With ihe completion of the primary structure of the 50,000-and 19,000-dalton fragments of human ceruloplasmin [ferroxidase; iron(il):oxygen oxidoreductase, EC 1.16.3.11], over halfofthe covalent structure ofthe single polypeptide chain ofthis protein is known. Visual and computer analysis of the sequence of the 564 amino acid residues in the two fragments gives clear evidence of statistically significant internal homology suggestive of evolutionary replication of two smaller units. Two homology regions, each composed of 224 residues, were defined by an intrasequence alignment that required only three gaps in each 224-residue segment. The two homology regions exhibited 43% identity in sequence, and 13% of the remaining positions had similar residues. The sequence ofa 160-residue segment in ceruloplasmin exhibits significant homology to the active (copper-binding) sites of blue electron-transfer proteins such as azurins and plastocyanins and multicopper oxidases such as cytochrome oxidase and superoxide dismutase. It is proposed that a primitive ceruloplasmin gene was formed by the fusion of two genes coding, respectively, for proteins about 160 and 190 amino acid residues in length and that this precursor gene coding for about 350 amino acids was later triplicated to form the gene for the present-day ceruloplasmin molecule of about 1050 amino acids.Analytical and statistical comparison of amino acid sequences has been a powerful method for detection and estimation of structural similarities among and within proteins, for evaluation of structure-function relationships in a family of proteins, and for study of genetic change in the course of evolutionary development of a class of proteins (1-3). The comparison may be done by visual alignment oftwo sequences for best fit as judged both by identity and similarity of paired amino acids at corresponding positions; gaps in the sequences are sometimes inserted to maximize the fit. Statistical methods for assessing relatedness of proteins are based on computer programs that accumulate pair scores for two amino acids, using a mutation data matrix based on physical and structural parametersg and codon differences for all possible pairs ofamino acids (4) or from actual accepted point mutations accumulated from related sequences (5). By such means the degree of intersequence or intrasequence homology may be assessed; here "homologous" is used to mean matching (identical or similar) in structure, position, physical characteristics, and codons.With the recent completion of the primary structures of the 50,000-dalton (50-kDal) (6) and 19,000-dalton (19-kDal) (7-9) fragments ofhuman ceruloplasmin [ferroxidase; iron(II):oxygen oxidoreductase, EC 1.16.3.1], 564 residues of the amino acid sequence of the single polypeptide chain of this protein are known. These fragments are from the carboxyl terminus of the molecule and account for more than halfofthe primary structure of this blue copper oxidase. During our study we identified a remarkable degree of internal homology in c...