Presently, a rather large number of ADAM10 substrates have been identified in different experimental settings (e.g., reviewed for proteomic approaches in Müller et al., 2016). Of notice, ADAM10 substrates belong to type I as well as type II transmembrane but also Glycosylphosphatidylinisotol (GPI)-anchored proteins, indicating a considerable flexibility of the protease with regard to substrate recognition. Consensus cleavage motifs for proteases are commonly deduced from the amino acids surrounding the naturally occurring cleavage sites within protein substrates. This approach failed in the case of ADAM10 because it lacks a well-defined consensus sequence: for ADAM10 leucine was found to be preferred (and tyrosine accepted) in the P1' position (immediately downstream of the cleavage site) in an investigation using oriented peptide mixture libraries which gives evidence of a shallow or deep S1' site (John et al., 2004). ADAM10's preference for larger residues at P1' has been confirmed but acceptance of aromatic amino acids and even glutamine were also reported (Caescu et al., 2009). This tolerance for aromatic residues in P1' may be the most relevant difference in cleavage site specificities between ADAM10 and its close relative ADAM17 (Tucher et al., 2014). In the early investigation tyrosine was found to be favored at P1 (immediately upstream of the cleavage site; John et al., 2004). In contrast to this, selectivity for small residues such as alanine at the P1 was described by Caescu et al. (2009) and specificities for proline and basic residues were recently reported (Tucher et al., 2014). In sum, these reports show the methodological limitations and uncertainties involved in pinpointing cleavage site specificities from linear unmodified peptide libraries. In addition, the activity state of the cell may also influence shedding capacity, as is the case for NG2 (Sakry et al., 2014) as well as the synaptic marker neuroligin 1 (Suzuki et al., 2012), further complicating work in this direction.
The earliest study on the distribution of ADAM10 at synapses was based on immunocytochemistry and suggested that ADAM10 co-localizes with the postsynaptic scaffold protein Synapse-associated protein 97 (SAP-97) but not with the presynaptic vesicle protein synaptophysin (Marcello et al., 2007). However, a more recent study using the sensitive proximity ligation assay reported proximity of the enzyme with synaptophysin in mouse primary hippocampal neurons (Lundgren et al., 2015). This suggests that ADAM10 can be present in both parts of a synapse. One example where this could be functionally relevant is the neurexin-neuroligin-interaction: neurexins and neuroligins are cell-adhesion molecules which form transsynaptic complexes (e.g., Tsetsenis et al., 2014). They appear to be important for normal synapse specification and function (Jedlicka et al., 2011, 2015). For the postsynaptic protein Neuroligin 1, ADAM10 has been found to act as the major sheddase, as could be shown by pharmacological and genetic means in primary rat cortical neurons (Suzuki et al., 2012). NMDA receptor activation as well as prolonged epileptic seizure condition increased shedding, suggesting a role for neuronal activity in this context. Interestingly, shedding of Neuroligin 1 could be induced by soluble neurexin 1α or β derived from overexpressing HEK293 cells (Suzuki et al., 2012), indicating that ligand binding at the cell surface also regulates Neuroligin 1 shedding. Similar observations have been made for the Notch-Delta complex where Notch1 cleavage by ADAM10 is induced by Delta binding (e.g., reviewed in Van Tetering and Vooijs, 2011). Intriguingly, Notch 1 as well as its ligands - Delta or Jagged - have been found to be cleaved by ADAM10 (for example: Pan and Rubin, 1997; Lavoie and Selkoe, 2003). A recent publication regarding systemic characterization of ADAM10 substrates from neurons highlighted that ADAM10 is also in principle capable of shedding the Neuroligin ligands Neurexins 2 and 3, although deletion of the proteinase resulted only in a comparably mild reduction of the shedding (Kuhn et al., 2016). If this role for ADAM10 in the cleavage of major anchoring proteins can be verified in vivo and in human brain, interfering with ADAM10 activity may indeed be a powerful tool to influence synaptic structure and function.
cleavage cg rar
The retinoic acid receptor (RAR) family is particularly interesting with regard to ADAM10 regulation because of its therapeutic potential. Both, RAR alpha and beta are capable of inducing human ADAM10 promoter activity (Tippmann et al., 2009). Moreover, the commercially available drug acitretin which intracellularly liberates retinoic acid (Ortiz et al., 2013), shifts APP processing in AD model mice toward the alpha-secretase cleavage pathway (Tippmann et al., 2009). The neuroprotective property of RARalpha agonists has been shown in cortical cultures, an AD mouse model (Tg2576 mice) (Jarvis et al., 2010), as well as in hippocampal tissue of aged SAMP8 mice (Kitaoka et al., 2013). Cilostazol-stimulated N2A cells with overexpression of human mutated APP also displayed ADAM10 elevation which was significantly attenuated by a RARbeta inhibitor and RARbeta-gene silencing (Lee et al., 2014). The effect of cilostazol on ADAM10 expression could be antagonized by sirtinol and by Sirtuin 1 (SIRT1)-gene silencing, suggesting that RARbeta and this class of deacetlyase together act on the ADAM10 promoter.
The ADAM10 zymogen is cleaved by proprotein convertases within the secretory pathway to yield the active enzyme (see paragraph 1). Removal of the prodomain of ADAMs likely involves a canonical consensus site for the proprotein convertase Furin (Roebroek et al., 1994), which is located between the pro- and the catalytic domain of ADAM10 (Anders et al., 2001). More recently, a novel cleavage site upstream of the prodomain has been identified (Wong et al., 2015). ADAM10 has four potential N-glycosylation sites of which three are located in the metalloprotease domain (N267, N278, and N439) and one in the disintegrin domain (N551). In bovine ADAM10 all four have been found glycosylated and required for full in vivo activity (Escrevente et al., 2008).
Binding of ADAM10 to synapse associated protein 97 (SAP97) is required for inserting ADAM10 into the synaptic membrane (Marcello et al., 2013). Interaction of SAP97 with ADAM10 is mediated via a protein kinase C (PK C) phosphorylation site within the SAP97 SRC homology domain (Saraceno et al., 2014). Removal of ADAM10 from excitatory synapses occurs by clathrin-mediated endocytosis in human hippocampal tissue (Marcello et al., 2013). This is mediated by the clathrin adaptor protein AP2 which interacts with the ADAM10 C-terminal domain. In addition to control of surface concentrations of ADAM10 by transport mechanisms, further cleavage events may occur: the ectodomain of ADAM10 can be processed by ADAM9/15 or gamma-secretase (Cissé et al., 2005; Parkin and Harris, 2009; Tousseyn et al., 2009). Using recombinant mouse ADAM9 prodomain as a competitive inhibitor of ADAM9, Moss et al. demonstrated an increase of ADAM10-dependent APP processing in human neuronal SH-SY5Y cells (Moss et al., 2011). However, a truncated soluble ADAM10 construct was incapable of shedding cell-associated amyloid precursor protein while earlier reports described that shedded ADAM10 had the ability to cleave endogenous Prion protein in fibroblasts (Cissé et al., 2005).
The intensity of ADAM10 cleavage may further depend on the cytoskeleton: a dominant negative dynamin I mutant not only increased surface expression of both, immature, and mature ADAM10 but also strongly increased the amount of the C-terminal cleavage product of ADAM10 (Carey et al., 2011). In addition to its role as a protease acting at the cell surface it has been speculated that the soluble ADAM10 C-terminus could act as a signaling molecule, facilitating nuclear entry of other proteins (Endsley et al., 2014). A protein class which is deeply involved in for example cytoskeletal anchoring and protein-trafficking is the tetraspanin family (Charrin et al., 2014). Several tetraspanins have been identified to interact with ADAM10: tetraspanin 12 (Tspan 12) binds to ADAM10 in a palmitylation-dependent mechanism and increases non-amyloidogenic shedding of APP by increased enzymatic maturation of the protease (Xu et al., 2009). Co-immunoprecipitation experiments also identified specific ADAM10 interactions with Tspan5, Tspan10, Tspan14, Tspan15, Tspan17, and Tspan33/Penumbra (Haining et al., 2012), which all led to enhanced enzyme maturation. Interestingly, only overexpression of Tspan15 resulted in a reduction of ligand-induced Notch-1 processing by ADAM10 (Jouannet et al., 2016). This led to the assumption that the tetraspanins might differentially influence compartimentalization of ADAM10. Indeed, the apparent diffusion coefficient of ADAM10 was higher in cells overexpressing Tspan15 as compared to control cells or Tspan5 overexpressing cells and also decreased the co-immunoprecipitation of proteins of the tetraspanin web with ADAM10 (Jouannet et al., 2016). Tspan12 and 17 also seem to stabilize a high molecular weight protein complex that tethers ADAM10 to the gamma-secretase allowing rapid sequential processing of substrates (Chen et al., 2015).
Of particular interest in this context is again the link between ADAM10 and APP. APP and in particular its cleavage product APPs-alpha have been shown to regulate dendritic complexity as well as spine numbers of hippocampal neurons (Lee et al., 2010; Tyan et al., 2012; Weyer et al., 2014). This effect appears to be age-dependent: whereas young APP-deficient mice had normal spine numbers, older APP-deficient mice showed a decrease in their spine density (Tyan et al., 2012). It may also depend on the brain region, since APP levels may show regional variations (Del Turco et al., 2016). Since APPs-alpha is generated by ADAM10 cleavage of APP, it is likely that some of the structural effects on spines seen in conditional ADAM10 knock-out mice (Prox et al., 2013) are the result of reduced APPs-alpha levels. Indeed, Prox et al. (2013) reported a reduction of APPs-alpha in brain of conditional ADAM10 knock-out mice to 5% of control levels. Since aging is also associated with reduced dendritic complexity and spine densities (Dickstein et al., 2007), it is attractive to speculate that reduced ADAM10 levels/activity and reduced APPs-alpha levels could play a role in this context (Lannfelt et al., 1995; Sennvik et al., 2000). 2ff7e9595c
Comments