Alzheimer's disease

A technical KO of amyloid-β peptide

Article metrics

Research into the molecular mechanisms of Alzheimer's disease — the most common late-life dementia — continues to illuminate intriguing issues in protein biology, while seeking to identify therapeutic targets. These twin outcomes are exemplified by the exciting results of De Strooper and co-workers1, reported on page 387 of this issue. The authors show that the presenilin-1 (PS1) gene, mutations in which cause an aggressive form of early-onset familial Alzheimer's disease2,3, regulates the unusual intramembranous proteolysis of the amyloid precursor protein (APP). PS1 might, therefore, serve as an unexpected target for drugs against the disease.

In families with autosomal dominant Alzheimer's disease, mutations in three different genes — APP, PS1 and PS2 — have been identified2,3. Missense mutations in the APP gene are located at, or close to, the recognition sites of the APP-cleaving secretases (α, β and γ; Fig. 1). These mutations lead to increased production of amyloid-β peptide, particularly its amyloidogenic 42-residue form, the amyloid-β(1-42) peptide, which accumulates in the amyloid plaques of both ‘sporadic’ and familial cases of Alzheimer's disease3. Whereas mutations in APP are extremely rare, numerous mutations have been observed in the PS1 and PS2 genes2,3 (which are homologous to each other). All of the presenilin mutations analysed so far increase the levels of secreted amyloid-β(1-42) peptide, even presymptomatically, supporting a central role for amyloid-β peptide in the pathogenesis of Alzheimer's disease2,3.

Figure 1: The influence of presenilin-1 (PS1) on proteolytic processing of the amyloid precursor protein (APP).

β-secretase cleaves APP at the amino terminus of the amyloid-β (Aβ)peptide domain to generate a carboxy (C)-terminal fragment containing the entire amyloid-β peptide. Generation of amyloid-β peptide is decreased by α-secretase, which cleaves in the middle of the amyloid-β peptide domain to create a shorter carboxy-terminal fragment. De Strooper et al.1 have found that both fragments accumulate in the PS1−/− mice. The carboxy-terminal fragments are the substrate for the γ-secretase(s), which liberate amyloid-β peptide from the β-secretase-cleaved fragment, and p3 from the α-secretase-cleaved fragment. (Asterisks indicate the approximate locations of the APP mutations.)

Presenilin proteins span the membrane six to eight times. They are proteolytically processed to produce stable amino- and carboxy-terminal fragments2, which form a heterodimeric complex that seems to be biologically relevant (Fig. 1)4. Work on the nematode Caenorhabditis elegans indicates that presenilins are involved physiologically in cell-fate decisions through the Notch signalling pathway2. In support of this, deletion of the PS1 gene in mice produces a fatal developmental defect that closely resembles the phenotype seen in Notch knockout mice5,6.

De Strooper et al.1 studied mice lacking both PS1 alleles (PS1−/−). Previous work by this group7 showed that, in mouse cells, endogenous APP is poorly processed into amyloid-β peptide, whereas the human APP sequence is efficiently cleaved. To allow generation of amyloid-β peptide in the PS1−/− background to be analysed quantitatively, the authors infected hippocampal neurons cultured from the knockout embryos with a recombinant Semliki Forest virus encoding human APP. Strikingly, these cells produced substantially (80%) less amyloid-β peptide, and the related peptide p3, than did the same construct infected into cells expressing PS1. (The residual amyloid-β peptide/p3 may have resulted from continued expression of the homologous PS2 gene.)

In contrast to the presenilin mutations that are responsible for familial Alzheimer's disease — and specifically increase generation of amyloid-β(1-42) peptide — the lack of PS1 reduced production of the 40- and 42-residue forms of amyloid-β peptide equally (Fig. 1). Carboxy-terminal fragments of APP beginning at the β- and α-secretase sites accumulated to high levels in lysates of the PS1−/− neurons. But, in neurons from normal mice, these fragments were efficiently cleaved by γ-secretase, giving rise to amyloid-β peptide and p3.

These observations could indicate that PS1 is the γ-secretase, although this seems unlikely because the protein has no sequence homology to known proteases. Instead, De Strooper et al.1 discuss the exciting possibility that PS1 activates γ-secretase cleavage of APP, similar to the SREBP-cleavage activating protein (SCAP)8. SCAP facilitates cleavage of the sterol regulatory element binding protein (SREBP) to liberate its transcription-factor domain, which migrates to the nucleus and regulates the expression of genes involved in cholesterol biosynthesis and uptake. Cleavage of SREBP occurs in two steps (Fig. 2). Cleavage 1 cuts the lumenal loop between the regulatory and transcriptional domains, and this cleavage is facilitated by SCAP. Cleavage 2 then liberates the membrane-bound transcription factor. Like the γ-secretase cut of APP, cleavage 2 occurs within the membrane.

Figure 2: Model for the two-site cleavage of the sterol regulatory element binding protein (SREBP).

a, Cleavage 1 cuts the lumenal loop between the regulatory and transcriptional domains. This cleavage is somehow facilitated by the SREBP cleavage activating protein (SCAP), and it occurs in the lumen of the endoplasmic reticulum. b, Cleavage 2 occurs in a trans-membrane domain, liberating the membrane-bound transcription factor (TF). This cleavage is similar to the γ-secretase cleavage of APP.

At present, γ-secretase and the SREBP protease are the only known proteolytic activities that cleave in an environment which, according to textbook knowledge, should preclude proteolysis. There are other remarkable similarities between the two systems. Both SREBP and the presenilins are located in the endoplasmic reticulum2, and the γ-secretase cleavage after amino acid 42 could occur, in part, within the endoplasmic reticulum and Golgi9,10,11. Furthermore, the SCAP proteins are predicted to form six to eight transmembrane domains — a structure similar to that of the presenilins.

Based on these analogies, PS1 might be expected to regulate γ-secretase, and De Strooper et al.1 now prove it. This physiologically normal activity is presumably disturbed by presenilin mutations, leading to increased cleavage of APP carboxy-terminal fragments after amino acid 42 of amyloid-β peptide. There is, however, one big difference between SCAP-activated SREBP proteolysis and presenilin-activated APP cleavage by γ-secretase: SCAP is currently known to activate only cleavage 1 within the lumen of the endoplasmic reticulum8, whereas γ-secretase cleavage of APP occurs within the transmembrane domain.

An alternative molecular explanation for the reduced production of amyloid-β peptide in PS1−/− mice is that PS1 regulates the transport of membrane proteins from the endoplasmic reticulum to other cellular compartments. This explanation could support the observed phenotype of mutations in the C. elegans presenilin homologue, which could be due to altered sorting of membrane proteins within the endoplasmic reticulum. This could lead to reduced transport of proteins such as APP and the Notch receptor through the Golgi and onto the cell surface. In this case, mutant presenilins might retain APP within the endoplasmic reticulum and early Golgi, leading to increased production of amyloid-β(1-42) peptide9,10,11. The lack of PS1 expression would affect targeting not only of APP, but also of other membrane proteins in the endoplasmic reticulum.

Regardless of which molecular mechanism is responsible for the sharp reduction in generation of amyloid-β peptide in PS1−/− mice, PS1 could serve as an unexpected therapeutic target for Alzheimer's disease. Inhibition of presenilin synthesis late during ageing, when the disease normally begins, would avoid elimination of the essential function of presenilin proteins during development5,6. Expression of PS1 could be repressed at the transcriptional level or by interfering with the post-translational processing of presenilin and presenilin complex formation (Fig. 1)4. And if levels of PS1 could be decreased enough in the brain, production of amyloid-β would be decreased and, therefore, accumulation of senile plaques could be retarded.


  1. 1

    De Strooper, al. Nature 391, 387–390 (1998).

  2. 2

    Haass, C. Neuron 18, 687–690 (1997).

  3. 3

    Selkoe, D. J. J. Biol. Chem. 271, 18295–18298 (1996).

  4. 4

    Capell, A. et al. (in the press).

  5. 5

    Wong, P. al. Nature 387, 288–292 (1997).

  6. 6

    Shen, al. Cell 89, 629–639 (1997).

  7. 7

    De Strooper, al. EMBO J. 14, 4932–4938 (1995).

  8. 8

    Brown, M. S. & Goldstein, J. L. Cell 89, 331–340 (1997).

  9. 9

    Hartmann, al. Nature Med. 3, 1016–1020 (1997).

  10. 10

    Cook, D. al. Nature Med. 3, 1021–1023 (1997).

  11. 11

    Wild-Bode, al. J. Biol. Chem. 272, 16085–16088 (1997).

Download references

Author information

Rights and permissions

Reprints and Permissions

About this article

Further reading


By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.