br Origin of the vertebrate
Origin of the vertebrate hexokinase gene family Multiple hexokinases have been found in vertebrates, plants and yeast (Wilson, 1995, Cárdenas et al., 1998). A hexokinase in both vertebrates and yeast has, unfortunately, been named glucokinase. The sharing of the name glucokinase has suggested to some that the vertebrate and yeast glucokinases are more closely related to each other than to other hexokinases (see Cárdenas et al. (1998) for examples), however, analyses of hexokinase and glucokinase sequences from mammals, yeast, and plants show that mammalian glucokinase is more closely related to other mammalian hexokinases and yeast glucokinase is more closely related to other yeast hexokinases than the two glucokinases are to each other (Griffin et al., 1991, Bork et al., 1993, Wilson, 1995, Cárdenas et al., 1998). Hexokinases from non-vertebrate species have a molecular weight of about 50kD, indicating that the 100kD mammalian hexokinases originated on the vertebrate lineage (Cárdenas et al., 1998). It was suggested that a emca receptor of the hexokinase domain occurred within an ancestral hexokinase gene in a pre-vertebrate ancestor (Ureta, 1982). The intragenic duplication of the hexokinase domain also occurred at about the same time as the gene duplications that generated the diverse hexokinase genes (Ureta, 1982). How glucokinase was related to the 100kD hexokinase could not be resolved, thus Ureta (1982) suggested two possible evolutionary relationships: (1) glucokinase diverged prior to the duplication of the hexokinase domain or (2) glucokinase lost a domain after duplication (see Fig. 3A). With the cloning and sequencing of mammalian hexokinase genes (Andreone et al., 1989, Magnuson et al., 1989, Schwab and Wilson, 1989, Schwab and Wilson, 1991, Printz et al., 1993a) it became possible to better investigate the origin and evolution of these genes. Griffin et al. (1991) compared the sequences of the two halves of mammalian hexokinase I to mammalian glucokinase and yeast hexokinases, and showed that mammalian glucokinase is more closely related to mammalian hexokinase I than to any of the yeast hexokinases (Griffin et al. (1991)). This conclusion is in agreement with the multiple mammalian and yeast hexokinases have independent origins (Griffin et al., 1991, Bork et al., 1993, Wilson, 1995, Cárdenas et al., 1998). The phylogeny presented by Griffin et al. (1991) was consistent with a relatively ancient origin of the 100kD hexokinase, and with the glucokinase sequence being more closely related to the C-terminal, rather than the N-terminal, portion of hexokinase I. The study by Griffin et al. (1991) implied that the duplication of the hexokinase domain occurred before the divergence of glucokinase from hexokinase I (Griffin et al., 1991), consistent with first possibility suggested by Ureta (1982) (see Fig. 3A left panel). Bork et al. (1993) followed this with a study that used the sequences of all four mammalian hexokinases, and while again they were able to show independent origins for the mammalian and yeast hexokinase gene families, they were unable to resolve the relationships among the mammalian sequences. Cárdenas et al. (1998) unfortunately used the UPGMA (unweighted pair-group method with arithmetic mean (Sokal and Michener, 1958)) method to calculate a tree for mammalian hexokinases, using protein sequences distances calculated by Fothergill-Gilmore and Michels (1993). As acknowledged by the authors (Cárdenas et al., 1998), UPGMA assumes that the sequences have evolved at the same rate, but it was already clear from the distance data (Fothergill-Gilmore and Michels, 1993) that the N-terminal domains of hexokinases evolve more rapidly than the C-terminal domains. The UPGMA distance tree generated by Cárdenas et al. (1998), unsurprisingly, yielded an unexpected tree where the two halves of hexokinase III have different origins. The authors attempted to rationalize their finding by placing glucokinase as the earliest diverging lineage, a phylogeny that is in accord with the second model presented by Ureta (1982) (see Fig. 3A, right panel).