Abstract
The processing of the amyloid-β protein precursor (AβPP) by β- and γ-secretases is a pivotal event in the genesis of Alzheimer’s disease (AD). Besides familial mutations on the AβPP gene, or upon its overexpression, familial forms of AD are often caused by mutations or deletions in presenilin 1 (PSEN1) and 2 (PSEN2) genes: the catalytic components of the proteolytic enzyme γ-secretase (GS). The “amyloid hypothesis”, modified over time, states that the aberrant processing of AβPP by GS induces the formation of specific neurotoxic soluble amyloid-β (Aβ) peptides which, in turn, cause neurodegeneration. This theory, however, has recently evidenced significant limitations and, in particular, the following issues are debated: 1) the concept and significance of presenilin’s “gain of function” versus “loss of function”; and 2) the presence of several and various GS substrates, which interact with AβPP and may influence Aβ formation. The latter consideration is suggestive: despite the increasing number of GS substrates so far identified, their reciprocal interaction with AβPP itself, even in the AD field, is significantly unexplored. On the other hand, GS is also an important pharmacological target in the cancer field; inhibitors or GS activity are investigated in clinical trials for treating different tumors. Furthermore, the function of AβPP and PSENs in brain development and in neuronal migration is well known. In this review, we focused on a specific subset of GS substrates that directly interact with AβPP and are involved in its proteolysis and signaling, by evaluating their role in neurodegeneration and in cell motility or proliferation, as a possible connection between AD and cancer.
AβPP AND γ-SECRETASE
The amyloid-β protein precursor (AβPP) is a type I membrane protein with a complex and intriguing proteolytic processing. Firstly it is cleaved at the luminal side by β-secretase (memapsin 1 or BACE1) [1, 2] to form membrane-bound C-terminal fragments know as C99 and C89 and, alternatively, by α-secretases (the most studied are members of the ADAM family) to form C83 fragments [3]. C-terminal fragments C99-C89 and C83 (collectively as CTFs) are then cleaved by γ-secretase (GS): a high molecular weight protein complex, with specific “intramembrane-cleaving aspartyl protease activity” (I-CLiPs) or “regulated intramembrane proteolysis” (RIP) [4, 5]. GS’s cleavage leads to the liberation of the amyloid intracellular domain (AICD), with a yet unclear transcriptional role at nuclear level [6], and to the formation of soluble amyloid-β peptides (Aβ), which are released as monomers or oligomers into the lumen. The soluble forms of Aβ (mainly those isoforms ending at residue 42, Aβ42) initiate and contribute to Alzheimer’s disease (AD) pathogenesis [7, 8]. The “amyloid hypothesis” (and its recent modifications) states that formation of soluble, oligomeric Aβ is the main cause of AD pathology and that neurofibrillary tangles, cell loss, vascular damage, and dementia follow as a direct result of this deposition [9, 10]. Therefore, in this view, to block GS and the aberrant formation of Aβ oligomers is an important therapeutic target to be pursued. The constituents of the GS complex are four transmembrane proteins: heterodimeric presenilins (either PSEN1 or PSEN2 which are 65% homologous), nicastrin (NCSTN), the anterior pharynx-defective 1 (APH-1), and the presenilin enhancer 2 (PEN-2).
NCSTN, a glycosylated 130 kDa type I protein, is involved principally in substrate recognition and selectivity [11]; APH-1, which regulates GS activity, is positioned at a cleavage site on AβPP and has a crucial role in substrate selectivity [12, 13]; PEN-2 is functionally required for endoproteolysis and for stabilization of the most important components: PSENs, which represent the catalytic core of the GS complex [14, 15].
During searches for genes responsible for familial early onset AD (EOAD) [16, 17], PSEN genes were the first identified. Currently, over 180 AD-causing mutations have been identified and most of them are localized on the PSEN1 gene (chromosome 14), while only a few are located in the PSEN2 gene (chromosome 1) [18]. Although debated, some PSEN mutants favour an increment of soluble Aβ42 over Aβ40 [19, 20]: this event is considered central in AD genesis, according to the “amyloid hypothesis”. However, familial patients with very early onset and bearing specific PSEN1 mutations have also a peculiar brain pattern of soluble Aβ42, having increasing amounts of N-terminal truncated and pyroglutamate-modified peptides such as AβN3pE-42 and AβN11pE-42 in comparison to sporadic cases [21]. These shortened peptides are more neurotoxic and more prone to oligomeric assembly than Aβ40/42 peptides [22].
PSENs are synthesized as full length proteins but their structure seems to be unstable and they are quickly either endoproteolysed or degraded. This intramolecular autocatalytic event, which induces a cleavage between residues N292 and V293 on PSEN1, is required for GS activation upon formation of a stable heterodimer between the amino-terminal fragment (NTF 30 kDa, transmembrane domain, TMD, 1-6) and the carboxyl-terminal fragments (CTF 20 kDa, TMD 7-9) [23, 24]. The intramolecular re-arrangement of TMD 6 and TMD 7 represent the biologically active structure of this protein, bearing in close proximity two aspartic acids (D257 and D385 residues), which are essential for the catalytic activity of PSENs [25].
To date, therapies based on GS inhibitors or through an immunological control of Aβ peptides failed at the pre-clinical or clinical level [26, 27], strongly suggesting that other conditions control AD development in parallel to AβPP processing by GS. Some authors also claim that Aβ has a protective role at a vascular level and that it may be an important mediator of long-term potentiation, thus implying that a “therapeutic” reduction of its physiological production might even be harmful[28–30].
The relevant role of PSENs/GS in AD and in AβPP processing is clear, and mainly linked to familial EOAD development. However, GS is also directly involved in a plethora of biological processes depending on cell type, interactors, and substrates [31, 32]. Its role in cancer development is widely recognized and extremely relevant, since some of its substrates are directly linked to carcinogenesis or metastasis control [31–33]. Therapeutic approaches to hamper GS activity, using specific inhibitors or “modulators”, is a scientific endeavor in the field of cancer [34, 35].
In this scenario, the contribution of AβPP itself to neuronal migration, cell motility, and proliferation [36–39] is relevant, considering the number of signaling adaptors of AβPP on the cytoplasmic site [40, 41]. Therefore, some authors suggest that, in abnormal circumstances, AβPP and GS might be modulators of aberrant attempts of neuronal cells to reenter into the cell cycle [42, 43]. It is interesting to note that, from this point of view, both AβPP and its CTFs are important cell cycle enhancers, modulators of MAPT phosphorylation [44], and, possibly, of neurodegeneration as well [38, 45].
Here we examine and discuss the role of specific GS substrates in AD development, exploring their physiopathological role and reciprocal interaction with AβPP, with respect to amyloid formation and their involvement in cell signaling pathways, focusing on cell proliferation, migration, and cancer development.
AβPP AND “THE OTHERS”
AβPP is the best studied among GS targets; however, the list of putative substrates includes more than 90 proteins [33]. Most of them are type-I transmembrane cell surface adhesion molecules and receptors, with a cytoplasmic C-terminus often involved in the initiation and mediation of intracellular signaling concerning: cell differentiation, adhesion, migration, neurite outgrowth, axon guidance, or formation and maintenance of synapses.
Among different GS substrates, we here focus on those that directly interact with AβPP and that have a putative role in amyloid formation; we explore, in parallel, the opposite hypothesis that GS (or AβPP) may hamper the function of each single substrate (Table 1).
Summary table about the main physiological roles of GS substrates in brain and the effects caused by their interaction with AβPP and/or its fragments
Alcadein α/Calsyntenin-1
The neural type I membrane protein Alcadein α (Alcα, Uniprot O94985), also known as Calsyntenin-1, is a member of the alcadein family: a subset of the cadherin superfamily, that binds to kinesin-1 light chain (KLC) to activate kinesin-1-dependent transport of Alcα-containing vesicles. Alcα is highly expressed in most cerebral cortical neurons in which mediate the axonal anterograde transport of different vesicles [46]. This protein is firstly cleaved by α-secretase (ADAM10 and ADAM17, primarily) to generate a membrane-associated carboxyl-terminal fragment (Alcα-CTF), which is further cleaved by GS to secrete p3-Alcα fragments producing an intracellular cytoplasmic domain fragment (Alcα-ICD) [47]. Alcα and AβPP are co-transported along axons to early endosomes in the central region of growth cones. Loss of Alcα, by RNAi, disrupts AβPP axonal transport, an early pathological feature in AD, at the same time enhancing AβPP processing and Aβ production [48, 49]. In fact, Alcα interacts with AβPP and with the cytoplasmic adapter protein X11L (all are co-localized in dystrophic neurites in senile plaques) resulting in the formation of a tripartite complex, which stabilizes AβPP, delaying its intracellular maturation [50]. X11L and Alcα form a complex also with C99, evading GS cleavage and therefore reducing Aβ and Alcα-ICD formation [50].
Recently, has been described a new role of Alcα-ICD in the regulation of membrane protein trafficking, facilitating AβPP transport into the late secretory pathway and enhancing Aβ generation [51]. The GS-dependent product p3-Alcα has been proposed as a biomarker for sporadic AD cases [52], since its plasma levels are enhanced in cohorts of clinically diagnosed AD patients or in patients with concomitant decrease in the Mini-Mental State Examination score, when compared with age-matched non-demented subjects [53]. In addition, p3-Alcα levels are also increased in the cerebrospinal fluid of patients with mild cognitive impairment or with clinical dementia [54].
Finally, GS cleavage of Alcα is crucial to limit the inappropriate peripheral retention of KLC that would lead to hampered axonal transport [55]. Altogether, a controversial and dual role of Alcα emerges: on one side, a regulated physiological processing of Alcα by GS seems required for the proper maintenance of axonal transport; on the other, an increment of its processing seems linked to a parallel increment of the amyloidogenic pathway. Therefore, a putative “loss of function” of PSEN/GS could only be explained, in the wake of the “amyloid hypothesis”, by an “erroneous cut” that favors the parallel processing of both Alcα-CTF and of AβPP-CTFs. In this sense, if we consider the increase of p3-Alcα found in AD patients to be valid, it is inconceivable that a putative PSEN/GS “loss of function” would instead increase the levels of full length Alcα in AD.
Deleted in colorectal cancer
Deleted in Colorectal Cancer (DCC, Uniprot P43146) is a cell surface receptor for Netrin-1 and regulates cell and axonal migration during nervous system development [56]. Netrin-1 was the first molecule known to be involved in neuronal guidance, and its expression in the neural tube is needed to guide and attract neurons during nervous system development [57]. It is well known that GS is needed for a proper development of axonal and dendritic processes, and that several substrates of GS are involved in synaptic physiology, brain plasticity, and neurite outgrowth [33]. In particular, the processing of DCC is required for the proper control of axons to midline guidance signals. In fact, its membrane-associated carboxyl terminal fragment (DCC-CTF), likely produced through an ADAM-like activity [58], induces neurites outgrowth in mouse neuroblastoma cells; on the contrary, the product of GS activity (DCC-ICD) inhibits neurites outgrowth in parallel with downregulation of DCC-mediated intracellular signaling [59]. In this context, the pharmacological inhibition of GS might lead to the accumulation of DCC-CTF at the plasma membrane, triggering neurites growth.In C. elegans, the overexpression of a DCC-CTF variant that is insensitive to GS cleavage maintains motor neurons responsive to Netrin-1, keeping them into the neural tube. The abnormal increment of DCC-CTF hampers the response to repellent signals which normally stop the migration at this step [60]. Similarly, in a PSEN1 mutant murine model that caused the accumulation of DCC-CTF within the membrane, an abnormal growth of motor axons occurs at the midline of the floor plate, with abnormal attraction to Netrin-1 and insensitivity to repulsive signals [61].
More recently, it has been demonstrated that AβPP expressed at the growth cone interacts with DCC as well, acting as a co-receptor, to mediate axon guidance during commissural axon navigation. Inactivation of AβPP is associated with abnormal distribution of commissural neurons in the developing neural tube and is associated with reduced commissural axon outgrowth [62]. Silencing of AβPP in mice embryo reduces the migration of neurons from the ventricular zone to the cortical plate, likely via Dab1 [37, 63]. From the other side, Netrin-1 modulates AβPP signaling, triggering AICD-dependent gene transcription, and suppressing Aβ production, at least in AD transgenic (Tg) mice [64].
In addition to its role in nervous system development, DCC has an obvious role in cancer: decreased DCC expression (linked to the loss of heterozygosity of chromosome 18q, where DCC is located), was observed in a large cohort of colorectal and other cancers. In particular, in most reports, allelic loss of 18q are infrequent in early stage of tumors, but are common in primary colorectal carcinomas and in nearly 100% of hepatic metastases arising from colorectal primaries. This implies that chromosome 18q loss of heterozygosity may contribute more to the progression than to the initiation of colorectal cancer [65]. Netrin-1 and DCC have been reported to promote tumorigenesis in many types of cancers, and interestingly, the DCC-ICD produced by GS seems a transcriptional co-activator, mediating cell-cycle regulation in colorectal cancer [66]. Therefore, there is a dual crosstalk between AβPP and netrin/DCC pathways: on one side, we can consider DCC (and its proteolysis by GS) involved in AβPP signaling and Aβ formation; on the other AβPP (and its processing by GS) might influence DCC. Whether these events may occur in neurodegeneration or in cancer development (or both) is still unclear.
LDL receptor family
The LDL receptor gene family is an apparently homogeneous group of cell surface receptors involved in a wide range of cellular signaling pathways [67]. Most of these receptors are processed through iCliPS by GS, and some of them interact with AβPP: in particular, very low density lipoprotein receptor (VLDLR), low density lipoprotein receptor-related protein 8 (LRP1), low density lipoprotein receptor-related protein 1 (LRP8), Sortilin-related receptor 1 (SorL1), and Sortilin (SORT1). They share with AβPP common cytosolic signaling adaptors such as FE65, X11/Mint, and Disabled-1 (Dab1), and all are receptors for apolipoprotein E (ApoE) [67]. APOE is a polymorphic gene known to be the most relevant risk factor for developing late-onset Alzheimer’s disease (LOAD) [68, 69]. The ApoE ɛ4 allele is genetically associated to familial and sporadic LOAD, with increased risk for developing AD from 20% to 90%, depending on ɛ4 allele load [69]. The mechanisms by which ApoE influences AD onset and development is still unclear and, to date, two opposite hypotheses prevail: either ApoE, in its various isoforms, is involved in complexes with Aβ whose relative stability affects their toxicity or clearance [70–72]; or ApoE outcomes descend from different and parallel biological processes, in which the family of LDL receptors plays a predominant role [73–77].
Some LDL receptors may modulate, at the same time, formation and vascular clearance of Aβ [78–81], synaptic plasticity and dendritic spine formation [82] or may control neuronal migration and formation of cortical layers [83, 84]. The latter feature is triggered by Reelin and F-spondin, which are soluble ligands for LRP1 (Uniprot Q07954), LRP8 (Uniprot Q14114), and VLDLR (Uniprot P98155), via phosphorylation of Dab1 [85, 86], the same adaptor that binds the cytosolic portion of AβPP. Reelin facilitates the interaction of LRP8 and AβPP with Dab1 and, in parallel, triggers the processing of both receptors by GS [87]. Apparently, NMDA receptors are also conditioned by Reelin and LDL receptors which recruit Dab1 and facilitate Ca2 + entry through NMDA channels [74]. This activity is needed for learning and memory function; it is negatively modulated by ApoE ɛ4 and arises from the regulated proteolysis of LRP8 or LRP1 by GS [74, 89]. Interestingly, LRP8-ICD is a negative regulator of Reelin, thus suggesting a feedback regulatory system triggered by GS cleavage of LRP8 [90].
However, as far as neuronal migration concerns, Reelin signaling is apparently intact in PSEN1-/- primary neuronal cultures, where GS activity is not required for Reelin-induced phosphorylation ofDab1 [91].
The crosstalk between LDL receptors and AβPP is even more complex if we look at another common cytoplasmic interactor: FE65. In fact LRP1, LRP8, and VLDLR bind to this cytoplasmic adapter protein, forming different tripartite complexes [92] which may modulate AβPP endocytic trafficking, Aβ production, AICD nuclear trafficking, and DNA protection [93–95]. It is worth to remember that AICD, along with FE65 and Tip60/Kat5, associates into AFT complexes in nuclear transcription factories, and that the C-terminal portions of LDL receptors may be recruited at nuclear level upon GS cleavage [88, 96]. At the same time AICD, along with FE65 and Tip60/Kat5, blocks LRP1 transcription [97], suggesting that AβPP proteolytic processing by GS reduces LRP1 brain levels possibly hampering both Aβ clearance and Reelin signaling. In fact LRP1 neuronal KO induces progressive, age-dependent, synapse loss and neurodegeneration [98]. Finally, a recent report suggests that ApoE (mainly ApoE ɛ4), via its receptors, activates a non-canonical MAP kinase cascade that enhances AβPP transcription and Aβ synthesis [77].
In summary, it is likely that any transcriptional/transductional activity mediated by C-terminal regions of both AβPP and LDL receptors is reciprocally conditioned upon GS cleavage. In this scenario, it is very difficult to understand the potential effect that a gain/loss of function on GS would exert on LDL receptor biology, since GS processing is apparently required for their function in learning processes. One question that needs to be answered is whether AβPP and LDL receptors, when they interact with each other, are co-processed by GS or if each receptor can be individually and separately cleaved also considering that GS exists in different combinations and variants [13, 99]. In the first instance, a possible loss of function on GS may involve both AβPP and the receptor coupled to it. In this case, we could observe at the same time an increment of Aβ42 (if “loss” means “erroneous”) and an hampered processing of LDL receptors, resulting in loss of their function (linked to Reelin or FE65/Tip60, for example). Otherwise, if GS cleavage occurs independently, we should observe opposite effects on each protein: either a normal or slightly increased processing on a substrate and a reduced cleavage on the second.
SorL1 (Uniprot Q92673), also known as LR11 or SorLA, is another member of the family, that acts as a sorting and signaling receptor expressed in neurons of the central and peripheral nervous system [100, 101]. SorL1 is genetically associated with development of LOAD: inherited variants in SorL1 occur in at least two different clusters of intronic sequences within the gene, regulating tissue-specific expression of SorL1 [18, 102]. SorL1 genetic variants apparently direct trafficking of AβPP into recycling pathways resulting in an increased secretion Aβ40 and Aβ42 [103]. SorL1 acts also as a retention factor of AβPP in the trans-Golgi, and drives AβPP to secretory and endocytic organelles [104–106]. Also SORT1 (Uniprot Q99523), another member of the Vps10 and the LDLR receptors family, is a GS substrate [101] and interacts with AβPP in the neurites of hippocampal neurons. There are, however, differences about the role of SorL1 and SORT1 in AβPP proteolysis and more generally in AD: 1) apparently SorL1 colocalizes with AβPP in neuronal soma, while SORT1 prefers a neuritic localization; 2) SORT1 promotes the α-secretase cleavage of AβPP rather than β-secretase cut [107]; 3) like SorL1 also SORT1 interacts with BACE1; however, the overexpression of SORT1 in cultured cells increases BACE1-mediated cleavage of AβPP, enhancing C99 and Aβ levels [108]; while SorL1, on the contrary, significantly reduced BACE-AβPP interactions in Golgi, preventing BACE-AβPP interactions and AβPP cleavage [109]. Furthermore, SORT1 effects on Aβ levels might be also explained by the fact that, as ApoE receptor, it may facilitate Aβ clearance [110].
Some evidences propose a protective role of SorL1 in AD pathogenesis, upon the observation that the over-expression of SorL1 causes a redistribution of AβPP to the Golgi apparatus and reduces Aβ formation; conversely SorL1 deletion increases amyloidogenic processing of AβPP and aggravates Aβ plaque deposition in Tg AβPP mice [111]. Furthermore, SorL1 levels are reduced in the Golgi apparatus and in early endosomal compartments in AD brains, in particular in limbic and occipital regions [112, 113]. Contrasting data emerge about quantitative analysis of SORT1 levels in human brain [114, 115].
Interestingly, a non-coding (nc) RNA mapping in antisense configuration to intron 1 of the SorL1 gene is upregulated in cerebral cortices from individuals with AD. This specific ncRNA, named 51A, triggers an alternative splicing shift of SorL1, resulting in reduced full-length SorL1, impaired processing of AβPP, and enhanced Aβ formation. In that work, the authors propose the triggering of 51A ncRNA overexpression as an upstream regulatory event in Aβ generation further strengthened by the demonstration of the upregulation of 51A in AD postmortem brain samples [116]. In this context, it should be noted that other four pol III-transcribed ncRNAs were discovered as regulators of proteins involved in pathways altered in AD thus suggesting a key role for pol III-transcribed ncRNAs in the disease [117–120].
As above described, SORT1 has multiple roles in protein transport, including in cancer cells. For example, SORT1 participates in breast tumor aggressiveness and its expression is downregulated in different type of cancers [121]. Curiously, SorL1 levels are apparently upregulated in some cancers [122]. In addition, SORT1 participates in complexes with p75-NTR (see below) in the apoptotic control of cancer cells, and regulates exosome biogenesis controlling tumor progression and angiogenesis [123].
Altogether, it is clear that the LDL receptor family, beside their neuronal function, are associated with the development of different forms of cancer and most of the adaptors previously described in AβPP/LDL-dependent signaling, such as FE65/Tip60, Dab1, X11, and others, are often pivotal in cancer development. It is therefore difficult to delineate the role of GS processing in cells “committed to proliferate” versus the post-mitotic setting of neuronal cells. Even when apparently similar interactions may occur among different GS substrates.
Notch receptors
Notch receptors (Notch 1–4, Uniprot P46531, Q04721, Q9UM47, and Q99466) are type I transmembrane receptors, pivotal components of the evolutionarily conserved signaling mechanism involved in cell fate specification and differentiation in various systems, especially during neuronal development. In the trans-Golgi network, Notch precursors are cleaved at Site-1 (S1) by a furin-like convertase. This cleavage is expected to stabilize the receptor cell surface expression and signaling. A second cleavage, ADAM-dependent or S2 cleavage (ADAM10 and ADAM17), occurs at the juxtamembrane extracellular domain close to the transmembrane domain. S2 fragments are further processed by GS at S3 site. Notch was the second GS substrate identified and all four Notch receptors are GS substrates [124, 125]. ADAMs metalloproteases and GS cleave Notch receptors releasing into the cytosol their intracellular domain (NICD), which is a well-known transcription factor for several genes, mostly involved in cellsurvival and differentiation [126]. Immunoprecipitation experiments showed that AβPP interacts with Notch 1, and that their transmembrane domains and short sequences near them are sufficient for this interaction [127]. Interestingly, AβPP-derived AICD functions as a negative regulator of Notch 1 signaling through the promotion of NICD degradation [128].
Notch, in the adult brain, is highly expressed in the hippocampus [129] and its action is related to the control of several functions, including neural stem cell maintenance [130]. In fact, it keeps the balance between neural stem cells and progenitors [131, 132] which are able to give rise to adult neurons, and this event is correlated to spatial memoryimprovement [133].
In AD patients, there is a reduction in neurogenesis, and GS, with its substrates, is central in the fine regulation of this process. In particular, the conditional knockout (KO) of PSEN1 in the forebrain reduces neurogenesis, while the conditional KO of Notch determines a depletion of the progenitor pool [134, 135]. An important role of Notch is related to the differentiation of glial progenitors into astrocytes, inhibiting the differentiation of progenitors into neurons [136]. On the contrary, GS inhibitors are apparently able to drive the differentiation of induced pluripotent stem cells (iPSCs) into neurons [137, 138]. Furthermore, Notch-dependent signaling is decreased with age, mainly in the hippocampus, and Notch levels are lower in mature neurons [139]. Abnormal Notch processing has been implicated in the development of a variety of cancers, with accumulating preclinical and clinical evidence delineating, in several solid tumors, a pro-oncogenic and pro-angiogenic role of Notch receptors, even in cancer stem cells, through their signaling [140, 141]. Therefore the development of GS inhibitors that specifically affect GS cleavage either in Notch (for cancer) or in AβPP (to treat AD), represents a relevant scientific endeavor of the last decade [142]. However, while clinical results in the AD field are to-date negative and disappointing (even for those molecules theoretically more specific in blocking AβPP cleavage than Notch processing) [143, 144], in cancer studies, along with failures and negative results, there are potential shots of light essentially for therapies combining GS inhibitors with other anticancer treatments [145, 146]. The major problems in AD trials is probably that GS inhibitors are not so selective as expected; the main complication being that, alongside AβPP, GS inhibitors would hamper four Notch receptors (and possibly other substrates) characterized by different levels of expression and localization, likely explaining the abundance of undesired effect. In any case, it is disappointing to observe that, to date, clinical data in AD experience with GS inhibitors or GS modulators, did not show reliefs or reduction of cognitive impairment in treatedpatient [144].
p75-neurotrophin receptor
p75-neurotrophin receptor (p75NTR, Uniprot P08138) is the founding member of the large tumor necrosis factor receptor superfamily and presents a death domain that is required for the regulation of many critical processes in the nervous system, ranging from apoptosis to synaptic plasticity. During the development of the sympathetic nervous system, p75NTR has dual opposite functions: the promotion of the neuronal survival with TrkA, in response to the nerve growth factor (NGF), and induction of cell death upon binding with pro or mature brain-derived neurotrophic factor (BDNF). In detail, p75NTR is involved in the neurotrophin-mediated death signaling in neurons [147] since, in sympathetic neurons, it is cleaved by GS generating p75NTR-ICD; particularly in response to pro-apoptotic ligands. This cleavage is necessary for the nuclear translocation of the neurotrophin receptor interacting factor (NRIF). Under physiological conditions, the production of p75NTR is enhanced during brain development, and it is gradually reduced in the adulthood. However, upon cellular stress and insults, as it possibly may occur in AD, p75NTR may be re-expressed [148]. Interestingly, hippocampal p75NTR levels are significantly upregulated in both AD brain samples and Tg mice [149].
p75NTR interacts with AβPP and C99, but not with C83, revealing that the α-secretase cleavage precludes this interaction: interestingly, the expression of p75NTR influence the processing of AβPP, leading to a reduction in sAβPPα production [150].
Aβ induced neuronal loss in the basal forebrain may require the direct binding to p75NTR, the level of which is highly expressed in this area [151]. p75NTR-/- mice display a significant increase in the size of basal forebrain cholinergic neurons, suggesting that p75NTR negatively regulates cholinergic neuronal phenotype, including cell size, target innervation, and neurotransmitter synthesis [152]; altogether with improved spatial memory and hippocampal long-term potentiation [153]. Tg 2576/p75NTR+/- mice (obtained by crossing AD Tg2576 mice, which overexpress an EOAD mutant form of AβPP, with p75NTR-/-) express lower p75NTR levels and do not develop deficits in learning, memory and hippocampal function, which are typical of Tg 2576 mice. This functional cognitive recovery may be a consequence of a reduction in Aβ accumulation through the inhibition of β-secretase and GS activities [154, 155].
In summary, the numerous interactions of p75NTR represent a biological cluster in several signaling platforms, with an apparent activity both in neurons and in cancer cells; even through a complex crosstalk with other receptors target of GS activity (see above SORT1 and p75NTR) [151, 157]. Also in this case, considering the plethora of possible interactions and downstream signaling, it is difficult to decipher the involvement of AβPP/GS contribution to thesesignals.
DISCUSSION
The activity of GS modulates the cleavages of a number of receptors which often interact each other, sharing signaling machinery through common adaptors and similar enzymatic cleavages. Most, if not all, the receptors here analyzed, besides being AβPP interactors, are processed in a very similar manner:in a first instance by metalloproteases (primarily by ADAM10/17), subsequently by GS. The latter cleavage usually triggers a biological cascade and a signaling activity, while sometimes (or rather, at the same time), blocks or hamper a biological function.
In this scenario, we must acknowledge the paucity of information about three main aspects relatively to the parallel processing of many GS’s substrates along with AβPP: 1) we have limited information about the co-distribution in specific tissues or cells of each single substrate with AβPP, in respect to GS activity; 2) we have few information about timing and stoichiometry of co-processing of specific substrates pairs/clusters cleaved by GS [158] together with AβPP; 3) we have a rather incomplete knowledge about the potential selectivity of GS subtypes [99, 159] on different substrates, including AβPP itself (see also Fig. 1). Therefore, it is very difficult to discriminate between “positive” or “negative” dynamics induced by GS cleavage: blocking the formation of NICD from Notch, and AICD from AβPP is always “positive”?; and to impede the formation of DCC-ICD or P75NTR-ICD is “negative”? Is it maybe “positive” in neurons and “negative” in cancer cells? We think that the issue is more complicated than it first appeared and, after all, it is quite surprising that many clinical trials have been settled trying to inhibit rather a-specifically GS, in AD and in cancer fields, in this context of uncertainty.
Intermembrane sequences and regions near them (from Uniprot) of GS substrates described in this review. Labelled black arrows indicate known GS cleavage sites.
AβPP, as most of its binding partners here described, is a pleiotropic receptor physiologically involved in cell motility and proliferation, even in cancer development [160] and GS, of course, can modulate these activities. This is particularly evident if we consider not only Notch (and clinical trials with GS inhibitors), but also DCC, the LDL family and even ApoE. In fact, recent data highlight the role of ApoE as metastatic blocker, and indicate LRP1 and endothelial LRP8 as modulators of angiogenesis and metastatic behavior in melanoma cells [161].
It is not surprising therefore that, among different unwanted effect that GS inhibitors treatment encountered in AD trials, we can enumerate also the abnormal occurrence of skin tumors and worsening of cognitive functions [26].
Understanding the GS’s “loss/gain of function” in AD development is a rather complicated issue in light of the ample network of imaginable interactions between AβPP and GS’s substrates. Does a loss of function on C99/C83 occurs equals on Alcα/DCC/LDL ... etc.? Or rather a loss of function on a substrate means an increment of the cleavage on another one? Or are there peculiar and definite clusters of receptors and GS, specifically, reserves a finely tuned cleavage for each single cluster in a tissue-dependent manner?
The good news is that exists the possibility that when we will understand the mechanics and stoichiometry of the receptor clusters processed by GS (and likely destroyed in course of AD), we will probably also comprehend which are the anomalous pathways GS-dependent in many tumors.
Unfortunately, a reliable cellular and animal model, in which recapitulate the “real” biological architecture and the sequence of neuronal events involving AβPP, GS (which isoform?) and specific signaling partners (which one?) deregulated in AD, is the scarlet pimpernel of AD research today.
Authors’ disclosures available online (http://j-alz.com/manuscript-disclosures/17-0628r1).
Footnotes
ACKNOWLEDGMENTS
This work was supported by the grant “Early diagnosis in Alzheimer’s disease to optimize pharmacological therapy and clinical assistance”- National Priority and National Priority Objectives 2013 – Molise Region to CR.