Refolding of Amphioxus Insulin-like Peptide: Implications of a Bifurcating Evolution of the Different Folding Behavior of Insulin and Insulin-like Growth Factor 1 ² Shuai Wang, ‡,§ Zhan-Yun Guo, ‡,| Lu Shen, | Ying-Jiu Zhang, § and You-Min Feng* ,| Key Laboratory of Proteomics, Institute of Biochemistry and Cell Biology, Shanghai Institutes for Biological Sciences, Chinese Academy of Sciences, Shanghai 200031, China, and Life Science College, Jilin UniVersity, Changchun 130061, China ReceiVed April 21, 2003; ReVised Manuscript ReceiVed June 12, 2003 ABSTRACT: Insulin and insulin-like growth factor 1 (IGF-1) share high sequence homology, but their folding behaviors are significantly different: insulin folds into one unique thermodynamically controlled structure, while IGF-1 folds into two thermodynamically controlled disulfide isomers. However, the origin of their different folding behaviors is still elusive. The amphioxus insulin-like peptide (ILP) is thought to be the common ancestor of insulin and IGF-1. A recombinant single-chain ILP has been expressed previously, and now its folding behavior is investigated. The folding behavior of ILP shows the characteristics of both insulin and IGF-1. On one hand, two thermodynamically controlled disulfide isomers of ILP have been identified; on the other hand, the content of isomer 1 (its disulfides are deduced identical to those of swap IGF-1) is much less than that of isomer 2 (its disulfides are deduced identical to those of native IGF-1); that is, more than 96% of ILP folds into the native structure. The present results suggest that the different folding behaviors of insulin and IGF-1 are acquired through a bifurcating evolution: the tendency of forming the thermodynamically controlled non-native disulfide isomer is diminished during evolution from ILP to insulin, while this tendency is amplified during evolution from ILP to IGF-1. Moreover, the N-terminal Gln residue of ILP can spontaneously form a pyroglutamate residue, and its cyclization has a significant effect on the folding behavior of ILP: the percentage of isomer 1 is approximately 2-fold that of isomer 1 of the noncyclized ILP; that is, isomer 1 becomes more favored when the N-terminal residue of ILP is cyclized. So, we deduce that the N-terminal residues have a significant effect on the folding properties of insulin, IGF-1, and ILP. In the 1960s Anfinsen and co-workers first demonstrated the three-dimensional structure of a globular protein is uniquely determined by its amino acid sequence (1). Since then, significant advances have been made in the understand- ing of protein folding through experimental and theoretical approaches. For small proteins with two-state folding, topology is a major determinant of the folding rate and greatly influences the structure of the transition-state en- semble (2-4). Studies on the disulfide-coupled folding of some small globular proteins, such as BPTI, 1 RNase A, and EGF, have revealed a sequence of preferred kinetic inter- mediates which define a folding pathway (5-12). In vivo the protein folding is assisted by the molecular chaperones, especially for large proteins (13-15); some chaperones even can provide the missing steric information for protein folding (16). Insulin is an extensively studied small globular protein with A- and B-chains linked by three disulfides (one intrachain bond, A6-A11; two interchain bonds, A7-B7 and A20-B19). Its three-dimensional structure has been well solved by X-ray crystallography (17, 18) and NMR (19- 21) since the 1970s. Although the separate A- and B-chains of insulin can be recombined successfully in vitro (22), a single-chain polypeptide (pre-proinsulin) is synthesized in vivo. When B29Lys and A1Gly were linked together by a peptide bond directly, the mini-proinsulin still retained the three-dimensional structure identical to that of insulin (23, 24). A single-chain insulin (PIP) can fold correctly and can be secreted efficiently from transformed yeast cells (25). In vitro PIP can efficiently fold into the native state with correct pairing of its three disulfides through a defined folding pathway (26). It can reasonably be presumed that the three- dimensional structure of PIP is identical or very similar to that of insulin/mini-proinsulin. ² This work was supported by grants from the Chinese Academy of Sciences (KJ951-B1-606) and the National Foundation of Nature Science (30170209). * Corresponding author. Tel: (86) 021-64374430. Fax: (86) 021- 64338357. E-mail: fengym@sunm.shcnc.ac.cn. ‡ These authors contributed equally to this work. § Jilin University. | Chinese Academy of Sciences. 1 Abbreviations: PIP, recombinant porcine insulin precursor in which the C-terminus of porcine insulin B-chain and the N-terminus of porcine insulin A-chain were linked together by a dipeptide, Ala-Lys; ILP, recombinant single-chain amphioxus insulin-like peptide in which the C-terminus of the B-chain and the N-terminus of the A-chain of the amphioxus insulin-like peptide were linked together by a tripeptide, Ala-Ala-Lys, and the B29Arg was replaced by Lys residue; IGF-1, insulin-like growth factor 1; IGF-2, insulin-like growth factor 2; BPTI, bovine pancreatic trypsin inhibitor; RNase A, ribonuclease A; EGF, epidermal growth factor; GSH, reduced glutathione; GSSG, oxidized glutathione; EDTA, ethylenediaminetetraacetic acid; HPLC, high- performance liquid chromatography; TFA, trifluoroacetic acid; PAGE, polyacrylamide gel electrophoresis; NMR, nuclear magnetic resonance; CD, circular dichroism. 9687 Biochemistry 2003, 42, 9687-9693 10.1021/bi0346289 CCC: $25.00 © 2003 American Chemical Society Published on Web 07/18/2003