Evaluation of cathepsins D and G and EC 3.4.24.15 as candidate beta-secretase proteases using peptide and amyloid precursor protein substrates.

No single protease has emerged that possesses all the expected properties for beta-secretase, including brain localization, appropriate peptide cleavage specificity, and the ability to cleave amyloid precursor protein exactly at the amino-terminus of beta-amyloid peptide. We have isolated and purified a brain-derived activity that cleaves the synthetic peptide substrate SEVKMDAEF between methionine and aspartate residues, as required to generate the amino-terminus of beta-amyloid peptide. Its molecular size of 55-60 kDa and inhibitory profile indicate that we have purified the metalloprotease EC 3.4.24.15. We have compared the sequence specificity of EC 3.4.24.15, cathepsin D, and cathepsin G for their ability to cleave the model peptide SEVKMDAEF or related peptides that contain substitutions reported to modulate beta-amyloid peptide production. We have also tested the ability of these enzymes to form carboxyl-terminal fragments from full-length, membrane-embedded amyloid precursor protein substrate or amyloid precursor protein that contains the Swedish KM --> NL mutation. The correct cleavage was tested with an antibody specific for the free amino-terminus of beta-amyloid peptide. Our results exclude EC 3.4.24.15 as a candidate beta-secretase. Although cathepsin G cleaves the model peptide correctly, it displays poor ability to cleave the Swedish KM --> NL peptide and does not generate carboxy-terminal fragments that are immunoreactive with amino-terminal-specific antiserum. Cathepsin D does not cleave the model peptide or show specificity for wild-type amyloid precursor protein; however, it cleaves the Swedish "NL peptide" and "NL precursor" substrates appropriately. Our results suggest that cathepsin D could act as beta-secretase in the Swedish type of familial Alzheimer's disease and demonstrate the importance of using full-length substrate to verify the sequence specificity of candidate proteases.