The physiology of the β-amyloid precursor protein intracellular domain AICD

Authors

  • Raphaëlle Pardossi-Piquard,

    1. Université de Nice Sophia-Antipolis, Institut de Pharmacologie Moléculaire et Cellulaire UMR6097 CNRS, Equipe labellisée Fondation pour la Recherche Médicale, Sophia-Antipolis, Valbonne, France
    Search for more papers by this author
  • Frédéric Checler

    1. Université de Nice Sophia-Antipolis, Institut de Pharmacologie Moléculaire et Cellulaire UMR6097 CNRS, Equipe labellisée Fondation pour la Recherche Médicale, Sophia-Antipolis, Valbonne, France
    Search for more papers by this author

Address correspondence and reprint requests to F. Checler, Université de Nice Sophia-Antipolis, Institut de Pharmacologie Moléculaire et Cellulaire UMR6097 CNRS, Equipe labellisée Fondation pour la Recherche Médicale, 660 route des Lucioles, Sophia-Antipolis, 06560 Valbonne, France. E-mail: checler@ipmc.cnrs.fr

Abstract

J. Neurochem. (2012) 120 (Suppl. 1), 109–124.

Abstract

The amyloid-β precursor protein (βAPP) undergoes several cleavages by enzymatic activities called secretases. Numerous studies aimed at studying the biogenesis and catabolic fate of Aβ peptides, the proteinaceous component of the senile plaques that accumulate in Alzheimer’s disease-affected brains. Relatively recently, another secretase-mediated β-APP-derived catabolite called APP IntraCellular Domain (AICD) entered the game. Whether AICD corresponded to a biologically inert by-pass product of βAPP processing or whether it could harbor its own function remained questionable. In this study, we review the mechanisms by which AICD is generated and how its production is regulated. Furthermore, we discuss the degradation mechanism underlying its rapid catabolic fate. Finally, we review putative AICD-related functions and more particularly, the numerous studies indicating that AICD could translocate to the nucleus and control at a transcriptional level, the expression of a series of proteins involved in various functions including the control of cell death and Aβ degradation.

Abbreviations used
βAPP

β-amyloid precursor protein

AD

Alzheimer’s disease

AICD

APP IntraCellular Domain

CTF

C-terminal fragment

FAD

familial Alzheimer’s disease

MED12

Mediator complex subunit 12

NEP

neprilysin

NICD

Notch Intracellular Domain

PTB

phosphotyrosine binding

PS

presenilins

In Alzheimer’s disease (AD)-affected brains, one of the main anatomical hallmarks is the widespread cortical and sub-cortical dissemination of senile plaques (Selkoe 2001; Gandy 2005; Duyckaerts et al. 2009). These extracellular lesions are mainly composed of Aβ peptides (Glenner and Wong 1984; Masters et al. 1985). Because mutations associated with early onset AD invariably lead to a modulation of Aβ load (Tanzi and Bertram 2001), it has been proposed that if not the primum movens, Aβ likely plays a key role in AD etiology (Hardy and Higgins 1992). Accordingly, it has been postulated that interfering with Aβ production could represent a beneficial strategy for AD treatment and huge efforts have been devoted to better understanding of Aβ-producing mechanisms.

Aβ derives from the proteolytic attack of a transmembrane precursor, the β-amyloid precursor protein (βAPP), that undergoes two main proteolytic cleavages by β- and γ-secretase that liberate N-terminal and C-terminal moieties, respectively (Checler 1995). The nature and the physiology of these two enzymes will be extensively described elsewhere in this special issue and will therefore not be detailed here. Briefly, γ-secretase that has been mainly characterized as a high-molecular weight complex (Takasugi et al. 2003) in which presenilins (PS) could harbor the catalytic activity (Wolfe et al. 1999b), conditions the nature of Aβ released that can be of 40 or 42 amino-acids long (Checler 1995). This is not anecdotic since the latter species is more prone to aggregation (Burdick et al. 1992), more toxic (El Khoury et al. 1996; Yan et al. 1996) and often selectively modulated by PS-associated pathogenic mutations (Suh and Checler 2002).

The original seducing view that AD pathology could be triggered by the sole Aβ42 species was, as is often the case in science, contradicted and very complicated along recent years. Thus, anatomical, biological and genetic clues and evidences concur to suggest that the generic term of Aβ indeed covers a series of N-terminally and C-terminally truncated Aβ species (Iwatsubo et al. 1996; Thal et al. 1999; Takeda et al. 2004) that are at least as toxic as their “full length counterparts”(Tekirian et al. 1999). Furthermore, as will be described elsewhere in this special issue, it is clear that the nature and biophysical state of the various Aβ species, as well as their intracellular versus extracellular distribution could govern their cellular toxic potential (Ohyagi et al. 2005; Walsh and Selkoe 2007; Ohyagi 2008).

The field has been even complicated by the arrival of a newcomer on stage. Thus, γ-secretase not only liberates Aβ but also yields its C-terminal intracellular counterpart called APP Intracellular Domain (AICD) (Passer et al. 2000; Gu et al. 2001; Sastre et al. 2001; Weidemann et al. 2002). The importance of this C-terminal fragment has been under-estimated, likely because its cellular instability precluded its deep investigation but a network evidences indicate that its putative function could underlie some of the stigmata observed in AD neurodegeneration. Here, we describe the molecular processes responsible for AICD formation and degradation and we review the intracellular modulators of AICD production. Furthermore, we discuss putative AICD-mediated functions with special emphasis on its transcription factor properties.

AICD production and degradation

At first, AICD was named AID. Thus, D’Adamio and colleagues were the first to demonstrate that γ-secretase-mediated cleavage of βAPP, together with Aβ peptides, could release an additional yet unknown C-terminal fragment (CTF) product (Passer et al. 2000). This observation was rapidly confirmed by several groups who documented a novel cleavage site of βAPP taking place in the cytosol close to the inside membrane leaflet (Gu et al. 2001; Sastre et al. 2001; Weidemann et al. 2002). This novel type of hydrolysis was referred to as ε-cleavage of βAPP. Why was this ε-derived product of βAPP called AICD? This came from the previous demonstration that Notch undergoes a similar cleavage (at S3 site) yielding an intracellular fragment called Notch Intracellular Domain (NICD) (Artavanis-Tsakonas et al. 1995; Kopan 2002). Immediately, several questions arose from this novel observation. Is there only one type of AICD? How are kinetically related γ- and ε-cleavages? Are these cleavages triggered by the same proteolytic activities?

The main species detected corresponds to AICD50 (sequence of CTF 50-99) that was identified by mass spectroscopy in neuroblastoma cells (Yu et al. 2001). Alternative cleavage could occur as suggested by the detection of AICD48/51(Yu et al. 2001; Sato et al. 2003). It appears that ε-cleavage occurs before γ-cleavage as AICD59 and AICD57 that are the C-terminal counterparts of Aβ40/42 have not been detected and indeed, Aβ48 and Aβ49 could be transformed into Aβ40 and Aβ42 in cells (Funamoto et al. 2004). Finally, AICD undergoes rapid cleavage by caspase-activity yielding a fragment called C31 (Lu et al. 2000).

It has been suggested that both γ- and ε-cleavages could be due to the same proteolytic activity. First, equimolar amounts of Aβ peptide and AICD are recovered from C99, the βAPP fragment derived from the sole action of β-secretase (Kakuda et al. 2006). The fact that PS-dependent γ-secretase could account for both γ and ε-secretase cleavages was supported by the observations that most of PS1 and PS2 mutations trigger enhanced Aβ42 and AICD productions (Sato et al. 2003). This formation was prevented by the PS-directed inhibitors DFK167 (Sato et al. 2003) and L685,458 (Wolfe et al. 1999a; Qi-Takahara et al. 2005) and by PS deficiency (Weidemann et al. 2002). However, several independent works suggested that γ- and ε-cleavages could be independent. Thus, mutational analysis indicated that certain PS mutations could trigger opposite effects on Aβ and AICD productions. Indeed, several mutations located at various parts of PS N- and C-terminal domains as well as within its intracellular loop, increase Aβ production while they inhibit AICD formation (Chen et al. 2002). Also, the aggressive L166P-PS1 mutation affected differently Aβ and AICD productions (Moehlmann et al. 2002). In agreement with this set of data, Hecimovic et al. (2004) demonstrated that unlike is the case for Aβ production, some mutations located on βAPP itself could not affect AICD production. In addition, three studies concerning the regulation of the γ-secretase complex suggested that Aβ and AICD could be differently regulated. Thus, the protein cargo TMP21 (21kDa transmembrane protein) lowers Aβ production without affecting the ε-cleavage (Chen et al. 2006). More recently, He et al. (2010) identified a γ-secretase activating protein able to increase Aβ production with concomitant AICD reduction. Finally, interestingly with respect to the regulation of AICD production (see below), the βAPP-binding protein Fe65 appears to increase AICD production while it concomitantly reduces Aβ42 formation (Wiley et al. 2007).

There is a consensus view concerning the fact that ε-cleavage of βAPP and S3-hydrolysis of Notch are triggered by similar activities. Thus, in the above-cited studies, although familial Alzheimer's disease (FAD) mutations could discriminate between Aβ and AICD productions, they never differently affect AICD and NICD formations (Chen et al. 2002; Moehlmann et al. 2002). It is therefore noteworthy that several pharmacological studies clearly showed that Aβ and NICD productions could be discriminated. Thus, we have demonstrated that a series of isocoumarin compounds (Petit et al. 2001) that do not interact with PS (Esler et al. 2002), indeed prevent Aβ production without interfering with NICD production (Petit et al. 2001). Accordingly, non-steroidal anti-inflammatory drugs (Weggen et al. 2001), the antidepressant agent lithium (Phiel et al. 2003), and the tyrosine kinase inhibitor imatinib (gleevec) (Netzer et al. 2003), all inhibit Aβ production without affecting NICD production.

Overall, it is clear that ε-cleavage is a limiting step for the occurrence of γ-cleavages. At first sight, ε-cleavages on βAPP and Notch are apparently triggered by PS-dependent proteolysis. However obviously, blocking the primary PS-dependent cleavages could explain the observation of a strict PS dependency for subsequent Aβ-production. The fact that in some experimental conditions, Aβ and AICD/NICD productions could be differently affected could suggest the occurrence of distinct proteolytic activities selectively responsible for γ- and ε-cleavages, the latter enzyme conditioning the proteolytic attack triggered by the former enzyme (that can be compared with the case of β-secretase that generates C99 which is the only βAPP fragment that behaves as a γ-secretase substrate and that is fully required for Aβ production). In this context, it is noteworthy that several PS-independent activities have been documented to display the ability to generate Aβ (Armogida et al. 2001; Wilson et al. 2002, 2003; Lai et al. 2006), some of them being even sensitive to PS-directed inhibitors as has been recently documented (Yagishita et al. 2008; Sevalle et al. 2009). Alternatively, PS-dependent γ-secretase complex could be indeed responsible for both γ- and ε-cleavages that would be finely regulated in response to membrane fluidity alterations, accessibility and topological exposition of substrates and FAD type of mutations.

Besides β-secretase cleavage that yields C99, βAPP also undergoes an additional cleavage within the Aβ sequence by α-secretase that generates the C-terminal fragment C83 (Suh and Checler 2002). Both C83 and C99 undergo subsequent cleavage by γ-secretase and therefore, behave as potential precursors of AICD. This means that both α-secretase-mediated physiological and β-secretase-mediated pathological pathways can theoretically produce AICD. Whether a given pathway preferentially produces AICD could therefore give insights on putative AICD-associated physiological or potentially toxic AD-related functions.

Data available on this specific question concur to propose that AICD is mainly produced via the amyloidogenic pathway. First, it has been shown that the ε-cleavage could be modulated by experimental conditions such as pH changes (Fukumori et al. 2006). This data agree well with the observation that AICD and its precursor C99 could accumulate in neuroblastoma cells treated with alkalizing drugs (Vingtdeux et al. 2007). This is an important point because intracellular compartments display distinct pH and this could condition the nature of activities involved as well as the rate of their enzymatic efficiencies. For instance, it is clear that acidic proteases as is the γ-secretase complex, are likely more active in a low pH environment than within a physiological environment close to neutral pH (Checler 2001). This suggests that AICD production is more likely associated with endosomal/lysosomal pathway than with direct release from plasma membrane-attached βAPP. Indeed, Goodger and Colleagues recently elegantly delineated the AICD-producing pathways in various cell systems. By means of combined approaches and by following AICD localization in nuclear AFT (AICD-Fe65-Tip60) complexes where AICD likely triggers its transcriptional factor properties (Konietzko et al. 2010) (see below chapter on AICD function), the authors showed that AICD production mostly occurs through the amyloidogenic pathway (Goodger et al. 2009). Thus, it is generally admitted that BACE1 mediated cleavage of βAPP occurs after endocytosis of membrane-embedded βAPP (Rajendran et al. 2006). Accordingly, the blockade of endocytosis by a dominant negative mutant of dynamin can fully prevent nuclear translocation of AICD that requires prior retrograde transport of AICD from the membrane (Goodger et al. 2009). Interestingly, when interfering with α- and β-secretase activities by both pharmacological and genetic approaches, Goodger and Colleagues clearly demonstrated that AICD routed through the endosome-mediated β-secretase-linked pathway. Thus, nucleus targeting of AICD was reduced by β-secretase inhibitors but not by α-secretase blockers (Goodger et al. 2009). This was corroborated by the observation that BACE1 deficiency almost totally abolished AICD signaling (Goodger et al. 2009). Finally, the authors took advantage of the previous demonstration that the Swedish mutation responsible for a subset of FAD triggered increased levels of Aβ peptides (Citron et al. 1994; Haass et al. 1995). This occurs likely because BACE1 displays higher affinity for the mutated than the parent sequence (for review on BACE, see this issue and Vassar et al. 2009). Thus, Goodger et al. (2009) showed that Swedish-mutated βAPP-expressing cells produced increased amounts of C99. In agreement with this set of data, Belyaev et al. (2010), using another functional readout (AICD-regulated neprilysin activity, see below in AICD function chapter) confirmed that β- and γ-secretase inhibitors but not α-secretase blockers abolished AICD-mediated function. It is interesting that in its first report on the generation of AICD (called AID at that time, see above), Passer et al. (2000) already showed that both Swedish-mutated βAPP- and C99-expressing cells produced larger amounts of AICD. Furthermore, they showed that cells expressing the α-secretase-derived C-terminal stub C83 did not produced AICD (Passer et al. 2000). The fact that conversely Kume et al. (2004) reported on an α-secretase-dependent production of AICD could be seen as paradoxical. However, one could envision that during the routing of C99, part of the substrate is targeted by α-secretase-like protease, thereby generating a subset of C83 fragment that could serve as substrate for AICD production. This hypothesis is supported by data from our laboratory (Flammang et al, in preparation).

The fact that AICD has been evidenced much later than Aβ peptides can likely be accounted for its extreme instability and susceptibility to proteolysis. Accordingly, AICD is usually poorly detectable in cells and human tissues. Nevertheless, the first report on AICD showed that this fragment was indeed detectable by MALDI mass spectroscopy in sporadic AD-affected brains (Passer et al. 2000). This has been confirmed and extended recently in a work where Pimplikar and Suryanarayana (2011) documented a technical procedure enabling the detection of AICD in brain lysates.

The catabolic fate of AICD is likely the result of a two-steps proteolytic cascade that conditions the fate of both AICD and its precursor C99. The latter is rapidly broken down by the proteasome machinery (Nunan et al. 2001), which therefore, acts as an upstream regulator of AICD levels. Consistent data indicate that once produced, AICD undergoes rapid inactivation by the mainly cytosolic and endosomal insulin-degrading enzyme insulysin. Thus, in vitro data showed that cytoplasmic fractions prepared from various cell types, including neurons, harbor an AICD-degrading metalloprotease that was identified by competition experiments as insulysin (Edbauer et al. 2002). Furthermore, the over-expression of wild-type insulysin, but not a catalytically inactive mutant, accelerated AICD degradation (Edbauer et al. 2002). This was corroborated by the analysis of the degradation of AICD by purified insulysin. Thus, Venugopal et al. (2007) demonstrated that insulysin cleaves AICD at multiple sites, yielding numerous small peptides, without any obvious apparent selectivity for a given site. Finally, Farris et al. (2003) brought the in vivo evidence that insulysin could contribute to the catabolism of AICD, the levels of which are augmented in insulysin-depleted mice.

Finally, βAPP is processed at its C-terminal tail by caspase 3 (Lu et al. 2000) that releases a cytotoxic fragment referred to as C31 (Lu et al. 2000; Dumanchin-Njock et al. 2001). It is not yet clear whether only βAPP, eventually its C99 membrane-attached fragment or free AICD could undergo such a cleavage but it is possible that this proapoptotic cleavage contribute to the overall catabolic fate of AICD.

Although both proteasome and caspase 3 could, to a certain extent, contribute to the degradation of AICD, it remains that consistent data concur to propose insulysin as the main AICD-degrading enzyme. In this context, it is interesting to note that insulysin levels appear to decrease with aging in the hippocampus of human brains (Caccamo et al. 2005) as well as in transgenic mice models of AD (Hwang et al. 2005). As several reports have consistently indicated that BACE1 activity augments with age (Nistor et al. 2007) and in AD-affected brains (Cai et al. 2010), it emerges from these observations that the conjunction of increased β-secretase-mediated biogenesis of its precursor C99 and lowered insulysin-mediated degradation, would lead to increased AICD production and suggest along with Aβ, it could contribute to AD pathology.

Regulation of AICD production and function

The C-terminal intracellular βAPP moiety harbors several conserved domains that could act as possible linkers of adaptors proteins. These adaptors could be seen as means to modulate AICD production and/or to increase its cellular lifetime. The difficulty in studying the regulation of AICD production and therefore its signaling function (see below) consists in the fact that it is hard to estimate whether identified regulatory domains bind free intracellular AICD or interact with the βAPP C-terminal fragment and thereby modulate AICD production in integrated systems. Possibly, these regulators could act selectively on βAPP-CTF or AICD or alternatively, could bind both βAPP-CTF and AICD and thereby, trigger sustained interaction even after ε-cleavage liberating AICD.

The intracellular βAPP-CTF exhibits a YENPTY sequence between residues 682 and 687 that contains a consensus sequence for phosphotyrosine binding (PTB) domain interaction and therefore, the putative phosphorylation of βAPP-CTF and/or AICD could condition the ability of a regulator to bind to these species. Indeed, it is interesting to note that tyrosine 682-phosphorylated AICD is one of the AICD phosphorylated species recovered as revealed by general mass spectrometric technology (Olsen et al. 2006).

About 20 proteins belonging to eight distinct families have been documented as AICD interacting proteins. The best-characterized group of adaptors corresponds to the Fe65 and Fe65-like family members. These proteins that contain two PTB domains have been shown to physically interact with the βAPP-CTF (Fiore et al. 1995; Borg et al. 1996; Bressler et al. 1996; Guénette et al. 1996; Mcloughlin and Miller 1996; Duilio et al. 1998). The importance of βAPP-CTF phosphorylation at tyrosine 682 phosphorylation is discussed. Thus, two studies indicated that βAPP-CTF-Fe65 interaction remained unaffected by the phosphorylation of this residue (Borg et al. 1996; Zambrano et al. 1998) while a recent works support the view that tyrosine 682 phosphorylation precluded physical interaction between these two proteins (Zhou et al. 2009).

The functional influence of AICD-Fe65 interaction is also a matter of discussion and will be documented later (see chapter on AICD function). Briefly, Fe65 could directly stimulate AICD formation (Wiley et al. 2007) but does not always lead to increased levels of AICD fragment (Huysseune et al. 2007). Furthermore, Fe65, the histone acetyltransferase Tip60 and AICD form a tripartite complex (Cao and Südhof 2001) that is thought to drive AICD cellular compartmentalization and AICD-mediated function. Inhibition of this interaction by tyrosine phosphorylation (Zhou et al. 2009) or Fe65 sequestration by the estrogen α receptor (Bao et al. 2007) all lead to the abolishment of AICD-mediated signaling. Conversely, Nakaya and Suzuki (2006) proposed that AICD increased Fe65-mediated nuclear signaling but that the two proteins did not traffic together to the nucleus.

Members of the Mint/X11 family (Miller et al. 2006; Rogelj et al. 2006) physically interact with AICD in a PTB domain-dependent manner (Borg et al. 1996; Mcloughlin and Miller 1996) and modulate AICD-mediated transcription in an isoform-specific manner. Thus, Mint 1 and Mint 2, but not their parent member Mint 3, drastically reduce AICD-induced transactivation (Biederer et al. 2002).

Finally, several other PTB-containing proteins could modulate AICD-mediated gene transactivation (for review see Raychaudhuri and Mukhopadhyay 2007; Müller et al. 2008; Chang and Suh 2010). Thus, members of the JIP family (c-jun-N-terminal kinase interacting protein, JIP1b and JIP2) activate AICD-mediated signaling but interestingly, do not modulate transcription associated with ICDs of other members of the βAPP family (Scheinfeld et al. 2003). Briefly, members of the Shc family (ShcA and ShcB), mammalian disabled (mDab1), Numb, Abl, and Grb2 also interact with the YENPTY domain of AICD in both phosphorylation-dependent and -independent mechanism (Raychaudhuri and Mukhopadhyay 2007; Müller et al. 2008; Chang and Suh 2010).

Overall, the above data indicate that AICD biogenesis is tightly regulated by a battery of adaptors that are able to precisely tune the production and thereby, the function of AICD.

AICD as a transcription factor: no longer controversial

As stated above, βAPP-CTF were first named AICD because this βAPP-derived fragment was, at first sight, analogous with NICD that is the intracellular fragment released from Notch (Gu et al. 2001; Sastre et al. 2001; Weidemann et al. 2002). This analogy is not restricted to βAPP-AICD and Notch-NICD. Thus, numerous studies later revealed that presenilins-dependent γ-secretase trigger regulated intramembrane proteolysis of various transmembrane substrates besides βAPP and Notch (Brown et al. 2000; Weihofen and Martoglio 2003; Beel and Sanders 2008; Lleo 2008) (see Table 1) as has been extensively reviewed elsewhere (Haapasalo and Kovacs 2011).

Table 1.   γ-Secretase substrates
SubstratePhysiological functionICDsGene transcriptional regulationReferences
  1. ADAM10, A disintegrin and metalloprotease; ApoER2, apolipoprotein E receptor-2; APP, amyloid precursor protein; APLP, amyloid precursor like-protein; CSF-1R, colony-stimulating factor 1; CXCL, chemokine (C-X-C motif) ligand; DCC, deleted in colorectal cancer; DNER, Delta/Notch-like EGF-related receptor; DSG2, desmoglein-2; GluR3, glutamate receptor 3; HLA-A2, human leukocyte antigen; IL-1R 1/2, interleukin-1 type 1 and 2 receptor; LAR, leukocyte-common antigen-related; LDLR, low-density lipoprotein receptor; LRP, low-density lipoprotein receptor-related protein; NCAM L1, neural cell adhesion molecule L1; NRG-1, neuregulin-1; NRADD, neurotrophin receptor alike death domain protein; NRXN-3β, neurexin-3β; PAM, peptidylglycine α-amidating monooxygenase; PLXDC2, plexin domain-containing protein; RPTP, receptor protein tyrosine phosphatase; VLDL, very low density lipoprotein; VEGFR, vascular endothelial growth factor.

APPCell adhesion and migration, neurite outgrowth, synaptogenesisAICDYesCao and Südhof 2001
APLP1/2Cell adhesion and migration, neurite outgrowth, synaptogenesisALID1/2YesScheinfeld et al. 2002; Pardossi-Piquard et al. 2005; Xu et al. 2006
Notch1/2/3/4Signaling receptor, cell–cell communication, cell differentiation during embryonic and adult lifeNICDYesReviewed in Bray (2006)
JaggedNotch signaling ligandJICDYesLaVoie and Selkoe 2003; Ikeuchi and Sisodia 2003
Delta 1Notch signaling ligandDII1ICYesSix et al. 2003; LaVoie and J. 2003; Ikeuchi and Sisodia 2003; Bland et al. 2003
DNERNeuronal notch signaling ligand, cerebellar developmentDNER ICDUnknownHemming et al. 2008
E/N-CadherinCell adhesionE/N-Cad-CTF2Yes (indirect)Marambaud et al. 2003
VE-CadherinCell adhesionunknownUnknownSchulz et al. 2008
γ-ProtocadherinsCell adhesionPcdhγ-CTF2YesHass and Yankner 2005; Hambsch et al. 2005
AlcadeinCadherin-related post-synaptic Ca2+-binding protein AlcICDYes (indirect)Araki et al. 2004
Ire1 α/βER transmembrane protein, unfolded protein response signalingUnknownYesNiwa et al. 1999
CD44Cell adhesion and migration, lymphocyte activation, wound healing, tumor cell growth and metastasisCD44-ICDYesOkamoto et al. 2001
CD43Immune function, T-cell activation, cell–cell communication in leucocytesUnknownUnknownAndersson et al. 2005
IFNaR2Subunit of the type 1 IFN-α receptorIFNaR2 ICDYesSaleh et al. 2004; El Fiky et al. 2005
HLA-A2MHC class 1 protein, T-cell development, immune cell responseHLA-A2 ICDUnknownCarey et al. 2007
CXCL16/CX3CL1Transmembrane chemokine ligands, cell migration and cell–cell communicationUnknownUnknownSchulte et al. 2007
IL-1R 1/2Cytokine receptor, inflammation and immune cell responseIL-1R1 ICD IL-1R2 ICDUnknownKuhn et al. 2007; Elzinga et al. 2009
Nectin-1αCell adhesion, cell–cell junction, synaptic contactNE-ICDUnknownKim et al. 2002
ErbB-4Receptor tyrosine kinase, receptor of growth factorsE4ICDYesNi et al. 2001
NRG-1Growth and differentiation factor, ligand for ErbB receptorsNrg-ICDYesBao et al. 2003
Beta cellulin (BTC)Cell growth, ligand for ErbB receptorsBTC-ICDUnknownStoeck et al. 2010
CSF-1RReceptor tyrosine kinase, cell proliferation and differentiationCDUnknownWilhelmsen and van der Geer 2004
Tie1Receptor tyrosine kinase, vascular developmentUnknownUnknownMarron et al. 2007
IGF-1RInsulin-like growth factor 1, receptor tyrosine kinase, cell proliferationIGF-IR ICDUnknownMcElroy et al. 2007
Insulin receptor (IR)Receptor tyrosine kinase, growth factor receptor, glucose metabolismIR-ICDUnknownKasuga et al. 2007
MetReceptor tyrosine kinase, hepatocyte growth factor receptor, cell proliferationMet-ICDUnknownFoveau et al. 2009
LARReceptor tyrosine phosphatase, cell adhesion, synapse formation and neurite outgrowthLICDYesHaapasalo et al. 2007
RPTPkReceptor tyrosine phosphatase, cell adhesionPICYesAnders et al. 2006
EphrinB1/B2Cell–cell communication, cell migration, vascular and nervous system development, axon guidance, synaptogenesis, ligand for Eph receptorseB1ICD, EphrinB2-CTF2UnknownTomita et al. 2006; Georgakopoulos et al. 2006
EphB2Receptor tyrosine kinase, neuritogenesis, angiogenesisEphB2-CTF2UnknownLitterst et al. 2007
EphA4Ephrin receptor, neuritogenesis, spine formationEphA4 ICDUnknownInoue et al. 2009
GluR3Ionopropic glutamate receptor, neural communication, memory and learningUnknownUnknownMeyer et al. 2003
p75NTRNeurotrophin receptor, axonal outgrowth, cell survivalp75ICDYesKanning et al. 2003
NRADDNeurotrophin receptor, death receptor like proteinNRICDUnknownGowrishankar et al. 2004
DCCNetrin-1 receptor, apoptosis, commissural axon guidance, conditional tumour suppressorDCC-ICDYesTaniguchi et al. 2003
LRP1LDL superfamily receptor, endocytic and signaling receptorLRP1 ICDYesMay et al. 2002Kinoshita et al. 2003
Megalin (LRP2)LDL superfamily receptor, endocytosis and transportMICDYesZou et al. 2004; Li et al. 2008
ApoER2 (LRP8)LDL superfamily receptor, neuronal migration, synaptic plasticity, LTPapoER2 ICDYesMay et al. 2003; Hoe and Rebeck 2005
VLDLRLDL superfamily receptor, endocytic and signaling receptorUnknownUnknownHoe and Rebeck 2005
LDLRLipoprotein receptorUnknownUnknownHemming et al. 2008
Syndecan-3Heparan sulfate proteoglycans, receptor of cytokines, growth factors and extracellular matrix componentsSICDYesSchulz et al. 2003
GHRGrowth hormone receptorGHR-stubUnknownCowan et al. 2005
VEGFR-1Growth factor receptor, angiogenesisUnknownUnknownCai et al. 2006
KlothoHormone metabolism, anti-aging proteinUnknownUnknownBloch et al. 2009
VGSCbVoltage-gated sodium channel β2-subunit, cell adhesion and migrationβ2-ICDUnknownKim et al. 2005; Wong et al. 2005
Tyr, Tyrp1/2Pigment synthesisUnknownUnknownWang et al. 2006
DSG2Structural components of desmosomes, cell adhesion, tumorigenesisDSG2 ICDUnknownHemming et al. 2008
Dystroglycan (DG)Member of the multiprotein dystrophin–glycoprotein complexDG ICDUnknownHemming et al. 2008
betaglycanType III TGF-β receptor, cell growth, differentiation, and adhesionUnknownUnknownBlair et al. 2011
NPR-CNatriuretic peptide receptor-CUnknownUnknownHemming et al. 2008
PLXDC2Nervous system proteinUnknownUnknownHemming et al. 2008
VasorinTGF-β inhibitor, vascular remodelingVasorin ICDUnknownHemming et al. 2008
NCAM-L1Cell adhesion and migration, neurite outgrowth, neuronal survivalL1-CTF2UnknownMaretzky et al. 2005
Polyductin/fibrocystinCell receptor, renal tubular formationPICDUnknownKaimori et al. 2007
SorLAVps10p receptors, sorting and intracellular traffickingSorICDYesBohm et al. 2006; Hermey et al. 2006; Nyborg et al. 2006
SortilinVps10p receptors, sorting and intracellular traffickingUnknownUnknownHermey et al. 2006; Nyborg et al. 2006
SorCS1-3Vps10p receptors, sorting and intracellular traffickingUnknownUnknownHermey et al. 2006; Nyborg et al. 2006
ADAM10Metalloprotease, ectodomain sheddingADAM10 ICDYesTousseyn et al. 2009
NRXN-3βCell adhesion, synapse formationNRXN3β-ICDYesBot et al. 2010
PAMSecretory granule membrane proteinPAM sf-CDYesRajagopal et al. 2010

As NICD functions as a transcription factor (Kopan et al. 1996; Kopan 2002), this theoretical ground led to propose that AICD could also signal to the nucleus to harbor such type of function. Such concept implies that AICD should either traffic to the nucleus or could be constitutively present in this organelle. Rapidly, several works confirmed the presence of AICD in the nucleus by anatomical approaches. (Kimberly et al. 2001; von Rotz et al. 2004; Pardossi-Piquard et al. 2005; Chang et al. 2006; Bao et al. 2007; Goodger et al. 2009). More precisely, Muresan and colleagues showed that endogenous phosphorylated AICD localized in subnuclear compartment, the protein markers of which indicated that it could correspond to splicing factor compartments (Muresan and Muresan 2004). Interestingly, over-expressed AICD distributes evenly in the cells in constitutive conditions, with part of it translocates into the nucleus where it co-localizes with Fe65. More recently, Konietzko et al. (2010) demonstrated by confocal analysis that nuclear export blockade allows revealing the nuclear localization of endogenous AICD at the level of nuclear transcription territories. Of most interest, AICD and NICD co-localize at these anatomical sub-domains. This is very interesting with respect to the previous demonstration of a putative functional dialogue between βAPP and Notch signaling. Along with this hypothesis, it is striking that both AICD and NICD can compete for binding to some of the above-described adaptors (Roncarati et al. 2002), and thereby, functionally interfere with their respective function. Overall, both anatomical and functional evidence leave very few doubts about the capacity of AICD to function as a transcription factor, thereby regulating a huge variety of cellular functions listed in Table 2. These doubts are totally dissipated when one examines the abundant literature providing multiple examples of γ-secretase substrates and PS-mediated generation of ICDs (see Table 1) that also act as transcription factors. Thus, out of more than 70 substrates, about 30 generate ICD, the activity as transcription factor of which has been documented (Table 1). Most of these ICDs act in concert with additional cofactors and behave as transcription factors although two of them regulate gene transcription indirectly, namely NCad/CTF2 that binds the transcription factor CBP (CREB-binding protein) and promotes its proteasomal degradation and AlcICD that competes with AICD for binding to Fe65.

Table 2.   AICD target genes
AICD target genesPhysiological functionUp- or down-regulationPromoter binding (gel shift or ChIP experiments)References
  1. APP, amyloid precursor protein; BACE, beta-site APP cleaving enzyme 1; CHOP, C/EBP homologous protein; EGFR, epidermal growth factor receptor; GSK3β, glycogen synthase kinase; IGFBP3, insulin growth factor-binding protein 3; LRP1, low-density lipoprotein receptor-related protein 1; NEP, neprilysin; Ptch1, Shh receptor patched 1; RA, retinoic acid; Shh, Sonic Hedgehog; VGLUT2, vesicular glutamate transporter 2.

RA-responsive genesCell–cell communication and developmentDownUnknownGao and Pimplikar 2001
KAI1/CD82metastasis suppressor, apoptosisUpYesBaek et al. 2002; von Rotz et al. 2004
GSK3βKinase, glycogen metabolism, cell development, cell cycle regulation, proliferation and apoptosisUpUnknownKim et al. 2003; von Rotz et al. 2004; Ryan and Pimplikar 2005
APPCell adhesion and migration, neurite outgrowth, synaptogenesisUpUnknownvon Rotz et al. 2004
BACEAspartyl protease, APP cleaving enzymeUpUnknownvon Rotz et al. 2004
Tip60Histone acetyltransferaseUpUnknownvon Rotz et al. 2004
NEPMetallopeptidase, Aβ degrading enzymeUpYesPardossi-Piquard et al. 2005; Belyaev et al. 2009; Xu et al. 2011
p53Tumor suppressor, apoptosisUpYesAlves da Costa et al. 2006; Ozaki et al. 2006
α2-ActinOrganization and dynamics of the actin cytoskeletonUpUnknownMuller et al. 2007
TransgelinOrganization and dynamics of the actin cytoskeletonUpUnknownMuller et al. 2007
IGFBP3Transport of insulin growth factorUpUnknownMuller et al. 2007
EGFRCell cycle, proliferation, differentiation, survival functionDownYesZhang et al. 2007
LRP1Endocytic and signaling receptor, transport of ApoE, ApoE/cholesterol metabolismDownYesLiu et al. 2007
Cyclins B1 and D1Cell cycle activationUpUnknownAhn et al. 2008
VGLUT2Transport of glutamate, neurotransmissionUpUnknownSchrenk-Siemens et al. 2008
CHOPER stress, unfold protein response (UPR), apoptosisUpYesTakahashi et al. 2009
Aquaporin 1Water channelUpUnknownHuysseune et al. 2009
S100a9Inflammation-associated calcium-binding proteinUpUnknownHa et al. 2010b
ApoJ/clusterinLipoprotein, transport of lipid, survival functionDownUnknownKogel et al. 2011
Ptch1Shh pathway, cell division, proliferation, brain developmentUpYesTrazzi et al. 2011

AICD function: focus on the control of cell death and neprilysin-mediated Aβ degradation

AICD has been implicated in various cellular functions that have been reviewed recently (Müller et al. 2008; Chang and Suh 2010). Briefly, AICD modulates the intracellular homeostasis of calcium and ATP (Hamid et al. 2007) and can control neuronal networks (Vogt et al. 2011). AICD could also control cellular trafficking and cytoskeletal dynamics. Thus, Ghosal et al. (2009) showed that transgenic mice expressing both AICD and Fe65 show increased transcription of glycogen synthase kinase-3β that was accompanied by enhanced phosphorylation of Tau, a protein involved in microtubule stabilization. This agrees well with the observation that in transgenic flies, the over-expression of βAPP but not of its C-terminally deleted counterpart (i.e. lacking the AICD domain), triggers axonal transport impairment (Gunawardena and Goldstein 2001).

AICD has been implicated in the control of cell death. Indeed, the first clue concerning the putative function of AICD was brought by the initial report describing its identification. Thus, Passer et al. (2000) showed that AICD strongly potentiated FADD (Fas domain associated protein)-induced cell death in Jurkat cells. Kinoshita et al. (2002) corroborated this observation by showing that AICD-modulated cell death was Tip60-dependent and two studies demonstrated that AICD-mediated apoptosis could be neuron-specific (Nakayama et al. 2008; Ohkawara et al. 2011). More recently, it was demonstrated that cAbl increased AICD and that this correlated with a decreased cellular viability (Vazquez et al. 2009). Finally, it was shown that the interaction of cytoplasmic inclusions called Hirano bodies with AICD/Fe65 could reduce the initiation of apoptosis (Ha et al. 2010a).

AICD-associated cell death can likely be accounted for its propensity to trans-activate the promoter of the pro-apoptotic tumor suppressor p53. Thus, we showed that AICD-mediated cell death was abolished by p53 deficiency and we demonstrated that AICD physically interacted with p53 promoter (Alves da Costa et al. 2006). Ozaki et al. (2006) corroborated this set of data and confirmed that AICD expression enhanced p53-dependent cell death.

Another set of data clearly demonstrated that AICD could contribute to the inactivation of Aβ. Thus, we first demonstrated that AICD controlled the promoter transactivation of the Aβ-degrading enzyme neprilysin (NEP) (Pardossi-Piquard et al. 2005, 2006). Combined gene-inactivation, pharmacological, biochemical and enzymatic approaches all converge to propose that AICD acts as a transcriptional activator of NEP. Thus, PS-, nicastrin- and βAPP-depleted cells (i.e. cells unable to produce AICD) all display drastically reduced levels of NEP protein and mRNA expressions and reduced NEP promoter transactivation, a phenotype that could be reversed by PS/nicastrin or AICD complementation (Pardossi-Piquard et al. 2005, 2006). Interestingly with respect to AICD regulation, NEP mRNA levels were increased by AICD/Fe65/Tip60 cDNA transfection while X11α/β/γ over-expression or endogenous Fe65 depletion reduced NEP activity (Pardossi-Piquard et al. 2005). It was interesting to note that AICD did not affect the levels of the other putative Aβ-degrading enzymes insulysin and endothelin-converting enzyme (Pardossi-Piquard et al. 2005). Finally, the ability for AICD to bind to NEP promoter was suggested by gel shift experiments (Pardossi-Piquard et al. 2005). These new data were disputed (Chen and Selkoe 2007; Pardossi-Piquard et al. 2007) but several subsequent works confirmed our findings. First, Eisele and Colleagues showed that the tyrosine kinase inhibitor imatinib (gleevec) that was previously shown to decrease Aβ levels (Netzer et al. 2003), indeed dose-dependently increases both AICD and NEP activity and mRNA levels (Eisele et al. 2007). We recently comforted these data by showing that gleevec was unable to modulate NEP activity in βAPP-deficient cells, thereby demonstrating that gleevec-mediated modulation of NEP activity was indeed mediated by AICD (Bauer et al. 2011). More recently, Belyaev et al. (2009) demonstrated the direct physical interaction of AICD with NEP promoter by ChIP analysis. Interestingly, AICD-mediated control of NEP and the extent of such modulation appear to be cell specific. Thus, Hong et al. (2011) demonstrated that neither AICD-induced NEP activity nor AICD-NEP physical interaction could be detected in prostate cells. This agrees well with our observation that unlike fibroblasts and HEK293 cells, blastocysts display weak measurable AICD-regulated NEP (Pardossi-Piquard et al. 2005). This likely explains the difficulty of Chen and Selkoe to reproduce our data since they used blastocysts as the choice model to study AICD-mediated regulation of NEP (Chen and Selkoe 2007).

Interestingly, Xu and Colleagues delineated the precise mechanisms by which AICD could control NEP. It appears that Mediator complex subunit 12 (MED12) that belongs to a transcriptional co-activator complex named mediator (Lewis and Reinberg 2003), indeed physically interacts with Fe65 and Tip60 only in the presence of AICD (Xu et al. 2011). Interestingly, although pharmacological inhibition of AICD formation abolished MED12 binding to NEP promoter, siRNA-targeting of MED12 did not affect AICD interaction with NEP promoter (Xu et al. 2011). This indicates that AICD first interacts with NEP promoter and then, recruits MED12 to AICD-responsive promoter elements. Finally, another independent study further documented AICD-mediated control of NEP. Huysseune et al. (2009) showed by a microarray approach that NEP and aquaporin 1 were the only proteins up-regulated (increase of log ratios by about 2, 5- and 6-fold, respectively), in PS-deficient fibroblasts.

Conclusion

Owing to the above-described data, it appears crystal-clear that AICD could function as a transcription factor. It is interesting to emphasize the fact that AICD could control Aβ genesis at various steps. Thus, AICD-mediated increase in βAPP and BACE1 promoter transactivation (von Rotz et al. 2004) theoretically triggers more substrate for increased levels of β-secretase. Thus, the enhancement of C99 is therefore the source of increased levels of γ-secretase-mediated production of Aβ. Furthermore, AICD-mediated increase of NEP could alter the ratio of Aβ42 over Aβ40 because NEP was shown to degrade the latter peptide more efficiently. This could have pathological consequences since small changes in Aβ42/Aβ40 ratio could lead to the exacerbation of cellular toxicity (Kuperstein et al. 2010). Finally, we recently documented the fact that p53, that is an AICD transcriptional target (Alves da Costa et al. 2006; Checler et al. 2007), could control Pen-2 expression and thereby, γ-secretase activity (Dunys et al. 2009; Checler et al. 2010). Together with the fact that AICD is mostly produced through the amyloidogenic pathway, AICD appears as a multifunctional factor affecting several physiological processes but also likely contributing to Alzheimer’s disease pathology.

Acknowledgements

This work was supported by the Fondation pour la Recherche Médicale and by the Conseil Général des Alpes Maritimes.

Conflicts of interest

All authors declare no conflict of interests.

Ancillary