|Home | About | Journals | Submit | Contact Us | Français|
The Alzheimer’s disease (AD)-associated amyloid-β protein precursor (AβPP) is cleaved by α-, β-, and presenilin (PS)/γ-secretases through sequential regulated proteolysis. These proteolytic events control the generation of the pathogenic amyloid-β (Aβ) peptide, which excessively accumulates in the brains of individuals afflicted by AD. A growing number of additional proteins cleaved by PS/γ-secretase continue to be discovered. Similarly to AβPP, most of these proteins are type-I transmembrane proteins involved in vital signaling functions regulating cell fate, adhesion, migration, neurite outgrowth, or synaptogenesis. All the identified proteins share common structural features, which are typical for their proteolysis. The consequences of the PS/γ-secretase-mediated cleavage on the function of many of these proteins are largely unknown. Here, we review the current literature on the proteolytic processing mediated by the versatile PS/γ-secretase complex. We begin by discussing the steps of AβPP processing and PS/γ-secretase complex composition and localization, which give clues to how and where the processing of other PS/γ-secretase substrates may take place. Then we summarize the typical features of PS/γ-secretase-mediated protein processing. Finally, we recapitulate the current knowledge on the possible physiological function of PS/γ-secretase-mediated cleavage of specific substrate proteins.
Since the original discovery that amyloid-β protein precursor (AβPP) is a substrate for presenilin (PS)-dependent γ-secretase (PS/γ-secretase), at least 90 additional proteins have been found to undergo similar proteolysis by this enzyme complex [1, 2]. PS/γ-secretase is a multiprotein complex consisting of four essential components: presenilin 1 (PS1) or presenilin 2 (PS2), nicastrin, Aph-1 (anterior pharynx defective-1), and Pen-2 (presenilin enhancer-2) [3, 4]. These four components are required for the formation of a functional PS/γ-secretase complex, even though several other proteins have been shown to also associate with the complex [3, 5]. Together with PS/γ-secretase, site-2 protease (S2P), signal peptide peptidases (SPPs), and rhomboids constitute a novel family of intramembrane cleaving proteases, I-CLiPs . PS/γ-secretase typically cleaves type-I transmembrane proteins, S2P cleaves the sterol regulatory element binding protein in the Golgi membrane, and SPPs are GxGD-type aspartyl proteases cleaving type-II transmembrane proteins in membranes along the secretory pathway, endosomes, and lysosomes. Rhomboids regulate different processes, such as epidermal growth factor receptor signaling, mitochondrial dynamics, or apoptotic stimuli, even though the identities of their substrates are unknown. All the I-CLiPs enzymatically cleave their substrate proteins within the plane of a membrane in a process termed regulated intramembrane proteolysis (RIP) [6–8].
Amyloid-β (Aβ) peptides form plaques in the cortical brain areas of patients with Alzheimer’s disease (AD). The proteolytic processing steps involved in the cleavage of AβPP and generation of Aβ have been characterized (Fig. 1). AβPP first undergoes ectodomain shedding at two alternative sites: one cleavage is mediated by α-secretase activity of ADAM (a disintegrin and metalloproteinase) family metalloproteases, and the other by β-secretase or BACE1 (β-site AβPP cleaving enzyme 1). These cleavages result in the release of secreted AβPP ectodomains (sAβPP) from the membrane and formation of two major membrane-associated C-terminal fragments (CTFs) termed C83 and C99. The subsequent PS/γ-secretase-mediated cleavage of C83 or C99 at the ε-site releases the AβPP intracellular domain (AICD). PS/γ-secretase also cleaves at the γ-site(s), releasing either a small soluble p3 peptide in the case of C83, or alternatively, Aβ peptides of different lengths after cleavage of C99 .
Over the past decade, a growing number of additional proteins have been identified as substrates for the three secretases. Similarly to AβPP, the ectodomain shedding is followed by RIP, which releases the soluble C-terminal domains of the proteins from the membrane . It is intriguing that these PS/γ-secretase substrate proteins play a vast variety of functions in both developing and adult tissues. Therefore, PS/γ-secretase-mediated cleavage may regulate a large range of cellular events. Despite the diversity of the thus far characterized PS/γ-secretase substrates in their structure and physiological functions, the majority of these proteins share several common features (see below) [2, 10, 11].
The wide variety of cellular processes and signaling events regulated by PS/γ-secretase via the cleavage of its multiple substrate proteins creates challenges for the development of therapeutic γ-secretase inhibitors for AD. Thus, it will be of utmost importance to design inhibitors that would specifically target only Aβ generation and not interfere with normal processing of other PS/γ-secretase substrates. Moreover, specific targeting of PS/γ-secretase, and not the other I-CLiPs, will be essential in drug development for AD. However, this may be complicated by the fact that the substrates of the other I-CLiPs remain thus far largely unidentified. In this review, we focus on PS/γ-secretase substrates and the functional consequences of their PS/γ-secretase-mediated cleavage. The currently characterized PS/γ-secretase substrates and the suggested role of their PS/γ-secretase-mediated cleavage are summarized in Tables 1 and and22.
AβPP is a type-I transmembrane protein suggested to play a role in cell adhesion, protein transport, synapse formation, neurite extension, and neuroprotection [12–18]. Understanding the sequential proteolytic processing of AβPP by secretases is of high importance given that one of its products is the pathogenic Aβ peptide. These peptides excessively accumulate in the brains of individuals affected by AD and majorly contribute to dysfunction and degeneration of neurons that result in progressive decline of cognitive functions. The proteolytic processing of AβPP is well characterized, but factors regulating its proteolysis by the different secretases are a subject of intense study. AβPP can undergo proteolysis by two alternative routes. The non-amyloidogenic pathway is more prevalent and excludes Aβ production, whereas activation of the amyloidogenic route generates the detrimental Aβ peptides (Fig. 1).
The non-amyloidogenic proteolytic pathway takes place at or near the cell surface and is initiated by α-secretases such as the ADAM-family metalloproteases ADAM10 or ADAM17 (or TACE; tumor necrosis factor-α converting enzyme) [19, 20]. This first cleavage step releases sAβPPα, the large soluble AβPP ectodomain, with a possible neuroprotective function . It also produces the membrane-anchored AβPP-CTF, C83. C83 is further cleaved by PS/γ-secretase at the AβPP ε-site to release AICD, and at the γ-site to generate a small p3 peptide [1, 22–26] (Fig. 1). Whether p3 plays a physiological function is thus far unknown. AICD assembles in a complex with the adaptor proteins Fe65 and Tip60, translocates to the nucleus, and regulates the transcription of genes such as KAI1, glycogen synthase kinase-3β, neprilysin, p53, and AβPP, and apoptosis [24–32]. However, AICD-mediated transcriptional regulation has been challenged by recent reports. Using pharmacological inhibition of PS/γ-secretase and PS/γ-secretase- or AβPP-deficient models, these studies have shown that AICD may not be directly required for the transcriptional control of its suggested target genes and that its interaction with Fe65 may not lead to nuclear translocation [33, 34]. Moreover, the expression of AβPP N-terminal domains alone rescued the physiological and learning deficits in AβPP knock-out mice, whereas the AβPP C-terminus was not required . This further suggests that AICD-mediated gene transcription may not be physiologically essential, even though this matter remains controversial. Furthermore, it was recently suggested that AICD may account for some of the deleterious effects previously associated with Aβ as nuclear localization of AICD is perhaps enhanced during the amyloidogenic processing of AβPP [36, 37].
The amyloidogenic pathway is initiated by β-secretase or BACE1, cleaving AβPP N-terminally from the α-secretase cleavage site [38–41]. This clip may take place at or near the cell surface or after AβPP internalization along the endocytic pathway. BACE1-mediated cleavage releases the soluble sAβPPβ and produces the membrane-bound C99 AβPP-CTF. Subsequent PS/γ-secretase-mediated cleavage of C99 at the γ-site(s) produces Aγ peptides of different lengths along with AICD released by cleavage at the ε-site (Fig. 1).
Aβ levels in the brain are regulated by its rate of generation, but also by its clearance or degradation by different Aβ-degrading enzymes, such as neprilysin and insulin-degrading enzyme (IDE). Their function has been recently reviewed elsewhere [42, 43]. An increase in BACE1 expression in specific AD patients results in elevated BACE1-mediated cleavage of AβPP . Familial AD-associated (FAD) mutations in the PSEN genes encoding for PS1 and PS2 enhance the generation of the more fibrillogenic 42-amino acid long Aβ (Aβ42), therefore increasing the ratio of Aβ42 to the shorter, more soluble Aβ40 [22, 45]. These mutations also affect the exact cleavage site in the Aβ-region of AβPP, determining the C-terminus and length of the Aβ-peptide generated. Complete loss of PS function or lack of PS/γ-secretase activity prevents production of Aβ and results in increased levels of AβPP-CTFs [46–48].
PS/γ-secretase is a high molecular weight enzymatic complex, whose estimated size has been reported to vary between 250 and 2000 kDa. The components of the active γ-secretase complex were revealed by a genetic screen in Caenorhabditis elegans (C. elegans) and search for PS-binding proteins in mammalian cells. Three proteins have been identified, in addition to PS: Aph-1, a multipass transmembrane protein essential for C. elegans embryogenesis; the type-I transmembrane protein nicastrin; and Pen-2, a hairpin-structured membrane protein enhancing sel-12/PS function in C. elegans (Fig. 2). Aph-1 and Pen-2 were later found to be required for Notch signaling, PS1 accumulation, and γ-secretase activity [49–56]. Reconstitution of an active PS/γ-secretase complex in Saccharomyces cerevisiae yeast, which does not endogenously express any of the four complex components, elegantly confirmed that PS, nicastrin, Aph-1, and Pen-2 were necessary and sufficient for creating a functional PS/γ-secretase complex .
PS contains the catalytic site of the PS/γ-secretase complex. This site consists of the critical aspartate residues at amino acid positions 257 and 385 in the 6th and 7th transmembrane domains, respectively [57, 58]. Nicastrin was initially suggested to function as a receptor using its Glu-333 to bind short N-termini of substrates generated by metalloprotease-mediated ectodomain shedding. As a second step, nicastrin was supposed to recruit the substrate proteins for PS/γ-secretase-mediated cleavage . However, a more recent report indicated that mutagenesis of Glu-332 in the mouse nicastrin ectodomain (corresponding to human nicastrin Glu-333) affected PS/γ-secretase complex maturation, but did not impede substrate recognition or PS/γ-secretase activity. This refuted the idea of nicastrin functioning as a specific receptor for PS/γ-secretase substrates . Furthermore, the PS1/Pen-2/Aph-1a trimeric complex lacking nicastrin can function as an active PS/γ-secretase with a substantially reduced activity and stability . Recent data have implied that nicastrin directly participates in PS/γ-secretase activity through its Glu-333 . Thus, nicastrin appears to stabilize PS/γ-secretase and may participate in its activity, but may not be indispensable for substrate recognition.
Increasing evidence implies that assembly of the active PS/γ-secretase complex takes place in a chronologic order. First, Aph-1 and nicastrin would form an intermediate complex. Aph-1 and Pen-2, in turn, have been suggested to function as chaperones assisting in PS/γ-secretase complex assembly, trafficking, and PS endoproteolysis . In addition, Aph-1 was recently suggested to be involved in PS/γ-secretase activity and to directly associate with PS/γ-secretase substrates [63, 64]. Following the intermediate Aph-1 and nicastrin complex formation, PS1 holoprotein via its C-terminus binds to the transmembrane domain of nicastrin. Finally, Pen-2 binds to the PS transmembrane domain inducing simultaneous PS1 endoproteolysis . Levels of any of the complex components affect the levels and stability of the other components, indicating that assembly of the active PS/γ-secretase complex is tightly regulated in an interdependent manner. Moreover, the four core components have been shown to exist in the complex in 1 : 1:1 : 1 stoichiometry [3, 66, 67].
The PS/γ-secretase complex composition and function are further complicated by the existence of different isoforms of Aph-1 and PS. In humans, two Aph-1 genes, APH-1a and APH-1b have been reported. Of these, the former one is also alternatively spliced to generate two isoforms . In addition, PS has two forms, namely PS1 and PS2, encoded by two individual genes, PSEN1 and PSEN2, respectively [69, 70]. Different combinations of these proteins may create from four to six alternative PS/γ-secretase complexes with possibly diverse activities or subcellular localizations. PS/γ-secretase complexes containing Aph-1b were recently shown to contribute to the generation of Aβ in human brain, but not to be essential for Notch cleavage . In addition, it has been suggested that PS1-containing γ-secretase complexes have a significantly higher specific activity than PS2-containing ones [71, 72]. Indeed, PS1 knock-out mice show severe developmental abnormalities and embryonic lethality, while PS2-deficient mice are viable, have a milder phenotype, and display unaltered AβPP processing [73, 74]. Interestingly though, two recent reports suggest that there are PS/γ-secretase functions that are specifically dependent on PS2, but not PS1. A novel PS/γ-secretase substrate, EpCAM, was shown as cleaved specifically by PS2-dependent γ-secretase activity . Furthermore, PS2-dependent γ-secretase activity released AICD regulating aquaporin 1 expression, while PS1-dependent activity was dispensable for this function .
Even though PS, nicastrin, Aph-1, and Pen-2 suffice to create the functional, fully active PS/γ-secretase complex, several other proteins have also been suggested to associate with the complex. The transmembrane glycoprotein CD147 was reported to associate with the PS/γ-secretase complex in several different cell lines. CD147 decreased Aβ production and was proven dispensable for the PS/γ-secretase activity . Another protein suggested to associate with and modulate PS/γ-secretase function is TMP21, a member of the p24 cargo protein family. TMP21 selectively modulated γ-, but not ε-cleavage of AβPP and suppressed Aβ generation . Additional binding partners keep emerging, some of which affect Aβ production . However, characterization of the PS/γ-secretase complex-associating proteins is hampered by artificial interactions during the purification process. Also, association of the interacting proteins with PS/γ-secretase complex in vivo lacks compelling evidence. It is possible that these proteins do not function as essential core components for the PS/γ-secretase activity. Instead, they may act as modulators of the PS/γ-secretase complex assembly, activity, or substrate selectivity, perhaps depending on the tissue, cell type, or subcellular compartment. The presence of loosely interacting proteins may explain the variations in the reported size of the PS/γ-secretase complex.
PS/γ-secretase substrate proteins first undergo ectodomain shedding at their N-termini. As seen above, this cleavage is mediated by either α- or β-secretase. The α-secretase-mediated ectodomain shedding is thought to take place mostly at the cell surface, even though there is evidence for α-secretase-mediated cleavage of AβPP in the trans-Golgi network (TGN) [80–83]. β-cleavage is suggested to occur in the Golgi apparatus or TGN, at the plasma membrane, and in endocytic compartments [40, 84–88]. PS/γ-secretase complex components localize to different cellular compartments, such as plasma membrane, early and late endosomes, autophagic vacuoles, lysosomes, mitochondria, Golgi, and endoplasmic reticulum (ER), creating a number of possible cellular locations for Aβ generation [89–95]. Blockade of endocytosis markedly decreases Aβ generation, implying that endocytosis of AβPP is critical for its processing . Interestingly, altered neuronal endocytosis is one of the earliest neuropathological changes occurring in AD [97, 98]. In the amyloidogenic pathway, AβPP is most likely endocytosed to allow for β-cleavage at the optimal acidic pH for BACE1 function in the endocytic compartments [40, 85, 88, 99]. AβPP-CTFs may then recycle back to the plasma membrane for PS/γ-secretase cleavage or alternatively, this cleavage step may also take place in the endocytic compartments [93, 100, 101]. We have shown evidence that PS/γ-secretase activity regulates AβPP clearance specifically from the endocytic recycling compartment . In summary, the current understanding denotes that amyloidogenic processing of AβPP mainly occurs in the TGN, plasma membrane, and endocytic compartments as AβPP traffics through the secretory and endocytic pathways. AβPP reaching the plasma membrane, however, is preferentially cleaved by α-secretase activity and undergoes non-amyloidogenic processing.
Little is known about the localization of PS/γ-secretase-mediated cleavage of substrate proteins other than AβPP. Given that ADAM family metalloproteases are generally responsible for ectodomain cleavage of the additional PS/γ-secretase substrates, their α-cleavage most likely takes place at the cell surface. The remaining CTFs may then be processed by PS/γ-secretase on the plasma membrane or alternatively after endocytosis [101–103]. Surface expression of many substrate proteins is decreased after ectodomain shedding and their CTFs are found in intracellular compartments, suggesting that endocytosis occurred prior to PS/γ-secretase-mediated cleavage. Indeed, Notch cleavage by PS/γ-secretase requires monoubiquitination of a lysine present in the transmembrane domain at the PS/γ-secretase recognition site and endocytosis . Nevertheless, further studies are warranted to fully understand the role of subcellular compartments in PS/γ-secretase-mediated cleavage of its different substrate proteins.
Even though PS/γ-secretase substrates are diverse in their structure, localization, and physiological functions, the majority of these proteins share several common features [2, 10, 11] (Table 3). First, most substrate proteins are type-I transmembrane proteins. Typically, they harbor a large ectodomain often containing cell adhesion molecule-like domains, a single-pass transmembrane domain, and a cytoplasmic C-terminus frequently capable of initiating or mediating intracellular signaling. Second, the characterized proteins appear to function as signaling proteins and regulate cellular events such as cell fate determination, adhesion, migration, neurite outgrowth, axon guidance, or formation and maintenance of synapses. Interestingly, many of these same events are disrupted during neurodegeneration in AD [105, 106]. Third, PS/γ-secretase preferentially cleaves membrane-bound protein stubs after ectodomain shedding of the full-length substrate . α-secretase mediated ectodomain shedding can be constitutive, but it may also be induced by several stimuli, such as ligand binding, protein kinase C (PKC) activation by phorbol esters, or Ca2+ influx. Some of the currently characterized PS/γ-secretase substrate proteins, such as lipoprotein-related protein-1 (LRP) or voltage-gated sodium channel β2 subunit (Navβ2), have been shown to undergo an alternative N-terminal cleavage by BACE1 in a similar manner to AβPP [107, 108]. Fourth, the PS/γ-secretase-mediated γ-like cleavage (corresponding to the ε-cleavage in AβPP, which releases AICD) takes place at or near the boundary of the transmembrane and cytoplasmic domains. The ε-like cleavage site flanks a stretch of hydrophobic amino acid sequence rich in lysine and/or arginine residues. It appears that PS/γ-secretase cleavage is not dependent on a specific amino acid target sequence at or adjacent to the cleavage site, but rather perhaps on the conformational state of the transmembrane domain . Fifth, PS/γ-secretase-mediated cleavage releases the intracellular domain (ICD) of each substrate protein to the cytosol. Several substrate ICDs translocate to the nucleus and have been suggested to play a role in transcriptional regulation in a similar manner to AICD. In addition to the released ICDs, some substrate proteins, such as alcadeins or Notch, release Aβ-like peptides as a consequence of PS/γ-secretase cleavage, even though it is not clear whether these peptides exhibit any pathogenic functions [109, 110]. Sixth, generation of the ICDs is prevented upon inhibition of PS/γ-secretase activity by pharmacological γ-secretase inhibitors or lack of functional PSs. Concomitantly, γ-secretase inhibition results in CTF accumulation in an analogous manner to the AβPP-CTFs. Seventh, the ICDs are often labile and accumulate in the presence of proteasomal inhibitors, suggesting that the proteasome at least partially contributes to their degradation. Together with the proteasome, PS/γ-secretase may therefore function as a “proteasome of the membrane” that removes the C-terminal stubs generated after ectodomain shedding. This function may also implicate a role for PS/γ-secretase as a terminator of signaling for particular substrate proteins . On the other hand, some substrate ICDs, such as those of amyloid β-protein precursor-like proteins (APLP1 and APLP2), have been shown to undergo degradation by IDE in a similar manner to AICD .
To date, 91 proteins, incuding AβPP, have been identified as PS/γ-secretase substrates [112, 113]. The majority of these proteins are type-I transmembrane proteins, such as Notch (Table 1) [1, 114]. In addition, one type-II membrane protein, β1,6-acetylglucosaminyltranferease (GnT-V), residing in the Golgi apparatus  and one polytopic transmembrane protein, the glutamate receptor subunit 3 , have also been reported to undergo PS/γ-secretase-dependent proteolysis (Table 2). However, understanding the functional significance and regulation of PS/γ-secretase-mediated cleavage of these atypical substrate proteins requires further studies. In terms of function, it appears that some of the ICDs released by PS/γ-secretase-mediated cleavage perform independent signaling functions apart from the function of the full-length form of the protein, such as regulation of gene transcription. On the other hand, generation of some substrate ICDs may operate as means to terminate or antagonize full-length substrate protein signaling, such as in cell adhesion or neurite outgrowth.
The thus far characterized PS/γ-secretase substrates and the suggested functional effects of their PS/γ-secretase mediated cleavage are listed in Tables 1 and and22 and discussed in some detail below.
The intracellular domain of AβPP forms a complex with adaptor proteins Fe65 and Tip60. Many studies have suggested that upon PS/γ-secretase cleavage this complex translocates to the nucleus and regulates gene transcription [24–32]. However, recent research has contradicted these results and suggested that AICD may not be involved in transcriptional control and that it may not translocate to the nucleus, but rather it remains associated with membranes or free in the cytoplasm [33, 34]. Moreover, studies in knock-in mice suggested that secreted sAβPPα, but not the AβPP C-terminus, was required to mediate the physiological effects of AβPP . Therefore, the relevance of AICD-mediated gene transcription still remains controversial. In spite of this, APLP1 and APLP2 have also been implicated in Fe65-dependent transcriptional regulation [29, 111, 117]. Together, it still appears plausible that the full-length forms of all AβPP family members may restrain Fe65 from entering the nucleus, while their ICDs interact with Fe65 and enter the nucleus to regulate transcription [25, 117, 118].
Several additional PS/γ-secretase substrate ICDs have also been implicated in transcriptional regulation, often suggested by their nuclear localization and ability to affect transcription in reporter gene assays. In some cases, the substrate ICDs have been shown to control transcription of endogenous target genes by associating with different transcription factors. Among others, specific ICDs that have been reported to affect gene transcription include the following: the ICDs of alcadeins ; CD44 ; DCC (deleted in colorectal cancer) ; Notch [121–125] and its ligands Delta and Jagged [126–129]; E- and N-cadherin [130–133]; and receptor-like protein tyrosine phoshatases (RPTP) RPTPκ  and LAR (leukocyte-common antigen related) . The confirmed or putative function of these ICDs in transcriptional regulation constitutes an independent signaling function, which may be different from the full-length forms of the proteins. Furthermore, ICD generation may serve as a regulatory switch in the signaling of these particular substrates. For example, the full-length forms of cadherins or RPTPs are intimately involved in cell adhesion, whereas their respective ICDs have been shown to translocate to the nucleus and regulate transcription [130–135].
Notch family proteins mediate short-range signaling controlling cell-fate decisions, patterning of gene expression, and maintenance of stem cell populations during development. Notch signaling in these events requires Notch activation and subsequent PS/γ-secretase-mediated release of Notch-ICD (NICD), which regulates gene transcription [125, 136].
In addition to Notch, some other substrates have been implicated in the regulation of cell fate as well. ErbB4 cleavage by PS/γ-secretase results in the release of ErbB4-ICD (E4ICD) . E4ICD has been suggested to regulate gene transcription through several transcription factors and to confer pro-apoptotic activity of ErbB4 [138–142]. In addition, ErbB4 proteolysis is involved in cell fate determination in the brain. Inhibition of PS/γ-secretase was shown to block E4ICD nuclear translocation and subsequent expression of proteins associated with myelination. One report indicated that maturation of oligodendrocytes, but not neurons or astrocytes, was thereby prevented and oligodendrocytes maintained their juvenile state . Sardi and colleagues  have shown that PS/γ-secretase-dependent nuclear signaling of ErbB4 regulates the timing of astrogenesis in vivo in the developing mouse brain. Their data suggested that PS/γ-secretase-mediated ErbB4 cleavage is important for maintaining the neuronal precursor cells in a neuro-genic state by blocking astrocytic differentiation .
Another event involving PS/γ-secretase-mediated cleavage of several substrate proteins is regulation of cell death signaling. LRP1 has been previously shown to facilitate AβPP endocytosis and enhance Aβ production and its cellular uptake . On the other hand, PS/γ-secretase-mediated processing of LRP1 is suggested to mediate ischemic cell death . Middle cerebral artery occlusion (MCAO) in mice was shown to induce activation of PS/γ-secretase in the ischemic hemisphere and nuclear translocation of LRP1-ICD in neurons within the ischemic penumbra. This was precluded by intracortical injection of a γ-secretase inhibitor. Concomitantly, PS/γ-secretase inhibition significantly decreased MCAO-induced caspase-3 activation, diminished the volume of the ischemic lesion by 50%, and improved the neurological outcome of the mice. This led the authors to propose that PS/γ-secretase-mediated proteolysis of LRP1 potentiates cell death during cerebral ischemia .
Neurotrophin signaling is usually considered to promote cell survival. However, p75 neurotrophin receptor (p75NTR), another PS/γ-secretase substrate, has been widely implicated in neurotrophin-mediated death signaling in neurons . It was shown that pro and mature forms of brain-derived neurotrophic factor induce PS/γ-secretase-mediated generation of p75NTR-ICD and apoptosis by causing nuclear translocation of neurotrophin receptor interacting factor (NRIF), a DNA-binding protein essential for p75NTR-mediated apoptosis. p75NTR proteolysis and NRIF nuclear translocation were also shown to take place in vivo during naturally occurring cell death of superior cervical ganglion neurons . Moreover, Aβ has been shown to induce p75NTR-mediated cell death signaling [149, 150]. An early event in AD is the loss of basal forebrain cholinergic neurons, which express high levels of p75NTR. The levels of p75NTR-CTF, p75NTR-ICD, and pro-nerve growth factor (pro-NGF), another pro-neurotrophin capable of inducing p75NTR-mediated death signaling, are increased in AD brain [151–156]. Furthermore, pro-NGF isolated from AD brain, but not from control brain, was shown to induce apoptosis in different cells in a PS/γ-secretase-dependent manner . Together, these findings strongly point to a central role for p75NTR and its proteolysis in neuronal death.
A recently identified PS/γ-secretase substrate protein is the receptor for advanced glycation end products (RAGE). RAGE is a cell surface receptor shown to mediate Aβ transport to the brain across the blood brain barrier and potentiate Aβ-induced disturbancies in neuronal function in transgenic mice [157–159]. Overexpression of RAGE-ICD (RICD) was shown to decrease cell viability, increase the number of apoptotic cells, and activate the stress-related kinases, p38 and JNK (c-Jun N-terminal kinase), implying that RICD may be involved in apoptosis . Interestingly, all PS/γ-secretase substrate proteins involved in cell death signaling appear to interact with Aβ, suggesting that PS/γ-secretase-mediated cleavage may be an important factor controlling their interaction with Aβ and mediating subsequent adverse effects.
Neurite outgrowth, axonal pathfinding, and cell adhesion are key events during development. These same events may also take place in the adult brain during neuroplasticity when new connections are established between neurons, for example during learning or in response to injury. Large body of evidence suggests that PS/γ-secretase plays a fundamental role in controlling these events by proteolytic cleavage of its substrate proteins.
DCC (deleted in colorectal cancer) functions as a cell surface receptor for netrin-1, a protein regulating cell and axonal migration during development . While full-length DCC was observed to extensively induce neurite outgrowth in mouse neuroblastoma cells, DCC-ICD, released by PS/γ-secretase activity, inhibited neurite outgrowth and associated with down-regulation of DCC-mediated intracellular signaling .
Eph receptors and their ephrin ligands are important players in synapse formation and plasticity in the central nervous system. Recent data have demonstrated that Eph receptors and ephrin signaling are regulated by PS/γ-secretase-mediated cleavage. Similarly to DCC, PS/γ-secretase-mediated processing of ephrinB1 prevented outgrowth of actin-enriched cellular protrusions in COS cells . EphB-receptor has been shown to induce proteolytic processing of another EphB ligand, ephrinB2. EphrinB2-ICD, generated as a result of ephrinB2 processing by PS/γ-secretase, was found to activate Src-mediated endothelial cell sprouting. Interestingly, ephrinB2-ICD generation was demonstrated to inhibit PS/γ-secretase-mediated processing of its own precursor, ephrinB2. This finding suggests that ephrinB2-ICD generation functions as an inhibitory feedback mechanism to limit Eph-ephrinB2 signaling . Recently, PS/γ-secretase was also shown to have an integral role in the formation of cell-matrix focal adhesions by transcriptional regulation of c-Src via ephrinB2 cleavage . Furthermore, termination of the ephrinB2 receptor-EphB2 signaling has been proposed to convert the initial cell-cell adhesion to cell dissociation . Finally, activation of EphA4 receptor by ephrin normally results in the retraction of dendritic spines. However, when PS/γ-secretase in hippocampal synapses released EphA4-ICD, this led to enhanced spine formation through activation of the Rac signaling pathway. Moreover, synaptic activity enhanced EphA4 cleavage by PS/γ-secretase, indicating that processing of EphA4 plays a role in the synaptic activity-dependent morphogenesis of dendritic spines . These reports collectively imply that the bidirectional ephrin-Eph receptor signaling is regulated by PS/γ-secretase activity and as a result, the initial signaling of the full-length proteins is converted to opposite signaling or even terminated after receptor proteolysis to generate the cognate ICDs.
PS/γ-secretase-mediated proteolytic processing of several substrates also plays a role in the regulation of cell adhesion. PS/γ-secretase-mediated cleavage of E-cadherin has been shown to result in the disassembly of adherens junctions in epithelial cells by inducing disruption of the E-cadherin-β-catenin adhesion complex [130, 168]. On the contrary, our data suggested that in the case of nectin-1α, PS/γ-secretase inhibition results in the disassembly of adherens junctions and release of β-catenin to the cytoplasmic pool . Whether PS/γ-secretase functions to stabilize or disassemble the junctions appears to depend on the substrate protein to be cleaved. We also found that whereas the full-length voltage-gated sodium channel β2 subunit (Navβ2) effectively induced cell adhesion and migration, γ-secretase inhibition completely blocked these events . These results support the notion that PS/γ-secretase plays a role in the regulation of cell adhesion.
PS/γ-secretase activity is also implicated in the regulation of angiogenesis and tumorigenesis. Vascular endothelial growth factor receptor-1 (VEGFR-1) regulates angiogenesis stimulated by its ligand VEGF. Recently, VEGFR-1 was shown to undergo RIP by PS/γ-secretase, which resulted in the inhibition of angiogenesis . Pigment epithelium-derived factor (PEDF), a potent inhibitor of formation of new vessels from endothelial cells, was found to enhance PS/γ-secretase-mediated cleavage of VEGFR-1. Furthermore, PEDF treatment of endothelial cells significantly increased PS/γ-secretase activity, PS1 endoproteolysis, and PS1 translocation to the cell membrane. Whereas full-length VEGFR-1 localized in the cytoskeletal and nuclear fractions, PEDF and VEGF induced the release of VEGFR-1-ICD to the cytoplasm . Inhibition of PS/γ-secretase activity completely abolished the PEDF-mediated inhibitory effect on VEGF-induced angiogenesis and prevented VEGFR-1-ICD release to the cytosol . These results suggest that PEDF functions as an inhibitor of VEGF-induced angiogenesis by inducing PS/γ-secretase-mediated proteolysis of VEGFR-1.
PS/γ-secretase has also been suggested to participate in tumorigenesis through the release of CD44-ICD, which may promote cell transformation . As already discussed above, ErbB4-ICD (E4ICD) released by PS/γ-secretase activity may regulate gene transcription through several transcription factors, but also confer pro-apoptotic activity of ErbB4 [138–142]. On the other hand, E4ICD (but not the membrane-associated forms of ErbB4) has been shown to increase the levels of the anti-apoptotic protein p53 and its transcriptional target p21 in response to neuregulin . These events may be important in tumorigenesis, as nuclear ErbB4 has been reported in several human tumors and it predicts poor diagnosis in breast cancer .
Recent evidence indicates that PS/γ-secretase not only regulates Navβ2-mediated cell adhesion, but also voltage-gated sodium channel levels and activity . We have found that BACE1 and PS/γ-secretase cleavages of Navβ2 release a Navβ2-ICD, which modulates the levels of the pore-forming Nav1.1 α-subunit intra-cellularly and on the cell surface . Since four auxiliary subunits of the voltage-gated potassium channel are also cleaved by BACE1 and PS/γ-secretase , it appears that these two enzymes are involved in the regulation of ion channel function. Interestingly, voltage-gated sodium and potassium channels together are the key components of action potential generation and propagation.
Yet another cellular function involving PS/γ-secretase mediated cleavage is pigmentation. Tyrosinase (and its related proteins Tyrp1 and Tyrp2) produces eumelanin in melanosomes (endosomal/lysosomal-related organelles) of vertebrate pigment cells during melanin synthesis . Tyrosinase is transported in vesicles from TGN to melanosomes and its correct processing and trafficking is essential for pigment synthesis . However, impaired PS/γ-secretase function in mice resulted in defective pigmentation of the eyes and subcellular mislocalization of tyrosinase. Accordingly, treatment of mouse primary melanocytes with γ-secretase inhibitors blocked pigmentation . Together, these data suggest that PS/γ-secretase is required for the correct targeting of tyrosinase to melanocytes to catalyze the production of melanin .
To date, 91 PS/γ-secretase substrate proteins have been identified. Whether processing of any of these proteins results in the generation of toxic or pathogenic cleavage products similar to Aβ is thus far unknown. However, it is apparent that PS/γ-secretase-mediated processing affects the normal biological signaling of its substrate proteins. Therefore, an imbalance or potential impairment of PS/γ-secretase function might result in aberrant signaling and contribute to neurodegeneration during the course of AD pathogenesis. Presently, two different γ-secretase inhibitors are in phase I and two others in phase II and III clinical trials (http://www.alzforum.org; http://www.clinicaltrials.gov; ). However, the latest drawback for the potential γ-secretase inhibitor therapy in AD occurred recently when the Semagacestat (LY450139) γ-secretase inhibitor program by Eli Lilly and Company was halted in Phase III trials. The drug did not slow the cognitive decline in patients with mild to moderate AD, but in fact made them worse. Most of the commonly used experimental γ-secretase inhibitors block PS/γ-secretase-mediated processing of the substrate proteins in general (even though some substrate specificity or preference may take place) and augment substrate CTF levels while preventing ICD release . Consequently, PS/γ-secretase inhibition may distort substrate signaling. For these reasons, wide-range γ-secretase inhibitors may be limited in their therapeutic potential. This has become evident in drug trials, as many of the γ-secretase inhibitors have caused side effects. Therefore, development of specific and safe γ-secretase inhibitors for clinical use faces great challenges. Encouraging evidence shows that substrate-specific γ-secretase inhibitors or modulators can inhibit Aβ production without affecting Notch cleavage [180–182]. In addition, partial inhibition of PS/γ-secretase activity might provide sufficient therapeutic outcome and be considered as a form of therapy targeting PS/γ-secretase . Nonetheless, even though PS/γ secretase is an appealing drug target, more preclinical and clinical studies are required before therapeutic γ-secretase inhibitors or modulators reach the market.
This work is supported by the Health Research Council of the Academy of Finland (AH) and two grants from the NIH/NIA (DMK).
Authors’ disclosures available online (http://www.jalz.com/disclosures/view.php?id=731).