Abstract
As a member of the A Disintegrin And Metalloproteinase (ADAM) family, ADAM10 has been identified as the constitutive α-secretase in the process of amyloid-β protein precursor (AβPP) cleavage and plays a critical role in reducing the generation of the amyloid-β (Aβ) peptides. Recent studies have demonstrated its beneficial role in alleviating the pathologic impairment in Alzheimer’s disease (AD) both in vitro and in vivo. However, the role of ADAM10 in AD and the underlying molecular mechanisms are still not well established. Increasing evidence indicates that ADAM10 not only reduces the generation of Aβ but may also affect the pathology of AD through potential mechanisms including reducing tau pathology, maintaining normal synaptic functions, and promoting hippocampal neurogenesis and the homeostasis of neuronal networks. Mechanistically, ADAM10 regulates these functions by interacting with postsynaptic substrates in brain, especially synaptic cell receptors and adhesion molecules. Furthermore, ADAM10 protein in platelets seems to be a promising biomarker for AD diagnosis. This review will summarize the role of ADAM10 in AD and highlight its functions besides its role as the α-secretase in AβPP cleavage. Meanwhile, we will discuss the therapeutic potential of ADAM10 in treating AD.
Keywords
INTRODUCTION
Alzheimer’s disease (AD) is the most common neurodegenerative disease, which is characterized by progressive impairment of memory and cognitive function [1]. The fundamental neuropathological changes of AD contain extracellular accumulation of amyloid-β (Aβ), intracellular deposition of neurofibrillary tangles (NFT), neuronal loss, and synaptic dysfunction [2]. Despite the tremendous progress that has been achieved over the past twenty years, the pathogenesis of AD is still not fully clear. Among all the hypotheses of AD, the amyloid cascade hypothesis established by Hardy in 1992 is the most widely accepted one, which highlights the central role of Aβ in the pathogenesis of AD. Aβ is a 38–43 amino acid hydrophobic peptide that tends to self-aggregate as oligomers, comprising hallmark of AD lesions in amyloid plaques. Furthermore, Aβ oligomers can trigger a series of pathological changes associated with AD, including the formation of NFT, synaptic dysfunction, and neuronal loss [3–5]. Abnormal processing of the amyloid-β protein precursor (AβPP) is responsible for excess deposition of Aβ. ADAM10, a member of the A Disintegrin And Metalloproteinase (ADAM) family, has been identified as the primary α-secretase of AβPP, and its beneficial role in alleviating the Aβ burden in AD has been revealed both in vivo and in vitro [6, 7]. As a transmembrane protease, ADAM10 cleaves ectodomain part of membrane-bound proteins including cell adhesion molecules and cell surface reporters. This process is also known as “ectodomain shedding” [8, 9]. Membrane-bound proteins are essential for intracellular signaling events, intercellular communication, cell differentiation, and apoptosis [10, 11]. Thus, it is important to regulate the structure and the levels of membrane-bound proteins, which consequently regulate their physiological functions. As important protease for ectodomain shedding of membrane-bound proteins in the brain, ADAM10 sheds more than 90 neuronal substrates and many of them are essential for brain development and normal physiological functions [12]. All these facts indicate a fundamental role of ADAM10 in central nervous system (CNS). Furthermore, recent studies indicate that ADAM10 not only act as a major α-secretase for AβPP, but may also affect tau pathology, synaptic functions, hippocampal neurogenesis, and gliogenesis by interacting with its substrates in the brain. Beyond the four established AD genes (APP, Presenilin-1, Presenilin-2, and Apolipoprotein E) [13], increasing evidence indicates that ADAM10 is a promising AD susceptibility gene [14, 15]. This review will summarize the role of ADAM10 in AD and highlight its functions besides as a major α-secretase for AβPP in the brain. In addition, we will discuss therapeutic potential of ADAM10 in AD therapy.
STRUCTURE AND NEUROBIOLOGY OF ADAM10
ADAM10 protein was initially purified from bovine brain for its ability to cleave the myelin basic protein (MBP) from myelin membranes and was cloned in 2006 [16, 17]. Human ADAM10 is composed of 748 amino acids (729 amino acids without signal peptide) with a long glycosylated ectodomain, a transmembrane domain, and an intracellular domain [18]. The extracellular part of nascent ADAM10 contains a prodomain, a catalytic site in metalloprotease domain, and a disintegrin domain and cysteine-rich domain [8]. ADAM10 is produced as a zymogen in endoplasmic reticulum (ER) and the prodomain keeps ADAM10 inactive by interacting directly with its catalytic site [19, 20]. ADAM10 matures into an active protease after the removal of the prodomain by furin and/or prohormone convertase (PC) 7 during pass through the Golgi compartment [9] (Fig. 1). Nascent ADAM10 in ER contains a PC7 and furin recognition sequence between the metalloprotease domain and the prodomain, and the proteolytic processing of this site transforms ADAM10 zymogen into an active form. A novel PC processing site within prodomain of ADAM10 has been identifed that is equally important for the activation of ADAM10. Mutations in this upstream PC recognition sequence led to significantly reduced catalytic activity of ADAM10, but did not affect proteolytic processing of the previously identified furin recognition sequence [21]. The prodomain also acts as an intramolecular chaperone, facilitating the proper folding of enzyme’s various domains [14]. The metalloprotease domain of ADAM10 contains a zinc-binding amino acid motif HEXXHXXGXXH and a methionine turn in the active site helix, which is homology to other ADAM family members [22]. The disintegrin and cysteine-rich domain of ADAM10 is composed of several stabilizing disulfide bonds and hydrophobic parts, forming a curved and elongated structure [22]. The transmembrane domain of ADAM10 is essential for the stimulation of ADAM10-mediated cleavage. In contrast, the intracellular domain of ADAM10 inhibits ADAM10-mediated cleavage through an arginine-rich ER retention motif. Mutations within ER retention motif increased the expression of ADAM10 at cell membrane [23]. In addition, the Src homology 3-binding (SH3) domain within intracellular domain is critically required for intracellular trafficking of ADAM10. Mutagenesis of the SH3 domain led to disordered distribution of ADAM10 at polarized cell surface [24]. Several ADAM10-binding SH3 domain proteins, including adaptor proteins and non-receptor tyrosine kinases, were identified in recent studies. These proteins play important roles in the regulation of the levels and activity of ADAM10 [25]. A more detailed description of the structure of ADAM10 can be found in recent reviews [9, 18].
The structure of ADAM10 and its role in AβPP processing. ADAM10 consists of an ectodomain, a transmembrane domain and an intracellular domain. The ectodomain of nascent ADAM10 contains a prodomain, a catalytic site in metalloprotease domain as well as a disintegrin domain and cysteine-rich domain. ADAM10 is produced as a zymogen in the endoplasmic reticulum (ER) and matures into an active protease after the removal of its prodomain by furin or prohormone convertase (PC) 7 during pass through the Golgi compartment. There are two pathways for AβPP cleavage, known as amyloidogenic pathway and non-amyloidogenic pathway. The amyloidogenic pathway is mediated by BACE1 and γ-secretase, leading to amyloid-β (Aβ) secretion. The non-amyloidogenic pathway is mediated by ADAM10 and γ-secretase, generating secreted AβPP ectodomain (sAβPPα) and p3 peptide.
In addition to being a major α-secretase for AβPP, ADAM10 also acts as an essential surface protease of other transmembrane proteins. Ectodomain shedding of transmembrane proteins is an important way to regulate their biological activities and functions. A classic example of ADAM10-mediated shedding is the Notch-1 signaling pathway. ADAM10 sheds both Notch-1 and its ligand near the cell membrane, followed by γ-secretase-mediated cleavage at intramembrane domain, releasing the intracellular domain of Notch-1 into the nucleus as a transcriptional activator for Notch-1-targeted genes [26, 27]. Besides AβPP and Notch-1 mentioned above, ADAM10 sheds more than 90 membrane proteins in CNS, and many of them are critical to brain development and normal physiological functions [12]. Some of these substrates, such as Neuroligin-1, N-Cadherin, and neural cell adhesion molecule (NCAM), have neuronal and synaptic functions [28–30]. ADAM10 conditional knock-out (cKO) mice showed an altered morphology of postsynaptic structures in the brain accompanied with epileptic seizures and learning disabilities, indicating that ADAM10 may play an important role in synaptic and neuronal network functions [31]. ADAM10 itself undergoes shedding by ADAM9 and ADAM15 at ectodomain and by γ-secretase at the transmembrane domain of ADAM10. The intracellular domain of ADAM10 may function as a signaling protein and regulate gene expression [32, 33].
It should be noted that the ADAM10-mediated cleavage not only happens on the same cell membrane but may also happen between neighboring cell surface [22]. However, whether the cleavage in trans is really happening has been debated and the theory has not been corroborated by other researchers. Furthermore, the activity of ADAM10 is not limited to the plasma membrane. After the removal of the prodomain in Golgi, nascent ADAM10 has transformed into an active protease that may exert proteolytic activity in its trafficking route to the plasma membrane. The same holds true that ADAM10 is still functioning at the early stage of its endocytosis when ADAM10 is not functionally degraded [18]. New functions of ADAM10 have been revealed in recent studies, indicating its potential role in alleviating AD pathology through regulating tau pathology, synaptic functions, hippocampal neurogenesis, and gliogenesis, which will be described in detail below.
ADAM10 IN AD
Genetics of ADAM10 in AD
ADAM10 has been identified as the major α-secretase in the process of AβPP cleavage in recent studies [6, 34], and its involvement in AβPP processing provides insights into its gene mutations associated with AD. Kim et al. tested nine single nucleotide polymorphisms (SNPs) in ADAM10 gene in over 400 AD families and identified the best association SNPs rs2305421 (p = 0.003). The result was more pronounced when stratified by the Apolipoprotein E (ApoE) ɛ4 status (p = 0.0005). Then they identified two rare ADAM10 prodomian mutations (Q170H, R181G) from seven late-onset AD (LOAD) families in their followed studies in 2009 [35]. The two ADAM10 prodomain mutations attenuated α-secretase activity of ADAM10 significantly and generated more Aβ peptides in cell-based studies, leading to AD-related pathology [14, 35]. The result was further validated both in cell-culture experiments and in transgenic mice [15, 35]. It is the first time that provides evidence to identify ADAM10 as a new candidate AD susceptibility gene, challenging the theory that only common variants such as APOE ɛ4 is associated with LOAD [36]. However, Cai et al. did not reproduce the genetic finding in a large-scale resequencing study in LOAD and they did not observe the Q170H and R181G variations in the prodomain of ADAM10. Different ancestry information and familial cases in Kim’s study may affect the frequencies of these variations [37]. In the following genome-wide association study (GWAS), the association of SNP rs2305421 with AD was further confirmed [38]. Studies performed in German and China drew a different conclusion that there was no association between rs2305421 and AD [39, 40]. Nevertheless, the allele and genotype of rs2305421 showed an association with AD (p = 0.037, p = 0.035, respectively) if stratified by the APOE ɛ4 levels in Northern Han Chinese population [40]. Furthermore, it was reported that genetic variations rs514049 and rs653765 within ADAM10 promoter were associated with the expression of ADAM10 and cerebrospinal fluid (CSF) secreted AβPP ectodomain (sAβPPα) both in normal controls and AD patients, and the levels of CSF sAβPPα in cognitively normal controls were higher than that of AD patients. However, there was no evidence that the two variations were associated with AD in the Caucasian population [41, 42], and the same result was confirmed in a recent research performed in Chinese population [43].
Because of the “dark matter” of GWAS, the genetic background of AD remains unclear. The currently known susceptibility gene derived from GWAS, including APP, Presenilin-1 (PSEN1), and Presenilin-2 (PSEN2), only confer less than 0.15-fold increase in AD risk [36]. Even the most well established LOAD associated gene, APOE ɛ4, confers less than 50% susceptibility of common LOAD [13]. This may be explained by copy number variants, rare sequence variants and mutations in introns. As discussed above, whether ADAM10 gene is a LOAD susceptibility gene is still debated. With the progress of next-generation sequencing and the establishment of large scale clinical database, more rare variants associated with AD, such as Q170H, R181G in ADAM10 gene, will be identified [44].
Amyloid-β-dependent roles for ADAM10 in AD
Amyloid cascade hypothesis, first established by Hardy in 1992, is the most widely accepted one, which highlights the central role of Aβ peptides in the pathogenesis of AD [45]. Aβ peptides are hydrophobic and tend to aggregate as oligomers. Aβ oligomers are the primary components of the amyloid plaques and trigger a series of pathological events, contributing to the lesions observed in the brains of AD patients [46, 47]. Aβ oligomers also aggravate the pathological impairment in AD via inducing the hyperphosphorylation of tau and synaptic dysfunction [48, 49]. Aβ oligomers may enhance BACE1 levels through altering its subcellular distribution at post-translational level, leading to a vicious circle of Aβ generation and aggravating the pathological change in AD [50]. Given the pivotal role of Aβ in AD, the mechanism of Aβ generation and its metabolism are hotspots and attract a lot of attention in the understanding of the pathogenesis of AD as well as providing therapeutic strategies [5].
Increasing evidence indicates that abnormal processing of AβPP is responsible for excess deposition of Aβ. There are two classic pathways in AβPP processing and three proteases, namely α-, β-, and γ-secretase, are involved. β-secretase, also referred to as BACE1, is a membrane-bound aspartyl protease. γ-secretase is a hetero-tetrameric protease complex composed of Aph1, PEN2, nicastrin, and presenilin [51]. BACE1-meidated AβPP cleavage occurs at the N-terminus of Aβ domain, leading to the release of the soluble AβPP ectodomain (sAβPPβ) and C-terminal AβPP fragment of 99 amino acids (C99). C99 is further cleaved by γ-secretase, releasing the AβPP intracellular domain (AICD) and Aβ peptide. Alternatively, the majority of AβPP in cell lines are processed by α-secretase within the Aβ domain at ectodomain, releasing a soluble fragment sAβPPα and C-terminal fragment of 83 amino acids (C83). C83 is further cleaved by γ-secretase, releasing AICD and the secreted p3 peptide [9] (Fig. 1). The role of sAβPPα is still not completely established, but it has been reported that sAβPPα can improve the phenotype of AβPP deficiency mice and has neurotrophic and neuroprotective properties [52, 53]. Importantly, recent studies reported a novel η-secretase-mediated AβPP cleavage pathway in vivo. η-secretase-mediated cleavage occurs at ectodomain of AβPP, releasing a short ectodomain and a high molecular mass C-terminal fragment (CTF-η) of AβPP. The levels of CTF-η were increased in dystrophic neurites both in the brains of AD patients and AD mice models. CTF-η is further cleaved by ADAM10 and BACE1 to generate Aη-α and Aη-β peptide, respectively. Furthermore, the fragment Aη-α is assumed to inhibit neuronal activity in the hippocampus [54]. Thus, the role of ADAM10 in AβPP processing is complicated, and more relevant researches are needed to clarify itsfunctions.
ADAM10 is identified as the α-secretase
The identity of α-secretase in AβPP possessing has been controversial for years. It is partly due to the α-secretase-mediated cleavage of AβPP occurring at constitutive or regulated conditions and different proteases may be involved, respectively. The constitutive α-secretase-mediated shedding refers to the shedding of AβPP without stimulator. In contrast, the α-secretase-mediated cleavage can be stimulated by stimulator, and the process is known as regulated α-secretase-mediated cleavage. The α-secretase in AβPP shedding was reported to be a member of metalloproteases and several metalloproteases candidates including ADAM9, ADAM10, and ADAM17 were suggested as major constitutive α-secretases [55]. Until recently, several studies identified ADAM10 as the major component of constitutive α-secretase in several cell lines [6, 34]. This is corresponding with the previous studies that ADAM10 overexpression enhanced α-cleavage of AβPP and reduced the secretion of Aβ, underlying the potential of stimulating ADAM10 as a therapeutic target [7]. Conversely, decreased activity of ADAM10 was responsible for excess Aβ production, which was observed in ADAM10 knockdown primary neurons [6]. A catalytically-inactive ADAM10 mutant (E384A) and two rare ADAM10 prodomain mutations (Q170H and R181G) attenuated the catalytic activity of ADAM10, leading to increased Aβ levels and less sAβPPα secretion in transgenic mice [14, 56]. Compared with the constitutive α-secretase, the main component of the regulated α-secretase is still unclear. The main component of regulated α-secretase may change with different stimulators. ADAM10 can act as regulated α-secretase after specific stimulator like pituitary adenylate cyclase-activating polypeptide (PACAP) [9].
As the constitutive α-secretase, ADAM10 may compete with BACE1 in the process of initial AβPP cleavage. This assumption can explain ADAM10 overexpression promotes α-cleavage of AβPP and the reverse situation in ADAM10 mutant transgenic mice as discussed above. The competition is the basis for the potential of stimulating ADAM10 as a therapeutic target for AD [51]. It should be noted that this competition is not observed under all kinds of cell lines. The stimulation or overexpression of ADAM10 led to increasing sAβPPα and decreased Aβ levels in most cell lines, but the inhibition of ADAM10 activity was not positively correlated with Aβ levels in several cell lines [18]. This is partly due to the different components of AβPP secretase in different cell lines. For example, human hippocampal neurons mainly express BACE1 and γ-secretase instead of ADAM10, the inverse situation has been observed in human non-neuronal cells. It also should be noted that ADAM10 sheds more than 90 neuronal substrates in brain, and the side effects of mild ADAM10 activation should be carefully evaluated [12].
Regulation of ADAM10-mediated AβPP cleavage
ADAM10-mediated cleavage is tightly regulated at the transcriptional, translational, and posttranslational level. To date, cellular control mechanisms of ADAM10-mediated cleavage are still not completely clear, but recent studies have revealed several pathways involved in its regulation [57]. In addition, increasing evidence indicates that intracellular trafficking is one of the most important pathway for the regulation of ADAM10-medaited proteolysis [58]. A more detail description about the regulation of ADAM10 can be found in a recent review [57], and we will mainly talk about part of the regulation of ADAM10 that has therapeutic potential in ADtherapy.
The expression of ADAM10 is tightly controlled at the transcription level by several transcription factors. Retinoic acid (RA) is the best learned transcription activator in ADAM10 upregulation. Both all-trans-RA (atRA) and cis-RA most likely bind to their respective cognate receptors RAR and RXR, then interact with the retinoic acid responsive elements (RARE) within ADAM10 promoter, promoting ADAM10 transcription and leading to increased sAβPPα levels and less Aβ generation [59, 60]. Donmez et al. have reported that SIRT1,a NAD-dependent deacetylase that has antiaging and stress protective properties in vivo, reduced the production of Aβ and promoted the ADAM10 gene expression through coactivation of RAR. Because aerobic glycolysis depletes NAD in cells and the distributions of Aβ deposition correlate with increased aerobic glycolysis, it is speculated that downregulation of the NAD-dependent SIRT1 pathway promotes the generation of Aβ [61, 62]. Additionally, SIRT1 may promote the Notch signaling pathway through ADAM10 activation [62]. Although this study has been retracted for some reason [63], the ADAM10 activation property of SIRT1 has been confirmed in other studies [64–66]. ADAM10 promoter also harbors peroxisome proliferator-activated receptor α (PPARα) response elements. PPARα promotes ADAM10 expression through the RA pathway since it interacts with RXR to form a heterodimeric structure that binds to RARE. The expression of ADAM10 was decreased in PPARα knockdown neurons. In contrast, lentiviral overexpression of PPARα improved ADAM10 expression levels. All these facts indicate that PPARα is an important transcription factor regulating neuronal ADAM10 expression [67]. Besides the retinoic RA signaling pathway, X-box binding protein-1 (XBP-1), a transcription factor, was reported to promote ADAM10 expression through regulating the unfolded protein-response pathway. The activity or presence of XBP-1 was reduced in AD patients, so was the expression of ADAM10. ADAM10 expression was dependently regulated by the spliced XBP-1 variant dose, which could be synergistically enhanced by insulin [68]. Moreover, N-methyl-D-aspartate receptors (NMDAR) activation upregulated the expression of ADAM10 in mouse primary cortical neurons, and the Wnt/MAPK signaling pathway might be involved in the upregulation. Beside the transcription activators mentioned above, pro-inflammatory cytokine IL1, activated protein C (APC), Sex-determining region Y related high mobility group Box 2 (Sox2), and melatonin have been reported as transcription activators in recent studies [69–72] (Fig. 2A).
At the translational level, the expression of ADAM10 is inhibited through the 5’-untranslated region (5’-UTR) of ADAM10 mRNA. ADAM10 translation rate is repressed by the GC-rich 5’-UTR region, and knock down of the 5’-UTR results in an apparent increase of ADAM10 levels [73]. The 5’-UTR region contains a G-quadruplex motif, which is responsible for the repression of ADAM10 translation. Therefore, drugs that mitigate the suppressive effect of the RNA G-quadruplex motif by interacting with this region may be a new approach for ADAM10 upregulation. Furthermore, there is a microRNA recognition sequence at the 3’-UTR of ADAM10 mRNA. MicroRNA 144, 451, 103, 107, and 1306 inhibit the expression of ADAM10 by interacting with 3’-UTR of ADAM10 mRNA, suggesting its possibilities to develop novel strategy for ADAM10 upregulation by interfering with this interaction [74, 75] (Fig. 2B).
The regulation of ADAM10. A) The ADAM10 promoter region contains several transcription factor binding-sites, including retinoic acid-responsive elements (RARE), X-box binding protein-1 (XBP-1), NMDA and others. These transcription factors can activate transcription of ADAM10 by binding to corresponding region of ADAM10 promoter. B) At the translational level, the translation of ADAM10 is suppressed through the GC-rich 5’-untranslated region (5’-UTR) of ADAM10 mRNA. The 3’-untranslated region (3’-UTR) of ADAM10 mRNA contains a recognition element for microRNA. MicroRNA 144, 451, 103, 107, and 1306 inhibit the translation of ADAM10 by interacting with 3’-UTR of ADAM10. C) Multiple mechanisms control ADAM10-mediated cleavage at the post-translational level, including an activation of certain receptors, a heterogeneous group of molecules, intracellular trafficking and endocytosis of ADAM10.
Multiple mechanisms control ADAM10-mediated cleavage at the post-translational level, including an activation of certain receptors, a heterogeneous group of molecules, intracellular trafficking of ADAM10 and endocytosis [9, 77] (Fig. 2C). Receptors such as muscarinic acetylcholine (mAChRs), serotonin type 4 receptors (5-HT4Rs), and PACAP receptors can stimulate the activity of ADAM10 at posttranslational level [78–80]. Activation of mAChRs has shown a beneficial effect to alleviate AD pathology, and the M1 mAChRs is the specific subtype for this regulation. The M1 mAChRs knock-out (KO) mice showed decreased neuroprotective sAβPPα levels and increased Aβ levels in neurons. However, expression of M1 mAChRs could rescue this phenotype, indicating that M1 mAChRs might promote the ADAM10-mediated AβPP processing [80]. The neuropeptide PACAP38 promotes the ADAM10-mediated shedding of AβPP through interacting with its specific G protein-coupled receptor (GPCR) PAC1. This stimulatory effect can be inhibited by PAC1 antagonist and ADAM10 inhibitor [81]. The level of ADAM10 was increased at the postsynaptic membrane when treated primary hippocampal neurons with PACAP38, leading to a reduction of the dendritic spine head width and glutamate receptors [78]. Furthermore, another GPCR receptor, 5-HT4Rs, can also promote the activity of ADAM10 and sAβPPα secretion by directly interacting with the mature ADAM10. This activation may be mediated through cAMP/Epac signaling pathway [82]. The activity of ADAM10 has been reported to be regulated through MAPK, protein kinase C (PKC) signaling pathway, cAMP, and PI-3 kinase signaling pathway [9, 82]. A heterogeneous group of molecules are involved in the regulation of ADAM10-mediated shedding, including tissue inhibitor of metalloproteases 1 (TIMP1) and TIMP3, secreted frizzled-related proteins (Sfrp), and membrane cholesterol concentration [83–85]. TIMP1 and TIMP3 were initially identified as inhibitors of matrix metalloproteinases (MMPs) and showed strong inhibition to ADAM10 [86]. TIMP1 shows ADAM10-specific inhibitory properties, which can be used as a valuable tool to discriminate ADAM10 from other ADAM members. Sfrp is another inhibitor that downregulates the activity of ADAM10 [85]. In addition, the activity of ADAM10 is regulated by membrane cholesterol concentration. Cholesterol reduction promotesthe α-secretase-mediated AβPP processing and the generation of sAβPPα [84]. Cholesterol-reducing drugs such as statins can increase the activity of ADAM10 by their anti-cholesterol properties. In contrast, the activity of ADAM10 is inhibited if ADAM10 is targeted to cholesterol-rich membranes via a glycosylphosphatidylinositolanchor [87].
Intracellular trafficking is one of the most important pathway for the regulation of ADAM10-medaited proteolysis [77]. The regulation of ADAM10 trafficking is exemplified by synapse-associated protein-97 (SAP97) and tetraspanin (Tspan) protein family, both of which play critical roles in ADAM10 trafficking. SAP97, a cytoplasmic protein, regulates ADAM10 trafficking from Golgi compartment to synaptic membranes [88]. SAP97 promotes ADAM10 trafficking in synapses by binding to the intracellular domain of ADAM10 through its SH3 domain, improving its levels and activity in synaptic membranes [88]. ADAM10 trafficking from Golgi outposts to synaptic membranes requires PKC phosphorylation in the SH3 domain of SAP97, and the PKC phosphosite was altered in the brains of AD patients [89]. The SAP97-ADAM10 complex increased its appearance at the postsynaptic membrane [90]. Interruption the association between SAP97 and ADAM10 in the brain led to a non-transgenic mice model of AD. In the brains of AD patients, less SAP97-ADAM10 complex was found compared with healthy controls [90, 91]. The Tspan protein family plays a critical role in membrane compartmentalization and cluster of certain cell surface molecules, leading to a dynamic network of interactions called “Tspan web” or “Tspan-riched microdomains” [92, 93]. ADAM10 trafficking is controlled by several Tspan proteins including Tspan12 and TspanC8 (Fig. 3). Overexpression of Tspan12 promoted ADAM10-mediated shedding of AβPP in MCF7 and SH-SY5Y cell lines and also promoted ADAM10 prodomain maturation [94]. Furthermore, TspanC8, a subgroup of Tspan (including Tspan5, 10, 14, 15, 17, and 33), are essential for the regulation of ADAM10 maturation and cellular trafficking [95]. The role of TspanC8 in promoting ADAM10-dependent Notch-1 was revealed through a genetic screen in Drosophila [96]. The expression levels of membrane surface ADAM10 are positively regulated by Tspan5, 14, 15, and 33, and consequently promote the cleavage of ADAM10 substrates including AβPP, N-cadherin, and Notch-1 [97]. The extracellular loop of Tspan14 is responsible for interaction with the disintegrin and cysteine-rich domain of ADAM10, and the interaction between Tspan14 and ADAM10 promotes ADAM10 maturation and trafficking [98]. Tspan3 is recently identified as a novel ADAM10 interaction partner in complex with AβPP and the γ-secretase protease presenilin. Different from Tspan12 and TspanC8, Tspan3 did not affect ADAM10 trafficking from ER to plasma membrane, but Tspan3 could increase ADAM10-mediated shedding of AβPP and reduce Aβ liberation. Tspan3 may act as a stabilizing factor of active ADAM10, AβPP, and the γ-secretase complex in concert with other tetraspanins at the plasma membrane [99].
Endocytosis is another pathway to regulate the ADAM10 levels in synaptic membrane. There are several critical elements of the endocytic machinery in dendritic spines including AP2, clathrin, and dynamin-2 [100]. The endocytosis of ADAM10 at cell surface is mediated by clathrin, and the clathrin-mediated endocytosis of ADAM10 requires clathrin adaptor AP2, which interacts with the atypical motif domain within the intracellular domain of ADAM10 (Fig. 3). Compared with normal controls, the association of ADAM10 and AP2 was increased in the brains of AD patients [101]. Moreover, it was reported that both long-term depression (LTD) and long-term potentiation (LTP) could regulate the activities and levels of ADAM10. LTP in hippocampal neuronal cultures promoted ADAM10-AP2 association, resulting in ADAM10 endocytosis from synaptic membrane. However, LTD induced an opposite situation that both ADAM10 levels and activity were increased at synaptic membrane. The SAP97-ADAM10 interaction is essential in ADAM10 trafficking induced by LTD and LTD maintenance and spine morphology changes [101]. The localization and activity of ADAM10 at synapses can be regulated through its interaction with SAP97 and AP2, and this regulation is the basis for synaptic functions modulation [101, 102].
ADAM10 trafficking and synaptic functions. The trafficking of ADAM10 requires synapse-associated protein-97 (SAP97) and the tetraspanin (Tspan) proteins. SAP97 promotes ADAM10 trafficking in synapses by binding to the intracellular domain of ADAM10 through its Src homology 3 (SH3) domain. Tspan proteins including Tspan12 and TspanC8 regulate ADAM10 trafficking. Endocytosis is another pathway to regulate the ADAM10 levels in the synaptic membrane. The clathrin-mediated endocytosis of ADAM10 requires clathrin adaptor AP2. ADAM10 shapes the postsynaptic surface via shedding of its substrates to regulate synaptic plasticity and synaptogenesis.
Amyloid-β-independent roles for ADAM10 in AD
Intracellular NFT, another characteristic pathological change in AD, is mainly composed of hyperphosphorylated tau protein. Experimental evidences have shown that NFT and Aβ generation are intimately related, not as independent mechanisms respectively, and one of the links between them was soluble Aβ [103, 104]. Given the key role of ADAM10 in reducing Aβ generation, it is speculated that ADAM10 may influence tau pathology indirectly by changing the metabolism of AβPP and its fragments Aβ or sAβPPα. After exposure to Aβ oligomers, differentiated primary hippocampal neurons occurred as early localized changes that endogenous tau was missorted into the dendritic compartment, where localized calcium ion increased significantly, accompanied with destruction ofmicrotubules and spines [103]. Incubation of organotypic hippocampal culture with Aβ also induced a significant upregulation in tau phosphorylation [105]. Although recent studies have shown that Aβ peptides aggravate tau pathology in vivo [106], the mechanisms linking Aβ and tau pathology are still partly known. Mitochondria may play an important role in the association of Aβ and tau phosphorylation. Aβ peptides induce an increase of reactive oxygen species (ROS) generation from mitochondria through direct interacting with the heme groups [107], and oxidative stress lead to mitochondrial morphology defects, which are associated with tau hyperphosphorylation [108]. An animal model lacking in mitochondrial superoxide dismutase showed increased levels of hyperphosphorylated tau, suggesting that mitochondrial oxidative stress is closed related with the pathological features of AD [109]. Aβ-induced oxidative stress activates regulator of calcineurin 1 (RCAN1) and p38, which inhibit the activity of the tau phosphatase and promote tau phosphorylation respectively. Both of the two processes lead to tau hyperphosphorylation [110]. Furthermore, glycogen synthase kinase 3β (GSK3β), cyclin-dependent kinase 5 (CDK5), and Pin1 are also involved in the link between Aβ and tau pathology [110, 111] (Fig. 4).
ADAM10 in tau pathology. ADAM10 may compete with BACE1 in the first step of AβPP cleavage, leading to more sAβPPα but less Aβ generation. Aβ can cause an increase of mitochondrial generation of reactive oxygen species (ROS), leading to phosphorylation of tau through activating regulator of calcineurin 1 (RCAN1) and p38. Aβ can also activate phosphorylation of tau through cyclin-dependent kinase 5 (CDK5), glycogen synthase kinase 3β (GSK3β), and Pin1 pathway. Moreover, sAβPPα can inhibit the activity of BACE1 directly. Thus, ADAM10 may reduce tau phosphorylation indirectly through non-amyloidogenic AβPP processing.
It has been reported that sAβPPα not only has neuroprotective and neurotrophic properties, but also improves learning and memory abilities [112, 113]. Furthermore, sAβPPα may inhibit the activity of BACE1, and consequently reduce Aβ generation and promote α-secretase-mediated cleavage of AβPP [114]. Inhibition of BACE1 reduced GSK3β-mediated tau phosphorylation [115]. Therefore, sAβPPα, the ADAM10-mediated AβPP fragments, may reduce tau phosphorylation by inhibiting the activity of BACE1 (Fig. 4). This hypothesis has been demonstrated both in cell lines and mice in a recent study [116]. Taken together, ADAM10 may reduce tau phosphorylation indirectly through AβPP processing. Further research is required to investigate the potential pathways between ADAM10 and tau pathology.
Given the importance of synaptotoxic nature of Aβ and the role of ADAM10 in reducing Aβ generation, ADAM10 is speculated to play a critical role in alleviating synaptic degeneration. It has been observed that both ADAM10 localization and its activity were reduced in synapses in the brains of AD patients, especially in hippocampus, indicating that an appropriate level of ADAM10 in synapses is indispensable to regulate its activity and maintain normal synaptic functions [90]. ADAM10 is abound in the trans-Golgi network (TGN), cell membrane, and postsynaptic density [117, 118], but the majority of BACE1 is confined to the TGN and endosome [119]. Because the two proteases are differentially segregated in cells, protein trafficking and subcellular localization of the two proteases have a great influence on Aβ generation. Appropriate ADAM10 trafficking and insertion at synaptic membrane are crucial for ADAM10-shedding activity and synaptic functions.
ADAM10 also plays a critical role in shaping the postsynaptic surface to regulate synaptic plasticity (Fig. 3). Because the early lethality of the ADAM10 cKO mice prevented investigating the functions of ADAM10 at later stages, ADAM10 cKO mice using a CaMKIIα-Cre deleter strain were generated to investigate the functions of ADAM10 in postnatal mice [31]. These mice model showed an altered morphology of postsynaptic structure and a reduction of NMDAR in brain with learning deficits and epileptic seizures [31]. The lack of ADAM10-mediated cleavage of postsynaptic membrane proteins including AβPP, Neuroligin-1 (NLG1), and N-Cadherin may explain this phenotype. As a postsynaptic adhesion molecule, NLG1 plays a critical role in the development of synaptic structure through binding to its presynaptic ligand neurexin [120]. NLG1 is one of the primary substrates of ADAM10, and the ADAM10-mediated NLG1 shedding can be promoted by soluble neurexin ligands or NMDAR activation. However, inhibition of NLG1 shedding led to increasing dendritic spines in neuronal cultures, indicating NLG1 shedding might negatively regulate the spine remodeling. N-cadherin is another primary substrate of ADAM10 in synapses. The downregulation of ADAM10 at synaptic membrane inhibited the ADAM10-mediated N-cadherin cleavage, leading to N-cadherin accumulation, spine head width increase and the change of AMPA type glutamate receptors [121].ADAM10-mediated N-cadherin shedding plays an important role in controlling neurite outgrowth in primary cultured retinal cells [122]. Dendritic spine morphology in hippocampal neurons was modulated by PACAP38 through ADAM10-N-cadherin signaling pathway. The modulation was prevented by either ADAM10-specific inhibitor or interruption of ADAM10-mediated N-cadherin cleavage [78]. These facts indicate a key role of N-cadherin in regulating synaptic plasticity through a complex sequence of ADAM10-mediated events. N-cadherin signaling initiated by ADAM10 promoted neural progenitor cells (NPCs) recruitment and migration from the adult subventricular zone into demyelinated lesions, contributing to repair of the injured section in the brain [29]. Furthermore, NCAM is an important regulator of neuronal development. NCAM undergoes ectodomain shedding by ADAM10 and generates soluble NCAM, which plays a critical role in neurite branching and outgrowth [30]. The ADAM10-mediated NCAM shedding is induced by ephrinA5 and EphA3. The shedding of NCAM by the ephrin5/EphA3/ADAM10 mechanism may affect synapse development and regulate cone collapse in neurons [123]. Recently, Kuhn et al reported a novel ADAM10 substrate NgCAM-related cell adhesion molecule (NrCAM). Reduced cleavage of NrCAM and other ADAM10 substrates led to mistargeted axons in the olfactory bulb of ADAM10 cKO mice, indicting ADAM10 may play an important role in axon target and brain development [12].
It has been reported that ADAM10 played an important role in synaptogenesis through the analysis of ADAM10 overexpression transgenic mice, which showed increased GABAergic glutamatergic and cholinergic presynaptic bouton densities in cortex. Injection of sAβPPα into the cortex of wild type mice got the similar positive effect, indicating that ADAM10-mediated AβPP shedding was responsible for this neuro-trophic effect [124]. However, an opposite result is observed in Fragile X Syndrome (FXS). In FXS, sAβPPα signals activate the MAP kinase pathway through the metabotropic receptor, leading to synaptic and behavioral deficits. Therefore, synaptic development and synaptic functions need precise control of the ADAM10-mediated AβPPcleavage [125].
ADAM10 in neurogenesis and gliogenesis
Adult neurogenesis was reported to be affected by all early-onset familial AD genes and by Aβ in AD mouse models. Dysfunctional neurogenesis may increase the neuronal susceptibility to AD and aggravate memory impairment, but improved neurogenesis is beneficial for brain repair mechanism [126]. The role of ADAM10-mediated Notch-1 signaling pathway in neurogenesis and gliogenesis has been revealed in recent studies. The classic ADAM10 KO mice died at the early embryonic stage with disturbed somitogenesis and severe vascular defects, due to lack of ADAM10-mediated shedding of Notch-1 receptor and its ligands [27]. Studies investigating the functions of ADAM10 in vivo had been precluded for the early lethality of ADAM10 KO mice. To investigate the role of ADAM10 in brain in vivo, Jorissen and his colleges generated nestin-driven ADAM10 cKO mice that ADAM10 was confined to inactivation both in NPCs and its derived neurons [34]. The precocious neuronal differentiation led to a disrupted cortex in brain and an obvious reduction of ganglionic eminence, and all these changes resulted in perinatal death of cKO mice. Dysfunction of Notch-1 dependent signaling pathway is liable for this neurogenic phenotype, indicating ADAM10-Notch-1 signaling is critical for neuronal differentiation and cerebral cortex development [34]. Immunohistochemistry and in situ hybridization methods showed that ADAM10-Notch-1 signaling pathway plays critical roles in both neuronal maturation and gliogenesis during the late embryonic stage [127]. To investigate its role in the adult brain, cortex and hippocampus ADAM10 deficiency cKO mice model were generated. The premature neuronal differentiation led to significant reduction of proliferating NPCs and a significant increase of postmitotic neurons both in the subgranular zone and the hippocampal dentate gyrus (DG) in the brains of ADAM10 cKO mice. These ADAM10 cKO mice showed memory and learning abilities decline in Morris water maze [128]. Jaehong Suh analyzed the proliferation of NPCs by injection of BrdU into nascent cells in the hippocampus of adult mice. It was observed that the levels of NPCs proliferation and differentiation in LOAD mutant ADAM10 mice (R181G and Q170H) was lower compared with ADAM10 wild type (WT) mice, but the levels of glial differentiation had no significant differences in different mouse groups [14]. Furthermore, the levels of gliosis in mice brains were correlated well with amyloid plaques, indicating that ADAM10-meidated non-amyloidogenic processing of AβPP reduces reactive gliosis in AD pathogenesis [14].
The AβPP fragment sAβPPα has been reported as a proliferation factor of NPCs both in vitro and in vivo [129]. Inhibition the activity of ADAM10 reduced proliferation of NPCs and this reduction could be recovered by additional sAβPPα. The sAβPPα-induced proliferation of NPCs may occur through ERK and MAPK signaling pathways. Importantly, ADAM10 and sAβPPα are enriched in the subventricular zone of adult mice, indicating that ADAM10 and AβPP processing may be critical for the proliferation of NPCs [130]. To test the effects of AβPP processing on hippocampal neurogenesis, Jaehong Suh measured sAβPPα and sAβPPβ levels in the hippocampal lysates in ADAM10-WT mice and LOAD mutant ADAM10 mice. An increased ratio of sAβPPα/sAβPPβ was observed in the hippocampus of ADAM10-WT mice compared to the LOAD mutant ADAM10 mice. However, the expression levels of Notch-1showed no difference between different mice groups [14].
Taken together, the proliferation of NPCs and their differentiation into neurons are regulated by ADAM10 both in embryo and adult mice. Although the role of ADAM10 in gliogenesis is still controversial, the ADAM10-meidated AβPP processing has been reported to reduce reactive gliosis in AD pathogenesis [14]. ADAM10 may affect neuronal proliferation and differentiation through AβPP processing and the Notch-1 signaling pathway, and it is a promising target to relieve the pathology in AD by improving hippocampus neurogenesis.
THE POTENTIAL OF ADAM10 AS A BIOMARKER FOR AD
In addition to abundant in CNS, ADAM10 protein also exists in platelets. It has been reported that the levels of ADAM10 in platelets were reduced in AD patients [131, 132]. Positive correlations between ADAM10 levels in platelets and the clock drawing test as well as the Mini-Mental State Examination were shown in recent studies [133, 134]. However, the ADAM10 gene expression in the platelets of AD patients had no significant differences compared with either mild cognitive impairment group or normal controls [135]. Interestingly, the levels and activity of ADAM10 in cognitively healthy individual are increased in an age-dependent manner [136]. The mechanisms that regulate the levels of ADAM10 protein in platelets are still unclear, and the expression of ADAM10 in platelets may be affected by medicine such as serotoninergic antidepressants [137]. According to a relevant study performed in Brazilian, the ADAM10 levels were decreased in platelets in 30 AD patients compared to 25 healthy controls, and disease progression could intensify this reduction. The research group pointed out that the reduction of ADAM10 in platelets may increase the occurrence probability of AD [132]. All these facts suggest that ADAM10 has potential to be a biomarker for AD diagnosis at an early stage and more relevant studies are required to confirm the theory.
ADAM10-TARGETED THERAPEUTICS
Previous studies have shown that mild ADAM10 overexpression led to a reduced AD pathology, and a moderate increase of ADAM10 expression was sufficient to change the major pathological features in AD [7]. The overexpression of ADAM10 led to more than 300 genes mild expression changes, but Notch signaling was not significantly altered in adult mice [138]. These facts indicate that ADAM10 is a promising target for AD therapy, and a proper upregulation of ADAM10 may be therapeutically acceptable [18, 139]. ADAM10 activation has multiple beneficial effects as indicated above. Firstly, ADAM10 reduces the generation of neurotoxic Aβ and increases the generation of the neuroprotective and neurotrophic sAβPPα fragments [18]. Secondly, ADAM10 is critical to for synaptic plasticity and synaptic functions, which are impaired at the early stage of AD. Thirdly, ADAM10 plays an important role in hippocampal neurogenesis and gliogenesis, which may alleviate the pathologic changes of AD (Fig. 5). All these beneficial effects of ADAM10 may delay the pathogenic process of AD, indicating the potential of ADAM10 as a target for drug development. The underlying pathways that increase the expression or activity of ADAM10 in the brain may contribute to alleviating AD pathology [6].
The role of ADAM10 in AD. ADAM10 plays multiple beneficial roles in mitigating the pathologic impairment in AD. As the major α-secretase, ADAM10 may compete with BACE1 for the AβPP cleavage, and consequently reduces the production of neurotoxic Aβ and increases the generation of neuroprotective sAβPPα. ADAM10 also contributes to alleviating the pathology of AD by reducing tau pathology, maintaining proper synaptic functions, promoting neurogenesis and gliogenesis, and thereby maintain the homeostasis of neuronal networks.
The expression of ADAM10 and its activity is tightly regulated at transcription level. Acitretin, a vitamin A analog, can increase ADAM10 transcription by acting on RARE in the ADAM10 promoter [59]. Increased CSF sAβPPα level in AD patients was observed after treating with acitretin in a recent clinical trial. Compared with the placebo group, the acitretin group showed a significantly increasing sAβPPα levels in CSF (p = 0.035) within a 4-week treatment period. Importantly, the synthetic retinoid acitretin was overall safe and well tolerated [140]. SIRT1 upregulated ADAM10 gene expression via coactivation of the RAR transcription factor, leading to reduced production of Aβ and plaques in mice model [66]. Additionally, SIRT1-mediated ADAM10 activation also increases the Notch signaling pathway, which plays an important role in brain injury repair [62]. These findings suggest that SIRT1 activation has therapeutic potential for AD through upregulation ADAM10. A very recent study showed that osmotin, a mammalian adiponectin analogue, can promote ADAM10 expression in an AMPK/SIRT1-dependent manner, leading to reduced Aβ generation and improved clinical symptoms in AD mice model [141]. Melatonin, a regulator of physiological functions, not only acts as an antioxidant to reduce Aβ-induced oxidative stress, but also inhibit BACE1 expression and promote ADAM10 expression [142]. Melatonin improves the activity and transcription of ADAM10 through activating melatonin receptor, and consequently results in transactivation of ADAM10 promoter in different cell lines [70].
ADAM10 translation rate is repressed by the 5’-UTR of ADAM10 mRNA, and knock down of the 5’-UTR results in an apparent increase of ADAM10 levels [73]. Therefore, drugs that mitigate the suppressive effect of the 5’-UTR by interacting with this region may be a new strategy for AD treatment. A recent study showed that the compound 24 of methylquinolinium derivatives is selective affinity for the 5’-UTR region and significantly promote the translation of ADAM10. After treatment with compound 24, the secretion of sAβPPα was significantly increased and Aβ in cellar was decreased [143]. These results indicate that drugs alleviate the inhibition of the 5’-UTR in ADAM10 mRNA may be therapeutic potential for AD.
It has been reported that exogenous neurotrophic factors could change AβPP metabolism and led to less generation of Aβ in mouse models. For instance, fibroblast growth factor-2 (FGF2) treatment inhibited the generation of Aβ and the activity of BACE1 in AβPP23 mice [144]. Furthermore, a recent study shows that FGF2 and nerve growth factor secreted by C6 glioma cells could reduce Aβ generation and promote ADAM10-meidated AβPP cleavage in vitro [145]. Astrocytic neurotrophic factors play a critical role in ADAM10 upregulation and may be therapeutic potential for AD. However, the side effect of astrocytic neurotrophic factors should be carefully evaluated. For example, FGF2 is able to increase the risk of cancer and promote cancer progression [146]. As mentioned above, the activity of ADAM10 is also regulated by membrane cholesterol concentration. Cholesterol reduction promotes the ADAM10-mediated AβPP processing and the generation of sAβPPα [84]. Cholesterol-reducing drugs such as statins can increase the activity of ADAM10 by their anti-cholesterol properties. Recent studies showed that lovastatin, a statin, was sufficient to inhibit cholesterol biosynthesis and activate ADAM10 at a low concentration [87]. Given their anti-cholesterol properties, statins may be beneficial for AD patients, especially for those with cerebral vascular diseases.
In summary, ADAM10 is a promising therapeutic target for AD. Drugs improve the activity or expression of ADAM10 may alleviate Aβ burden, synaptic dysfunction, and neuronal loss in brains of AD patients. All these beneficial effects may compensate for the pathological lesion in AD and slow down AD progression. However, as a primary protease for membrane proteins in brain, ADAM10 sheds more than 90 neuronal substrates. Thus, side effects of mild ADAM10 activation should be carefully evaluated [9, 12]. Although many ADAM10 stimulators have been identified, their side effects have not been carefully evaluated. Thus, ADAM10-specific stimulators may be more suitable for ADAM10-targted therapy. More basic studies and clinical trials are needed to investigate the therapeutic potential of ADAM10-targeted AD therapy.
CONCLUSIONS AND FUTURE PERSPECTIVES
As the major constitutive α-secretase for AβPP ectodomain shedding, ADAM10 plays a critical role in reducing the generation of Aβ peptides. Recent studies have shown its beneficial role in alleviating the pathologic impairment in AD both in vitro and in vivo. However, the role of ADAM10 in AD and the underlying molecular mechanisms are still not well established. Mechanistically, ADAM10 regulates its functions through interacting with postsynaptic substrates such as synaptic cell receptors and adhesion molecules. With the identification of the new substrates in CNS, new ADAM10 functions will be identified. This review summarized the recent work on ADAM10 functions andhighlighted the potential role of ADAM10 in reducing tau pathology, maintaining synaptic functions, promoting hippocampal neurogenesis, and reducing reactive gliosis, and thereby promotes the homeostasis of neuronal networks. Given its critical role in AD, ADAM10 may be a potential therapeutic target for AD. However, it should be noted that ADAM10 is widely distributed as a transmembrane protease, which is responsible for the cleavage of various transmembrane proteins. ADAM10 is indispensable in cell differentiation, proliferation, and cell adhesion, and ADAM10 activation may disturb the normal physiological functions. Thus, it is necessary to evaluate the side-effect and consequences induced by ADAM10 activation. Furthermore, the main component of the regulated α-secretase is still unclear and it may change with different stimulators. ADAM10 activation may not influence the cleavage of AβPP in the regulated α-secretase process. The processing of AβPP is more complicated than what it has been reported and more relevant studies are needed to clarify the underlying cellular pathways in vivo, especially in the brains of AD patients. Although it has been debated, ADAM10 gene may be a LOAD susceptibility gene and ADAM10 protein in the platelets seems to be a promising biomarker for AD diagnosis at an early stage. More relevant studies are needed to clarify the molecular mechanism of ADAM10 activation and its potential application in AD therapy.
Footnotes
ACKNOWLEDGMENTS
This work was supported by grants from the National Natural Science Foundation of China (81471309, 81371406, 81571245, and 81501103), Shandong Provincial Collaborative Innovation Center for Neurodegenerative Disorders and Taishan Scholars Program of Shandong Province.
