Memapsin 2 is the protease known as beta-secretase whose action on beta-amyloid precursor protein leads to the production of the beta-amyloid (Abeta) peptide. Since the accumulation of Abeta in the brain is a key event in the pathogenesis of Alzheimer's disease, memapsin 2 is an important target for the design of inhibitory drugs. Here we describe the residue preference for the subsites of memapsin 2. The relative k(cat)/K(M) values of residues in each of the eight subsites were determined by the relative initial cleavage rates of substrate mixtures as quantified by MALDI-TOF mass spectrometry. We found that each subsite can accommodate multiple residues. The S(1) subsite is the most stringent, preferring residues in the order of Leu > Phe > Met > Tyr. The preferences of other subsites are the following: S(2), Asp > Asn > Met; S(3), Ile > Val > Leu; S(4), Glu > Gln > Asp; S(1)', Met > Glu > Gln > Ala; S(2)', Val > Ile > Ala; S(3)', Leu > Trp > Ala; S(4)', Asp > Glu > Trp. In general, S subsites are more specific than the S' subsites. A peptide comprising the eight most favored residues (Glu-Ile-Asp-Leu-Met-Val-Leu-Asp) was found to be hydrolyzed with the highest k(cat)/K(M) value so far observed for memapsin 2. Residue preferences at four subsites were also studied by binding of memapsin 2 to a combinatorial inhibitor library. From 10 tight binding inhibitors, the consensus preferences were as follows: S(2), Asp and Glu; S(3), Leu and Ile; S(2)', Val; and S(3)', Glu and Gln. An inhibitor, OM00-3, Glu-Leu-Asp-LeuAla-Val-Glu-Phe (where the asterisk represents the hydroxyethylene tansition-state isostere), designed from the consensus residues, was found to be the most potent inhibitor of memapsin 2 so far reported (K(i) of 3.1 x 10(-10) M). A molecular model of OM00-3 binding to memapsin 2 revealed critical improvement of the interactions between inhibitor side chains with enzyme over a previous inhibitor, OM99-2 [Ghosh, A. K., et al. (2000) J. Am. Chem. Soc. 14, 3522-3523].