The neural cell adhesion molecule (NCAM) is the prototypic member of the immunoglobulin family of cell adhesion molecules, whose trans-interactions initiate co-signaling through fibroblast growth factor receptors and the non-receptor tyrosine kinase fyn to stimulate neurite outgrowth (Maness and Schachner 2007) (Fig. 1a). However, NCAM is far from adhesive in many cases. In the embryonic brain, most if not all NCAM is covered by the glycan polysialic acid (polySia), which reduces cell surface contact and impedes binding of NCAM and other cell adhesion molecules, while creating a different molecular interface for other interactions such as with heparan sulfate proteoglycans (for a comprehensive recent review, see Hildebrandt and Dityatev 2013). Moreover, besides the well-known transmembrane and glycosylphosphatidylinositol-anchored isoforms of NCAM, there is a secreted isoform and NCAM, either polysialylated or not, can be cleaved off by metalloprotease activity leading to release of its extracellular domain (Diestel et al. 2005; Hübschmann et al. 2005; Hinkle et al. 2006; Kalus et al. 2006). As of today the physiological relevance of soluble NCAM and particularly the role and regulations of the enzymatic shedding mechanism is open. However, the importance of a tight control of NCAM interactions as well as the potential of soluble NCAM fragments to trigger or to interfere with essential developmental processes is evident from at least three different experimental approaches.
Already in 1996, it has been demonstrated that, in stark contrast to a complete genetic deletion, the replacement of NCAM by a secreted form of its extracellular domain results in dominant embryonic lethality (Rabinowitz et al. 1996). In 2005, studies on mice with a complete ablation of polySia synthesis corroborated the devastating impact of deregulated NCAM interactions. In these mice the absence of polySia leads to an excess of polySia-negative NCAM and this is the cause of severe defects in brain development (Weinhold et al. 2005). In the same year, the group of Patricia Maness presented mice over-expressing a soluble extracellular domain fragment of NCAM under the neuron-specific enolase promoter (Pillai-Nair et al. 2005). These mice display reduced inhibitory perisomatic innervation from basket cells and, as shown in subsequent studies, disrupted neurite branching of interneurons and impaired synaptic plasticity in the prefrontal cortex (Brennaman and Maness 2008; Brennaman et al. 2011).
Reciprocal to constrained formation of inhibitory synapses in mice with excessive soluble NCAM expression, NCAM null mice show signs of increased perisomatic inhibition (Brennaman et al. 2013b). Notably, the same phenotype was found in ephrinA or EphA3 deficient mice, and NCAM, mainly the transmembrane isoform NCAM-140, could be detected in EphA3 immunoprecipitates from forebrain lysates of wild-type mice (Brennaman et al. 2013b). These findings suggest that the role of NCAM in restricting inhibitory synapse formation or remodeling may be linked to the well-known function of ephrinA/EphA3 complexes in axon repulsion (Janes et al. 2012). Interestingly, incubation with ephrinA5 caused growth cone collapse in wild-type but not NCAM null mutant neurons and a slight decrease of perisomatic innervation in organotypic slice cultures of the prefrontal cortex obtained from wild-type but not NCAM null mice (Brennaman et al. 2013b). This decrease was abolished after enzymatic removal of polySia indicating that the polysialylated form of NCAM is responsible for the ephrinA5-induced restrictions.
In this issue of the Journal of Neurochemistry, Maness and colleagues now report on a possible mechanism of how EphA3 signaling in response to ephrinA5 may be linked to NCAM shedding (Brennaman et al. 2013a) (Fig. 1b). The study starts by describing ectodomain cleavage of NCAM, if transfected together with kinase-active but not kinase-inactive EphA3 into HEK cells. Shedding is prevented by co-transfection with a dominant-negative version of the metalloprotease ADAM10, which has been previously suggested as the major enzyme responsible for NCAM shedding (Hinkle et al. 2006; but see Kalus et al. 2006). In assays with primary cultured cortical neurons from wild-type mouse embryos, the authors were able to demonstrate an increase of a soluble NCAM fragment in response to stimulation with ephrinA5. As a logical consequence of the fact that all NCAM in these cells is modified with polySia, the soluble fragment could be detected in immunoblots with NCAM- and with polySia-specific antibodies and migrated as a high molecular weight smear typical for polysialylated NCAM. If the cultured neurons were treated with endosialidase prior to ephrinA5 stimulation, sharp bands of NCAM-140 and NCAM-180 isoforms were detected in the cell lysates, indicating successful removal of polySia from these two NCAM isoforms. Strikingly, in these cultures, the release of soluble NCAM was completely abolished. This seems contradictory to the shedding of polySia-negative NCAM obtained with transfected cells but raises the possibility that only polysialylated NCAM can be released in response to ephrinA5 under physiological conditions. This is remarkable, because another metalloproteinase, ADAM17, that has been previously implicated in NCAM shedding (Kalus et al. 2006), cleaves irrespective of the presence or absence of polySia, further pointing toward a regulation of NCAM versus polySia-NCAM shedding specificity. Further studies are necessary to dissect the determinants of this specificity.
The study by Brennaman et al. (2013a) proceeds by identifying an ADAM10 cleavage site within the second fibronectin-type III domain of NCAM. Experiments in cultured neurons from NCAM knockout mice revealed that ephrinA5-induced growth cone collapse can be restored by transfection with NCAM-140 but not with NCAM mutated at the ADAM10 cleavage site. This effect was particularly prominent in the GABA-positive fraction of neurons. As indicated by a previous study, the ADAM10-dependent shedding mechanism may limit the extent of interneuron axon arborization and restrict the number of inhibitory synapses (Brennaman et al. 2013b). Thus, in addition to elucidating the regulation of ephrinA5-induced growth cone collapse by ADAM10 and polySia-NCAM, the current study by Brennaman et al. (2013a) also suggests a role of ADAM10 in regulating perisomatic inhibition via soluble polySia-NCAM-dependent mechanisms.
ADAM10-dependent shedding of polySia-NCAM could also play a role in synaptic plasticity. The injection of recombinant polySia-NCAM ectodomain into hippocampal slices from wild-type mice was found to impair long term potentiation (LTP) in the CA1 region, whereas the same manipulation led to a rescue of CA1 LTP in slices from NCAM deficient mice (Senkov et al. 2006). Similarly, application of polySia-NCAM ectodomain to the dorsal hippocampus impaired formation of hippocampus-dependent contextual fear memory in wild-type mice and partially rescued contexual memory in NCAM deficient mice (Senkov et al. 2006). Thus, there is an optimal range of polySia-NCAM ectodomain expression that is necessary and sufficient for synaptic plasticity in the CA1 region, but elevated or too low levels of soluble polySia-NCAM are detrimental. The positive effects of soluble polySia-NCAM on LTP and contextual memory are mediated by inhibition of extrasynaptic NMDA receptors containing the glutamate receptor NMDA subunit GluN2B (reviewed by Hildebrandt and Dityatev 2013) (Fig. 1c), whereas the mechanism underlying the negative effects of excessive levels of soluble polySia-NCAM remains unknown. The present study of Brennaman et al. (2013a) suggests that conditions leading to either reduced or elevated activity of ADAM10/EphA3 may cause impaired synaptic plasticity via deregulating soluble polySia-NCAM levels.
As increased levels of soluble NCAM were found in Alzheimer's disease and schizophrenia patients and because Brennaman et al. (2013a) identified a putative ADAM10 cleavage site, their study stimulates further interest to measure the levels of NCAM cleaved by ADAM10 in these patients. In Alzheimer's disease, increased activity of ADAM10 toward amyloid precursor protein might be beneficial to generate soluble amyloid precursor protein and decrease the levels of beta-amyloid generation, but, as shown now, it also may cause excessive shedding of polySia-NCAM with a negative impact on synaptic plasticity. This study also stimulates an interest to generate NCAM knockin mice with the mutated ADAM10 cleavage site, which would provide an excellent option to understand the role of NCAM shedding in terms of brain development and synaptic plasticity, also in the context of schizophrenia and Alzheimer's disease. In summary, although the physiological significance of the new mechanism of polySia-NCAM- and ADAM10-dependent Eph3A signaling and growth cone collapse in vivo remains elusive, it clearly opens up new vistas on possible roles of NCAM shedding in health and disease.