J. Neurochem. (2010) 115, 1299–1308.
In recent decades, the study of the amyloid precursor protein (APP) and of its proteolytic products carboxy terminal fragment (CTF), APP intracellular C-terminal domain (AICD) and amyloid beta has been mostly focussed on the role of APP as a producer of the toxic amyloid beta peptide. Here, we reconsider the role of APP suggesting, in a provocative way, the protein as a central player in a putative signalling pathway. We highlight the presence in the cytosolic tail of APP of the YENPTY motif which is typical of tyrosine kinase receptors, the phosphorylation of the tyrosine, serine and threonine residues, the kinases involved and the interaction with intracellular adaptor proteins. In particular, we examine the interaction with Shc and Grb2 regulators, which through the activation of Ras proteins elicit downstream signalling events such as the MAPK pathway. The review also addresses the interaction of APP, CTFs and AICD with other adaptor proteins and in particular with Fe65 for nuclear transcriptional activity and the importance of phosphorylation for sorting the secretases involved in the amyloidogenic or non-amyloidogenic pathways. We provide a novel perspective on Alzheimer’s disease pathogenesis, focussing on the perturbation of the physiological activities of APP-CTFs and AICD as an alternative perspective from that which normally focuses on the accumulation of neurotoxic proteolytic fragments.
APP intracellular C-terminal domain
protein kinase B
amyloid precursor protein
carboxy terminal fragment
extracellular signal regulated kinase
growth factor receptor bound
glycogen synthase-kinase 3
C-Jun-N-terminal kinase-interacting protein 1
jun N-terminal kinase
lipoprotein receptor-related protein
Src and collagen homologue
tyrosine kinase receptor
Alzheimer’s disease (AD) is a neurodegenerative disorder with progressive decline of cognitive functions that lacks an effective therapy. The neuropathology is characterised by cortical neurodegeneration, reactive gliosis and dystrophic neurons with neurofibrillary tangles and deposition of amyloid plaques. Amyloid beta (Aβ) is the main component of amyloid plaques. This 40–42 aa peptide is derived from the proteolytic processing of the amyloid precursor protein (APP), a type 1 transmembrane protein whose physiological role is yet to be fully defined. The different isoforms of APP are characterized by an extracellular domain, a transmembranous segment and a cytoplasmic domain. There are many proteolytic enzymes capable of cutting APP; the most relevant are α secretase (ADAM10, TNFalpha converting enzyme or TACE) (Buxbaum et al. 1998; Allinson et al. 2003), β secretase (BACE1, 2) (reviewed in Vassar 2004) and γ secretase, a protein complex composed of Presenilin 1–2 (PS1–2), nicastrin, Anterior Pharinx – Defective 1 (APH1) and Presenilin enhancer 2 (Pen2) (De Strooper 2003). It is generally recognized that Aβ results from the sequential cleavage by β and γ secretases, and is a major factor responsible for the development of AD (Selkoe 2002). The proteolytic cleavage of APP by the α, β and γ secretases generates a soluble ectodomain (sAPP), carboxy terminal fragments (CTFs) and an APP intracellular C-terminal domain (AICD) (Fig. 1). β secretase cleaves APP at position 1 or 11 of the Aβ sequence, α secretase cleaves at position 17, while the γ secretase cleavage site is intramembranous, at position 40–42. Beta secretase cleavage generates β-CTFs fragments, named C99/100 and C89, α cleavage generates the fragment C83 and γ cleavage gives rise to fragments C59-57, also called AICD. All the CTF fragments are named according to their aa lengths. Beta and α-CTF can be cleaved intramembranously by γ secretase to generate Aβ peptides of 40–42 aa or the shorter non-amyloidogenic P3 fragment, respectively (Fig. 1).
As this review focuses on the peptides generated through the post-translational modifications and proteolytic processing of APP, we consider APP-CTF-AICD pathway as as a continuum. For instance, the end products (CTF or AICD) and their downstream functions can derive from particular post-translational modifications of APP. These post-translational changes can regulate the selection of a specific type of processing, such as β cleavage rather than α cleavage, the interaction with a specific adaptor protein and the activation of a physiological or pathological pathway.
The hypothesis that APP-CTF-AICD, through processing and interactions with adaptor proteins that regulate gene transcription, cell signalling systems and in turn a variety of cellular functions, is compelling. Indeed, the alteration of these functions may induce cell-cycle abnormality, apoptosis, neuronal death and/or gliosis. Thus, the identification of the physiopathological role of APP-CTF-AICD could permit a better understanding of the neurodegenerative processes in AD and how it relates to Aβ formation.
The sequence of the APP C-terminal domain contains a Tyr residues at position 653 and an YENPTY motif, spanning from aa 682 to aa 687 (referring to APP 695 numbering). YENPTY is a typically motif present in many Tyr-kinase (TK) receptors and non-receptor TKs; it is generally phosphorylated and represents the docking site for multiple interacting proteins involved in cell signalling and gene transcription.
Ser and Thr residues are present in positions 654, 655 and 668 in the cytosolic domain of APP. Phosphorylation of these residues, and the resulting cellular effects, are very important for the fate of APP-CTF-AICD processing, its interactions with other proteins and how it functions.
Phosphorylation of APP and its relevance to CTF-AICD formation and function
The cytoplasmic tail of APP undergoes post-translational modifications because of phosphorylation (Tarr et al. 2002a; da Cruz e Silva and da Cruz e Silva 2003). As APP, all the CTFs and AICD fragments contain 682YENPTY687 sequence which can be bound by different adaptor proteins, also regardless of Tyr-phosphorylation. In cell-culture studies, Tyr682 can be phosphorylated by the over-expression of the nerve growth factor receptor TrkA (Tarr et al. 2002b), by a constitutively active form of the tyrosine kinase Abl (Zambrano et al. 2001) or by Src kinase (Zhou et al. 2004). Site-direct mutagenesis of Tyr682 to phenylalanine, but not of Tyr653 or 687, abrogates APP phosphorylation by Abl (Zambrano et al. 2001). cAbl can also regulate AICD formation and the modulation of AICD-dependent cellular responses, such as transcriptional induction and apoptotic cell death (Vazquez et al. 2009). In postmortem brain of both normal and AD subjects, the detection of phosphorylated full-length APP is more difficult than in cultured cells. In contrast, most β-CTFs are Tyr-phosphorylated in the brain of both AD and age-matched normal subjects (Russo et al. 2001, 2002). In vitro, thrombin treatment of cultured astrocytes induces the formation of Tyr-phosphorylated β-CTF but not of Tyr-phosphorylated full-length APP. Importantly, α-CTF is phosphorylated in neither AD brain nor cultured astrocytes treated with thrombin (Russo et al. 2002). Contradicting these observations, Takahashi et al. (2008) reported that in HEK293 cells the phosphorylation of APP at the other Tyr residue, Tyr687, is important for its processing by α and γ secretases, increasing α-CTF and AICD generation. As the processing of APP occurs at several subcellular localisations and α-CTF fragments are produced at the cell surface level, it is suggested that the alternative phosphorylation of amino acid residues can deliver APP to different subcellular compartments. This occurrence can be very important in preparing the APP substrate for different enzymatic processing steps that may or may not lead to the formation of amyloidogenic fragments. These findings indicate that alternative phosphorylation of Tyr682 or Tyr687, directing APP to different subcellular sites, could be relevant for sorting α or β-secretase cleavage. The resulting CTFs could either play a functional role as a whole or be further processed by γ secretases to yield Aβ or P3 fragments and AICD. AICD fragments, however, are generally detected in both phosphorylated and non-phosphorylated forms (reviewed in Raychaudhuri and Mukhopadhyay 2007). Phosphorylation is also a cell-cycle-dependent process and may influence AICD interactions with adaptors (Ando et al. 2001; Tarr et al. 2002b). These observations are important for understanding the role of APP-CTF-AICD Tyr-phosphorylation. This phosphorylation process triggers as yet unknown signals involved in cellular physiopathology or influences post-translational modifications capable to sort different secretases for cleavage of APP-CTF substrates. Therefore, the phosphorylation of APP-CTF-AICD may be very tightly regulated, as the kinases involved can affect not only cell signalling but also the amyloidogenic pathway. Non-phosphorylated AICD may derive from non-phosphorylated APP or preferentially from α-cleaved CTFs (non-amyloidogenic); it can bind different adaptor proteins and be recovered in the nucleus, both ex vivo and in vitro, where it can exert transcriptional regulation (reviewed in Raychaudhuri and Mukhopadhyay 2007). Phosphorylated AICD is mainly produced from β-CTF and is scantily present in the nucleus, although Tyr-phosphorylated α-CTF would yield phosphorylated AICD fragments (Takahashi et al. 2008) detected in the nucleus. β-CTFs also bind to the adaptor proteins ShcA (Src and collagen homologue) and Grb2 (growth factor receptor bound) which are involved in the activation of signal transduction pathways such as MAPKs, but neither are present in the nucleus. Consistent with this hypothesis ERK1–2 activation has been observed both in AD brain and in thrombin-stimulated astrocytes (Russo et al. 2002). Thus, Tyr-phosphorylated β-CTFs, interacting with ShcA and Grb2, can stimulate a signalling pathway. The triggering of this pathway can contribute to neuronal death, glial proliferation and tau phosphorylation, with neurofibrillary tangle formation in AD brain (Roder et al. 1993; Guise et al. 2001; Hardy and Selkoe 2002; Zhu et al. 2002). Alternatively, phosphorylated β-CTFs can undergo γ-secretase cleavage to form Aβ peptides and phosphorylated AICD. While there is no clear evidence that phosphorylated AICD interacts with ShcA and/or Grb2 and activates the Ras-MAPK pathway, AICD certainly binds to other adaptor proteins and can control transcriptional regulation. As reported by Takahashi et al., in HEK cells, phosphorylated AICD can also results from α-secretase cleavage of APP phosphorylated at the Tyr687 residue. Therefore, the issue becomes even more complex as the possible sorting between β and α-secretase cleavage may depend not only on the phosphorylation status but also on the Tyr residue (682 vs. 687) involved. However, as the search for phosphorylated AICD remains elusive, it is not clear whether its functions are mediated by both phosphorylated and non-phosphorylated forms. To shed more light on this issue, AICD-over-expressing transgenic mice were generated. They showed abnormal activation of glycogen synthase-kinase 3 (GSK-3) β, a pathological feature of AD, with hyperphosphorylation and aggregation of tau, neurodegeneration and working memory deficits. These defects were prevented by treatment with lithium, a GSK-3β inhibitor (Ghosal et al. 2009). The AICD over-expressing mice also showed abnormal neuronal networks and increased seizure susceptibility (Vogt et al. 2010). Furthermore, the observation that AICD levels are increased in the brain of AD patients is consistent with AICD having a potential neurodegenerative role, as supported by the transgenic mouse studies.
Nerve growth factor receptor TrkA, Abl and Src kinases phosphorylate Tyr residues of APP-CTFs in vitro, but the mechanism of recruitment to the subcellular compartments to phosphorylate APP-CTFs is unknown. The characterisation of other proteins interacting with APP-CTFs and the identification of other Tyr kinases that phosphorylate APP and the Tyr residues is important. Understanding these issues may provide important clues to determining if and how phosphorylation is crucial for the generation of Aβ, and whether the inhibition of a specific Tyr-phosphorylation process or Tyr kinase can abrogate Aβ formation.
Ser and Thr phosphorylation
Ser655 can be phosphorylated by protein kinase C and calcium-calmodulin dependent-kinase II (Suzuki et al. 1994), while Thr654 is phosphorylated by calcium-calmodulin dependent-kinase II (Gandy et al. 1988) and during the G2/M phase of the cell cycle by Cdc2-kinase (Suzuki et al. 1994, 1997) and Cdk5 (a neuronal homolog of Cdc2) (Iijima et al. 2000). Thr668 is constitutively phosphorylated in adult rat brain (Oishi et al. 1997), specifically in neurons (Suzuki et al. 1994), by GSK-3, c-Jun-N-terminal kinase 3 (c-Jun-NTK) (Inomata et al. 2003), Cdk5 or Cdc2 protein kinase. Thr668 is also phosphorylated by c-Jun-NTK 1–2 in a c-jun amino-terminal kinase – interacting protein 1 (JIP1)-dependent (Inomata et al. 2003; Schenfield et al. 2003) or independent manner (Schenfield et al. 2003), both in vivo and in vitro. The physiopathological relevance of these post-translational modifications and how they relate to Aβ formation and AD development have received particular attention, without really clarifying the issues. Lee et al. (2003) reported that Thr668 phosphorylation enhanced APP processing by β-secretase, increasing Aβ formation; in contrast, Feyt et al. (2007) showed that Thr668 phosphorylation inhibited γ-secretase cleavage of C99 fragment and actually reduced Aβ generation. Thus, given these contradictory results, the role of APP phosphorylation at Thr668 in Aβ production is yet to be defined. Moreover, the role of Thr668 phosphorylation in Fe65 binding and nuclear translocation of AICD or AICD/Fe65 complex is controversial. Indeed, the observation that Thr668 mutation is ineffective on AICD nuclear transport and Fe65 association is in contrast to data showing that phosphorylation of Thr668 appears indispensable for nuclear translocation and Fe65 binding (Ando et al. 2001; Chang et al. 2006; Nakaya and Suzuki 2006).
APP-CTF-AICD adaptor protein interaction
APP-CTF and ShcA, Grb2 adaptor interactions
The Tyr residues are present in the YENPTY motif in both receptor and non-receptor TK, in full-length APP and its C-terminal fragments CTF and AICD. Upon phosphorylation, they become docking sites for intracellular signalling proteins containing specific SH2 (Src homology 2), SH3, PH (pleckstrin homology) and PTB (phosphotyrosine binding) domains (reviewed in Cattaneo and Pelicci 1998). Phosphorylated Tyr residues of TKR bind to SH2/PTB-containing Shc proteins and the SH2/SH3-containing Grb2 proteins. The binding triggers signalling cascades that lead to the control of different cell functions including cell growth and cell proliferation and of gene transcription (Cattaneo and Pelicci 1998). Three genes encode different Shc proteins, ShcA, ShcB and ShcC. The ShcA gene encodes three isoforms (p46, p52 and p66) with similar structural characteristics but different functions (Migliaccio et al. 1997; Ventura et al. 2002). All three isoforms contain SH2 domain at the C-terminus, a PTB at the N-terminus and a central glycine- and proline-rich region CH1 (collagen homology region 1). ShcA isoforms regulate major functions such as growth (p46, p52), apoptosis and lifespan (p66) (Luzi et al. 2000). Grb2 via its SH2 domains can interact with TKRs or with the pYENPTY motif of APP-CTFs (Lowenstein et al. 1992; Puto et al. 2003; Zhou et al. 2004; Nizzari et al. 2007).
Among the APP-CTF-interacting proteins, only ShcA and Grb2 adaptor proteins require the phosphorylation of the Tyr682 of APP-CTFs (Tarr et al. 2002b; Zhou et al. 2004). CTFs are Tyr-phosphorylated in human brain (Russo et al. 2001, 2002), generating a pYXXP motif that can be recognised by SH2 domains of adaptor proteins such as ShcA or Grb2, or by the PTB domain of ShcA (Russo et al. 2002; Zhou et al. 2004). Immunoprecipitations of human brain tissues with ShcA, Grb2 or phospho-Tyr antibody, followed by hybridisation with specific CTF antibody, or vice-versa, revealed ShcA or Grb2 binding only to phospho-Tyr residues in β-CTFs but not to α-CTFs (Russo et al. 2002). The recruitment of ShcA to activated TKRs or phosphorylated APP-CTFs is accompanied by the phosphorylation of ShcA Tyr residues Tyr239, Tyr240 and Tyr317, which provide binding sites for Grb2 SH2 domains. Thus, Grb2 directly binds to TKRs or is indirectly recruited by ShcA (Dankort et al. 2001). These events represent the trigger to activate signalling pathways such as Ras-ERK and/or PI3K-AKT (Cattaneo and Pelicci 1998; Dankort et al. 2001; Pelicci et al. 2002) involved in cell differentiation and apoptosis, neuronal development and oncogenic proliferation (Cattaneo and Pelicci 1998; Pelicci et al. 2002; Saucier et al. 2002; Trinei et al. 2002).
While Grb2 may or may not require ShcA to bind to APP-CTFs, the phosphorylation of Tyr 682 in APP seems to be essential (Tarr et al. 2002b; Venezia et al. 2004a; Zhou et al. 2004). Indeed, the interplay and the switch between Grb2 and ShcA in regulating different cellular conditions comes from the studies on the SH-SY5Y neuroblastoma cell line. During normal cell culture conditions, phosphorylated APP 695 interacts with Grb2 without the involvement of ShcA (Venezia et al. 2004b). Upon apoptotic induction, the complex between phosphorylated APP and Grb2 is rapidly processed, APP is partially cleaved and the APP-Grb2 interaction is substituted by a new complex between phosphorylated CTFs and ShcA, but still involving Grb2. These data suggest that APP and CTFs can play a dual role, mediated by alternative interactions with Grb2 or ShcA, which can trigger different signalling events and activate different cellular functions (Venezia et al. 2004a). The role of the APP-CTFs, ShcA-Grb2 interaction in the pathophysiology of AD, is further strengthened by the data showing that CTF-ShcA-Grb2 complex and ERK1,2 activation are strongly enhanced in AD brains, compared to age-matched control brains (Russo et al. 2002). The increased expression of ShcA in AD brains is detected in reactive astrocytes around plaques and vessels (Russo et al. 2002). In vitro it is possible to induce the formation of CTF-ShcA complexes in astrocytes, but not in neurones, challenged with mitogenic stimuli such as thrombin (Russo et al. 2002). The downstream phosphorylation of ERK1,2 in thrombin-activated astrocytes also suggests that phospho-CTFs are able to activate this MAPK pathway through ShcA. This observation can partially explain the phosphorylation of ERK1,2 often described in AD and Down Syndrome (DS) brains, which is consistent with the glial proliferation in AD (Ferrer et al. 2001; Russo et al. 2002, 2005) and with the hyperphosphorylation of Tau (Roder et al. 1993; Guise et al. 2001; Hardy and Selkoe 2002). The interaction of Grb2 with both APP and PS1 in vesicular structures at the centrosome of cells leads to the activation of ERK1,2 in mitotic centrosomes in a PS1- and APP-dependent manner, and seems to confirm that APP and also PS1 are involved in the signalling pathways that regulate ERK1,2 and the cell cycle (Nizzari et al. 2007).
Other proteins interacting with APP-CTFs-AICD
In addition to Shc and Grb2, the C-terminal region of APP, CTFs and AICD is also the docking site for interacting proteins such as Fe65 (Fiore et al. 1995; Borg et al.1996), X11 (Borg et al. 1996), mDAB (Howell et al. 1997), Numb (Roncarati et al. 2002), c-Abl (Zambrano et al. 2001) and JIP-1 (Scheinfeld et al. 2002). The aforementioned adaptor proteins are able to recognise, via their PTB domain, the NPXpY motif which is generated by phosphorylation. They can also bind the C-terminus of APP and related compounds regardless of the phosphorylation of the YENPTY motif.
Fe65 protein family
Three members, Fe65, Fe65-like-1 (Fe65L and L1) and Fe65-like-2 (Fe65L2), belong to the Fe65 family of proteins. They have a common structure containing an N-terminal WW domain and two PTB domains (PTB1 and PTB2) at the C-terminus. While Fe65L1 and Fe65L2 are ubiquitously expressed, the adaptor protein Fe65 is neuron-specific, interacts with the YENPTY motif of APP through the second PTB domain (PTB2) of Fe65 (Fiore et al. 1995; Borg et al.1996) and regulates APP trafficking (Guenette et al. 1999), APP processing and Aβ generation (Sabo et al. 1999; Ando et al. 2001; Santiard-Baron et al. 2005). The APP-Fe65 interaction does not require Tyr phosphorylation and prevents Fe65 nuclear translocation by anchoring it to the cytosolic domain of APP (Borg et al. 1996; Zambrano et al. 1998). Fe65 and AICD interaction is also controlled by Thr668 phosphorylation within the Fe65-binding region of AICD. Phosphorylation or mutation of Thr668 blocks the binding of Fe65, presumably through alteration of the conformation of AICD (Ando et al. 2001; Chang et al. 2006; Nakaya and Suzuki 2006). Upon γ secretase cleavage, the AICD-Fe65 complex, likely through the first Fe65 PTB domain (PTB1), binds to two transcription factors, CP2/LSF/LBP1 and the histone acetyl transferase Tip60. It then translocates to the nucleus and activates expression of APP, BACE and Tip60 (Zambrano et al. 1998; Cao and Sudhof 2001, 2004) and this may increase amyloidogenic processing. The expression of GSK-3 is also enhanced and this contributes to tau phosphorylation. The activation of the transcriptional function seems to require the recruitment of Fe65 to membrane-bound APP/CTFs, with the subsequent cleavage of APP to release Fe65 for nuclear translocation (Cao and Sudhof 2004). The peptide complex binds to an NPXY motif in the intracellular domain of the low-density-lipoprotein receptor-related protein (LRP) and couples LRP to APP in the trimeric complex LRP/Fe65/APP (Trommsdorff et al. 1998; Kinoshita et al. 2001; Pietrzik et al. 2004). However, uncertainty exists about the requirement of AICD for Fe65 nuclear translocation and gene transcription (Hebert et al. 2006; Nakaya and Suzuki 2006). APP and Fe65 can activate transcription either in presenilin-1/2-deficient cells or in the presence of a γ-secretase inhibitor, thus in the absence of AICD generation (Haas and Yankner 2005). Moreover, AICD expression did not activate transcription. The mechanism by which APP activates transcription resides in its ability to recruit Tip60 to the membrane and induce phosphorylation by Cdk, followed by nuclear translocation of Tip60 and Fe65.
Fe65, through the WW domain, binds to proline-rich sequences (Sudol et al. 2001), allowing its interaction with other partners such as Mena and Evl members of the Ena/vasodilator-stimulated phosphoprotein (VASP) family of actin cytoskeleton regulatory proteins, the c-Abl tyrosine kinase and the ionotropic P2X2 receptor subunit (Ermekova et al. 1997; Lambrechts et al. 2000; Zambrano et al. 2001; Masin et al. 2006). Together, APP and FE65 have been involved in the regulation of actin-based cell motility and colocalise in neuronal growth cones and interact in synaptic terminals, suggesting potential roles for APP/FE65/Mena in synaptic plasticity (Sabo et al. 2001, 2003). Therefore, the AICD, through its association with Fe65 adaptor protein, could be a constituent of a multimeric complex consisting of proteins with transcription-related activities.
mDAB1 and X11 proteins
mDab1 is another adaptor molecule that binds to the YENPTY motif of APP-CTF-AICD through its PTB domain. It functions as part of the Reelin signalling pathway and is active during embryogenesis to regulate the position of neurons in the laminar structures of the brain (Howell et al. 1997). Serine phosphorylation of mDab1 increases the cellular levels of mature APP and Aβ formation (Parisiadou and Efthimiopoulos 2007). In contrast, X11 stabilises APP and inhibits Aβ secretion in cultured cells (Borg et al. 1998; Sastre et al. 1998), interfering with the amyloidogenic pathway. The interaction with the YENPTY motif of APP is phosphorylation-independent. Most likely X11 reduces Aβ formation by impairing APP trafficking to sites containing active γ-secretase complexes (Rogelj et al. 2006). Munc 18a, in concert with X11, enhances the suppression of γ-secretase cleavage (Ho et al. 2002).
JIP protein family
The cytoplasmic tail of APP binds to JNK-interacting proteins, JIP1b and JIP2. This interaction involves the GYENPTY motif in the APP cytoplasmic domain and the JIP1b or JIP2 carboxyl-terminal PTB domain. JIPs are a scaffolding protein family of JNK pathway kinases that have been implicated in various signalling pathways, including neuronal apoptosis. AICD dimerisation triggers apoptosis signal-regulating kinase-1/JNK-mediated neuronal cell death. apoptosis signal-regulating kinase-1 forms a complex with AICD through JIP1b, which is associated with the JNK signalling pathway.
Another JIP function depends on its ability to phosphorylate APP at Thr668 through JNK activation; this phosphorylation is reportedly involved in modulating APP putative functions and metabolism (Taru et al. 2002a).
The expression of JIP1b stabilizes immature APP and abolish the production of sAPP ectodomain, the formation of a cleaved intracellular carboxyl-terminal fragment of APP and the production of β-amyloid 40 and 42. Deletion of the PTB domain or alteration of its amino acid residues prevents the interaction of JIP1b with APP and affects APP metabolism, but deletion of the JNK-binding domain of JIP1b has no effect. The weaker APP-binding protein, JIP2, does not affect the processing of APP, although both JIP1b and JIP2 equally modulate the JNK signalling cascade. These results suggest that JIP1b can directly regulate APP metabolism by interacting with the APP cytoplasmic domain, independently of its involvement in the JNK signalling cascade (Taru et al. 2002b).
Kinesin-1 protein family
Kinesin-1 may also play a role in the axonal transport of APP (Kamal et al. 2000). APP acts as a kinesin-1 membrane receptor and mediates the axonal transport of BACE1 and PS1 (Kamal et al. 2001). The APP cytoplasmic domain interacts with Go-α (Giambarella et al. 1997; Brouillet et al. 1999), suggesting its involvement in G-protein-mediated signal transduction with APP-binding protein 1 (Chow et al. 1996). APP-binding protein 1 regulates the cell cycle and together with the serine/threonine kinase p-21 activated kinase is involved in DNA synthesis and neuronal apoptosis (McPhie et al. 2003). Therefore, APP may be a component of a protein complex made of Go and p-21 activated kinase, that transduces extracellular signals to the cytoplasm (Neve and McPhie 2006).
This review gives some clues about the significance of APP phosphorylation and its interaction with adaptor proteins in regulating the functions of APP and/or its proteolytic fragments CTFs and AICD. The structure of the APP cytoplasmic tail includes three Tyr residues and five Ser/Thr putative phosphorylation sites. The in vitro evidence suggests that only two of the Tyr residues are phosphorylated, specifically, Tyr682 and Tyr687, while Tyr653 does not appear to be a relevant phosphorylation site. Ser/Thr phosphorylation, has only been reported to occur in vitro for Thr654–668 and Ser655. Ex vivo data indicates that some of these phosphorylation sites are actual kinase targets that can lead to physiopathologically relevant phosphorylations. For example, Tyr682 phosphorylation would be necessary for the binding of APP C-terminal domain with the adaptor proteins Shc and Grb2. This association has been found in brain samples from healthy and AD subjects. It is important to emphasise that the association of Shc and Grb2 with APP C-terminal domain has only been found on CTF derived from β-secretase cleavage and not from CTF derived from α-secretase cleavage. This suggests a dual scenario whereby the phosphorylation of APP C-terminal domain could be involved in diverting APP metabolism towards β-secretase cleavage or, alternatively, the phosphorylation of Tyr in the C-terminal domain could occur only after β-secretase cleavage. In the first case the phosphorylation can be important not only to select the β-secretase cleavage for its physiological functions, but also to direct the APP processing toward the amyloidogenic pathway. The lack of phosphorylated α-CTFs, as well as of the interaction of α-CTFs with ShcA-Grb2 in AD brains, seems to confirm the scenario that Tyr682 phosphorylation is related to the β-secretase cut. Accordingly, Tyr phosphorylation may shift APP towards a preferential β-secretase cleavage and formation of phosphorylated β-CTFs that can be involved in the activation of a signalling pathway. However, the increased availability of β-CTFs for γ-secretase cleavage may also sustain the amyloidogenic pathway in the context of impaired Aβ removal.
The identification of CTFs phosphorylated after secretases cleavage and the characterization of their physiopathological role are still matters of investigation. Although phosphorylation sites have been identified on APP and, upon phosphorylation, these sites can become docking sites for adaptor proteins, the sequence of events and their intracellular location may govern which secretase cleaves APP and therefore be important for Aβ generation and activation of signalling cascades.
More complex is the analysis of the data concerning AICD generation and functions. This fragment of APP, which is derived from further proteolytic metabolism of both α- and β- CTF by γ secretase, seems to have transcriptional rather than signalling functions. The transcription function is inferred from its nuclear localisation (controversial) its interaction with transcription associated adaptor proteins such as Fe65, rather than with Shc and Grb2. It has been suggested that the formation of AICD occurs at membrane levels as a consequence of Fe65 binding to the YENPTY domain of APP/CTFs (Cao and Sudhof 2001, 2004). The binding of Fe65 protects AICD from proteasome degradation (Ando et al. 2001) and favors the assembly of the Fe65-AICD and/or Fe65/Tip60-AICD complexes. After the intramembranous γ-secretase cleavage, the AICD-Fe65 or AICD-Fe65/Tip60 complexes are released in the cytosol to translocate to the nucleus. In the cytosol, AICD can be degraded by IDE (insulin degrading enzyme) while Fe65 or Fe65/Tip60 translocate to the nucleus to execute their transcriptional functions. This hypothesis is consistent with the work of Haas and Yankner (2005) which exclude a transcriptional role for AICD. Nakaya and Suzuki (2006), however, argue that AICD released in the cytosol can independently translocate to the nucleus where, assembling with Fe65/Tip60, acquires the faculty to regulate gene expression.
Although direct and complete experimental evidence is still lacking, it is tempting to suggest that phosphorylation can modulate CTFs/AICD formation and fate. Indeed, the outcome of Tyr phosphorylation may depend on the Tyr residues involved. Tyr682 phosphorylation may commit APP to β-secretase cleavage, formation of β-CTF, binding to Shc/Grb2 and subsequent signalling activation. β-CTF can also become γ-secretase substrate for Aβ generation. Alternatively, Tyr687 phosphorylation may favor α-secretase cut with formation of α-CTF, not bound to Shc/Grb2 and thus not involved in signalling. α-CTF, upon γ-secretase cleavage, generates not amyloidogenic P3 fragment and AICD. AICD, derived from this post-translational modification, could more likely interact with Fe65/Tip60 and be involved in transcriptional regulation.
The role of Ser/Thr phosphorylation is less clear, with the exception of Thr668 phosphorylation that can prevent Fe65 from binding to AICD and facilitate the β-secretase association with APP, with possible increased Aβ formation and AICD release.
Overall, the data discussed in this review underline the functional role of APP and its proteolytic products, CTFs and AICD, supporting their involvement in a complex set of post-translational modifications and protein/protein interactions, that can be involved in regulating transcriptional and signalling functions. Alterations to the potential functions controlled by APP-CTFs-AICD, and defining the mechanisms that lead to these changes, may contribute to a better understanding of the development of the pathological features, including Aβ formation, that can lead to AD.
[ Schematic representation of APP processing and intracellular domain interactions. The yellow stub identifies the AICD fragment. Upon binding to adaptor proteins as reported in the white box, AICD can be translocated to the nucleus to regulate gene transcription. This function does not necessarily require Tyr phosphorylation. In addition, phosphorylation of 682 and/or 687 Tyr residues can favour the binding of Shc and/or Grb2 adaptor proteins, leading to the activation of MAPK pathway. ]
We gratefully acknowledge the financial support of Alzheimer Association Grant 2002 IIRG-02-3976, Telethon Grant E1144, MIUR Prin 2003 to GS, Cariplo 2006 to MR and MIUR Prin 2007 HJCCSF to SG and GS. We gratefully thank ‘Anchor English’ and Mrs Teresa Gomez for carefully reviewing the manuscript.