Multicellular biological systems, particularly living neural networks, exhibit highly complex organization properties that pose difficulties for building cell-specific biocompatible interfaces. We previously developed an approach to genetically program cells to assemble structures that modify electrical properties of neurons in situ, opening up the possibility of building minimally invasive cell-specific structures and interfaces. However, the efficiency and biocompatibility of this approach were challenged by limited membrane targeting of the constructed materials. Here, we design a method for highly localized expression of enzymes targeted to the plasma membrane of primary neurons, with minimal intracellular retention. Next, we show that polymers synthesized in situ by this approach form dense extracellular clusters selectively on the targeted cell membrane and that neurons remain viable after polymerization. Last, we show generalizability of this method across a range of design strategies. This platform can be readily extended to incorporate a broad diversity of materials onto specific cell membranes within tissues and may further enable next-generation biological interfaces.
Multicellular biological systems, most notably living neural networks, exhibit highly complex physical organization properties that pose challenges for building cell-specific and biocompatible interfaces. We developed a novel approach to genetically program cells to chemically assemble artificial structures that modify the electrical properties of neurons in situ , opening up the possibility of minimally-invasive cell-specific interfaces with neural circuits in living animals. However, the efficiency and biocompatibility of this approach were challenged by limited membrane targeting of the constructed material. Here, we report a method with significantly improved molecular construct properties, which expresses highly localized enzymes targeted to the plasma membrane of primary neurons with minimal intracellular retention. Polymers synthesized in situ by this approach form dense clusters on the targeted cell membrane, and neurons remain viable after polymerization. This platform can be readily extended to incorporate a broad range of materials onto the surface membranes of specific cells within complex tissues, using chemistry that may further enable the next generation of interfaces with living biological systems.
Accurate prediction of protein-ligand complexes remains a central challenge in structural biology. Traditional methods are computationally inefficient and prone to local minima, whereas deep learning approaches struggle to capture structural flexibility and physical plausibility. We introduce PhysDock, a physics-guided diffusion model that uniquely integrates (i) all-atom diffusion to model ligand flexibility and protein precision-flexibility ( i . e ., subtle conformational adjustments); (ii) physical priors as diffusion conditioning, alongside two-phase physics guidance during the denoising diffusion to ensure physical plausibility. PhysDock demonstrates state-of-the-art performance in redocking benchmarks and excels in the more challenging cross-docking assessments. For practical utility, PhysDock (i) resolves cannabinoid receptor selectivity across diverse molecules, achieving accuracy comparable to experiments; (ii) distinguishes most drug candidates from weak binders in virtual screening of NTRK3 kinase, while uncovering novel candidates with structural insights. PhysDock serves as a versatile tool for protein-ligand complex prediction, with substantial potential to accelerate structure-based drug discovery.