APP processing and amyloid deposition in mice haplo-insufficient for presenilin 1

Abstract

More than 70 different mutations in presenilin 1 (PS1) have been associated with inherited early onset Alzheimer’s disease (AD). How all these different mutations cause disease has not been clearly delineated. Our laboratory has previously shown that co-expression of mutant PS1 in mice transgenic for amyloid precursor protein (APPswe) dramatically accelerates the rate of amyloid deposition in the brain. In our original animals mutant PS1 was substantially over-expressed, and the stabilized pool of mouse PS1 fragments was largely replaced by the human protein. In this setting the accelerated amyloid pathology in the double transgenic mice could have been due, in part, to decreased endogenous PS1 activity. To investigate this possibility, we generated APP transgenic mice with reduced levels of endogenous PS1. We find that mice harboring only one functional PS1 allele and co-expressing Mo/HuAPPswe do not develop amyloid deposits at ages comparable to mice expressing mutant PS1. We next tested whether hypo-expression of mutant PS1 could accelerate the rate of amyloid deposition using an unusual line of transgenic mice expressing PS1dE9 at low levels, finding no significant acceleration. Our findings demonstrate that the accelerated amyloid pathology, caused by so many different mutations in PS1, is clearly not a result of haplo-insufficiency that might result from inactivating mutations. Instead, our data are consistent with a gain of property mechanism.

Introduction

More than 70 different mutations in presenilin 1 (PS1), and its homologue PS2, can cause early onset familial Alzheimer’s disease (FAD) [30] (for review see [14]). Mutations occur within each of the eight transmembrane domains, at junctions of membrane and loop domains, and within a large intracellular loop domain [14]. How all these mutations act to cause disease is an area of extensive research and ongoing controversy. It is easier to imagine that a common feature of such a large number, and so scattered a distribution, of changes would be loss of function than it is to envision the diverse mutations all resulting in the same pathogenic gain of property. Consistent with this notion, were studies in which the egg-laying defects in Caenorhabditis elegans lacking PS1 (called sel-12) were rescued by FAD variants of PS1. Wild type human PS1 could completely reverse the developmental defects, whereas PS1 harboring familial AD mutations resulted in only partial rescue [1]. However, similar rescue experiments conducted in PS1 null mice point to the opposite conclusion because the transgenic expression of either wild-type or mutant PS1 resulted in complete rescue of embryonic lethality in mice homozygous for a targeted endogenous PS1 allele [9], [27]. The mouse experiments argue that in mammals, mutant human PS1 retains significant normal function. Moreover, to date only missense or in-frame deletion mutations in either PS1 or PS2 have been associated with FAD; no premature termination or frameshift-mutations have been identified. Collectively, these latter studies are consistent with a gain-of-property mechanism to explain the association of mutant PS1 with FAD.

Studies in both mice and cultured cells suggest that one of the gained properties of mutant human PS1 is to alter APP processing and amyloid deposition. In transgenic mice and human mutant PS1 gene carriers, mutations associated with FAD affect the processing of APP by γ-secretase, causing a shift in the ratio of Aβ40:42 to favor the production Aβ42 [5], [8], [11], [29], [37]. Moreover, more than 25 PS1 mutations have been tested in cell culture; compared to wild-type presenilin all elevate the ratio of Aβ42 relative to Aβ40 [24], [25].

Still, how so many different mutations cause the same shift in APP processing remains unclear. Studies of presenilin biology in both vertebrates and invertebrates suggest that PS1 and PS2 compete for inclusion into a larger protein complex that contains nicastrin, Aph-1, and PEN-2 [7], [12], [17], [20], [22], [32], [35], [43]. This complex, now designated the γ-secretase complex, participates in the proteolytic processing of APP to generate Aβ peptides which deposit in senile plaques characteristic of AD [10], [16]. Thus, the current thinking is that γ-secretase complexes that contain mutant PS1 acquire a greater preference for cleaving APP at the C-terminus of Aβ42.

However, the unique biology of PS1 makes it difficult to rule out the possibility that the net effect of expressing human PS1–FAD variants in transgenic mice or cultured cells is to decrease PS1 function. As mentioned above, PS1 and PS2 compete for inclusion in a multi-protein γ-secretase complex. Moreover, transgene-derived mutant human PS1 competes with endogenous PS1 for inclusion in this complex. When human PS1 is hyper-expressed in mice, the endogenous mouse protein is replaced by the human variant [35]. Thus, in any transgenic study with mutant PS1, a net reduction of endogenous PS1 function will occur. It could be argued that FAD variants of PS1 are less functional than wild-type, and that when mutant transgenes are expressed in mice they diminish the level, and therefore the function, of the endogenous protein.

We tested genetically whether partial loss of PS1 function could account for the accelerated amyloid deposition caused by co-expression of mutant human PS1 in APPswe transgenic mice. A targeted deletion of the endogenous PS1 gene was introduced into APPswe transgenic mice, to determine if PS1 haplo-insufficiency would generate the same early onset pathology as over-expression of FAD–PS1 variants. We find that lowering endogenous PS1 by 50% does not accelerate amyloid deposition. We also studied a line of mice that expresses a very low level mutant human PS1 (HuPS1dE9) to explore how the dose of mutant protein influences amyloid deposition. Our results are consistent with the hypothesis that FAD-associated mutations in PS1 alter, rather than decrease, its function.