Several molecules have been identified as synapse-inducing proteins that mediate the recruitment of pre- and postsynaptic molecules important for the function of an active synapse. A number of CAM's have been shown to perform this function. Specifically, the neuroligins and SynCAM have been demonstrated to induce presynaptic morphology at contact points between axons and non-neuronal cells expressing either of these proteins. Both have been shown to stimulate the recruitment of synaptic proteins to the site of contact (Biederer et al., 2002, Dean et al., 2003, Scheiffele et al., 2000). This suggests that some cell adhesion molecules are, at least when over-expressed, sufficient to initiate synaptogenesis.
Neuroligins were first identified for their ability to bind, in a Ca++-dependent manner, all three isoforms of β-neurexin, a presynaptic transmembrane protein (discussed below) (Ichtchenko et al., 1995). The interaction between neuroligin and β-neurexin (Figure 1 and Figure 3) has been shown to increase the size and number of presynaptic terminals (Levinson et al., 2005, Prange et al., 2004), as well as potentiate the clustering of the postsynaptic protein PSD95 under some circumstances (Scheiffele et al., 2000). The converse has also been shown. β-neurexin presented on beads or expressed in heterologous cells, when contacting dendrites, can recruit postsynaptic proteins to the contact point (Graf et al., 2004, Nam and Chen, 2005).
SynCAM is a transmembrane member of the Ig superfamily that shows homophilic Ca++-independent binding. SynCAM contains 3 Ig-domains, and an intracellular C-terminal PDZ-binding motif that binds to the synaptic scaffolding proteins CASK and syntenin. SynCAM overexpression in cultured neurons promotes synapse formation, and overexpression in heterologous cells stimulates development of presynaptic specializations in contacting axons (Biederer et al., 2002).
Artificial presynaptic terminals induced by both SynCAM and neuroligins on heterologous cells have been shown to be identical in all parameters measured. When over-expressed in vivo, however, SynCAM and neuroligins do not have identical effects. Expression of SynCAM increased synaptic function as determined by electrophysiology, whereas overexpression of neuroligin increased synapse number as determined through morphological analyses (Sara et al., 2005). However, other studies have used electrophysiological methods to show that there is increased synaptic function in neuroligin over-expressing neurons (Levinson et al., 2005, Prange et al., 2004), and that knockdown of neuroligin diminishes synaptic function and activity (Chih et al., 2005). These differences are puzzling, but may be explained by developmental and cell-type differences of the neurons studied.
Many studies have investigated the sequential order for the recruitment of synaptic proteins, and asked whether they are recruited as preassembled complexes or as individual molecules (Zhen and Jin, 2004, Bury and Sabo, 2010). Recruitment of structural proteins probably contributes significantly to presynaptic formation. Studies from several laboratories have suggested, however, that delivery of transport packet vesicles may also play an important role. The fusion of dense core vesicles has been observed shortly after initial contact (Bresler et al., 2004). These vesicles may deliver structural elements of the presynaptic active zone, such as Piccolo and Bassoon. A different vesicle population also arrives at the presynaptic area shortly after initial contact (Ahmari et al., 2000). These transport packets have been shown to contain proteins which are important for neurotransmitter release. For instance, these cargos can include the v-SNARE VAMP and synaptophysin (Goldstein et al., 2008). The order of assembly of presynaptic precursor vesicles and how they are recruited is largely unknown. Cell adhesion complexes have often been implicated in their recruitment, but the sequence and mode of delivery for most adhesion molecules during synaptogenesis has also not been determined. For example, β-neurexin and SynCAM can induce, in vitro, active zone assembly. Their delivery may be an early event in vivo, resulting in the transformation of a nascent contact to a functional presynaptic zone. However, it is not yet clear whether these proteins are already diffuse in the plasma membrane and cluster through lateral movement, or whether they are delivered as preassembled clusters by the vesicular transport systems. Overexpression of PSD95 enhances recruitment of diffuse postsynaptic neuroligin-1 to nascent glutamatergic synapses. This overexpression of PSD95 caused an increase in the size of the presynaptic terminals (Prange et al., 2004). These data suggest that clustering of postsynaptic elements may enhance recruitment of the presynaptic release machinery. The formation of chemical synapses involves reciprocal induction and independent assembly of pre- and postsynaptic structures, however, and presynaptic assembly proceeds faster than postsynaptic assembly (Fox and Umemori, 2006). This process may be required for stabilization of newly formed contacts. Neuroligin-1 associates with PSD95, even in the absence of an active presynaptic terminal. Time-lapse images revealed the presence of preformed complexes containing scaffold proteins and neuroligin-1 before the recruitment of presynaptic proteins (Gerrow et al., 2006). Other studies, however, have demonstrated that the recruitment of PSD95 occurs after the establishment of an active presynaptic terminal (Ahmari et al., 2000, Friedman et al., 2000, Bresler et al., 2004). None of these studies looked at the role of neuroligin-1 in the recruitment of PSD95, but that has been suggested as one of neuroligin-1’s roles. Other studies examined new axonal contacts, and demonstrated the rapid recruitment of NMDA-type glutamate receptors without the involvement of PSD95 (Washbourne et al., 2002). This recruitment was also shown to take place before the establishment of an active presynaptic terminal.
Glycine and γ-aminobutyric acid (GABA) are the primary inhibitory neurotransmitters in the vertebrate CNS. The action of GABA is mediated by ionotropic (GABAA and GABAC) and metabotropic (GABAB) receptors. These receptors are ubiquitously expressed in the mammalian CNS (Barnard et al., 1998, Bowery et al., 2002). Modulation of GABAergic function is accomplished at multiple levels. This modulation includes transmitter synthesis by two isoforms of glutamic acid decarboxylase (GAD), vesicular storage, Ca++ dependent and independent release, neurotransmitter clearance by neurons and glia cells and the activation of receptors (Kneussel and Betz, 2000).
While hundreds of proteins have been identified at the PSDs of excitatory synapses (because these PSDs are relatively easy to isolate), (Peng et al., 2004, Walikonis et al., 2000) postsynaptic fractions of inhibitory synapses cannot be isolated selectively. Therefore, much less is known about the molecular constituents of inhibitory postsynapses. There is apparently an absence of proteins with PDZ domains mediating protein-protein interactions (Sheng and Sala, 2001). At GABAergic synapses, gephyrin is a postsynaptic scaffolding protein. Gephyrin is reported to be important for synaptic clustering of GABAA and glycine receptors at inhibitory synapses in sections of mammalian hippocampus (Kneussel et al., 1999b, Levi et al., 2004). However, gephyrin is apparently only partially required for receptor clustering in the retina and spinal cord (Kneussel et al., 2001, Fischer et al., 2000), and gephryn independent clustering of GABAA receptors has also been documented.
A number of CAMs have been shown to be involved in some manner in inhibitory synaptogenesis, particularly dystroglycan. Dystroglycan was the first CAM identified at mature GABAergic synapses. Dystroglycan was found to accumulate at nascent synapses, albeit after GAD, GABAA receptors, synaptic vesicles and gephyrin (Kneussel et al., 1999a, Levi et al., 2002). Dystroglycan binds several molecules involved in synaptogenesis, including agrin, laminin and β-neurexin (Sugita et al., 2001). Dystroglycan is not essential for synapse formation, since clustering of many proteins at inhibitory synapses is not affected in neurons obtained from mutant mice lacking dystroglycan (Levi et al., 2002) but these synapses did not contain detectable dystrophin, and thus is likely to function in modulating inhibitory synapses or conferring specialized properties on a subset of them. The timing of dystroglycan clustering suggests that it may play a more important role in synapse maturation and or maintenance.
The first direct indication of a CAM initiating inhibitory synapses formation was the observation of Neuroligin-1 induction of GABAergic presynaptic contacts when overexpressed in hippocampal neurons (Gilbert and Auld, 2005, Levinson and El-Husseini, 2005a, Prange et al., 2004). This was an unexpected outcome, since Neuroligin-1 is primarily concentrated at glutamatergic synapses, and has been shown to induce glutamatergic synapse formation. It has subsequently been shown that all members of the neuroligin family and their presynaptic binding partners the β-neurexins (Chih et al., 2005, Graf et al., 2004, Levinson et al., 2005) can induce all synapse types, and that their specificity may be simply a relative affinity. Endogenous Neuroligin-1 and -3, however, are enriched at excitatory synapses, while Neuroligin-2 is enriched at inhibitory synapses in vivo (Levinson et al., 2005, Song et al., 1999, Varoqueaux et al., 2004). The synaptogenic activity of neuroligins appears to require interaction with β-neurexin, since the development of both excitatory and inhibitory synapses in cultured neurons is inhibited by a soluble form of β-neurexin (Levinson et al., 2005, Scheiffele et al., 2000). Neuroligins can also interact with α-neurexin, a process that is regulated by alternative splicing of both proteins (Boucard et al., 2005, Missler et al., 2003). Specific combinations of neuroligins and neurexins, as well as other CAM interactions, therefore, may dictate the properties of neuronal contacts. This combinatorial diversity, as well as developmental programs, may contribute to the complexity of the mammalian CNS.
An example of the diversity of interactions by CAMs is illustrated by the cell adhesion molecule neurexin (Nrxn). Neurexins were first identified in rat brains in 1992 based on their affinity for α-latrotoxin, a component of black widow spider venom that triggers massive synaptic vesicle fusion and neurotransmitter release. During the following two decades, groundbreaking studies by Tom Südhof and colleagues described neurexins and a major class of their postsynaptic binding partners, the neuroligins (Ichtchenko et al., 1995, Ichtchenko et al., 1996, Boucard et al., 2005, Chih et al., 2006, Comoletti et al., 2006). The importance of the neuroligin-neurexin interaction was first demonstrated by Peter Scheiffele et al. (Scheiffele et al., 2000) when they co-cultured neurons with non-neuronal HEK-293 cells engineered to express either neuroligin-1 or neuroligin-2 (Scheiffele et al., 2000). The neuroligin on the surface of the HEK cells induced presynaptic differentiation in contacting axons. Electron microscopy and immunostaining revealed that these induced hemi-synapses were very similar to presynaptic structures seen in the CNS. Induction could be abrogated by adding a soluble neurexin derivative to the culture media. In other studies, the hemi-synapses were shown to be functional by adding either N-methyl-d-aspartate (NMDA) or α-amino-3-hydroxyl-5-methyl-4-isoxazole-propionate (AMPA) receptor subunits to the HEK-293 cells (Fu et al., 2003). This synapse-inducing activity requires only Ca++, and the extracellular portion of neuroligin-1 on a polystyrene bead was sufficient to induce presynaptic differentiation (Dean et al., 2003). However, such induction does require neuroligin oligomerization. The synapse induction activity of the neuroligin-neurexin interaction was extended to GABAergic synapses and to postsynaptic induction by neurexin (Graf et al., 2004). They also found that neuroligin-1 is primarily associated with induction of glutamatergic synapses, whereas neuroligin-2 is primarily associated with induction of GABAergic synapses. Neuroligin-1-induced postsynaptic hemi-synapses contained NMDA receptors, but not AMPA receptors. However, application of glutamate to the culture recruited AMPA receptors to the postsynaptic hemi-synapses. Time-lapse imaging showed that increased Nlgn1/Nrxn interaction transiently hyper-stabilized dendritic filopodia while knockdown of Nlgn1 blocked dendritic filopodia stabilization (Chen et al., 2010). This Nlgn1 over-expression-mediated filopodia stabilization requires NMDA receptor transmission.
As will be discussed below, neurexin has many possible isoforms. Neurexin are concentrated in the presynaptic terminal, but postsynaptic neurexin have also detected in vivo (Taniguchi et al., 2007, Scheiffele et al., 2000). Presynaptically, Nrxns bind through PDZ-domain interactions to the intracellular scaffolding proteins LIN10/Mint/X11alpha and LIN2/Cask, which in turn bind LIN7/Velis. Cask may exhibit a kinase activity that may phosphorylate neurexin. Through the Cask/Mint/Velis complex, neurexin may be coupled to presynaptic voltage-gated Ca++ channels (Cav2.1 and Cav2.2). Neurexin also engages in extracellular interactions with ligands other than the neuroligins: with neurexophilins, with leucine-rich repeat transmembrane proteins (LRRTM1 and LRRTM2) at glutamatergic synapses (Ko et al., 2009), and with α-dystroglycan at GABAergic synapses (Sugita et al., 2001). Neurexin has also been shown to interact directly with GABAA receptors and trans-synaptically with the glutamate receptor subunit GluRδ2 through cerebellin 1 (Uemura et al., 2010, Zhang et al., 2010). In vitro, specific neurexin splice variants exhibit differential ability to promote assembly of postsynaptic glutamatergic and GABAergic structures, but cell type-specific Neurexin isoform repertoires are currently unknown.
These early studies suggested that the role of neuroligin was to induce synaptogenesis. However, subsequent studies suggest that neuroligin functions chiefly in synapse maturation, rather than synapse formation per se. This view comes largely from the study by Varoqueaux and colleagues wherein they constructed a “true knockout” mouse lacking neuroligins-1, -2 and -3, and saw a reduction in presynaptic protein content, but not a change in synaptic number (Varoqueaux et al., 2006). These observations suggested that neuroligins are not responsible for synaptogenesis, but instead help facilitate synaptic maturation. Immature presynaptic boutons differ from mature boutons in several ways. For example, immature boutons of cultured hippocampal neurons require F-actin to prevent dispersal of their constituents (Zhang and Benson, 2001). Additionally, the number of synaptic vesicles that undergo exocytosis, the rate of exocytosis and the response to stimulation are all increased in mature synapses compared with immature synapses (Mozhayeva et al., 2002). It is largely unknown how these presynaptic maturation steps are controlled.
Knockout of specific neuroligins have different effects on excitatory and inhibitory synapses (Chubykin et al., 2007). An analysis of the effect of neuroligin-1 on the differences between immature and mature synapses by Wittenmayer (Wittenmayer et al., 2009) demonstrated that synapses in cultured hippocampal neurons derived from neuroligin-1 knockout mice remain structurally and functionally immature with respect to active zone stability and synaptic vesicle pool size. Blundell (Blundell et al., 2010) and Jedlicka (Blundell et al., 2009) showed that neuroligin-2 deficient mice exhibited increased dentate gyrus excitability, suggesting impaired inhibitory synapse transmission. Mutant mice lacking neuroligin-2 displayed decreased inhibitory synaptic currents. This phenotypic difference for neuroligin-1 and neuroligin-2 knockout mice is consistent with the observation that different neuroligins potentiate different types of synapses (Chih et al., 2005, Graf et al., 2004, Levinson et al., 2005, Prange et al., 2004). These results led to the model that neuroligins control the excitatory to inhibitory balance (Levinson and El-Husseini, 2005a, Levinson and El-Husseini, 2005b) through differential potentiation and control of synapse maturation.