Non-random Codon Usage of Synonymous and Non-synonymous Mutations in the Human HLA-A Gene.

The structure and function of human leucocyte antigen (HLA-A) is well known and is an extremely variable protein. From the public HLA-A database, we chose 26 high frequency HLA-A alleles (45% of sequenced alleles). Using five arbitrary references from these alleles, we analyzed synonymous mutations at the third codon position (sSNP 3 ) and non-synonymous mutations (NSM). Both mutation types showed non-random locations of 29 sSNP 3 codons and 71 NSM codons in the five reference lists. Most sSNP 3 codons show identical mutation types with many mutations resulting from cytosine deamination. We proposed 23 ancestral parents of sSNP 3 in five reference sequences using conserved parents in five unidirectional codons and 18 majority parents in reciprocal codons. These 23 proposed ancestral parents show exclusive codon usage of G 3 or C 3 parents located on both DNA strands that mutate to A 3 or T 3 variants mostly (76%) by cytosine deamination The sSNP 3 and NSM show clear separation of the two variant types with most sSNP 3 located in conserved areas in exons 2, 3 and 4, compared to most NSM appearing in two Variable Areas with no sSNP 3 in the latter parts of exons 2 (α1) and 3 (α2). The Variable Areas contain NSM (polymorphic) residues at the center of the groove that bind the foreign peptide. We find distinctly different mutation patterns in NSM codons from those of sSNP 3 . Namely, G-C to A-T mutation frequency was much smaller, suggesting that evolutional pressures of deamination and other mechanisms applied to the two areas are significantly different.