Evolutionary analysis of acyl-CoA oxidases and acyl-CoA dehydrogenases. Multiple sequence alignment of the protein sequences was performed using the pileup program (34) from the GCG package as well as with clustal w (35). The phylogenetic tree was constructed according to the neighbor-joining method (36), as implemented in the neighbor program while the distance matrices needed were calculated with protdist, both from the phylip package (37). The reliability of the tree was assessed by bootstrap analysis. Bootstrap values were in general higher than 80%. Whereas all known acyl-CoA oxidases were aligned, a selection of acyl-CoA dehydrogenases, possessing distinct substrate specificities, was made (data bank accession numbers are also given) as follows: hBRCACox/rTHCCox, X95189X95189; h/rPALMCox I, X71440X71440 and P07872P07872; rPRISCox, X95188X95188; CeACox (Caenorhabditis elegans acyl-CoA oxidase), P34355P34355; ScPXP1 (S. cerevisiae POX1 protein), P13711P13711; CtPXP2/4/5 (Candida tropicalis POX2/4/5 proteins), P06598P06598, P11356P11356, and P08790P08790; CmPXP4 (Candida maltosa POX4 protein), P05335P05335; mACDG (mouse glutaryl-CoA dehydrogenase), U18992U18992; h/rACDS (human and rat short-chain acyl-CoA dehydrogenases), P16219P16219 and P15651P15651; hACDB (human short branched-chain acyl-CoA dehydrogenase), P45954P45954; h/rACDM (human and rat medium-chain acyl-CoA dehydrogenases), P11310P11310 and P08503P08503; h/rACDL (human and rat long-chain acyl-CoA dehydrogenases), P28330P28330 and P15650P15650; rACDV (rat very-long-chain acyl-CoA dehydrogenase), P45953P45953; Ce/h/rIVD (C. elegans, human, and rat isovaleryl-CoA dehydrogenases), P34275P34275, P26440P26440, and P12007P12007. The h/rPALMCoxs II are not shown since the exon duplication, giving rise to the type I and II mRNAs, is a recent event and the branches are hardly visible. Similar topologies were seen in trees obtained by other programs, the tree-construction option in clustal w (35) or the quartet puzzling method (38) (data not shown).