Background

Neurons communicate via synapses that constitute the presynaptic bouton, synaptic cleft, and postsynaptic termini. Postsynaptic density (PSD) refers to a densely packed, protein-enriched area that locates just beneath the postsynaptic membrane. It contains thousands of proteins such as transmembrane receptors and channels, scaffold proteins, protein kinases and other enzymes, and cytoskeletal proteins. PSD serves as a signaling hub that converges signals received from the presynaptic termini and translates them into a series of downstream cellular processes, including dynamic translocation of receptors, re-organization of actin cytoskeleton, and regulation of protein degradation machinery, ultimately leading to altered synaptic morphology and function. The tiny compartmental size of neuronal synapses, the great heterogeneity across synapses, and the redundancy and compensation between different signaling pathways all make it more challenging to investigate mechanistic details behind synaptic organization via conventional imaging techniques. Recent studies have suggested that phase separation might provide an explanation to how synaptic compartments, including the presynaptic active zone and PSD, are assembled (Zeng et al., 2016, 2018 and 2019; Milovanovic et al., 2018; Wu et al., 2019 and 2021; McDonald et al., 2020; Pechstein et al., 2020; Bai et al., 2021; Cai et al., 2021; Hosokawa et al., 2021). In previously published studies, we demonstrated that major excitatory PSD (ePSD) scaffold proteins, when mixed in vitro, could readily condense into molecular assemblies via phase separation (Zeng et al., 2018; Feng et al., 2022). This minimal ePSD system reconstituted in vitro is reminiscent of the ePSD assemblies in vivo in many aspects. ePSD condensates could cluster receptors, exclude inhibitory PSD proteins, and be dispersed in the presence of negative regulators. The biochemically reconstituted minimal PSD system, therefore, provides a powerful platform to bridge in vitro observations to cellular functions in vivo. To better mimic the physiological context of semi-membrane-tethered ePSD assemblies, we developed methods for reconstituting microclusters using purified proteins assembled on supported lipid bilayers (SLBs) (Feng et al., 2022). We attached PSD-95 through the interaction of N-terminal His8 tag with Ni-NTA-functionalized lipids incorporated into the bilayer in order to mimic its membrane proximal localization via N-terminal palmitoylation in a physiological context. We also incorporated phosphatidylinositol 4,5-bisphosphate [PI (4,5) P2] lipids into the bilayer to enable insulin receptor substrate protein 53 (IRSp53), a major PSD scaffold protein as well as a PIP2 binder, to localize to the membrane via its Bin/Amphiphysin/RVS (BAR) domain known to bind negatively charged lipids. To enable visualization, we labeled proteins with amide/maleimide-conjugated fluorophores. PSD-95 was uniformly distributed on membranes and readily assembled into nanodomains when other PSD scaffold proteins, SH3 and multiple ankyrin repeat domains 3 (Shank3), guanylate kinase–associated protein (GKAP), IRSp53, and Homer3, were added to trigger phase separation. We incubated all the components with actin in the experimental system. We observed that rhodamine-labeled actin co-localized with the PSD condensates on the membrane and formed thin filament bundles. This reconstitution system allows us to investigate interactions between lipid membranes, membrane-proximal molecular assemblies, and actin cytoskeleton, as well as to recapture complex cellular behaviors observed in vivo.