Dynamic aspects of CNS synapse formation

Abstract

The mammalian central nervous system (CNS) requires the proper formation of exquisitely precise circuits to function correctly. These neuronal circuits are assembled during development by the formation of synaptic connections between thousands of differentiating neurons. Proper synapse formation during childhood provides the substrate for cognition, whereas improper formation or function of these synapses leads to neurodevelopmental disorders, including mental retardation and autism. Recent work has begun to identify some of the early cellular events in synapse formation as well as the molecular signals that initiate this process. However, despite the wealth of information published on this topic in the past few years, some of the most fundamental questions about how, whether, and where glutamatergic synapses form in the mammalian CNS remain unanswered. This review focuses on the dynamic aspects of the early cellular and molecular events in the initial assembly of glutamatergic synapses in the mammalian CNS.

Figure 1

CNS glutamarergic synapses are comprised of several major protein classes. Synaptic vesicles containing the neurotransmitter glutamate cycle at the active zone, which is composed of many kinds of proteins including presynaptic scaffolding proteins. The presynaptic terminal is separated from the postsynaptic dendrite by the synaptic cleft; a number of trans-synaptic adhesion molecules span this cleft, providing a molecular connection between the pre-and postsynaptic membranes. Glutamate receptors, including AMPA and NMDA receptors, are found in rhe postsynaptic membrane, where they are associated with a large number of scaffolding and signaling proteins that together comprise the postsynaptic density. Although glutamatergic synapses are usually located on dendritic spines in the adult, these synapses are more often found on dendritic shafts and filopodia in the CNS during the initial stages of synaptogenesis.

Figure 2

Multiple cellular mechanisms for CNS synaptogenesis. There appear to be multiple mechanisms for the recruitment and stabilization of pre-and postsynaptic proteins to new sites of axo-dendritic contact. (a) Glutamatergic synapses between axon and dendrite shafts of hippocampal neurons can form in about an hour of the initial accumulation of presynaptic vesicles. Presynaptic proteins, including synaptic vesicle precursors (STVs) and piccolo-transport vesicles (PTVs), are mobile in axons before synapses are formed (upper axon/dendrite pair). These precursors are the first proteins recruited to nascent synapses (second axon/dendrite pair). After ~30 min, PSD-95 accumulates at these sites (third pair) followed by glutamate receptors (fourth pair, Friedman et al. 2000). (b) In young cortical neurons, glutamatergic synapses can form even faster, on a timescale of several minutes. In these cells, STVs and NMDARs are both found in transport packers that are highly mobile in the axons and dendrites, respectively, before synapse formation (upper axon/dendrite pair). Both STVs and NMDAR transport packets cycle with the membrane during their transport (Washbourne et al. 2004, Sabo et al. 2006). Contact between an axonal growth cone filopodium and a dendrite (right), or between axon and dendrite shafts (left), leads to the rapid and simultaneous recruitment of STVs and NMDARs at nascent synapses within ~7 min of contact (second pair). PSD-95 is recruited to these sites with a variable time course, and AMPARs are recruited an hour following initial recruitment of NMDARs (third pair; Washbourne et al. 2002). (c) Glutamatergic synapses can also form at prespecified sites along the dendritic shaft of hippocampal neurons, defined by stable preformed scaffold complexes associated with neuroligin. In this scenario, complexes of scaffolding proteins (including PSD-95, Shank, and GKAP) are mobile within dendrites before synapses are formed (upper axon/dendrite pair). When these complexes associate with neuroligin, they often become stabilized in the dendritic membrane (second pair). A significant proportion of these complexes then recruit STVs to form synapses within 2 h of their stabilizarion (third pair; Gerrow et al. 2006). (d) There are also predefined sites along the axon shaft of cortical neurons where en passant synapses selectively form. These predefined sites are stable locations along the axon where STVs cycle with the plasma membrane (first axon/dendrite pair) and presumably release diffusible molecules before synapses are formed (second pair). Filopodia from dendritic growth cones (right) and presumably also dendritic shafts (lift) are selectively attracted to, and stabilized, at these sites (third pair). Following stabilization of this contact at this predefined site, the presynaptic terminal is farmed and additional pre-and postsynaptic proteins are recruited to form a nascent synapse (fourth pair; Sabo et al. 2006).

Figure 3

Possible mechanisms for recruitment of proteins to nascent synapses. (a) Current models, for CNS synaptogenesis suggest that synaptogenesis is initiated by the binding of trans-synaptic adhesion molecules, such as neuroligin and β-neurexin, across the synaptic cleft. Binding of these molecules then leads to the activation of intracellular signal transduction cascades, which somehow alter the trafficking of STVs and NMDAR transport packets, causing them no rapidly accumulate at nascent contact sites. This model implies that synaptic proteins are actively recruited to nascent synapses through intracellular signaling, (b) It is also possible that activation of synaptogenic molecules locally alters the cytoskeleton, leading to the passive capture of mobile precursors that get stuck at these sires, (c) In addition to adhesion molecules, diffusible molecules may also play instructive roles in glutamatergic synaptogenesis. Diffusible molecules released from cycling STVs, such as glutamate, may selectively attract d dendritic filopodia to form synapses at sites of their release (Sabo et al. 2006). (d) More speculative is the idea that activation of synaptogenic molecules could alter STV or NMDAR transport packer accumulation at nascent synapses through direct interactions between adhesion molecules in the membrane and intracellular transport packers. For example, TrkB is trafficked with STVs (Gomes et al. 2006) and NCAM may move with NMDA receptors (Sytnyk et al 2006). Dendritic release of BDNF may alter and recruit STVs to its source through a direct interaction with TrkB. Similarly, homophilic NCAM interactions across the synaptic cleft could lead to the accumulation of NMDARs at those sites owing to a direct interaction with the NMDA receptor transport packers.