06 Mar 2002

The identity of γ-secretase, the aspartyl protease responsible for snipping amyloid-β (Aβ) from β-secretase-processed amyloid precursor protein (AβPP), has remained elusive for some time. If pushed, most experts in the field would probably agree that presenilin-1 (PS1) is the most likely candidate, but the fact that γ-secretase is found in a large multiprotein complex has made this association difficult to prove. Researchers with Dennis Selkoe and Michael Wolfe at Brigham and Women's Hospital, Boston, have now narrowed the field of candidates to perhaps two proteins. Advance publication of their work, which appears in tomorrow's PNAS, is available online.

The researchers attempted to isolate γ-secretase by affinity-based purification using an enzyme inhibitor that binds tightly to the active site of the protease. When immobilized on a solid support, this analog was capable of retaining both N-terminal fragments (NTF) and C-terminal fragments (CTF) of presenilin 1 from a solubilized γ-secretase preparation. In addition, the inhibitor matrix quantitatively bound nicastrin, previously shown to be associated with the protease complex, suggesting that this protein may be an essential γ-secretase component. This idea was supported by the fact that treatments that inactivate γ-secretase, such as detergents or heat, abolish binding of nicastrin or PS heterodimers to the protease inhibitor.

In contrast, other proteins that have been suspected of playing a role in γ-secretase catalysis failed to bind to the affinity matrix. These included β-catenin, presenilin-associated protein, and calsenilin. Curiously, when the substrate analog was used to purify γ-secretase components, C83, the part of APP that serves as a substrate for the protease, co-purified with PS1 fragments and nicastrin. The authors suggest that this indicates a two-step mechanism for the interaction of protease and substrate; an initial binding event on the surface, followed by entry to the proteolytic site.

"Both presenilin fragments and nicastrin track quite nicely with the activity," says Wolfe, "and you can immunoprecipitate nicastrin and bring down presenilin and γ-secretase activity. So the data is certainly consistent with the idea that these proteins are part of the active complex. Although to prove it, one would need to purify the components to homogeneity and see if that's all there is, or put the [three components] together and see if you have activity."—Tom Fagan