AD (Alzheimer's disease) is a neurodegenerative disease characterized by a gradual loss of neurons and the accumulation of neurotoxic Aβ (amyloid β-peptide) and hyperphosphorylated tau. The discovery of mutations in three genes, PSEN1 (presenilin 1), PSEN2 (presenilin 2) and APP (amyloid precursor protein), in patients with FAD (familial AD) has made an important contribution towards an understanding of the disease aetiology; however, a complete molecular mechanism is still lacking. Both presenilins belong to the γ-secretase complex, and serve as the catalytic entity needed for the final cleavage of APP into Aβ. PSEN only functions within the γ-secretase complex through intra- and inter-molecular interactions with three other membrane components, including nicastrin, Aph-1 (anterior pharynx defective-1) and Pen-2 (PSEN enhancer-2). However, although the list of γ-secretase substrates is still expanding, other non-catalytic activities of presenilins are also increasing the complexity behind its molecular contribution towards AD. These γ-secretase-independent roles are so far mainly attributed to PSEN1, including the transport of membrane proteins, cell adhesion, ER (endoplasmic reticulum) Ca2+ regulation and cell signalling. In the present minireview, we discuss the current understanding of the γ-secretase-independent roles of PSENs and their possible implications in respect of AD.
- Alzheimer's disease
- calcium homoeostasis
- endo/lysosomal dysfunction
- familial Alzheimer's disease (FAD)
Protein transport and degradation
Abnormalities in neuronal endocytic pathways, as is evident by the accumulation of degradative and/or autophagic vacuole-like structures, has been shown to be one of the earliest neuropathological hallmarks in sporadic AD (Alzheimer's disease) patients . Although PSENs (presenilins) have been found to be involved in the maturation and post-Golgi trafficking/turnover of several membrane proteins [2,3], there is still a lot of speculation on the exact molecular mechanism.
Among the missorted proteins, we find several cell-surface receptors, including ICAM-5 (intercellular adhesion molecule-5, formerly known as telencephalin) , EGFR (epidermal growth factor receptor)  and β1 integrins . Surprisingly, these proteins do not serve as substrates for γ-secretase [4–6], underscoring the involvement of PSENs in protein trafficking irrespective of its catalytic activity. The defects in trafficking seem to be mainly in the final stages of protein degradation, resulting in the respective accumulation of full-length ICAM-5 , α- and β-synuclein  and EGFR  in intracellular compartments. In PSEN1−/− neurons or PSEN1/PSEN2−/− cells, ICAM-5 and α- and β-synuclein aggregates respectively were localized to enlarged autophagic vacuoles before their fusion with the endosomal/lysosomal system [4,7]. The abnormal trafficking of ICAM-5 in PSEN1−/− hippocampal neurons was rescued upon expression of WT (wild-type) and the catalytic dead (aspartate) mutant of PSEN1 . EGFR follows classical routes for internalization, but in PSEN1/PSEN2−/− cells, the defective fusion between endosomes and lysosomes results in delayed or blocked EGFR degradation . Increased EGFR levels in PSEN1−/− neurons were also observed by Zhang et al. ; however, they showed that EGFR up-regulation occurred at the mRNA level via AICD [APP (amyloid precursor protein) intracellular domain] production, thus in a γ-secretase-dependent manner . So far, the observations in PSEN-deficient cells or neurons point to a delay in, or an impaired capability of, late endosomal or degradative organelles in general to fuse with lysosomes. Although one can only speculate on the molecular machinery involved in these trafficking defects, some clues can be found from PSEN-interacting proteins reported so far.
For instance, PSENs appear to interact with several members and/or effectors of the Rab family of small GTPases. Many of these Rab proteins are the main regulators for proper endocytic and recycling pathways, sorting cargo to a range of subcellular compartments . In an FAD (familial AD)-associated mutant, PSEN1A260V, lowered Aβ (amyloid β-peptide) production and increased APP CTF (C-terminal fragment) levels are observed concomitantly with decreased expression of Rab8, a GTPase that mediates selective transport routes from the TGN (trans-Golgi network)/endosome to the cell surface. This indicates that certain PSEN1 mutations may perturb membrane vesicle transport, which indirectly affects APP processing . Rab11, which is mainly involved at the level of the recycling endosomes , may interact with both PSEN1 and PSEN2 . However, caution is required with such studies because the reported interaction domain (amino acids 374–400) is in conflict with the most recent updates on PSEN topology  as this region majorly corresponds to the transmembrane domain 7. Finally, PSEN1 binds to RabGDI (Rab GDP-dissociation inhibitor), which chaperones Rab-GTPases in the cytosol and targets them to their cognate membranes . This interaction can influence proper membrane recruitment of Rab6, a regulator of specific retrograde transport routes from the cell surface to the Golgi and ER (endoplasmic reticulum) . Although impaired fusion events with lysosomes are closely related to PSEN deficiencies, no Rab GTPases typically associated with late endosomes have been reported to interact with PSENs.
Alternatively, we can speculate that PSENs may interact or mediate the location/function of critical factors involved in endosomal acidification or Ca2+ storage as both are required for proper fusion events. A possible candidate might be the AQP1 (aquaporin 1) gene. AQPs are membrane proteins that serve in the transfer of water and small solutes across cellular membranes . In this respect, the Drosophila homologue Bib (Big Brain) has been shown to mediate endosome acidification and maturation, resulting in aberrant accumulations of internalized Notch and its ligand Delta independent of the γ-secretase complex . AQP1 expression was shown to be significantly increased, not only in Creutzfeldt–Jacob disease, but also in early and late stages of sporadic AD . Furthermore, AQP1 mRNA and protein levels were down-regulated in PSEN2−/− MEFs (mouse embryonic fibroblasts) [18,19]. Here, the authors argue for a PSEN2-dependent processing of APP CTFs as a major factor that affects the epigenetic control of AQP1 expression. This indicates once more that the γ-secretase-dependent and -independent roles of PSENs in endosomes remain difficult to discern, emphasizing the search and need for novel cellular models and organisms.
One such promising model organism is the early land plant Physcomitrella patens (a moss), which expresses PSEN, but lacks main substrates of γ-secretase, such as APP and Notch. Loss of PSENs in this model organism results in abnormally long and straight cytoskeletal filaments, and chloroplast redistribution problems in response to light. These defects may originate from a failure in cytoskeleton-driven vesicle/organelle transport. The ability of both WT and aspartate mutant PSEN1 to rescue P. patens PSEN knockdown suggests that the observed phenotypes are independent of any proteolytic activity and provide unbiased proof of an evolutionarily conserved function of PSEN1 in cytoskeletal organization and vesicular transport .
The observed transport deficiencies could also be caused by PSEN1 and/or PSEN2 affecting single or selective endocytic sorting routes, thereby jeopardizing the overall balance in the regulation of degradation against recycling. For instance, caveolin-mediated endocytosis is clearly compromised in PSEN1/PSEN2−/− blastocysts with an apparent lack of caveolae at the cell surface and a concomitant intracellular retainment of caveolin 1, the main structural component of caveolae . Interestingly, caveolin 1 regulates extracellular matrix remodelling via β1 integrin endocytosis , one of the aforementioned missorted membrane proteins, thereby linking PSEN function directly or indirectly not only to caveolin-dependent endocytosis, but also to cell migration and adhesion .
Cell adhesion and cell signalling
Several proteins, missorted due to PSEN deficiency have important functions in cell migration and/or cell adhesion [4,6]. We cannot exclude at this moment that the missorting might also result indirectly from γ-secretase-dependent effects on cell adhesion. In PSEN1/PSEN2−/− MEFs, focal adhesion site maturation is reduced via dephosphorylation of different scaffolding proteins. Importantly, and although no data are available on the aspartate mutant PSENs, most defects could not be mimicked by γ-secretase inhibitors , indicative of possible γ-secretase-independent effects. Another example is the well-documented relationship of PSEN1 with β-catenin, a member of the Armadillo family , which mediates cadherin-based cell–cell adhesion  and influences Wnt signalling . In this process, PSENs are negative regulators for β-catenin function, affecting its stabilization and degradation . However, although initial data supported the view that PSEN1 exerted this role in a γ-secretase-independent way through interacting directly with β-catenin, more recent studies suggest an indirect effect, controlled through the interaction and processing of cadherins [29,30].
As mentioned above, loss of PSEN1 and PSEN2 leads to enhanced maturation and cell-surface delivery of mature β1 integrins, a focal adhesion receptor, resulting in increased cell adhesion properties. PSEN1 was suggested to exert an inhibitory effect on the post-translational maturation of β1 integrin in the ER. As PSEN1 aspartate mutants could not restore proper β1 integrin maturation, and γ-secretase inhibitors were unable to mimic the PSEN1/PSEN2 phenotype , these defects can be attributed to γ-secretase-independent roles of PSEN1. Besides β1 integrin, ICAM-5 is also a neuronal cell adhesion molecule belonging to the ICAM family of intercellular adhesion proteins. It is uniquely expressed in neurons of the telencephalon, where it is confined to the somatodendritic plasma membranes. Functionally, ICAM-5 is seen as a negative regulator of dendritic spine maturation, implying a role in synaptogenesis and the establishment of functional neural circuitry in the developing brain [31,32]. In this respect, the delayed turnover of ICAM-5 in PSEN1−/− neurons, concomitantly with its intracellular localization [4,33], may distort this critical balance in spine formation. Indeed, in accordance with increased cell-surface expression of ICAM5, PSEN1−/− primary neurons present an increased number of filopodia and spine-like protrusions . Strangely, they also develop more extensive neurite outgrowth and higher spine densities, resulting in increased synaptic transmission , suggesting that other factors affected by PSEN1 deficiency contribute to the imbalance. For instance, δ-catenin, a known interactor of PSEN1, promotes dendritic arborization and spine maturation through its interaction with p190RhoGEF (p190 Rho guanine-nucleotide-exchange factor) . Alternatively, γ-secretase-dependent processing and downstream signalling of substrate proteins, such as DCC (deleted in colorectal cancer), cannot be excluded from affecting the neuronal circuit in PSEN−/− neurons. Indeed, the DCC CTFs accumulating at the membrane can affect the cAMP signalling cascade, thereby strengthening axon–dendritic connectivity and synaptic transmission .
PSENs are essential for the maintenance and function of cortical neurons, and also both PSEN1 and PSEN2 genes are important factors in neurodegeneration . PSEN2 signals via the PDGFR (platelet-derived growth factor receptor) to the MAPK (mitogen-activated protein kinase) signalling cascade, resulting in much lower ERK1/2 (extracellular-signal-regulated kinase 1/2) phosphorylation in PSEN2−/− and PSEN1/PSEN2−/− cells. ERK activation connects extracellular signalling to cell growth, proliferation, survival and differentiation. PSEN1, on the other hand, couples to Akt signalling via different cell-surface receptors, including TrkB (tropomyosin receptor kinase B) and cadherin . Although many signalling pathways that affect cell survival are misregulated upon PSEN loss, this does not result in a change in basal neuronal survival. Nonetheless, cell survival responses under high oxidative stress conditions are decreased via the p53 pathway in a γ-secretase-dependent and -independent manner . On the other hand, PSEN1 also promotes cell survival via PI3K (phosphoinositide 3-kinase)/Akt activation, whereas PSEN1 FAD mutants reduce the Akt pro-survival signalling .
Ca2+ release from internal stores
Defective Ca2+ signalling has been documented for both PSEN1 and PSEN2, but also for multiple PSEN FAD mutations in many different experimental systems . The first description of Ca2+ defects in fibroblasts derived from AD patients date even from before the discovery of the PSEN genes. Ito et al.  described greatly enhanced InsP3-mediated Ca2+ release in skin fibroblasts from a family of chromosome-14-linked FAD patients, which were later shown to harbour the PSEN1A246Q mutation. PSEN-mediated abnormalities in Ca2+ signalling have so far only been linked to ER Ca2+ homoeostasis (reviewed in [41,43]), obviously because the ER is the major Ca2+-storage compartment and a major part of the endogenous PSEN1 resides here [44,45]. On the other hand, the available studies often provided opposing theories on functional implications of PSENs in the mechanisms of ER Ca2+ homoeostasis.
In a multitude of organisms and cell types, PSEN1, PSEN2 and FAD-linked mutants are associated with exaggerated Ca2+ responses to agonists for both IP3Rs (InsP3 receptors) and RyRs (ryanodine receptors) [46,47]. Overexpression of PSEN1 WT and FAD mutants resulted in higher expression levels of RyR type 3, giving rise to increased Ca2+ release from caffeine-sensitive stores in cortical neurons and PC12 cells . Also InsP3-induced ER Ca2+ release has been shown frequently, but the molecular mechanism is more complicated and dependent on the experimental conditions used. First, PSEN1 and PSEN2 cause similar alterations in InsP3-mediated Ca2+ signalling, without changes in expression levels . Kasri et al.  demonstrated enhanced expression levels of the IP3R type 1 in PSEN-deficient fibroblasts, and the specific knockdown of this receptor caused a decrease in the passive Ca2+ leak from the ER, restoring normal ER Ca2+ levels. In contrast, in another study using the same cell lines, no alterations in the expression levels of IP3R type 1 were noticed . Moreover, others demonstrated an interaction of PSENs with both IP3R type 1 and type 3 in Sf9 insect cells overexpressing PSEN1 or PSEN2, again without changing expression levels of the IP3Rs. Here, the increased Ca2+ leakage from the ER is explained by an excessive channel gating activity . Yet another study points to affected SERCA (sarcoplasmic/endoplasmic reticulum Ca2+-ATPase) pumps as being the culprit in altered Ca2+ homoeostasis in PSEN−/− MEFs . In this case, PSENs may associate with SERCA2b, regulating its proper function, and thus contributing to the active filling mechanism of the ER. Overload of ER Ca2+ stores leads to enhanced Ca2+ release after cell stimulation, resulting in a higher susceptibility to apoptosis [39,46].
Finally, PSENs themselves were proposed to function as Ca2+ leak channels in the ER . Although appealing, this function could only be attributed to full-length unprocessed PSEN1. However, the holoprotein is short-lived and becomes stabilized as an N- and C-terminal heterodimer shortly after biosynthesis in the ER [53,54]. Since, in most cells, endogenous holoprotein levels are barely detectable, it remains to be investigated to what extent such endogenous levels can contribute to passive Ca2+ leakage. Also, how this function reconciles with several reports of PSEN-mediated InsP3-induced Ca2+ release remains a question of debate. Hayrapetyan et al.  argued that PSEN1 and PSEN2 NTFs (N-terminal fragments) can interact directly with RyRs, resulting in hypoxia-induced Ca2+ responses. They could show that Ca2+ currents are evoked by PSEN2 NTF overexpression in RyR-containing mouse brain microsomes . An identical method was used to support a role for PSENs as low-conductance cation channels , but currents shown here were much smaller compared with single-channel responses mediated by RyR  or IP3R .
So far, Ca2+ disruptions have been related to all major ER Ca2+ channels/pumps, including IP3R , RyR , SERCA pumps , or even the occurrence of PSENs as low-conductance ion channels [50,56], resulting in an overload of ER Ca2+ stores [47,50] or the opposite [51,52]. It cannot be excluded, however, that the basis of such conflicting data might originate from different methodological approaches. In most studies, data were mainly generated by inducing the release of Ca2+ from intracellular stores (ER and Golgi) via different ligands, and increases in cytosolic Ca2+ concentrations were measured with the ratiometric Ca2+ sensor Fura-2 [47,50–52]. Alternative strategies use recombinant ER-targeted aequorins that allow direct measurement of Ca2+ in the ER. In such studies, it was shown, for instance, that ER Ca2+ concentrations by themselves were clearly decreased in PSEN-deficient MEFs  as well as in different cell types overexpressing PSEN1, PSEN2 and FAD mutants [51,57]. Although a full mechanistic insight is still lacking, it is clear that PSENs play some roles in this process and that a failure to properly control intracellular Ca2+ responses during signal transduction has major implications in neuronal function and neurodegeneration .
In general, it remains extremely difficult to dissociate the different γ-secretase-dependent and -independent functions of PSENs. Hence the physiological outcome in PSEN-deficient cells/neurons is likely to be the result of both. PSENs may thus function as a focal point influencing a diverse array of signalling molecules and protein trafficking routes that can contribute to AD pathogenesis. Perturbations in these pathways may therefore be even upfront in the aetiology of AD, further potentiating the harmful effects of Aβ42 accumulation. In order to decipher the contributions of γ-secretase-dependent/independent functions to AD, one should explore PSEN function in more detail in cellular models such as the moss plant P. patens. Alternatively, (high-throughput) screens in PSEN-deficient cells aiming to rescue the observed transport deficiencies, but not γ-secretase activity, might identify novel components and pathways scrutinizing its γ-secretase-independent role(s).
This work is supported by Methusalem/KULeuven and VIB, the Federal Government [grant number IAP-P6/43], the Alzheimer's Association [grant number IIRG-08–91535], the SAO/FRMA (Stichting voor Alzheimer Onderzoek/Fondation pour la Recherche sur la Maladie d'Alzheimer) grant cycle 2008, and the FWO (Fonds voor Wetenschappelijk Onderzoek) grant numbers G.0663.09 and G.0.754.10.N]. K.C. is holder of a Ph.D. fellowship of the IWT-Vlaanderen.
Lysosomes in Health and Disease: A Biochemical Society Focused Meeting held at Charles Darwin House, London, U.K., 13–14 May 2010. Organized and Edited by Frances Platt (Oxford, U.K.) and Paul Pryor (York, U.K.).
Abbreviations: Aβ, amyloid β-peptide; AD, Alzheimer's disease; APP, amyloid precursor protein; AQP, aquaporin; CTF, C-terminal fragment; DCC, deleted in colorectal cancer; EGFR, epidermal growth factor receptor; ER, endoplasmic reticulum; ERK, extracellular-signal-regulated kinase; FAD, familial Alzheimer's disease; ICAM-5, intercellular adhesion molecule-5 (telencephalin); IP3R, InsP3 receptor; MEF, mouse embryonic fibroblast; NTF, N-terminal fragment; PSEN, presenilin; RyR, ryanodine receptor; SERCA, sarcoplasmic/endoplasmic reticulum Ca2+-ATPase; WT, wild-type
- © The Authors Journal compilation © 2010 Biochemical Society