9 June 2005. Presenilins make up the catalytic core of the γ-secretase protease complex, and are infamous for their role in processing the amyloid precursor protein to liberate toxic amyloid-β (Aβ) that contributes to Alzheimer disease (AD) pathology. The two mammalian presenilin genes (PS1 and PS2) have long been thought to be interchangeable when it comes to causing AD, because mutations in either can shift γ-secretase activity from production of Aβ1-40 (Aβ40) in favor of the more pathogenic Aβ1-42 (Aβ42). But when it comes to wild-type presenilin, PS1 and PS2 do not cover for each other, according to new research from David Westaway and colleagues at the University of Toronto.
Their study, slated for publication in the early edition of PNAS online, shows that a wild-type PS2 gene fails to rescue Aβ production in the CNS, or Notch signaling, for that matter, in PS1-deficient mice. Also, in the May 27 Journal of Biological Chemistry online, Jochen Walter and colleagues in Bonn, Germany, report that Aβ production can be attenuated in an entirely novel manner. By modifying lipid metabolism, they have been able to prevent the normal maturation of Aβ precursor protein (AβPP) and so attenuate Aβ production.
The demonstration that the two presenilin alleles are not functionally redundant contrasts with results from genetic reconstitution experiments that used cultured cells and blastocysts (see, for example, Kimberly et al., 2000). But in a surprising twist, the authors found that PS2 alleles bearing familial AD mutations gained the ability to compensate for PS1, increasing both Aβ42 production and Notch signaling. The authors suggest that cell context may be important for determining the substrate and cleavage site preference of γ-secretase in vivo.
To test the roles of PS1 and PS2 in vivo, first author Peter Mastrangelo and colleagues turned to genetic complementation, adding back both wild-type and mutant PS alleles to PS-deficient mice. Differences between the two alleles became apparent early on when the researchers generated transgenic mice carrying either wild-type or mutant forms of human presenilins. Expression of wild-type PS2 in a normal mouse background caused a decrease in CNS Aβ peptide levels, indicating that the allele might not be as efficient at APP cleavage as is PS1.
More differences showed up when the transgenic mice were bred with PS1-deficient mice. These crosses showed that wild-type PS2 failed to restore CNS Aβ production or rescue skeletal abnormalities arising from failed Notch signaling in PS1 hypomorphic mice. PS2 also could not reverse the accumulation of APP C-terminal fragments that occurs in the PS1-deficient mice, indicating that the PS2 was not efficient at cleaving the APP fragment. Unlike the hypomorphs, knockout mice that are truly null for PS1 die early in embryonic life, and wild-type PS2 did not rescue those either.
Introducing AD-causing mutations into PS2, however, changed the picture. Two different familial AD alleles were able to increase brain Aβ42 and Notch pathway function in PS1 hypomorphs, and supported survival of the PS1 homozygous knockout mice. Unlike PS1, though, neither wild-type nor mutant PS2 supported production of Aβ40, even though two-dimensional gel electrophoreses showed that transgenic PS1 and PS2 associated with the other γ-secretase subunits and were incorporated into high-molecular-weight structures. So the differences in PS protein function were not due to failure to form γ-secretase complexes.
In contrast to previous in vitro data, these in vivo results suggest that PS2 cannot process APP efficiently or participate in the Notch pathway unless it carries FAD mutations. Why the in vivo and in vitro observations differ is unclear, but the authors speculate that extrinsic components like tissue-specific accessory proteins, in combination with the core complex, may determine the preferred substrates for different γ-secretase complexes containing wild-type or mutant presenilins; the idea that there might be different γ-secretase has been about for some time (see ARF related news story).
The association of Aβ42 production and rescue of Notch signaling, observed with two different mutant PS2 alleles and PS1, as well, suggests that developing γ-secretase inhibitors that prevent Aβ production without attenuating the Notch pathway may not be an easy task (see ARF related news story). But the approach reported by Walter and colleagues provides a potential way around that problem. First author Tamboli et al. show that inhibition of glycosphingolipid biosynthesis prevents the maturation and cell surface transport of AβPP and subsequent secretion of Aβ peptides. They propose that enzymes involved in glycosphingolipid metabolism could be targets to inhibit production of Aβ.—Pat McCaffrey.
Mastrangelo P, Mathews PM, Azhar C, Schmidt SD, Gu Y, Yang J, Mazzella MJ, Coomaraswamy J, Horne P, Strome B, Pelly H, Levesque G, Ebeling C, Jiang Y, Nixon RA, Rozmahel R, Fraser P, St George-Hyslop P, Carlson GA, Westaway D. Dissociated phenotypes in presenilin transgenic mice define functionally distinct γ-secretases. PNAS Early Edition. May 30, 2005. Abstract
Tamboli IY, Prager K, Barth E, Heneka M, Sandhoff K, Walter J. Inhibition of glycosphingolipid biosynthesis reduces secretion of the β amyloid precursor protein and amyloid β peptide. J Biol Chem. 2005 May 27; [Epub ahead of print] Abstract