Partial amino acid sequences of six major subunits of bovine P-crystallin have been determined by automatic liquid-phase Edman degradation and the dansyl-Edman procedure, complemented by amino acid analyses of peptides. The results show that, including the previously established PBp sequence [H. P. Eur. J. Biochem. 121,, there exist at least seven primary gene products in bovine P-crystallin, which exhibit 40 % or more sequence homology. Two of the gene products are completely identical except for the presence in one of them of 17 additional residues at the N terminus, possibly caused by differential splicing of the same primary RNA transcript. The rate of evolutionary change of the fi chains (4 % sequence change per 100 x lo6 years) is about equally slow as that of a-crystallin, and the gene duplications giving rise to the different chains must have occurred very early in vertebrate evolution. The P chains can be divided into two groups, according to sequence homology and presence of deletions/insertions and C-terminal extension, on which basis a new, rational nomenclature for the /? subunits is introduced. The N-terminal extensions of all P chains are very different in length and sequence, even between homologous P chains in different species. Possible explanations for this finding are discussed.The evolutionarily highly conserved structural eye lens proteins, the crystallins, can be divided into four classes a, P, y and h-crystallin, of which P-crystallin is the most heterogeneous [I]. In bovine P-crystallin six or more chains are primary gene products, whereas some ten others arise by post-translational modification [2]. They can associate to oligomers, varying from dimers and trimers (filow) to octamers (fihigh) [2,3]. This diversity in P-crystallin subunits has also been found at the mRNA level in the chicken lens [4]. The primary structure of the predominant bovine P-crystallin chain, PBp, has been elucidated [5], while the protein structure of a major murine P-crystallin polypeptide (P 23) has been deduced from cDNA and genomic DNA analysis [6, 71. Several mRNAs of the rat lens crystallins have been cloned [8], and recently the nucleotide sequence of one of these has revealed the primary structure of the PBlchain (J. T. den Dunnen and J. G. G. Schoenmakers, unpublished data).Amino acid sequence determination of bovine y-crystallins [9, 101 and nucleotide sequence analyses of several y-crystallin cDNAs from rat [ll] and frog [12] have shown a considerable homology between P and y-crystallins. The three-dimensional structure of the bovine y-I1 chain has been determined [13] and a similar tertiary fold has been predicted for two / 3 chains, BBp [I41 and 823 [7]. The P and y-crystallins are built up of two domains, which show considerable sequence similarity, suggesting an intragenic duplication in the ancestral gene of these proteins. Each of the domains is again folded into two similar 'Greek key' motifs. A relationship between these structural motifs and the exons of the murine P23 gene has been suggeste...