Kinetic and structural characterization of amyloid‐β peptide hydrolysis by human angiotensin‐1‐converting enzyme

Angiotensin‐1‐converting enzyme (ACE), a zinc metallopeptidase, consists of two homologous catalytic domains (N and C) with different substrate specificities. Here we report kinetic parameters of five different forms of human ACE with various amyloid beta (Aβ) substrates together with high resolution crystal structures of the N‐domain in complex with Aβ fragments. For the physiological Aβ(1–16) peptide, a novel ACE cleavage site was found at His14‐Gln15. Furthermore, Aβ(1–16) was preferentially cleaved by the individual N‐domain; however, the presence of an inactive C‐domain in full‐length somatic ACE (sACE) greatly reduced enzyme activity and affected apparent selectivity. Two fluorogenic substrates, Aβ(4–10)Q and Aβ(4–10)Y, underwent endoproteolytic cleavage at the Asp7‐Ser8 bond with all ACE constructs showing greater catalytic efficiency for Aβ(4–10)Y. Surprisingly, in contrast to Aβ(1–16) and Aβ(4–10)Q, sACE showed positive domain cooperativity and the double C‐domain (CC‐sACE) construct no cooperativity towards Aβ(4–10)Y. The structures of the Aβ peptide–ACE complexes revealed a common mode of peptide binding for both domains which principally targets the C‐terminal P2′ position to the S2′ pocket and recognizes the main chain of the P1′ peptide. It is likely that N‐domain selectivity for the amyloid peptide is conferred through the N‐domain specific S2′ residue Thr358. Additionally, the N‐domain can accommodate larger substrates through movement of the N‐terminal helices, as suggested by the disorder of the hinge region in the crystal structures. Our findings are important for the design of domain selective inhibitors as the differences in domain selectivity are more pronounced with the truncated domains compared to the more physiological full‐length forms. DATABASE: The atomic coordinates and structure factors for N‐domain ACE with Aβ peptides 4–10 (5AM8), 10–16 (5AM9), 1–16 (5AMA), 35–42 (5AMB) and (4–10)Y (5AMC) complexes have been deposited in the Protein Data Bank, Research Collaboratory for Structural Bioinformatics, Rutgers University, New Brunswick, NJ, USA (http://www.rcsb.org/).