We previously derived the cDNA sequence of the desaturase (Thiede et al., 1986 ) and constructed an expression vector for the production of active desaturase in Escherichia coli (Strittmatter et al., 1988 ). DNA constructs expressing desaturase in mice (Kaestner et al., 1989 ), Saccharomyces cerevisiae (Stukey et al., 1990 ), and a soluble form of desaturase in plants (Thompson et al., 1991 ; Shanklin and Somerville, 1991 ) have also been reported. A form of desaturase up-regulated in response to cold has been described in fish (Tiku et al., 1996 ).

Previously, we describe the isolation, cDNA sequence, and bacterial expression of rat liver desaturase (Thiede et al., 1986 ; Strittmatter et al., 1988 ). The results described in the current report provide a glimpse of the proteolytic processing of this short-lived membrane protein. As seen in Figure ​Figure1,1, fasting and refeeding a fat-free, high-carbohydrate diet induced high levels of desaturase in the liver microsomal membranes. When the dietary regimen was stopped, desaturase levels rapidly decreased to levels not detectable by immunoblots (Figure ​(Figure1,1, lane 3). An in vivo half-life of about 2 h has been estimated for the desaturase (Oshino and Sato, 1972 ).

Surprisingly, a considerable amount of desaturase was present in the subcellular fractions P-1 and P-2, in addition to being present in the microsomes (Table ​(Table1).1). The postmicrosomal supernatant or the high-salt wash of the microsomes did not contain desaturase (Figure ​(Figure1,1, lane 6). While the P-1 fraction consisted essentially of nuclear material, P-2 subcellular fraction was heterogeneous organelle preparation. The outer nuclear membrane of hepatocytes is continuous with the endoplasmic reticulum, implying that some of the desaturase-containing membranes may traffick to the nuclear membrane. The amount of desaturase associated with P-1 and P-2 subcellular material was significant (Table ​(Table1)1) and unanticipated.

To determine the relationship between the microsomal and putative nuclear enzyme present in the P-1, P-2 subcellular fractions, desaturase from the latter fractions was purified for sequence analysis. N-terminal sequence analysis of the P-1, P-2 preparations showed an absence of three residues present in the microsomal protein, whereas the microsomal enzyme has a blocked N terminus (Table ​(Table2).2). Thus, the desaturase present in the nuclear fraction does not represent microsomal enzyme contamination, but appears to represent a specifically processed form of the enzyme. The relationship between the N-terminal processing and the nuclear localization remains to be elucidated, as discussed below.

Incubation of the microsomal membranes at 37°C resulted in the complete degradation of the desaturase, whereas in microsomes washed with a high-salt buffer, the degradation was incomplete as seen in Figure ​Figure4.4. The complete desaturase degradation, however, could not be restored by the addition of concentrated cytosol or high-salt wash fractions to the microsomes. Complete degradation of desaturase in the high-salt–washed microsomes could be restored by the addition of lipids, cytochrome b₅ and its reductase, which constitutes functional desaturase activity. In these experiments, omission of lipid or any of the protein components limited proteolysis. It appears that in high-salt–washed microsomes, only 30–40% of the desaturase is degraded (Figure ​(Figure4C).4C). Although the possible effects of the salt wash on the desaturase degradation could be the result of many factors, one explanation may involve the formation of high-salt wash–induced conformations in the desaturase population that are resistant to the protease action. The data in Figure ​Figure5C5C suggest that supplementation of salt-washed microsomes with the lipid, cytochrome b5 reductase, and cytochrome b5 renders the resistant form of the desaturase to further proteolysis. The formation of insoluble desaturase aggregates have been observed during centrifugation on glycerol gradient detergent-solubilized microsomal preparations in the presence of high salt. Some of these forms retain enzymatic activity whereas others do not. Notwithstanding, the degradation reconstitution experiments imply that the procedure used to reconstitute an enzymatically active desaturase system may also yield to protein conformations that are susceptible to the proteolysis of the enzyme.

Desaturase in P-1 or P-2 fractions was degraded completely. Premixing of the salt-washed microsomes with the P-1, P-2 fraction also resulted in a complete degradation of the desaturase. Desaturase antigen bands of lower molecular mass than the desaturase could not be detected in the degradation mixtures, although the desaturase antibody can recognize the bacterial synthesis product lacking some 30 residues from the N terminus, corresponding to a decrease of 3000–5000 Da (Strittmatter et al., 1988 ).

Hepatic lysosomal or endosomal proteases or their precursors are ubiquitous enzymes, and their presence in microsomes and in P-1 and P-2 fractions would not be surprising, since proteases such as the procathepsin B and L may exist in the microsomal membranes as latent precursors. To determine whether lysosomal proteases are involved in the desaturase degradation, several types of protease inhibitors were examined. Leupeptin and pepstatin, inhibitors of lysosomal and endosomal proteases, had no effect on the microsomal or the P-1, P-2 desaturase degradation (Figures ​(Figures66 and ​and7).7). The cysteine protease inhibitor ALLM and the serine protease inhibitor PSMF were also ineffective in blocking the desaturase degradation (Figure ​(Figure7).7). ALLM has been shown to inhibit the regulated degradation of microsomal HMG-CoA reductase (Inoue et al., 1991 ), and a serine protease has been implicated in the rapid degradation of unassembled Ig light chains in endoplasmic reticulum (Gardner et al., 1993 ). One nonlysosomal pathway present in the cytoplasm and nuclear components that mediates rapid elimination of proteins is the proteosome pathway. Multiple types of evidence suggest that the proteosome plays a key role in the processing of antigens for the major histocompatibility complex class I presentation (Chiechanover, 1994 ) and is involved in generating the active forms of molecules such as the production of the 50-kDa subunit of the transcription factor NF-κB from the 105-kDa precursor (Palombella et al., 1994 ). The proteosome is also thought to be responsible for the degradation of the HMG-CoA reductase (McGee et al., 1996 ) and of the cystic fibrosis gene product (CFTR) in the endoplasmic reticulum (Rock et al., 1994 ; Ward et al., 1995 ). The proteosome is a 26S (2000-kDa) complex, containing the 20S proteosome as a key proteolytic component (Rechsteiner et al., 1993 ; Jentsch and Schlenker, 1995 ; Lowe et al., 1995 ). The 20S (700 kDa) complex consists of seven different α-subunits and seven unrelated β-subunits with masses ranging from 24 to 32 kDa comprising about 1% of the protein in mammalian cells (Jentsch and Schlenker, 1995 ). None of the individual subunits of the proteosome have proteolytic activity or show relationship to any known proteases. Recently, a highly specific, irreversible inhibitor of the proteosome, a Streptomyces metabolite–lactacystin has been identified (Fenteany et al., 1995 ). Lactacystin modifies covalently the highly conserved N-terminal threonine of the mammalian proteosome subunit X, a close homologue of the LMP7 proteosome subunit encoded by the major histocompatibility complex (Fenteany et al., 1995 ). Lactacystin has not been found to inhibit any other known protease (Fenteany et al., 1995 ). In view of such a remarkable housekeeping proteolytic function of the proteosome, it was of interest to determine whether the proteosome is involved in the degradation of desaturase. Lactacystin had no effect on the microsomal desaturase degradation (Figure ​(Figure8).8). The experiment of Figure ​Figure88 also shows that lactacystin (100–350 μM) also failed to inhibit the desaturase degradation in the P-1 and P-2 subcellular fractions.

The studies reported here show that degradation of desaturase occurs in several subcellular fractions isolated by differential centrifugation. The degradation of desaturase, however, was insensitive to the lysosomal and proteosome inhibitors. If the lysosomal proteases or proteosome do not play a significant role in the desaturase degradation, what alternatives do we have to explain the desaturase degradation? Proteolytic activities such as the ER-60 protease have been detected in detergent-solubilized microsomal preparations (Otsu et al., 1995 ). The proteolytic activity of ER-60, however, is inhibited by leupeptin and ALLM (Otsu et al., 1995 ).

The observation that desaturase is present and readily degraded in subcellular fractions other than the microsomes implies that degradation of native desaturase may also involve targeting of the enzyme to compartments containing specific proteolytic machinery, which constitute a sorting pathway or the reverse process of protein targeting to the membranes. Of interest is that the amino acid sequence of desaturase has two segments that contain a potential nuclear localization sequence (NLS). In residues 33–36, Lys-Met-Lys-Lys and Arg-Lys-Lys-Val-Ser-Lys, residues 335–340 constitute potential consensus sequences for the import of proteins to the nucleus. Import of proteins to the nuclear pore complex is specified by short stretches of amino acids known as the NLSs (see review in Melchior and Gerace, 1995 ; Gorlich and Mattaj, 1996 ). Site-directed mutagenesis of desaturase in the two putative NLS segments should clarify the significance of this finding. Are posttranslation modifications involved in this process? Structure analysis of desaturase in the P-1 and P-2 fractions showed that it lacked three residues at the N terminus (Figure ​(Figure22 and Table ​Table2).2). The cDNA sequence predicts a Met-Pro-Ala sequence at the N terminus of the microsomal desaturase (Strittmatter et al., 1988 ). The N terminus of the enzyme present in microsomes is blocked, and the nature of the blocking group remains to be determined. The N-terminal–blocking groups of the two upstream essential components of the desaturase pathway, cytochrome b₅ and its reductase, are an acetyl and myristoyl residue, respectively (Ozols et al., 1984 ; Ozols, 1989 ). The presence of a myristoylated residue at the N terminus of desaturase is unlikely because of the absence of a consensus Gly residue in the proximity of its N terminus. The removal of an N-blocked terminus and ProAla segment from the native desaturase is of interest because we are not aware of any reports on hepatic aminopeptidases capable of cleaving residues from N-acetylated proteins. The hepatic acylpeptide hydrolase (E.C.3.4.19.1) acts only on N-acetylated peptides that are shorter than 10 to 15 residues (Tsunasawa et al., 1983 ). Cathepsins that function as aminopeptidases act only on proteins with a free N terminus.