Abstract
The functional human adenine phosphoribosyltransferase (APRT) gene is less than 2.6 kilobases in length and contains five exons. The amino acid sequences of APRTs have been highly conserved throughout evolution. The human enzyme is 82%, 90%, and 40% identical to the mouse, hamster, and Escherichia coli enzymes, respectively. The promoter region of the human APRT gene, like that of several other "housekeeping" genes, lacks "TATA" and "CCAAT" boxes but contains five GC boxes that are potential binding sites for the Sp1 transcription factor. The distal three, however, are dispensable for gene expression. Comparison between human and mouse APRT gene nucleotide sequences reveals a high degree of homology within protein coding regions but an absence of significant homology in 5' flanking, 3' untranslated, and intron sequences, except for similarly positioned GC boxes in the promoter region and a 26-base-pair region in intron 3. This 26-base-pair sequence is 92% identical with a similarly positioned sequence in the mouse gene and is also found in intron 3 of the hamster gene, suggesting that its retention may be a consequence of stringent selection. The positions of all introns have been precisely retained in the human and both rodent genes, as has an unusual AG/GC donor splice site in intron 2. Particularly striking is the distribution of CpG dinucleotides within human and rodent APRT genes. Although the nucleotide sequences of intron 1 and the 5' flanking regions of human and mouse APRT genes have no substantial homology, they have a frequency of CpG dinucleotides that is much higher than expected and nonrandom considering the G + C content of the gene. Retention of an elevated CpG dinucleotide content, despite loss of sequence homology, suggests that there may be selection for CpG dinucleotides in these regions and that their maintenance may be important for APRT gene function.