Conserved and variable repeat structures in the Balbiani ring gene family in Chironomus tentans.

The four Balbiani ring (BR) genes, BR1, BR2.1, BR2.2, and BR6 in the midge Chironomus tentans constitute a gene family encoding secretory proteins with molecular weights of approximately 10(6) daltons. The major part of each gene is known to consist of tandemly organized composite repeat units resulting in a hierarchic repeat arrangement. Here, we present the sequence organization of the 5' part of the BR2.2 and BR6 genes and describe the entire transcribed part of the two genes. As the BR1 and BR2.1 genes were also fully characterized recently, this allows the comparison of all genes in the BR gene family. All four genes share the same exon-intron structure and have evolved by gene duplications starting from a common ancestor, having the same overall organization as the BR genes of today. The genes encode proteins that have an approximately 10,000-amino acid residue extended central domain, flanked by a highly charged, approximately 200-residue amino-terminal domain and a globular 110-residue carboxy-terminal domain. Exons 1-3 and the beginning of exon 4 encode the amino-terminal domain, which throughout contains many regions built from short repeats. These repeats are often degenerate as to repeat unit and sequence and are present in different numbers between the genes. In several instances these repeat structures, however, are conserved at the protein level where they form positively or negatively charged regions. Each BR gene has a 26-38-kb-long exon 4, which consists of an array of 125-150 repeat units and encodes the central domain. The number of repeat units appears to be largely preserved by selection and all repeat units in the array are very efficiently homogenized. Occasionally variant repeats have been introduced, presumably from another BR gene by gene conversion, and spread within the array. Introns 1-3 at the 5' end of the genes have diverged extensively in sequence and length between the genes. In contrast, intron 4 at the 3' end is virtually identical between three of the four genes, suggesting that gene conversion homogenizes the 3' ends of the genes, but not the 5' ends.