Covalent attachment of ubiquitin marks substrates for proteolysis, but features that identify ubiquitination targets such as chicken egg white lysozyme are poorly understood. Recognition of lysozyme first requires reduction of Cys-6 Cys-127, one of its four native disulfide bonds, and Cys-6,Cys-127-carboxymethylated (6,127-rcm) lysozyme can mimic this three-disulfide intermediate. The 6,127-rcm form of lysozyme is known to retain a substantially native-like conformation in solution, and we demonstrate that it is this folded structure that is recognized for ubiquitination. Because native lysozyme is not a substrate, differences between the native and three-disulfide structures must include features responsible for selective ubiquitination. The 1.9-A resolution crystal structure of 6,127-rcm-lysozyme, reported here, affords a view of this ubiquitin-dependent degradation substrate. Two conformers of 6,127-rcm-lysozyme were obtained in the crystal. These differ uniquely from crystal forms of native lysozyme by displacement of the C-terminal residues. The structures suggest that localized unfolding at the C terminus of three-disulfide lysozyme allows the complex of E3a (ubiquitin-protein ligase) and E2 (ubiquitincarrier protein) to bind to a surface that includes Lys-1 and the putative ubiquitination site Lys-13. From this we infer that the N-terminal and internal substrate recognition sites on the E3a-E2 complex are separated by -20 A. Intracellular protein degradation is remarkable for its combination of extreme selectivity and the ability to accommodate an enormous variety of substrates. Protein half-lives in vivo span several orders of magnitude. Moreover, mutations, translational errors, mislocalization, and chemical damage all can lead to polypeptides that are rapidly degraded (1-5). How such proteins are distinguished from their long-lived counterparts is largely unknown. In eukaryotes, a major route for intracellular proteolysis involves covalent modification of protein lysine(s) with the protein ubiquitin (Ub) and subsequent degradation by the 26S Ub-dependent protease complex (6-8). Specificity resides, at least in part, with the Ub-protein ligase that marks a substrate for recognition by the protease. This process is best understood for the E3a ligase from rabbit reticulocytes and the related UBRIencoded enzyme from yeast (9-13).One important E3a recognition determinant is the substrate's N terminus, where only a subset of amino acids is permissive for ubiquitination (9-12). A permissive N terminus is not suffi'cient, however, and the relative orientations and distances that separate lysine ubiquitination sites from