Recent studies have implicated the amyloid Abeta peptide and its ability to self-assemble as key factors in the pathogenesis of Alzheimer's disease. Relatively little is known about the structure of soluble Abeta or its oligomeric state, and the existing data are often contradictory. In this study, we used intrinsic fluorescence of wild type Abeta-(1-40), fluorescence resonance energy transfer (FRET), and gel filtration chromatography to examine the structure of Abeta-(1-40) in solution. We synthesized a series of mono-substituted fluorescent Abeta-(1-40) derivatives to use as donors and acceptors in FRET experiments. We selected fluorescent peptides that exhibit aggregation properties comparable to wild type Abeta for analysis in donor-acceptor pairs; two labeled with 5-(2-((iodoacetyl)amino)ethyl)aminonaphthylene-1-sulfonic acid at Cys-25 or Cys-34 and fluorescein maleimide at Cys-4 or Cys-7. Another peptide containing a Trp substitution at position 10 was used as an acceptor for the intrinsic Tyr fluorescence of wild type Abeta-(1-40). Equilibrium studies of the denaturation of Abeta-(1-40) by increasing concentrations of dimethyl sulfoxide (Me2SO) were conducted by monitoring fluorescence, with a midpoint value for the unfolding transition of both the substituted and wild type peptides at among 40 and 50% Me2SO. Abeta-(1-40) is well solvated and largely monomeric in Me2SO as evidenced by a lack of FRET. When donor and acceptor Abeta derivatives are mixed together in Me2SO and then diluted 10-fold into aqueous Tris-HCl buffer at pH 7.4, efficient FRET is observed immediately for all pairs of fluorescent peptides, indicating that donor-acceptor dimers exist in solution. FRET is abolished by the addition of an excess of unlabeled Abeta-(1-40), demonstrating that the fluorescent peptides interact with wild type Abeta-(1-40) to form heterodimers that do not exhibit FRET. The Abeta-(1-40) dimers appear to be very stable, because no subunit exchange is observed after 24 h between fluorescent homodimers. Gel filtration confirms that nanomolar concentrations of 14C-labeled Abeta-(1-40) and fluorescein-labeled Abeta-(1-40) elute at the same dimeric position as wild type Abeta-(1-40), suggesting that soluble Abeta-(1-40) is also dimeric at more physiologically plausible concentrations.