STRUCTURE OF THE CHUM SALMON INSULIN-LIKE GROWTH FACTOR-I GENE

insulin-like growth factor I (IGF-1) plays a major role in development and metabolism. Currently, the cDNA-derived primary structure of IGF- I is known for some mammals and for chicken, frog, and salmon. Additio nally, the organization of the human, rat, and chicken IGF-I genes has been established. The investigation of IGF-I gene structure in fish w ould extend the evolutionary picture for this hormone and facilitate o ur understanding of the features of the IGF-I gene that are common to all vertebrate species. The cloned chum salmon IGF-I gene appears to b e much more compact than the mammalian and avian genes, being less tha n 20 kb in length. As in other species, however, the mature IGF-I pept ide appears to consist of 70 amino acids and is encoded by exons 2 and 3. Intriguingly, exon 1-encoded 5'-untranslated region sequences are highly conserved, while the coding sequences at the 3' end of the same exon are less conserved. The amino terminus of the signal peptide is four amino acids shorter than in the mammalian and avian peptides. The end of the B domain, the C, A, and D domains, and the first part of t he E peptide are encoded by exon 3, but the exon 3-encoded E peptide s equence is 27 amino acids longer than in other species. These extra 27 amino acids, encoded by both coho and chum salmon cDNAs, may be delet ed by alternative splicing, as suggested from the sequence of a coho s almon IGF-I cDNA. However, the substitution of AT for GT at the necess ary splice donor site in the chum salmon IGF-I gene does not allow for the deletion of the sequence encoding these amino acids. The presence of additional chum salmon cDNAs that do not encode this 27-amino-acid sequence suggests the presence of a second chum salmon IGF-I gene wit h a functional splice junction at this location. Indeed, preliminary m apping data and partial sequencing suggests the existence of a second IGF-I gene in chum salmon. These observations show that the IGF-I gene s in four classes of vertebrates have similar overall structures with regard to their common coding regions, but differ significantly in the ir 5' and 3' regions.