Molecular Evolution of Insulin in Non-Mammalian Vertebrates(1).

STRUCTURE-ACTIVITY RELATIONSHIPS

Elucidation of the three-dimensional crystal structure of pig insulin by X-ray analysis (Blundell et al., 1972) provided the first insight into the conformation of the protein and the nature of the surface in the molecule that interacts with the insulin receptor. The A-chain of insulin comprises two [Alpha]-helical segments (A2-A-8 and A13-A19). The B-chain can adopt two distinct conformations. In the T-state, residues B9-B19 form an [Alpha]-helix, residues B20-B23 constitute a [Beta]-turn and residues B24-B30 adopt an extended [Beta]-strand conformation. In the R-state, residues B1-B19 form a contiguous [Alpha]-helical region (Baker et al., 1988). Pig insulin readily associates into stable dimers and, in the presence of [Zn.sup.2+] ions, forms stable hexamers. The traditional view, based primarily on X-ray crystallographic data, is that the receptor-binding region of insulin is related to the dimer-forming surface and comprises a domain formed from B12, B16, B23-B26, A1-A5, A19 and A21. Residues B20 and B28 are also important in dimer formation and residues B6, B10, B14, B17, B18, A13, and A14 are involved in hexamer formation are (Baker et al., 1988).

Several pieces of evidence have suggested that neither of the conformations of insulin adopted in the crystalline state (T- and R-states) is the receptor binding conformation. The crystal structure of insulin from the Agnathan Myxine glutinosa (Atlantic hagfish) is very similar to that of pig insulin (Cutfield et al., 1979) yet hagfish insulin displays only 5% of the potency of pig insulin in stimulating lipogenesis in isolated rat fat cells (Emdin et al., 1977). In the crystal structure of pig insulin, the C-terminal region of the B-chain is located in close proximity to the N-terminal region of the A-chain. Chemical cross-linking of these domains by formation of a peptide bond between LysB29 and GlyA1 produces a molecule whose crystal structure and self-associative properties are very similar to native insulin but is biologically inactive (Derewenda et al., 1991). This observation implies that the active, receptor-binding conformation of insulin requires a separation of the C-terminus of the B-chain and the N-terminus of the A-chain. An alternative model has been provided by the results of a study by Kristensen et al. (1997) who used alanine scanning mutagenesis to identify specific side chains of insulin which strongly influenced binding to its receptor. Substitution of LeuB6, GlyB8, GlyB23, and PheB24, IleA2, Va1A3 and TyrA19 by Ala resulted in [is greater than] 20-fold decrease in binding affinity. In contrast, substitutions at LeuB16, TyrB26, GluA4, GlnA5, and AsnA21, residues formerly considered to part of the binding surface, had relatively minor effects on binding affinity. Substitutions at LeuB11, GluB13 and PheB25 resulted in approximately a 10-fold decrease in binding affinity but, unexpectedly, replacement of GlyB20, GluB21 and ArgB22 by Ala produced analogs with appreciably increased binding affinity (2-4-fold). These data have suggested the alternative model in which the receptor binding domain consists of five residues (IleA2, Va1A3, TyrA19, GlyB23, and Phe24) forming a patch on the surface of the molecule (Kristensen et al., 1997). Residues LeuB6, GlyB8, LeuB11, GluB13 and PheB26, although not part of the binding epitope, are presumably of importance in maintaining the overall receptor-binding conformation of insulin.

This review compares the amino acid sequences of insulins from a wide range of non-mammalian vertebrates with a view to correlating the residues in the molecule have been strongly conserved during evolution with those residues that structure-activity studies have shown are important in mediating its biological actions.

INSULINS FROM NON-MAMMALIAN AMNIOTA

The primary structures of reptilian insulins are known for a crocodilian, the American alligator Alligator mississipiensis (Lance et al., 1984); the chelonians, the red-eared turtle Psuedemys scripta (Conlon and Hicks, 1990) and the Brazilian slider turtle, Chrysemys dorbigni (Cascone et al., 1991); and for three species of snakes, the rattlesnake Crotalus atrox (Kimmel et al. 1976), the colubrid snake Zoacys dhumndades (Zhang et al., 1981) and the Burmese python Python molurus (Conlon et al., 1997b). These amino acid sequences are compared with the corresponding sequences of avian insulins (the common structures from chicken/turkey/ostrich and from duck/ goose [Evans et al., 1988]) and human insulin in Figure 1.

FIG. 1. A comparison of We primary structures of insulins from species of non-mammalian Amniota with human insulin. (-) denotes residue identity.