Alzheimer's disease (AD) is a complex disease, involving multiple factors such as the production of aggregation-prone amyloid beta (Abeta) peptides, the formation of fibrillarly tangles of microtubule-associating proteins, Tau, and the polymorphism of cholesterol binding protein, APOE4. While understanding the mechanism of AD and the involvement of key players should lead to rational drug discovery against this disease, a traditional screening approach should also work for identifying drugs using AD models. We have used a cellular AD model, in which a cell death was induced by AD-causing neurotoxicities, and then screened the genes, which rescued the cells from the cell death. This resulted in isolation of a gene encoding a novel 24-amino acid long peptide, termed Humanin (HN), which protected neuronal cells at approximately microM level. Surprisingly, these gene products and the synthetic peptides not only protected neurons from cell death induced by Abeta-related neurotoxicities, but also Abeta-unrelated neurotoxicities. While a broad range of activities of HN against AD-related insults is discovered, the detailed mechanism of its action is still obscure. Structure analysis of HN showed that it is largely disordered and flexible at low peptide concentrations and heavily aggregates at high concentrations. Interestingly, one of the HN analogs, which is 10000-times more active than the parent HN molecule (i.e. active below nM range), was found to be monomeric. Based on findings of structural analyses, we propose here that membrane environment may enable HN to achieve high affinity for target protein(s) with multiple-transmembrane domains, such as G-protein coupled receptors.