The betabellin target structure is a P-sandwich protein consisting of two 32 residue P-sheets packed against one another by interaction of their hydrophobic faces. The 32 residue chain of betabellin-15s (HSLTAKIpkLTFSIAphTYTCAV pkYTAKVSH, where p = DPrO, k = DLYS, and h = DHis) did not fold in water at pH 6.5. Air oxidation of betabellin-15s provided betabellin-15D, the 64 residue disulfide bridged two-chain molecule, which also remained unfolded in water at pH 6.5. By circular dichroic spectropolarimetry, the extent of / 3 structure observed for betabellin-15D increased with the pH and ionic strength of the solution and the betabellin-15D concentration. By electron microscopy, in 5.0 mM MOPS and 0.25 M NaCl at pH 6.9, betabellin-15D formed long narrow multimeric fibrils. A molecular model was constructed to show that the dimensions of these betabellin-15D fibrils are consistent with a single row of P-sandwich molecules joined by multiple intersheet H-bonds.Keywords: betabellin; P-sandwich; P-sheet; circular dichroism; de novo design; fibril formation; protein engineering One way to study protein folding is to start with an amino acid sequence and attempt to predict its folded protein structure. Another way is to start with a folded protein structure and attempt to predict an amino acid sequence that will fold into this structure. Most natural protein structures are too complicated to be easy folding targets. De novo protein engineering involves designing a simple folded protein structure from scratch, synthesizing a peptide chain having an amino acid sequence that may fold into the designed structure, and studying how it folds (DeGrado, 1997).This practical approach has provided many designed a-helical proteins (Hodges et al., 1988;Hecht et al., 1990;Engel et al., 1991;Handel et al., 1993;Bryson et al., 1995;Betz et al., , 1997Hodges, 1996;Lombardi et al., 1996;Dahiyat & Mayo, 1997;Kohn et al., 1997). Fewer proteins have been designed with the P-sheet as the main element of secondary structure (Kullmann, 1984;Moser et al., 1985;Osterman & Kaiser, 1985; Richardson et al., 1992;Pessi et al., 1993; Quinn et al., 1994;Wagner et al., 1994;Krause et al., 1996;Mayo et al., 1996;Nesloney & Kelly, 1996;Smith & Regan, 1997; Zappacosta et al., 1997). Because P-sheets are less modular than a-helices and tend to aggregate in solution, it is more Reprint requests to: Bruce W. Erickson, Department of Chemistry, The University of North Carolina at Chapel Hill, Chapel Hill, North Carolina 27599-3290; e-mail: bruce@peptide.chem.unc.edu. challenging to design a native-like /?-sheet protein (Hecht, 1994). Designing a water-soluble P-sheet protein involves understanding of the noncovalent interactions between P-strands and between P-sheets. Therefore, engineering of a P-sheet protein should increase our understanding of the interactions of p-turns, P-strands, and P-sheets. This paper describes the de novo design, solid phase synthesis, and biophysical characterization of betabellin-I5D, a 64 residue disulfide bridged two-c...