Cluster formation was impaired, though not abolished, in neurons expressing mutated subunits together with endogenous WT subunits. Consistent with this clustering deficit, single-molecule imaging revealed altered lateral diffusion of mutated channels. While WT channels remained largely immobile (stabilized), mutated forms exhibited intermittent diffusion punctuated by transient immobilization events. The percentage of stabilized trajectories was lower for mutated channels in the soma, dendrites and distal axons but not in the proximal portion of the axons. The differential diffusive behavior of WT and mutated channels was reproduced by diffusion-capture simulations considering channels with different numbers of PRC domains (i.e. heterotetramers with mutant subunits) and labile scaffolding interactions. Patch-clamp recordings revealed no significant difference in excitability between WT- and mutant-expressing neurons. However, we observed increased firing in both conditions compared to non-transfected neurons. Altogether, our results suggest that 1) mutated subunits form heterotetramers with endogenous WT subunits; 2) several PRC per channel are needed to efficiently immobilize Kv2.1 in large clusters; 3) increasing Kv2.1 channel expression level, independently of clustering, could drive neuronal hyperexcitability.
Ca2+ binding regulates muscle contraction through coupled structural and dynamical mechanisms, yet a unified description of mutation-induced perturbations in binding energetics remains incomplete. Mutational effects on proteins are often high-dimensional and difficult to interpret mechanistically. Here, we develop a physically interpretable low-dimensional reaction coordinate that captures coordination environment, electrostatic features, and dynamical coupling. Structural, electrostatic, and dynamical descriptors correlate with experimental binding free energies with Pearson coefficients of Rp = 0.83, 0.83, and 0.85, respectively. Integration of these domains improves agreement with experiment (Rp = 0.90), indicating substantial dimensional compression of the underlying descriptor space. This framework provides a physically grounded description of how mutations reshape Ca2+ binding energetics in a minimal EF-hand system.
The SARS-CoV-2 main protease (MPro) is an essential enzyme for viral replication and a primary target for antiviral drug development. Despite extensive structural and biochemical characterization, the allosteric mechanisms by which dimerization informs conformational changes at active site lack an explicit comparison across the different states that identify key residues that connect substrate binding, dimerization, and catalytic activation. Here, we integrate microsecond time scale all-atom molecular dynamics (MD) simulations with dynamical network analysis to characterize how ligand binding and dimerization modulate the allosteric communication landscape of MPro. We performed triplicate 1-μs simulations of MPro in the monomer and dimer states. For each of these states, we simulated MPro in the apo state, as well as bound to a natural peptide substrate (nsp 15/16), the covalent inhibitor nirmatrelvir (Paxlovid) and the noncovalent inhibitor ensitrelvir (Xocova). Dynamical network analyses from the resulting simulations reveal that dimerization redirects the highest correlated motions from the interdomain loop towards the domain II and III interface. At the dimer interface, we identified N-terminal and domain II β-hairpin residues that act as central communication hubs creating networks that connect both chains in the dimer. Small molecule binding to the active site further modulates these networks in distinct ways: nirmatrelvir and peptide substrate binding results in the formation of allosteric networks within the oxyanion loop, while ensitrelvir-bound monomeric MPro results in a dimer-like network, suggesting an inhibitory “allosteric switch” mechanism that may hinder dimerization upon binding. Across all systems, domain III emerges as an allosteric “pivot”, providing a platform that allows the most relevant networks to connect inter-chain communication to the active site upon dimerization. Together, these findings define how correlated motion networks couple active-site dynamics to dimerization and ligand binding, providing molecular insight into the principles governing allosteric regulation in MPro. This framework highlights potential avenues for developing antivirals that target not only the catalytic site but also the communication pathways sustaining dimer stability and enzymatic function.
Most cancer deaths result from metastasis, yet only a subset of tumor cells can complete this process. Polyaneuploid cancer cells (PACCs), which arise via endoreplication under stressors such as hypoxia, are implicated as metastatic drivers, but how they acquire this capacity remains unclear.Here, we show that under extreme self-generated hypoxia, the polyaneuploid subset of PC3 prostate cancer cells upregulates a hypoxia-responsive, HIF-1α-dependent RhoA signaling axis that promotes aerotaxis, a behavior predictive of intravasation during metastasis. We observe an enrichment of PACCs using a membrane-based culture system in which extreme, self-generated radial oxygen gradients emerge as cellular oxygen demand exceeds supply, recapitulating key features of the tumor microenvironment. Under these conditions, PACCs exhibited elevated HIF-1α and RhoA expression relative to non-PACCs. Disruption of either HIF-1α or RhoA attenuates aerotaxis, with RhoA expression suppressed upon HIF-1α inhibition, suggesting a functional HIF-1α- dependent RhoA pathway selectively enhanced in the PACC state.Quantitative analysis of nuclear morphology further reveals that nuclear circularity and solidity, morphology metrics associated with nuclear rounding and boundary organization, increase with and saturate at high nuclear area. Notably, the hypoxic PACC population exhibits significantly greater nuclear circularity and solidity compared to non-PACCs, suggesting increased rounding and stabilization of the enlarged nuclear architecture characteristic of PACCs under extreme hypoxia.Together, these findings reveal a new functional role for HIF-1α- dependent RhoA signaling in regulating aerotaxis in polyaneuploid cancer cells, which may enable aerotactic escape from hypoxic tumor cores and entry into oxygen-rich vasculature. Additionally, the association between elevated RhoA signaling and increased nuclear circularity at larger nuclear area suggests a potential link between hypoxia-responsive migration and maintenance of enlarged nuclear architecture in the PACC state. Collectively, these characteristics highlight PACCs as a hypoxia-adapted subpopulation with relevance for anti-metastatic therapeutic strategies.
