Molecular evolutionary conservation of the Drosophila alpha-amylase gene and its regulatory systems.

Alpha-amylase is a digestive enzyme involved in the metabolism of starch. It is found in a wide variety of organisms, and its expression is often regulated by carbon source. The D. melanogaster $\alpha$-amylase gene was used here to show the evolutionary conservation of both a structural gene as well as its glucose repression regulatory system. The repression machinery was shown to be functionally conserved between yeast and flies, as the $\alpha$-amylase promoter was used to express luciferase-based reporter constructs in a glucose repressible manner in Saccharomyces. The repression motifs of the $\alpha$-amylase promoter were mapped to a 126 bp fragment which contains two putative binding sites for MIG1, the yeast transcription factor involved in glucose repression. Further 5$\sp\prime$ deletions or removal of the 3$\sp\prime$ MIG 1 binding site eliminated glucose repression. Additional evidence for conservation of this machinery was provided by the cloning of Drosophila genomic and cDNA copies for SNF4, a subunit of the yeast derepression complex. This genes codes for a 684 aa protein which features extended carboxy and amino terminal sequences relative to the known yeast and mammalian homologues. Southern analysis suggested the presence of two gene copies in Drosophila, which would agree with the multiple cDNA isoforms seen in mammals. Inferred phylogenies indicated that the short isoform seen in yeasts and most mammalian sequences group separately from the long isoforms seen in Drosophila, C. elegans and a second human sequence--indicating the possible organization of orthologous gene copies. Aside from low level similarity to a handful of IMP dehydrogenases and IMPDH-like proteins, SNF4 is not significantly similar to other known proteins. Last, the evolution of the $\alpha$-amylase gene itself was examined with regards to nucleotide and amino acid biases. Nucleotide bias was observed at all three codon positions, primarily as a pyrimidine preference in the coding strand. Due to a nucleotide bias at nonsynonymous sites, there were significant content differences in GC-rich and AT-rich amino acids depending on the GC content of the gene. Although an amino acid bias might be expected to affect phylogenetic determination, with $\alpha$-amylase this bias was shown to not have a strong influence on inferred gene trees