During direct communication between two cells, the plasma membranes of each cell serve as a platform for ligand-receptor interaction initiating downstream signalling cascades. In immune cell signalling, this cell-cell interface – the immune synapse – is highly spatiotemporally organized. Multiple stimulatory and co-stimulatory signals need to be integrated over time to ensure proper immune cell function. This process is still not fully understood given the vast complexity of interactions between proteins, lipids, glycocalyx and associated cortical actin cytoskeleton. To examine the impact of a single component, the use of model membrane systems has increased. Here, we developed a fully artificial system to study the interface between two vesicles and a semi-artificial one between a live cell and a vesicle to reconstitute 3D contacts. We investigated the distribution and reorganization of immune cell proteins at artificial and semi-artificial contacts. Using our vesicle-vesicle system, we show the enrichment and depletion of different proteins in the synapse. Using the cell-vesicle system we showed how different peptides with varying affinity presented by the same MHC class I affect T cell activation. We further explored the distribution of glycocalyx elements at the cell-cell contact and showed differential partitioning of different sugar moieties in the interface. While we focused on the T cell interface here, our model systems are powerful tools to study distribution and reorganization of lipids, proteins and glycocalyx components at any cell-cell contact.
The Hidden Power of Normalization: Exponential Capacity Control in Deep Neural Networks
arXiv:2511.00958v1 Announce Type: cross Abstract: Normalization methods are fundamental components of modern deep neural networks (DNNs). Empirically, they are known to stabilize optimization dynamics and