Possible evolutionary relationship between mammalian alanine:glyoxylate aminotransferase 1 and the 42-kD subunit of cyanobacterial soluble hydrogenase.

Using the sequence comparison programs TFASTA / FASTA ( Pearson and Lipman 1988) to search the GenBank/EMBL DNA data bases, we have discovered that there is considerable amino acid sequence identity between mammalian alanine: glyoxylate aminotransferase 1 (AGT; E.C.2.6.1.44) and the 42-kD tritium-exchange subunit of cyanobacterial soluble hydrogenase (TES; E.C. 1.12.-.-/ 1.18.-.-) . Over a 36 l-amino-acid region (residues 24-384) of human liver AGT (Takada et al. 1990) and a 355amino-acid region (residues 6-360) of Anabaena cylindrica TES (Ewart et al. 1990)) 110 amino acids were found to be identical, and 90 were found to be similar (using the criteria advocated by Barker and Dayhoff 1972) (see fig. 1). When gaps are counted as mismatches, the level of sequence identity is 30. I%, and similarity is 54.8%. Similar results were obtained when comparisons were made between TES from Anabaena or Synechococcus (Van der Oost et al. 1989) and the AGT sequences obtained from rat ( Oda et al. 1987 ), marmoset, and rabbit (Purdue et al. 1992). Eightytwo (23.7%) amino acids are absolutely conserved in all six proteins (i.e., two TESS and four AGTs) (see fig. 1). In some regions there is considerably greater match between Anabaena TES and human AGT. For example, TES residues 66-98 and AGT residues 85-117 show 17/33 (51.5%) identity and 27/33 (81.8%) similarity. Two regions (TES 137-148/AGT 153-164, and TES 164-175/AGT 179-190) show 7 / 12 ( 58.3% ) amino acid identity and 12/ 12 ( 100% ) similarity. Whether these regions represent equivalent functional domains has yet to be determined. AGT, like all aminotransferases, requires pyridoxal phosphate as a cofactor. The binding site for this cofactor has been suggested to be Lys-209 (Oda et al. 1987). There is no evidence that hydrogenases require pyridoxal phosphate. Therefore, it is surprising to find that, not only is this Lys absolutely conserved in both TES sequences, in addition to all four AGT sequences, but the four preceding amino acids are conserved as well. However, this SerGlySerGlnLys sequence is not found in other aminotransferases (Mehta et al. 1989), and, therefore, its functional significance is unclear. As far as is currently known, there are no similarities in the catalytic or functional properties of mammalian AGT and cyanobacterial soluble hydrogenase. Nevertheless, they show considerably more sequence identity to each other than does either AGT to other aminotransferases or TES to other hydrogenases, indicating that, although AGT and TES might be functionally disparate, they are homologous proteins and that they may have evolved independently from other aminotransferases or hydrogenases. Whether mammalian AGTs and cyanobacterial soluble hydrogenases actually do share any functional/catalytic properties awaits investigation.