Formation of mammalian synapses entails the precise alignment of presynaptic release sites with postsynaptic receptors but how nascent cell-cell contacts initiate synapse assembly remains unclear. We found that Liprin-α proteins directly link trans-synaptic contacts to downstream assembly of presynaptic specializations. In human neurons lacking all four Liprin-α isoforms, nascent synaptic contacts are formed but recruitment of active zone components and accumulation of synaptic vesicles is blocked, resulting in ‘empty’ boutons and loss of synaptic transmission. Interactions with presynaptic cell adhesion molecules (CAMs) of either the LAR-RPTP family or Neurexins via CASK are required to localize Liprin-α to nascent synaptic sites. Assembly of human presynaptic terminals is thus governed by a hierarchical sequence of events in which the recruitment of Liprin-α proteins by presynaptic CAMs is an essential initial step.

It has been known for decades that chemical synapses of the CNS show tremendous functional and structural diversity. Understanding the molecular mechanisms underlying functional synaptic diversity is a major challenge and has been the focus of our research in the past two decades. We investigated the mechanisms underlying different release probabilities of cerebellar glutamatergic and GABAergic synapses and found that distinct nano-topologies of the presynaptic voltage-gated Ca2+ channels and docked synaptic vesicles (SV) could explain the functional differences. Another example of synaptic diversity is the so-called postsynaptic target cell type-dependent variations in the presynaptic glutamate release in cortical circuits. The probability of glutamate release from hippocampal pyramidal cell (PC) axons that innervate oriens-lacunosum-moleculare (O-LM) interneurons is 10-fold lower than that innervating fast-spiking interneurons (FSINs). Our high-resolution immunolocalization experiments revealed that the nano-topology of synaptic vesicles and voltage-gated Ca2+ channels is very similar in these two hippocampal synapses, revealing that the same mechanism that underlies functional heterogeneity in the cerebellum is not present here. Pharmacological experiments demonstrated that the major difference between these two hippocampal synapses is their differential sensitivity to PDBU, indicating differential priming states of the SV. A sequential two-step priming model of synaptic transmission predicted a 6.5-fold smaller fraction of properly primed SVs at PC – O-LM synapses compared to those at PC – FSIN synapses. Our modelling also predicted that the fusion probability of properly primed SVs is only 40% lower at PC – O-LM synapses. Our results demonstrate that incompletely primed SVs limit the output of PC – O-LM cell synapses and provide evidence for multiple presynaptic mechanisms underlying distinct presynaptic release properties.

Neurotransmitter release is mediated by fusion of synaptic vesicles at defined release sites within the presynaptic active zone. Nanoclusters of the priming protein Munc13 have been shown to correlate with the number of release sites. However, the molecular architecture of the active zone and its plastic regulation remain poorly understood. We therefore investigated the molecular nanostructure of the active zone and the electrophysiological properties of excitatory and inhibitory synapses in cultured neocortical neurons during homeostatic plasticity. We show that both the number of Munc13 nanoclusters and the functional number of release sites are similarly elevated in excitatory synapses during homeostatic plasticity. Our data demonstrate the plastic regulation of Munc13 nanoclusters and further support that synaptic vesicles fuse at Munc13 nanoclusters.