The multiple paradoxes of presenilins

One of the main histological hallmarks observed in the cortex of Alzheimer's disease-affected patients is the senile plaque, a proteinaceous deposit that derives mainly from the overproduction of a 39±43 amino-acid-containing peptide called b-amyloid (Ab). The most striking evidence that Ab overproduction is intimately linked to Alzheimer's disease pathology comes from observations that all but one (Ancolio et al. 1999) of the pathological mutations located on the genes responsible for early onset aggressive forms of the disease, lead to an increase in Ab production and, particularly, in the production of the more readily aggregable 42-amino-acid-long species (Van Broeckhoven 1995; Hutton and Hardy 1997). Whether Ab is, per se, the etiological cause of the disease, or whether it behaves as an intermediate effector of a pathological cascade of events is still a matter of discussion. However, whatever the case, it remains clear that Ab derives from its precursor, the b-precursor protein, by enzymatic digestion by so-called band g-secretases, two proteolytic activities responsible for the generation of the Nand C-termini of Ab, respectively (Selkoe 1991; Checler 1995). As the inhibition of these two activities is theoretically the easiest possibility of slowing down or blockading Ab-mediated Alzheimer neuropathology, several teams have targeted the band g-secretases for identi®cation and characterization. It is now generally believed that an unusual membrane-bound aspartyl protease (BACE; b-site APP cleaving enzyme, or memapsin2 or Asp2) is a b-secretase (Hussain et al. 1999; Sinha et al. 1999; Vassar et al. 1999; Yan et al. 1999; Lin et al. 2000). Concerning the elusive g-secretase, several groups suggest that presenilins could act as the genuine g-secretase, a hypothesis that is disputed by others. Here, we will present some objective experimental and theoretical arguments that lead us to argue against a direct role of presenilins as proteases. Presenilins 1 and 2 are homologous proteins that most likely play a central role in Alzheimer's pathology. First, when mutated, these proteins are associated with early onset forms of the disease and, indeed, appear responsible for most of the autosomal dominant familial cases of the disease (Van Broeckhoven 1995; Hutton and Hardy 1997). Similar to all but one of the mutations on bAPP, mutations on presenilins trigger the increased production of Ab42, and speci®cally increase the Ab42 : total-Ab ratio (Checler 1999a). It is easy to envisage that mutations in bAPP, all located near the g-secretase-targeted sequence, can introduce modi®cations of the recognition/catalytic parameters of g-secretase for bAPP. Therefore, this may explain the phenotypical alteration of bAPP maturation at the g-secretase site. At this stage, it is more dif®cult to conceive how mutations in presenilins, which appear widely distributed within the serpentine sequence of presenilins, i.e. in putative transmembrane domains as well as in the hydrophilic domains, could lead to the same selective increase in production of Ab42. Before there was any de®nite explanation for these observations, this ®nding alone indicated that there was likely to be a control of g-secretase-mediated bAPP maturation, mediated either directly or indirectly, by presenilins. An important breakthrough in the elucidation of this bAPP/presenilins link came from the deletion of presenilin genes. Bart De Strooper and colleagues elegantly demonstrated that the knock-out of the presenilin 1 gene led to drastic reduction in the g-secretase-mediated cleavage of bAPP (De Strooper et al. 1998). The residual secretion of Ab40/42 in presenilin 1 knock-out cells appeared likely to be caused by the contribution of endogenous presenilin 2, as the double knock-out of both presenilins 1 and 2 fully abolished Ab secretion (Herreman et al. 2000; Zhang et al. 2000b). These observations, however, did not elucidate the mechanism by which presenilins in uence g-secretasemediated bAPP cleavage.