We report the high-resolution crystal structures of an extensively simplified variant of bovine pancreatic trypsin inhibitor containing 20 alanines (BPTI-20st) and a reference single-disulfide-bonded variant (BPTI-[5,55]st) at, respectively, 1.39 and 1.09 Å resolutions. The sequence was simplified based on the results of an alanine scanning experiment, as reported previously. The effects of the multiple alanine substitutions on the overall backbone structure were surprisingly small (C ␣ atom RMSD of 0.53 Å) being limited to small local structural perturbations. Both BPTI variants retained a wild-type level of trypsin inhibitory activity. The side-chain configurations of residues buried in the hydrophobic cores (<30% accessible surface area) were almost perfectly retained in both BPTI-20st and BPTI-[5,55]st, indicating that neither multiple alanine replacements nor the removal of the disulfide bonds affected their precise placements. However, the side chains of three partially buried residues (Q31, R20, and to some extent Y21) and several unburied residues rearranged into alternative dense-packing structures, suggesting some plasticity in their shape complementarity. These results indicate that a protein sequence simplified over its entire length can retain its densely packed, native side-chain structure, and suggest that both the design and fold recognition of natively folded proteins may be easier than previously thought.native structure ͉ protein design ͉ protein folding ͉ protein structure ͉ sequence simplification T he protein folding problem (1), or how a protein structure is encoded in its amino acid sequence, remains fundamentally unsolved. Indeed, an astronomical number of sequences are generated by the combination of the 20 natural amino acids at each site of a sequence, but few of these can fold into native, fully functional protein structures. However, the information contained in natural amino acid sequences is redundant (2, 3); namely, a small fraction of the information can fully determine a protein structure (4, 5). Protein sequence simplification has emerged as a method of choice in deciphering the determinants of protein structures.Information redundancy can be conceptually classified into class and site redundancy (6). Class redundancy refers to the overlapping physicochemical properties of the amino acids, and it can be removed by designing proteins with a reduced amino acid alphabet (7). Class redundancy has been reported in the Src SH3 domain, for which a functional variant with 70% of its sequence encoded by only five amino acids types was identified by phage display (8). Similarly, simplified variants of chorismate mutase (9) and orotate phosphoribosyl transferase (10), with sequences predominantly encoded by only nine amino acid types, rescued auxotrophic host cells lacking the respective enzymes.Site redundancy refers to the occurrence of several residues in a protein's amino acid sequence that do not contribute to the formation of the structure, except that their backbone atoms act as l...