The amyloid precursor protein (APP) has been extensively studied for the proteolytic cleavages it undergoes to generate the Amyloid β (Aβ) peptide of Alzheimer's disease (Fig. 1) (Zhang et al. 2012). Cleavage of APP by the secretases (α-, β-, and γ-secretase), to release Aβ species, also generates the secreted ectodomain fragments sAPPβ and sAPPα, and the APP-intracellular domain (AICD) (Fig. 1). Non-secretase-mediated cleavages also occur well away from the Aβ region to release APP N-terminal fragments (APP-NTF) that contain the N-terminal cysteine-rich domain. An added layer of complexity arises from the alternative splicing that APP undergoes. The major isoforms: APP695, 751 and 770 (numbering refers to number of amino acids), are derived from splicing of exon 7 that encodes a Kunitz-protease inhibitor (KPI) domain and exon 8 that contains an OX-2 homology sequence (Zhang et al. 2012). APP695 lacks KPI and OX2, while APP-751 and APP770 contain the KPI and KPI plus OX-2, respectively. Although the role of the KPI is well understood as a regulator of platelet aggregation and hemostasis, the action of OX-2 remains poorly defined. The APP695 isoform is the most highly expressed isoform in neurons. This assortment of cleavage events and isoforms means the proteolytic processing of APP yields a diverse array of secreted and intracellular APP metabolites.
In deciphering the relationship between APP metabolism and function it is appropriate to consider if these APP metabolites can regulate their respective activities and/or levels. In addressing this point a very interesting interplay between AICD and Aβ has been described which is providing important, and broader, insights into the function of APP. AICD is an intracellular metabolite that is transported to the nucleus and acts as a transcriptional regulator. Around 20 AICD gene targets have been proposed to date and they encode proteins involved in the cytoskeleton, lipid metabolism, signaling, cell cycling, and protein metabolism (reviewed in; Pardossi-Piquard and Checler 2012). However, binding of AICD to the gene promoter has not been confirmed in all cases and some of these regulator effects may well occur via an indirect route.
Among the genes that AICD targets is the protease neprilysin (Belyaev et al. 2010). Neprilysin is a secreted protease that cleaves Aβ, and increasing neprilysin expression in the brain reduces neuronal Aβ levels. The ability of AICD to regulate neprilysin expression is pertinent since dysregulation of Aβ clearance, as a driver of Aβ accumulation, is seen as a key element in sporadic Alzheimer's disease (Mawuenyega et al. 2010). Therefore, understanding the regulation of molecules that modulate the deposition and/or removal of Aβ from the brain has direct clinical relevance. In this issue, Kerridge et al. (2014) have expanded the AICD:Aβ clearance repertoire by showing that transthyretin expression is also regulated by AICD. Transthyretin is an extracellular protein that binds Aβ and inhibits Aβ aggregation and neurotoxic activity, and promotes Aβ clearance from the brain.
Finding that AICD can influence Aβ levels shows that the impact of the amyloidogenic pathway on amyloidogenesis is not restricted to producing Aβ, but extends to the production of AICD. Moreover, these studies go beyond informing us about the regulation of Aβ metabolism and are also shedding novel insights into the complexity of the relationship between APP metabolism and APP function.
One of the more instructive outcomes from this work is finding that APP regulates neprilysin and transthyretin expression in an APP isoform-dependent manner (Belyaev et al. 2010; Kerridge et al. 2014). The KPI-less APP695 species is active while the KPI containing isoforms, APP751 and APP770, are inactive. However, not all APP-regulated genes were affected in an isoform-dependent manner. The expression of Glycogen synthase kinase-3β and aquaporin were up-regulated similarly by over-expressing either APP695, APP751, or APP770. However, the APP-mediated effect on Glycogen synthase kinase-3β (GSK-3β) and aquaporin was not via AICD binding directly to the promoter regions (Kerridge et al. 2014). Therefore, while the APP:KPI+ isoforms could regulate gene expression it did not involve AICD binding directly to the promoter, in contrast to what happens with APP695 on neprilysin and transthyretin, The APP695 isoforms regulate neprilysin and transthyretin expression by AICD displacing histone deacetylases and altering histone acetylation at the neprilysin and transthyretin promoters (Belyaev et al. 2010; Kerridge et al. 2014). This highlights the need to do a thorough promoter analysis to confirm the mechanism of action being employed.
The APP-mediated effects on neprilysin and transthyretin expression are modulated by β-secretase. This is consistent with the AICD fragment being primarily produced from the β-secretase cleaved APP C-terminal fragment, rather than α-secretase cleaved C-terminal fragment (Pardossi-Piquard and Checler 2012). The amyloidogenic processing of the APP:KPI isoforms is cell type dependent with comparable levels of β-secretase cleavage of APP695, APP751, and APP770 occurring in HEK293 cells, however, in neuronal cells β-secretase cleavage of APP751 and APP770 was significantly less than APP95 (Belyaev et al. 2010; Kerridge et al. 2014). However, the AICD-mediated up-regulation of neprilysin only occurred in neuronal cells and not in HEK cells, indicating that this activity is cell type restricted and presumably dependent upon the cellular environment and the availability of particular cofactors to form an active AICD complex.
What is it about APP695 that causes its AICD to target to the transthyretin and neprilysin promoters? Alternatively, what prevents APP751 and APP770 from possessing this activity? The sequence differences between APP695 and APP751/APP770 either at the splice junctions or the additional domains may be providing or omitting motifs that facilitate transport of APP695 along a distinct processing route that leads to β-secretase cleavage, while the APP:KPI isoforms are excluded from this pathway or reside in a distinct subcellular compartment. β-secretase cleavage occurs in the acidic endosome compartment resulting from endocytosed APP (Zhang et al. 2012). The sequence differences may modulate APP binding to different transport proteins involved in regulating APP transport and metabolism. While a large range of APP binding proteins are known, the identity of those which are involved in the regulation of the APP:KPI versus APP695 isoforms remains unclear.
Discovering that APP, via AICD, is regulating Aβ levels has broader implications regarding Aβ and implies that it has a biological function, rather than representing an unintended and non-functional toxic by-product arising from the cleavage of APP. While it may be coincidental that some of the genes targeted by AICD happen to also affect Aβ metabolism, the fact that cells have devised these pathways to modulate Aβ production and levels would suggest otherwise. The processing of APP is highly complex in terms of intracellular transport, proteolytic processing, metabolites produced and, importantly, the functional outcomes that follow. If the metabolites sAPPα, sAPPβ, APP-NTFs, and AICD are accepted as functional molecules, then why should not Aβ? The early studies on Aβ attributed a neurotrophic activity to Aβ at low concentrations (Yankner et al. 1990). Other activities ascribed to Aβ include regulation of cholesterol transport (Igbavboa et al. 2009) and kinase activity (Tabaton et al. 2010) and acting as an anti-microbial (Soscia et al. 2010) and metal homeostasis (Bush 2013).
Uncovering this interplay between AICD and Aβ has provided some valuable lessons in the deciphering the function of APP. In particular, we need to consider APP function in the context of specific isoforms and cell types, rather than any APP species being active in any cell. As we have seen this is a dangerous assumption to make and it highlights the need to explore more thoroughly the actions of APP and its metabolites based on isoform and cell type. Moreover, in defining the functions of APP we need to remember that APP is part of gene family with the orthologues Amyloid Precursor Like Protein 1 and Amyloid Precursor Like Protein 2. Since the orthologues have both redundant and unique activities compared to APP, they will impact on the actions of APP. Ultimately, a proper and thorough description of the function of APP must be a study into the entire APP family. The tale provided by APP's tail is proving to be a fascinating read.