The primary structure of the mitochondrial form of horse liver aldehyde dehydrogenase has been determined, utilizing peptide analyses and homology with other enzyme forms. The subunit exhibits N-terminal heterogeneity in size similar to that for the corresponding human mitochondrial protein, the longest form having 500 residues. Catalase was identified as a contaminant of the preparations. All four pairs within a set of aldehyde dehydrogenases can now be compared, including the same two species variants (horse and human) for both the cytosolic and mitochondrial enzyme, revealing characteristic differences although Cys-302 and other segments of presumed functional importance are unchanged. The cytosolic and mitochondrial enzymes are clearly different (172 exchanges in the horse pair; 160 exchanges in the human pair) and the mitochondrial forms are more conserved (28 exchanges of 500 residues) than the cytosolic ones (43 exchanges). Distributions of the residue substitutions also differ between the two enzyme types. These results suggest a comparatively distant separation of the cytosolic and mitochondrial enzymes into forms with separate functional constraints that are more strict on the mitochondrial than the cytosolic enzyme. Unexpectedly, positions with residues unique to one of the four enzymes are about twice as common in both of the horse proteins than in either of the human proteins. This difference may reflect a general pattern for human/non-human proteins, showing that not only functional properties of the protein, but also other factors, such as generation time (longer in man than in horse), are important for enzyme divergence.

The class II enzyme of human liver alcohol dehydrogenase was isolated, carboxymethylated, and cleaved with CNBr and proteolytic enzymes. Sequence analysis of peptides established structures corresponding to the pi subunit. Two segments from the C-terminal region unique to pi were selected for synthesis of oligodeoxyribonucleotide probes to screen a human liver cDNA library constructed in plasmid pT4. Sequence analysis of two identical hybridization-positive clones with cDNA inserts of about 2000 nucleotides gave the entire coding region of the pi subunit, a 61-nucleotide 5' noncoding region and a 741-nucleotide 3' noncoding region containing four possible polyadenylation sites. Translation of the coding region yields a 391-residue polypeptide, which in all regions except the C-terminal segment corresponds to the protein structure as determined directly by peptide analysis. With the class I numbering system, the exception concerns a residue exchange at position 368, the actual C-terminus which is Phe-374 by peptide data but a 12-residue extension by cDNA data, and possibly two further residue exchanges at positions 303 and 312. The size difference might indicate the existence of posttranslational modifications of the mature protein or, in combination with the residue exchanges, the existence of polymorphism at the locus for class II subunits. The pi subunit analyzed directly results in a 379-residue polypeptide and is the only class II size thus far known to occur in the mature protein.(ABSTRACT TRUNCATED AT 250 WORDS)