Toward â-Amino Acid Proteins: Design, Synthesis, and Characterization of a Fifteen Kilodalton â-Peptide Tetramer

Natural proteins are composed of linear chains of R-amino acid monomers that adopt complex folded structures stabilized primarily by noncovalent interactions.1 Within a single polypeptide chain, local interactions generate helix or sheet secondary structures and longer-range interactions generate tertiary structuressassemblies of multiple secondary structures.2 Furthermore, multiple discrete polypeptides can associate to form quaternary complexes. This higher order structure is nearly universally responsible for the sophistication of protein function. Non-natural polymers have the potential for structural complexity and sophisticated function, but the design of such molecules is even more challenging than protein design because there exist no natural templates to mimic.3,4 Our efforts to design higher order â-peptide folds began with the synthesis of the oppositely charged â-peptides Acid-1F and Base-1F, which assemble in aqueous buffer into a stable, octameric quaternary complex.5 To more fully characterize the structure and thermodynamics of Acid-1F/Base-1F association, we designed the dodecapeptide Zwit-1F, which we crystallized to obtain the first high-resolution images of a 14-helical â-peptide bundle.6 A detailed thermodynamic analysis established that the solution-phase behavior of Zwit-1F was consistent with an equilibrium between unfolded monomers and folded octamers with an association/folding constant of 4 × 1030 M-7 at 25 °C.7 The Zwit-1F octamer contains four copies of a parallel â-peptide dimer. Each parallel dimer associates in an antiparallel fashion with another dimer to form a tetramer, and two tetramers assemble with a 39° crossing angle to form the octamer (Figure 1A). The octamer core is composed entirely of solvent-inaccessible â3-homoleucine (â3L) residues; hydrophobic burial of their side chains presumably drives association. The solvent-exposed octamer surface is decorated by â3-homoglutamate (â3E), â3-homoaspartate (â3D), and â3homoornithine (â3O) residues that are charge-paired within the dimer unit and across the dimer/dimer or tetramer/tetramer interfaces. While we regarded the crystal structure as an important step toward the design of a higher order â-peptide fold, several factors make Zwit-1F nonideal. First, it has a low self-affinity: the concentrations at which 50% (C50) or 90% (C90) of Zwit-1F exists in the folded, octameric state are 50 and 350 μM, respectively. Second, because of the complete surface charge pairing and exclusively 14-helical structure, the Zwit-1F octamer contains no easily modified residues and no pocket to support ligand binding and catalysis. Last, the octameric stoichiometry complicates interpretation of self-association data and increases the chance that substitution of any single side chain will alter the Zwit-1F fold. We considered two strategies to identify longer â-peptides that could recapitulate the Zwit-1F fold with fewer subunits (Figure 1A). One, which was not pursued, involves joining the antiparallel strands of a single tetramer with an appropriate (albeit lengthy) linker (Strategy A); the strategy reported here reverses the relative orientation of the two internal â-peptide monomers, converting four copies of a noncovalent parallel dimer into four copies of a covalent antiparallel dimer, Z28 (Strategy B). To reverse the relative orientation of the two internal â-peptides in each tetramer of Zwit-1F, we used Spartan8 to convert each of the four parallel â-peptide dimers that comprise the octamer into an antiparallel dimer (Figure 1B). The side chains of the interior helix of each parallel dimer (helix 1) were disconnected, and the â-peptide backbone was flipped axially by 180°. The side chains were then reconnected and the backbone structure energy-minimized with the side chains fixed. The now antiparallel â-peptide dimers were then connected by âG linkers of varying lengths, and a four âG linker was determined to be the shortest that could span the two helices. Finally, the resulting 28 residue structure was energyminimized. This modeling exercise suggested that the antiparallel helix-loop-helix structure, which we call Z28 (Figure 1C), could replace each parallel dimer unit in Zwit-1F, effectively placing the side chains in the same regions of space as the parallel dimer from Zwit-1F (Figure S1 in Supporting Information). It has previously been shown that the purity of â-peptides prepared using solid-phase synthesis falls off precipitously with chain length.9,10 We sought methods that could yield â-peptide 28† Department of Chemistry. ‡ Department of Molecular, Cellular and Developmental Biology. Figure 1. (A) Two strategies for conversion of the Zwit-1F octamer into a tetramer. (B) Details of Strategy B. We began with a parallel dimer unit (helices 1 and 2) from the Zwit-1F crystal structure and proceeded as described in the text. Sphere representations of side chains are colored by residue type as in the sequence. (C) Sequence of Z28. â-Amino acids are represented by the single letter code of the equivalent R-amino acid. F* denotes p-iodophenylalanine and O denotes ornithine. âG is commonly referred to as â-alanine. Published on Web 01/01/2008