Biophysical journal最新文献

Subcellular localization prediction is crucial for understanding protein functions and cellular processes. Subcellular localization is dependent on tissue and cell lines derived from different cell types. Predicting cell line-specific subcellular localization using the information of protein-protein interactions (PPIs) offers deeper insights into dynamic cellular organization and molecular mechanisms. However, many existing PPI networks contain systematic errors that limit prediction accuracy. In this study, we propose a reinforcement learning approach, ProtLoc-GRPO, to enhance subcellular localization prediction by optimizing the structure of the underlying PPI network. ProtLoc-GRPO learns to rank and retain the most informative PPI edges to maximize the macro-F1 score for cell line-specific subcellular localization. Our approach yields a 7% improvement in macro-F1 score over the baseline. We further evaluate its robustness across various edge pruning rates and benchmark it against conventional pruning strategies. Results show that our proposed method consistently outperforms existing approaches. To our knowledge, this work represents the first sequence-based study to predict cell line-specific protein subcellular localization and the first application of the Group Relative Policy Optimization (GRPO) framework to a graph-based model for bioinformatics tasks.

Biomolecular condensates exhibit a wide range of viscoelastic properties, shaped by their molecular sequences and compositions. Coarse-grained molecular models of intrinsically disordered proteins are widely used to complement experimental data by revealing the structures and thermodynamics of condensates. However, fully flexible chain representations of intrinsically disordered proteins often fail to capture their complex viscoelastic behavior, instead predicting purely viscous responses. In this work, we demonstrate that introducing sequence-dependent chain rigidity enables us to reproduce the experimentally observed trends in the elastic and viscous moduli for A1-LCD condensates and their variants. Furthermore, we show that the frequency-dependent loss factor can be characterized by a single descriptor that correlates with viscosity across A1-LCD variants and diverse parameter settings within a single-bead, semiflexible coarse-grained model. We further find that increasing backbone rigidity expands the elastic-dominated frequency range and is accompanied by more extended condensate-phase conformations. Finally, we elucidate the microscopic origins of sequence-encoded viscoelasticity by demonstrating how it can be tuned through sequence rearrangements that promote the formation of sticker clusters.

When active, NMDA receptors pass Ca²⁺-rich excitatory currents that are essential for the normal development and function of the central nervous system. In turn, fluctuations in extracellular Ca²⁺ levels, as observed during synaptic activity and pathological states, affect the NMDA receptor gating kinetics and conductance. Here, we used patch-clamp electrophysiology, kinetic analyses, and mutagenesis to evaluate how changes in the ambient Ca²⁺ concentration affect the sensitivity of recombinant NMDA receptors to open-channel blockers. NMDA receptor currents are characteristically sensitive to voltage-dependent block by Mg²⁺, which endows them physiologically, with coincidence detection. This regulatory mechanism is shared with ketamine and memantine, two synthetic compounds that are clinically effective for treating depression and Alzheimer's disease, respectively. We found that extracellular Ca²⁺ increased the sensitivity of NMDA receptors to block by Mg²⁺ and memantine, but not by ketamine. Further, the effect of Ca²⁺ on block by memantine required intracellular Ca²⁺ and functional calmodulin, whereas the effect of Ca²⁺ on block by Mg²⁺ required the extracellular residue GluN1-D658. We conclude that extracellular Ca²⁺ fluctuations modulate the sensitivity of NMDA receptors to open-channel blockers by discrete mechanisms, which may explain the distinct clinical profiles of NMDA receptor blockers.

In 1952, Alan Turing showed that analog reaction-diffusion equations were extremely powerful models of biological development and of distributed cellular automata. Analog circuits have been shown to accelerate the simulation of chemical reactions by many orders of magnitude, including in stochastic (noisy) cytomorphic chips, which are useful for drug-cocktail formulation and in systems medicine. However, the simulation of the partial differential equations of diffusion is expensive to architect in analog systems. Here, we show how to simulate diffusion as though it were a chemical reaction such that reaction-reaction analog systems can simulate reaction-diffusion systems. As an example, we show that the BMP-SOX9-WNT reaction-diffusion system can be simulated in analog cytomorphic integrated circuits that only have reaction circuits but no explicit diffusion circuits. Experimental data from reaction-reaction circuits show excellent agreement with MATLAB and COPASI simulations of biological models. Even cytomorphic chips with relatively sparse sampling appear to demonstrate decaying and unstable growing waves. Our work is the first step toward large-scale simulations of spatiotemporal reaction-diffusion equations.

The apparent bending moduli (K_C) of bilayers composed of binary mixtures of lipids with different spontaneous curvatures have been obtained using x-ray diffuse scattering (XDS). The mixtures that were studied are POPC/POPE, POPC/POPA, POPC/POPS, and DLPC/DiPhyPC. The data are qualitatively consistent with what is expected from the theory of diffusional softening for lipids with different spontaneous curvatures. However, the derived spontaneous curvature differences are larger than those obtained from the hexagonal_II (H_II) phase and from a recent giant unilamellar vesicle (GUV) study. We propose that the interactions between lipids, which we have added to the theory, also play an important role in the values of K_C obtained at the short length scale of XDS. Inclusion of a mean field term in the analysis brings the calculated difference in spontaneous curvatures ΔC₀ of the two lipids closer to the values from the H_II and GUV methods. The use of XDS opens a new experimental window on diffusional softening and the interactions between lipids in mixtures.

Rapid diffusion of membrane lipids and membrane proteins in living cell plasma membranes demonstrates that the membrane is fluid. However, motion of membrane molecules is inhibited on one side by the cytoskeletal mesh and on the other by the glycocalyx, a layer of proteoglycans carrying polysaccharide chains that covers the membrane surface. Events such as blood circulation, cilia-driven flows, and the swimming motion of microorganisms apply shear stress to cell surfaces. Cell responses to these flows govern important physiological processes such as blood pressure and immune activation. The presence of the glycocalyx is generally thought to shield cell membranes from shear stress that arises from flow. However, here we show that two different proteins, each attached by a lipid anchor to the extracellular membrane surface of living COS-7 cells, formed reversible, cell-wide concentration gradients in the direction of applied flow. Protein redistribution occurred within minutes after we applied shear stress levels commonly found in animal cardiovascular systems. The dynamic and spatial features of these gradients were consistent with passive transport by flow. Passive flow transport could be a general mechanism for spatial organization of membrane proteins. This mechanism may explain protein patterning previously observed on flow-exposed cells and potentially forms an initial step in flow sensing.

Single-molecule manipulation techniques are used to elucidate mechanisms in biological systems. Optical tweezers are powerful tools because of their ease of use in combination with optical microscopy and appropriate torque range. However, the use of optical tweezers to generate rotational motion is difficult owing to the complexity of applying constant torque to a moving molecule. The magnitude of the torque applied with optical tweezers depends on the positional relationship with the trapping particle and requires positional feedback. In this study, we found that the adaption of optical vortices (OVs) generated by phase modulation of optical tweezers enabled quantitative mechanical manipulations. Moreover, optical tweezers with an OV could be applied to measure the torque generated by a molecular motor. We used an OV to apply torque via a precise handle consisting of a DNA origami rod to a rotating molecular motor, F₁-ATPase. Using the constant torque generated by the OV, we applied torques to F₁-ATPase and succeeded in stalling and reversing its rotation. This technique is useful for applying constant torque to biomolecules.

Cells sense substrate mechanical properties through the integrin-talin-F-actin linkage. Talin's N-terminal head domain binds β-integrin, whereas its C-terminal domain connects to F-actin directly via two actin-binding sites (ABSs) and indirectly through cryptic vinculin-binding sites (VBSs) within rod domain bundles. Force-induced unfolding of these α-helical bundles exposes VBSs, recruiting vinculin to strengthen the talin-actin bond. This system is sensitive to the loading rate and is influenced by rates of F-actin movement and substrate stiffness. Although the components of this pathway are well studied, how talin, vinculin, and actin synergize to mechanically buffer loads and mediate cellular stiffness sensing remains incompletely understood. We developed a multiscale stochastic finite element model to simulate talin unfolding during interactions with retrograde actin flows and analyzed the contributions of ABS2, ABS3, and VBSs to talin mechanosensitivity. Vinculin attachments strengthened the force-bearing capacity in talin, stabilized the actin-talin contact, and regulated binding site activity at RD3. Lifetime of the dynamic bond formed between talin and actin decreased with an increase in actin flow velocity. Higher substrate stiffness enhanced the lifetime at low actin flow velocity but negatively impacted it at higher velocities. ABS3 primarily mediated force transfer from actin to talin at rapid actin flows, whereas vinculin and ABS2 reinforced the F-actin bond under slower flows. Stiffer substrates enhanced force transmission through VBSs. Our results show that stretch rate modulates force feedback between the unfolding of talin rod domains and VBS attachments, driving the sensitivity of talin to substrate stiffness.