Acumen’s founders first proposed that the higher propensity of Aβ42 vs. Aβ40 to assemble into ADDLs could be attributed to the enhanced ability of the Aβ42 C-terminus to adopt a hairpin conformation consisting of two anti-parallel mini-β-strands (Klein, et al., 2003).  Gly-37 and Gly-38 were suggested to adopt a β-turn, with Ile-41 and Ala-42 conferring essential stabilization of the hairpin via cross-strand hydrogen bonds.  They further postulated that the alternating hydrophobic residues projecting from the top and bottom faces of this planar C-terminal hairpin or “hook” structure engaged in critical interactions with counterpart side-chains on adjacent Aβ42 hairpins, essentially driving assembly into the initial trimer intermediates now believed to be the basic subunits comprising the binding competent ADDL 12-mers.  The crucial role of Ile-41 was verified in cross-linking experiments, which showed that oligomer formation was greatly enhanced in Aβ41 compared with Aβ40.  Recent molecular dynamics simulations indicate that the C-terminal β-turn motif is readily adopted within an Aβ42 monomer and persists as a stable sub-structure (Urbanc, et al., 2004).

We hypothesize that small molecules capable of binding to the clustered hydrophobic residues comprising the C-terminal hook should be effective inhibitors of ADDL assembly, and the inhibitor discovery and optimization work in which we are now engaged provides an excellent framework for testing this hypothesis.  Several recent studies lend support to the concept that protein-protein interactions can be inhibited by small molecule binding to critical protein interaction motifs (Berg, 2003, Toogood, 2002, Arkin and Wells, 2004).

Acumen developed and implemented a four-tiered screening cascade, involving a combination of biochemical and cell-based functional assays.  1) a homogeneous, solution based FRET/FP assay for Aβ42 assembly; 2) an immobilized protein binding assay to determine the selectivity and affinity of chemical binding to Aβ40 and Aβ42 peptides; 3) an in vitro, ADDL binding assay to determine chemical inhibition of ADDL-binding to neuronal cells; and 4) an in vitro, neuronal cell viability assay to determine chemical toxicity and inhibition of ADDL-mediated toxicity.

Z’ coefficients for the biochemical assays are greater than 0.7, while Z’ coefficients for cell based function assays are greater than 0.4.  All assay conditions were fully characterized to ensure that ADDLs, but not fibrils, form under assay conditions.  The validity of the assays was confirmed using proprietary Aβ42 monomer- and ADDL-selective antibodies as positive controls.

A large small molecule diversity library was selected on the basis of “lead-drug like criteria” and predicted blood-brain-barrier permeation coefficients of greater than 0.7.  The diversity across all compounds was 0.83.  Screening and hit analysis revealed at least three distinct chemical classes that inhibited ADDL assembly.  The most potent hits had IC50 values < 5 µM and represent bona fide hits suitable for chemistry optimization.  Subsequent optimization has already produced leads that inhibit ADDL assembly with IC50 values of less than 100 nM.

We also screened a number of compounds reported to inhibit amyloid aggregation.  With only one exception, none inhibited ADDL assembly at less 30 µM.  Gossypol did inhibit with an IC50 ca. 30 µM, however, its binding to Aβ was irreversible, and it also bound covalently to reference proteins.  Thus, none of the reported inhibitors were deemed suitable templates for development into potent assembly blockers.