Most neurodegenerative disorders, such as Alzheimer's (AD), Parkinson's, Huntington's and Creutzfeldt-Jakob disease, are characterised by the accumulation of insoluble filamentous aggregates known as amyloid. These pathologies share common pathways involving protein aggregation which can lead to fibril formation and amyloid plaques. The 4 kDa Abeta peptide (39-43 amino acids) derived from the proteolysis of the amyloid precursor protein is currently a validated target for therapy in AD. Both active and passive immunisation studies against Abeta are being trialled as potential AD therapeutic approaches. In this study, we have characterised engineered antibody fragments derived from the monoclonal antibody, WO-2 which recognises an epitope in the N-terminal region of Abeta (amino acids 2-8 of Abeta). A chimeric recombinant Fab (rFab) and single chain fragments (scFvs) of WO-2 were constructed and expressed in Escherichia coli. Rationally designed mutants to improve the stability of antibody fragments were also constructed. All antibody formats retained high affinity (K(D) approximately 8 x 10(-9) M) for the Abeta peptide, comparable with the intact parental IgG as measured by surface plasmon resonance. Likewise, all engineered fragments were able to: (i) prevent amyloid fibrillisation, (ii) disaggregate preformed Abeta(1-42) fibrils and (iii) inhibit Abeta(1-42) oligomer-mediated neurotoxicity in vitro as efficiently as the whole IgG molecule. These data indicate that the WO-2 antibody and its fragments have immunotherapeutic potential. The perceived advantages of using small Fab and scFv engineered antibody formats which lack the effector function include more efficient passage across the blood-brain barrier and minimising the risk of triggering inflammatory side reactions. Hence, these recombinant antibody fragments represent attractive candidates and safer formulations of passive immunotherapy for AD.