Bacterial synthesis of active rat stearyl-CoA desaturase lacking the 26-residue amino-terminal amino acid sequence.

Two clones containing inserts in pBR322 that together include the entire 1074-base open reading frame coding for the 358 amino acids of rat liver stearyl-CoA desaturase have been used to construct expression vectors for residues 3-358 and 27-358 fused to the first 6 residues of beta-galactosidase and several amino acids of the multiple cloning site of pUC8. Growth of transformed Escherichia coli under conditions for suppression of the lac promoter, followed by subsequent induction of these cultures results in the synthesis of higher levels of desaturase proteins than those found in induced rat liver. The proteins are almost completely associated with the membrane fraction of cell homogenates. Posttranslational iron insertion into the apoproteins, either in vitro with membrane preparations or by iron addition during induction, results in the formation of active holoenzyme which can be reconstituted with NADH cytochrome b5 reductase and cytochrome b5 to form an active stearyl-CoA desaturase system. The deletion of the first 26 amino-terminal amino acid residues does not affect either enzyme activity or membrane binding. Therefore, the unusual sequence of 11 residues containing 10 amino acids with hydroxyl groups plays no apparent significant role in either protein insertion into membranes or iron chelation. Since the protein product for residues 3-358 is processed even further to delete the initial 33 amino-terminal residues, the limiting polypeptide primary structure required for an active membrane-bound catalyst is even smaller than this initial deletion mutation indicates.

Two clones containing inserts in pBR322 that together include the entire 1074-base open reading frame coding for the 358 amino acids of rat liver stearyl-CoA desaturase have been used to construct expression vectors for residues 3-358 and 27-358 fused to the first 6 residues of @-galactosidase and several amino acids of the multiple cloning site of pUC8. Growth of transformed Escherichia coli under conditions for suppression of the lac promoter, followed by subsequent induction of these cultures results in the synthesis of higher levels of desaturase proteins than those found in induced rat liver. The proteins are almost completely associated with the membrane fraction of cell homogenates. Posttranslational iron insertion into the apoproteins, either in vitro with membrane preparations or by iron addition during induction, results in the formation of active holoenzyme which can be reconstituted with NADH cytochrome b, reductase and cytochrome bs to form an active stearyl-CoA desaturase system. The deletion of the first 26 amino-terminal amino acid residues does not affect either enzyme activity or membrane binding. Therefore, the unusual sequence of I1 residues containing 10 amino acids with hydroxyl groups plays no apparent significant role in either protein insertion into membranes or iron chelation. Since the protein product for residues 3-358 is processed even further to delete the initial 33 amino-terminal residues, the limiting polypeptide primary structure required for an active membrane-bound catalyst is even smaller than this initial deletion mutation indicates.
Steuryl-CoA desaturase is the terminal component of the electron transport system of endoplasmic reticulum that utilizes cytoplasmic NADH and molecular oxygen to effect the A9 desaturation of CoA esters of long chain fatty acids (1). All three enzyme components of this sequence, NADH cytochrome bs reductase, cytochrome b5, and the desaturase, are oriented with their catalytic domains on the cytoplasmic side of this organelle (1,2). Both the cytochrome and desaturase are synthesized on cytoplasmic ribosomes and undergo posttranslational insertion into endoplasmic reticulum and binding of either heme or iron, respectively, to form active holoenzymes (3, 4). However, the details of the posttranslational *This work was supported by Public Health Service Research Grants GM-15924 (to P. S.) and GM-29595 (to J. 0.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
$ To whom correspondence should be addressed. iron binding and membrane insertion are not clear. We have previously exploited the fact that nutritional manipulation can be used to induce high levels of the desaturase ( 5 ) to characterize the induction and structure of desaturase mRNA (4). This 4,900-base mRNA was then used for primer extension of cDNA coding for the enzyme (6). Of the six overlapping cDNA segments required to sequence the entire cDNA, the first two 5' segments in plasmids pDsl and pDs2 contain the 1,074-base open reading frame coding for the 358 amino acids, corresponding to a molecular mass of 41,400 daltons. These two plasmids were therefore used here to construct expression vectors, which, under suitable conditions, synthesize desaturase peptides that undergo membrane insertion and iron binding to form active holoenzyme that can be reconstituted with mammalian cytochrome b5 and cytochrome b, reductase. To initiate a systematic examination of both protein-protein and protein-membrane interactions, as well as the functional roles of active-site residues by selected mutational deletions of peptide segments and sitedirected mutagenesis of specific residues, we have shown that the desaturase peptide with a 26-residue amino-terminal deletion will bind iron and can be reconstituted with cytochrome b5 and NADH cytochrome b5 reductase to form an active stearyl-CoA desaturase system.

EXPERIMENTAL PROCEDURES
Materials-The enzymes used in the construction and characterization of the plasmid vectors were obtained from either Bethesda Research Laboratories or New England Biolabs. Plasmids pDsl and pDs2 were constructed previously (6), pUC8 was the construction of Viera and Messing (7), and Escherichia coli strain UT-481 (A(hcpro),r-,m-,lacl",lacZ) was obtained from Dr. Gordon Carmichael of the Department of Microbiology of this institution. Cytochrome b, (8), cytochrome b, reductase (91, and desaturase antibody (4) were prepared as described previously.
Standard Procedures-Procedures such as plasmid isolation, restriction fragment isolation, plasmid constructions, transformations, colony hybridization and restriction characterization of plasmids were those published by Maniatis et al. (IO). Protein electrophoresis in sodium dodecyl sulfate and 12.5% polyacrylamide gels were as described by Laemmli (ll), and immunoblotting was done by the method of Towbin et al. (12).
Construction of pDs27-358"Initially a clone, pDs3, (Fig. 1A) containing the entire coding sequence for desaturase was identified from a ligated AuaI digest of a mixture of pDsl and pDs2 (6). This plasmid of pBR322 contains the insert at the PstI site of the plasmid with a short 5"noncoding sequence from pDsl and a similar 3' extension past the termination codon from pDs2. Further restriction characterization with PstI, RsaI, AuaI-PstI, AceI-PstI, and BglII-PstI were used to establish that pDs3 contains the complete 1935-base pair insert shown in Fig. 1A.
The PstI-Hind111 restriction fragment from pDs3 was then ligated to linear pUC8 produced by PstI-Hind111 digestion. Transformation of UT-481 cells produced clones on plates containing 0.5% glucose, to suppress expression, which contained the expected 1131 base pair insert. One of these, designated pDs27-358 ( Fig. l B ) , was shown to have the expected sequence from nucleotide 181-1131. Since ligation places this insert in frame with the first 6 residues of @-galactosidase and 5 residues of the multiple cloning site, the expected protein product will be a fusion protein with these 11 residues amino-terminal to residue 27 of the desaturase (6) and continuing to residue 358 as determined by the termination codon for desaturase. Construction of pDs3-358"Two restriction fragments were isolated from pDsl; a 384-base pair AuaI-AuaI segment after digestion with AuaI, and a second by NaeI digestion at 106 followed by addition of a 10 base pair BamHI linker to obtain the 453 base pair BamHI-AuaI restriction fragment. AuaI-Hind111 digestion of pDs2, and BamHI-Hind111 restriction of pUC8 were used to isolate the 367-base pair AuaI-Hind111 fragment and linear pUC8. Forced ligation, due to the sequence differences in the two AuaI sites, produced a high frequency of clones containing the expected 1211-base pair insert in pUC8 in frame with the sequence for the initial 6 residues of pgalactosidase and 4 residues of the multiple cloning site prior to the codon for Ala-3 of the desaturase. This was confirmed by extensive restriction characterization of one clone. For reasons not pertinent to this study, the Hind111 site following the termination codon was replaced by insertion of a Sac1 linker. The result was the plasmid pDs3-358 ( Fig. 1C) coding for a fusion protein with the 9 residues from pUC8 preceding residue 3 of the desaturase and continuing to the termination codon after residue 358.
Induction of Desaturase-UT-481 cells transformed with either pDs3-358 or pDs27-358 were grown in an enriched medium containing glucose (20 g of tryptone, 10 g of yeast extract, 10 g of NaCl and 10 g of glucose/liter) to stationary phase overnight to achieve maximum cell density with repression of the plasmid luc promotor in this hcIq strain. Cells were then washed and placed in a medium without added glucose, but including isopropylthiogalactoside and FeC13 (10 g of tryptone, 5 g of yeast extract, 10 g NaCl/liter, and 0.3 mM isopropylthiogalactoside and 50 p~ FeCl,) to induce synthesis of the plasmid protein product aerobically at 37 "C for 3 h.
Preparation of Membranes-Membrane fractions were isolated from either repressed or induced cells by centrifuging the cultures at 4,000 X g for 5 min, washing the packed cells first with cold 20 mM Tris acetate, 5 mM EDTA, 0.2 mM phenylmethylsulfonyl fluoride, pH 8.1, and then with 20 mM Tris acetate, 0.2 mM phenylmethylsulfonyl fluoride, pH 8.1, and employing a French press to achieve cell disruption in 2-3 volumes of the latter buffer per volume of packed wet cells. After dilution to 10 volumes with the same buffer, unbroken cells were removed at 2 "C by centrifuging for 10 min at 4,000 X g. The membrane fraction was obtained as a pellet by centrifuging for 50 min at 2 "C and 165,000 X g. The membrane fractions, designated MR or MI for membranes from repressed or induced cells, were resuspended in a volume equal to the original volume of packed wet cells by sonication in 20 mM Tris acetate, pH 8.1. To solubilize nonpolar membrane-bound proteins, including any desaturase peptide, the membrane suspensions were brought to 2.0% Triton X-100 and 0.4% sodium deoxycholate and sonicated intermittantly at 2 "c for 10 min before centrifuging for 45 min at 165,000 x g. The supernatant fluid containing detergent-extracted protein are designated M-T.
Reconstitution of the Stearyl-CoA Desaturase System-Prior to reconstitution with cytochrome b,, cytochrome b, reductase, and phospholipid, all preparations were brought to 2% Triton x-100 and 0.4% sodium deoxycholate. The reconstitution and assay are as described (8). Typically 17 p1 of 600 p M cytochrome b,, 6 pl of 110 p M NADH cytochrome b, reductase and 20-40 p1 of 30 mM egg phosphatidylcholine were added to 115 p1 of preparations containing detergent at 0 "C. After incubation for 1-2 h at 0 "C, enzyme activity was measured at 25 "C by the rate of NADH oxidation in the presence and absence of stearyl-CoA as described (8). To estimate the concentration of active desaturase a turnover number of 28/min was used (14).
Purification and Sequence Analysis of Desaturme-Desaturase was purified from the membrane fraction from pDs3-358 by modification of the procedure described previously for the rat liver enzyme (1). Here, initial deoxycholate extractions were with 1.0 and 1.5% detergent, followed directly by solubilization in 2.5% Triton X-100 and 10 mM CaC12. Micro sequence analysis of the desaturase protein band, eluted from a preparative electrophoresis gel with 40 mM NH4HC03, 0.02% sodium dodecyl sulfate buffer using the Elutrap chamber from Schleicher & Schuell, according to their instructions, was carried out on an Applied Biosystems Model 470A gas-phase sequenator, equipped with a Model 120A PTH analyzer, according to the manufacturer's instructions.  (lanes 3  and 5 ) . Clearly, the synthesized desaturase becomes associated with the membranes of these cells since the samples in Fig. 2 are detergent extracts of the membrane fractions. Moreover in preliminary experiments antigen was only found in membrane fractions and was not detectable in the supernatant fluid of homogenates following membrane sedimentation. From the immunoblots shown in Fig. 2, as well as other similar experiments, the concentration of desaturase peptides was approximately 1.25-2.5 pg/mg bacterial protein. This constitutes approximately 2-4% of the total protein of the inner membrane and must be considered a minimum value since transfer of the desaturase from acrylamide gels to the nitrocellulose acetate sheets may be less efficient in the membrane preparations that contain a spectrum of other proteins compared to transfer of the purified desaturase standards (lanes 1 and 6). This value for bacterial expression of the desaturase is actually higher than the levels estimated for induced rat liver (approximately 0.75 pg/mg liver protein) (8).

RESULTS AND DISCUSSION
There is proteolytic processing of the desaturase synthesized from pDs3-358 (lane 3) since, as a fusion protein with the initial 9 residues derived from pUC8, the expected molecular weight is actually slightly larger than purified rat liver desaturase (lanes 1 and 6). There is a faint antigen band in the position of full length product in lane 3, but the major desaturase antigen has an apparent molecular mass corresponding to a decrease of approximately 3000-5000 daltons. Although both EDTA and a serine protease inhibitor were included in the preparations of cell-free extracts at a low temperature, it is not clear whether this proteolysis occurs in situ or during membrane isolation. The plasmid coding for the product beginning at residue 27 of the desaturase, interestingly, is more resistant to proteolysis. The expected molecular mass for the pDs27-358 peptide, with 11 residues from pUC8 fused to Ala-27, should be approximately 2000 daltons lower than the rat liver standards and this is the major antigen detected in lane 5. A smaller amount of protein is processed further to a molecular weight similar to that of the major antigen from pDs3-358.
To determine the site of the processing of the desaturase from pDs3-358, the enzyme was purified from the membrane fraction as shown in Fig. 3. A major antigenic peptide ( A , lune  2) representing the dominant protein species ( B , lune 2 ) was isolated by preparative gel electrophoresis and subjected to sequence analysis (Table I). The sequence of the first 20 residues are identical to residues 34-53 of the primary structure of rat liver desaturase (6). The processed protein therefore represents a deletion of the amino-terminal 33 residues of the rat liver enzyme corresponding to 4200 daltons. The recent studies of Bachmair et al. (15) on the effect of the amino-terminal amino acid on the half-life of proteins in yeast have shown that amino-terminal Arg is particularly effective in destabilizing proteins to proteolysis. The coding sequence for the 4 amino acids preceding residue 3 of the desaturase in pDs3-358 yields an Arg-Gly-Ser-Arg sequence. Thus, any fortuitous initial processing that placed either arginyl residue amino-terminal might result in the rapid processing observed with this plasmid product.
Preliminary attempts to reconstitute the desaturase system by cytochrome b, and NADH cytochrome b5 reductase addition to membrane fractions isolated from cells transformed  with pDs27-358 and induced in the absence of iron yielded low levels of desaturase activity which were stimulated approximately 2-2.5-fold by preincubation in buffers containing 10 p~ concentrations of ferric glycinate. Thus, cells grown in low iron medium produced membrane-bound apoprotein of the desaturase in significant quantities. For this reason subsequent induction media for insuring posttranslational iron insertion to form the active holoenzyme included 50 PM FeC13. The ability to sequester undenatured apodesaturase in the bacterial membranes, however, does offer a method for determining whether iron insertion in mammalian tissues is spontaneous or requires an iron carrier or catalysis to effect iron insertion at the ambient intracellular iron concentrations. Whereas repeated previous attempts to prepare undenatured apoenzyme from purified desaturase have failed, the bacterial membrane fraction containing apodesaturase represents the ideal substrate for monitoring the mechanism of such iron binding. MI, membranes from induced cells; MI-T, detergent extracts of membranes from induced cells; and SI , the supernatant from MI preparations. These preparations, reconstitution of the desaturase system, and assay procedures are described under "Experimental Procedures." Activities are expressed as nanomoles of substrate desaturation per minute per aliquot of preparation derived from 10 ~1 of wet packed cells, and the amounts of active desaturase were calculated using a turnover number of 28/min.  Table I1 shows the levels of desaturase activity obtained with membrane fractions from cells induced in the presence of iron. The low levels of activity measured in preparations from repressed cells are near the limits of reliability of the assay method with crude enzyme preparations. Nevertheless, induced cells show at least an order of magnitude elevation of enzyme activities, concentrated almost completely in the membrane fractions. Virtually all of this activity is found in the Triton X-100 and sodium deoxycholate extract of membranes derived from induced cells as would be expected from the behavior of the apolar rat liver enzyme (8). Calculations of the amount of active desaturase in induced cells transformed with either plasmid using the turnover number of 28/ min (14) yields values that are similar to the levels of antigen estimated from the immunoblots. Thus, virtually all of the desaturase is converted to active holoenzyme during protein synthesis in an iron-enriched medium. The deletion of the 26-residue segment from the amino terminus of the desaturase represents the first mutational modification of the protein designed to identify the limiting sequences and particular residues required for membrane and iron binding to yield the substrate and cytochrome b6 interactions involved in stearyl-CoA desaturation. Clearly, the first 26 residues are not required for either membrane insertion or enzymatic activity. This peptide segment (6), which contains an unusual sequence of 11 residues (10-20) containing 10 residues with a hydroxyl group, therefore has no apparent functional significance. Moreover, the processing of the enzyme from pDs3-358 extends the limited catalytic significance of the amino-terminal segment of the desaturase through at least 33 residues.
The primary structure of the desaturase (6) is similar to that of one form of cytochrome P-450 (16). Both intrinsic membrane proteins contain several polar segments rich in cationic charges which are interrupted with several highly nonpolar sequences that may serve to orient the polar loops in the aqueous phase by insertion of the nonpolar segments in the hydrocarbon region of lipid bilayers. Extension of mutational deletions to selected "loop" or "anchoring" regions should identify domains that participate in specific binding or catalytic sequences. As the cationic residues involved in interactions with cytochrome b5 (14), the 2 tyrosyl residues involved in iron binding (17), and the 2 arginyl residues required for stearyl-CoA binding (17) are identified in the primary structure, selected site-directed mutagenesis will also address the questions of the functional roles of single amino acid residues.
In this respect, the recent success of Beck von Bodman et ai. (18) in constructing an expression vector for cytochrome bs is of particular interest. The application of mutational methods to a detailed examination of structure-function relationships in this amphipathic heme protein will not only provide comparative data on a second oxidative membrane protein, but the combined information on both proteins should be complementary in understanding the precise protein-protein interactions that govern electron transfer between them.