The human albumin gene. Characterization of the 5' and 3' flanking regions and the polymorphic gene transcripts.

We have characterized the 5' and 3' regions of the human albumin gene with respect to nucleotide sequences, mRNA transcription initiation and polyadenylation. There are at least two transcription initiation sites and two polyadenylation sites in the human albumin gene. The multiple transcription initiation sites are utilized almost equally. The two polyadenylation sites are used differentially, with the site proximal to the protein termination codon dominating. The "TATA" and the "CCAAT" boxes are present at about 30 and 90 nucleotides upstream of the putative cap sites. About 250 nucleotides immediately 5' to the transcription initiation sites are highly conserved among the human, rat, and chicken. Within the 1-kilobase pair 5' flanking region of the human albumin gene there are 8- and 35-base pair alternating purine/pyrimidine sequences, primarily consisting of dG and dT, and three nucleotide segments exhibiting homology to the proposed progesterone receptor-binding site. In these and other characteristics, the 5' flanking region of the human albumin gene is distinct from the corresponding region of the human alpha-fetoprotein gene. Each of the two polyadenylation sites of the human albumin gene is preceded by a AATAAA hexamer. In addition, the first polyadenylation site has several conserved sequences believed to play a role in the 3'-end formation. The second polyadenylation site lacks these signals which may explain why it is used less frequently. In spite of these differences the two polyadenylation sites show a high overall sequence similarity, suggesting that they arose by duplication. There is no major difference between the normal liver and hepatoma cells in the relative utilization of multiple sites for transcriptional initiation and polyadenylation of the human albumin gene, suggesting that the albumin mRNA polymorphism has no direct relevance to hepatocarcinogenesis.