beta-Amyloid (Abeta), the primary protein component of Alzheimer's plaques, is neurotoxic when aggregated into fibrils. We have devised a modular strategy for generating compounds that inhibit Abeta toxicity, based on linking a recognition element for Abeta to a disrupting element designed to interfere with Abeta aggregation. One such compound, with the 15-25 sequence of Abeta as the recognition element and a lysine hexamer as the disrupting element, altered Abeta aggregation kinetics and protected cells from Abeta toxicity [Ghanta et al. (1996) J. Biol. Chem. 271, 29525]. To optimize the recognition element, peptides of 4-8 residues composed of overlapping sequences within the 15-25 domain were synthesized, along with hybrid compounds containing those recognition sequences coupled to a lysine hexamer. None of the recognition peptides altered Abeta aggregation kinetics and only two, KLVFF and KLVF, had any protective effect against Abeta toxicity. The hybrid peptide KLVFF-KKKKKK dramatically altered Abeta aggregation kinetics and aggregate morphology and provided significantly improved protection against Abeta toxicity compared to the recognition peptide alone. In contrast, FAEDVG-KKKKKK possessed only modest inhibitory activity and had no marked effect on Abeta aggregation. The scrambled sequence VLFKF was nearly as effective a recognition domain as KLVFF, suggesting the hydrophobic characteristics of the recognition sequence are critical. None of the cytoprotective peptides prevented Abeta aggregation; rather, they increased aggregate size and altered aggregate morphology. These results suggest that coupling recognition with disrupting elements is an effective generalizable strategy for the creation of Abeta inhibitors. Significantly, prevention of Abeta aggregation may not be required for prevention of toxicity.