Molecular Cloning and Expression of Mammalian Peroxisomaltrans-2-Enoyl-coenzyme A Reductase cDNAs*

Chain elongation of fatty acids is an important cellular process and is believed to occur in the endoplasmic reticulum of all eukaroytic cells. Herein we describe the cloning and characterization of a peroxisomal NADPH-specifictrans-2-enoyl-CoA reductase, the key enzyme for a proposed peroxisomal chain elongation pathway. The reductase was solubilized and partially purified from guinea pig liver peroxisomes by affinity chromatography. On SDS-polyacrylamide gel electrophoresis, a 40-kDa band was identified as the enzyme, and its partial amino acid sequence (27 amino acids) was determined. A full-length cDNA for the reductase was cloned from a guinea pig liver cDNA library. The open reading frame of this nucleotide sequence encodes a 302-amino acid polypeptide with a calculated molecular mass of 32.5 kDa. Full-length mouse and human cDNA clones encoding homologous proteins have also been isolated. All of these translated polypeptides have the type I peroxisomal targeting signal, AKL, at the carboxyl terminus. The identity of the cloned enoyl-CoA reductase cDNAs was confirmed by expressing the guinea pig and human cDNAs in Escherichia coli. The His-tagged recombinant enzymes were found to have very high NADPH-specific 2-enoyl-CoA reductase activity with similar properties and specificity as the liver peroxisomal reductase. Both the natural and the recombinant enzyme catalyze the reduction oftrans-2-enoyl-CoAs of varying chain lengths from 6:1 to 16:1, having maximum activity with 10:1 CoA. Northern blot analysis demonstrated that a single transcript of 1.3 kilobases is present in most mouse tissues, with particularly high concentrations in liver and kidney.

In mammalian cells, the fatty acid chain elongation system has been shown to be present in both endoplasmic reticulum (ER) 1 and mitochondria (1). The ER pathway is similar to the fatty acid synthesis system. In this multistep process, malonyl-CoA first condenses with acyl-CoA to form a ␤-keto acyl-CoA with two additional carbons with the release of CO 2 . The ␤-keto acyl-CoA then undergoes reduction of the ketone group followed by dehydration to form a 2-enoyl-CoA, which is then reduced by NADH or NADPH to form the longer (ϩ2C) acyl-CoA. Chain elongation and other biotransformation of fatty acids have been shown to occur primarily in the ER (1,2). The mitochondrial chain elongation system is a reversal of the fatty acid ␤-oxidation system; thus, acetyl-CoA, instead of malonyl-CoA, is used for the condensation reaction, and the resulting ␤-keto acyl-CoA undergoes the same series of reactions as described above (3). The physiological function of the mitochondrial chain elongation system is not clear (4).
Contradictory findings have been reported regarding the presence of another cellular fatty acid chain elongation system in mammalian peroxisomes. Nagi et al. (5) initially reported the absence of an elongation system in rat liver peroxisomes. However, Horie et al. (6) later provided good evidence for the presence of an active peroxisomal chain elongation system in rat liver. This peroxisomal system was shown to be similar to the mitochondrial system in that acetyl-CoA is used as the carbon donor for chain elongation. In both, all of the chain elongation reactions are apparently catalyzed by the fatty acid ␤-oxidation pathway enzymes in the reverse directions except for the last step (i.e. the reduction of the 2-enoyl-CoA). This is because in the ␤-oxidation pathway the initial oxidation of acyl-CoAs to 2-enoyl-CoAs is an irreversible step catalyzed by either acyl-CoA dehydrogenase (mitochondria) or oxidase (peroxisomes). Therefore, a separate 2-enoyl-CoA reductase is necessary for the reduction of the double bond at C-2 at the last step of elongation. Mitochondria have been shown to contain such a specific 2-enoyl-CoA reductase, which has been purified to homogeneity as a 35.5-kDa protein (4). This purified enzyme only utilized NADPH as the coenzyme. Horie et al. (6) provided good evidence for the presence of a similar NADPH-linked enoyl-CoA reductase in rat liver peroxisomes, although the enzyme was not solubilized or purified.
During the course of purification of guinea pig liver peroxisomal acyldihydroxyacetone phosphate (DHAP) reductase (7), a contaminating 40-kDa protein was found to be associated with the reductase. This protein was found to bind to the NADPH affinity column along with the acyl-DHAP reductase. We report here that this protein is a 2-enoyl-CoA reductase. From its partial amino acid sequence information, we have cloned the cDNA of this guinea pig liver protein and also the corresponding mouse and human cDNAs. The guinea pig cDNA was expressed in Escherichia coli, and the recombinant protein has been found to have high trans-2-enoyl-CoA reductase activity with NADPH as the reductant.

EXPERIMENTAL PROCEDURES
Synthesis of trans-2-Enoyl-CoAs-trans-2-Hexenoic and octenoic acids were obtained from Aldrich. trans-2-Decenoic, 2-dodecenoic, and 2-hexadecenoic acids were kind gifts from Dr. Shuichi Horie (Teikyo University, Kanagawa, Japan). trans-2-Tetradecenoic acid was synthesized by condensing dodecyl aldehyde (Aldrich) with malonic acid in pyridine followed by decarboxylation as described by Lauer et al. (8). All-trans-2,4-decadienoic acid was prepared by oxidizing 2-trans,4trans-decadienal (Aldrich) by Ag 2 O as described (9). The CoA derivatives of all these fatty acids were prepared by first converting them to acyl chlorides, followed by condensing the acyl chlorides with CoASH (1.1:1.0 molar ratio) in carbonate buffer at pH 8.5 as described previously (10). The 12:1, 14:1, and 16:1 CoA were purified by perchloric acid precipitation followed by acetone and ether washings (10). The crude 6:1, 8:1, and 10:1 CoA preparations were purified by acidification to pH 3 followed by extraction with diethyl ether to remove the free fatty acids. The pH of the aqueous layers was then adjusted to 6.0, and the enoyl-CoA was lyophilized. The 10:1 CoA was further purified by hydrophobic chromatography on an octadecyl-silica column. The decenoyl-CoA dissolved in 0.1 M sodium phosphate buffer (pH 6.0) was loaded onto a 2-ml C-18 SPE-Bakerbond cartridge (J.T. Baker), and after washing the column with 8 ml of the same buffer, the enoyl-CoA was eluted with 4 ml of methanol. The methanol was removed at room temperature by blowing a stream of N 2 , and the residue was dissolved in 0.05 M sodium phosphate buffer (pH 6.0). The recovery of the enoyl-CoA was 80%. Thin-layer chromatography of the acyl-CoAs on Silica Gel 60 (Merck) using butanol/acetic acid/water (50:20:30) as the mobile phase (10) showed that the acyl-CoAs (R f ϭ 0.45) were pure, with only traces of free CoA (R f ϭ 0.16) and no free fatty acids (R f ϭ 0.82) present as contaminant. The enoyl-CoAs were also characterized by their spectral properties (10,11). Crotonyl-(trans-4:1⌬ 2 )-CoA was purchased from Sigma. The concentrations of the acyl-CoAs were determined by measuring the amount of CoASH released after alkaline hydrolysis, with 5,5Ј-dithio-bis(2-nitrobenzoic acid) (12).
Assay of Enoyl-CoA Reductase-Spectrophotometric and radiometric assays were used to measure the enzyme activity. In the radiometric assay, the incubation mixture contained phosphate buffer (50 mM, pH 7.4), S-[4-3 H]NADPH (70 M, 8000 dpm/nmol), fatty acid-poor bovine serum albumin (2 mg/ml), enoyl-CoAs (30 -40 M), and enzyme protein in a total volume of 0.3 ml. After incubation at 37°C for 5 min, the reactions were stopped by adding 50 l of 1 N NaOH. The mixture was incubated at 37°C for an additional 10 min to hydrolyze all acyl-CoAs and then acidified by adding 10 l of 6 N HCl followed by the addition of 0.75 ml of 2 M KCl containing 0.05 M H 3 PO 4 . The free fatty acids were extracted from the mixture with 1 ml of toluene, and the 3 H radioactivity in an aliquot of the toluene layer was determined by liquid scintillation spectrometry.
For the spectrophotometric assay, the incubation mixture was the same as above except that nonradioactive NADPH (0.l mM) was used instead of the [ 3 H]NADPH. The time course of oxidation of NADPH was monitored at 340 nm using a Beckman DU70 spectrophotometer. Controls containing all ingredients except the enoyl-CoAs were always run in parallel, and the control values were subtracted from the experimental ones.
Purification of Enoyl-CoA Reductase from Peroxisomes-Guinea pig liver peroxisomes, isolated as described (14), were suspended in 0.25 M sucrose, 10 mM TES, 1 mM EDTA buffer (pH 7.5) containing 0.1% Triton X-100 and protease inhibitors (1 M leupeptin, 0.2 mM phenylmethylsulfonyl fluoride). The mixture was stirred for 30 min at room temper-ature and then centrifuged at 100,000 ϫ g for 40 min. The enzyme from the pellet was solubilized in the above buffer containing also 1.0 M KCl and 0.3 mM NADPH. The mixture was centrifuged as above, and the supernatant was dialyzed overnight against a buffer (pH 7.4) containing 10 mM Tris, 2 mM dithiothreitol, 20% glycerol, 0.05% Triton X-100, and 0.2 M KCl. The dialyzed sample was centrifuged at 100,000 ϫ g for 40 min to remove the precipitated acyl-DHAP reductase (7), and the supernatant was loaded onto a 2-ml agarose gel column containing bound NADP ϩ (4 -5 mol/ml) prepared as described previously (7). The column was washed with 12 ml (six fractions of 2 ml each) of the above dialyzing buffer, and then the enzyme was eluted from the column by using 12 ml (in 2-ml fractions) of the same buffer but also containing NADPH (2 mM).
SDS-PAGE Analysis-SDS-PAGE and staining of the gel with Coomassie Blue G250 were as described before (15). Proteins from dilute samples were concentrated by trichloroacetic acid coprecipitation with carrier sodium deoxycholate (16) before electrophoresis. The precipitates were washed once with acetone to remove the trichloracetic acid, dissolved in 15 l of sample buffer (0.1 M Tris-HCl, pH 6.8, 2% SDS, 20% ␤-mercaptoethanol, 10% glycerol, and 0.02% bromphenol blue) by heating at 100°C, and loaded on the gel.
Characterization of the Reaction Product by Argentation-Thin Layer Chromatography-The AgNO 3 -containing thin layer chromatography plates were prepared by dipping the plates (Silica Gel 60; Merck) in a 30% AgNO 3 solution in acetonitrile. After air-drying, the plates were activated by heating in an oven at 110°C for 1 h, cooled, and used immediately. The 3 H-labeled product formed by incubating 10:1 CoA and S-[4-3 H]NADPH with the enzyme was converted to free fatty acids by hydrolysis as described above for the radiometric assay method. Fifty g each of decanoic acid and trans-2-decenoic acid were added to the toluene extracts containing the labeled free acids, and the toluene was removed by blowing nitrogen. The fatty acids were converted to the methyl esters by treatment with 2,2Ј-dimethoxypropane-methanol-HCl as described previously (17). The methyl esters, along with the methyl ester standards were spotted on the AgNO 3 -impregnated plate, and the plate was developed with hexane/diethyl ether/acetic acid (70:30:1). In this system, the methyl ester of the trans-2-enoic fatty acids migrates at a slower rate (R f ϭ 0.56) than the corresponding methyl esters of saturated fatty acids (R f ϭ 0.64), as also described by Wood and Lee (18). The methyl ester spots on the TLC plate were visualized under UV light after spraying with Primuline (19) and scraped out, and the esters were extracted by chloroform-methanol (1:1) followed by removal of the solvents by blowing air. The amount of 3 H in the extracts was quantified by liquid scintillation spectrometry. Partial Amino Acid Sequencing-Amino acid sequencing by Edman degradation was performed at the Protein Sequencing Facility of the University of Michigan. The partially purified enzyme was subjected to SDS-PAGE, the gel area corresponding to the 40-kDa band was excised and treated with CNBr (20), and the cleavage products were separated by SDS-PAGE followed by blotting on to a polyvinylidine difluoride membrane (Problott; Applied Biosystems). Two peptide bands from the blot, 3 and 6 kDa in size, were used directly for amino acid sequencing using an Applied Biosystems gas phase sequenator.
Screening Guinea Pig and Mouse Liver cDNA Libraries-Total RNA from tissues was isolated by using an acidic CHCl 3 -phenol extraction method using the Trizol reagent (Life Technologies, Inc.) following the manufacturer's protocol and then further purified by CsCl density gradient centrifugation (21). A guinea pig liver cDNA library in Zap Express vector was made by Stratagene from the total RNA isolated from livers of adult animals. An adult mouse liver cDNA library cloned into ZapII vector was purchased from Stratagene. The cDNA library screenings were done as described before (22). The filter paper blots of the colonies were hybridized with a mouse enoyl-CoA reductase cDNA probe labeled by using the random decamer priming method (Decaprime II; Ambion) and [␣-32 P]dATP. The template DNA used for making the 32 P-probe was a 424-base pair (Ϫ45 to ϩ378) mouse enoyl-CoA reductase cDNA fragment prepared by digestion of the plasmid isolated from a mouse EST clone (AA241896 Genome Systems) with EcoRI and SphI.
Expression and Purification of the Recombinant Proteins-A DNA fragment containing the open reading frame of the guinea pig liver reductase cDNA in Zap Express vector (clone 14) was excised by digestion with EcoRI and XhoI, purified by agarose gel electrophoresis (1.2-kb fragment), and then ligated into EcoRI-and XhoI-digested linearized 5.3-kb PET 28a(ϩ) vector (Novagen) using T4 DNA ligase (Promega) in the presence of ATP. The resulting plasmid encodes a protein in which an amino-terminal His tag from the vector is fused in frame to the enoyl-CoA reductase coding sequence. The plasmid was introduced into competent BL-21 cells by heat-shock treatment according to the manufacturer's (Promega) instructions, and the transformed cells were selected for by utilizing their resistance to Kanamycin. The plasmid-containing bacteria were grown in LB broth with Kanamycin (40 g/ml) at 37°C to the mid-log phase (A 600 ϭ 0.3-0.4), IPTG (final concentration 0.5 mM) was then added, and the cells were allowed to grow for another 4 -5 h. The cells were harvested by centrifugation and then lysed by treating with lysozyme (0.1 mg/ml) at pH 8.0 (50 mM Tris-HCl) in the presence of 0.1% Triton X-100 for 20 min at 30°C. The mixture, cooled in ice, was sonicated two times for 10 s each using a probe sonicator (Branson). The sonicated mixture was centrifuged at 12,000 ϫ g for 10 min, and the supernatant was collected. This supernatant was used both to measure the reductase activity and for the affinity purification of the recombinant His-tagged enzyme.
For the affinity purification of the recombinant enzyme, the bacterial extract (5-10 ml) in 20 mM Tris buffer (pH 8.0), 0.5 M NaCl, 5 mM imidazole, and 1.0 mM ␤-mercaptoethanol (1.0 mM) was loaded onto a 2.5-ml Ni 2ϩ -nitrilotriacetic acid-agarose (Qiagen) column. The column was washed first with 10 volumes of the above binding buffer (wash 1) and further washed (wash 2) with six volumes of the same buffer but containing a higher concentration of imidazole (60 mM). The enzyme was eluted with six volumes of the buffer containing 0.3 M imidazole (elute 1), followed by elution with four volumes of 1.0 M imidazole (elute 2).
Raising Monoclonal Antibodies against the Reductase-Monoclonal antibodies against the enoyl-CoA reductase were produced by the Hybridoma Core Facility (University of Michigan). The Balb/c mice were immunized against the partially purified enoyl-CoA reductase (see above), and their splenic cells were used to generate the hybridomas. The hybridomas were screened for antibody secretion by employing a dot blot enzyme-linked immunosorbent assay using the partially purified enzyme as the antigen. On that basis, six clones were selected and propagated as ascites cells in mice. One of the ascites fluid samples (isotype IgG1), which showed high specificity toward the enoyl-CoA reductase, was used in the Western blot analysis.
Western Blot Analysis of the Subcellular Fractions-Subcellular fractionation of guinea pig liver to isolate major cellular organelles including peroxisomes and the assay of marker enzymes were done as described previously (14). The Western blot analysis was done by subjecting the subcellular fractions to SDS-PAGE (15), transferring the separated proteins to a polyvinylidene difluoride membrane, probing the blot with the monoclonal antibody, and then detecting the antibodyreacting protein bands using the Protoblot Western blot alkaline phosphatase system kit (Promega).
Northern Blotting-Total RNA from guinea pig tissues was isolated and subjected to formaldehyde gel electrophoretic separation and transfer to Nytran membranes as described (21,22). The mouse multiple tissue mRNA blot was from CLONTECH. The blots were hybridized with the radioactive mouse reductase cDNA probe prepared as described above. Blots were prehybridized for 30 min at 68°C in Express Hyb (CLONTECH) and then hybridized for 60 min at 68°C in Express Hyb containing the labeled probe (3-4 ϫ 10 6 cpm/ml, 1-2 ϫ 10 9 cpm/g of template). The blots were washed three times at room temperature with a solution containing 2ϫ SSC and 0.05% SDS, followed by two washes with 0.1ϫ SSC and 0.1% SDS at 50°C. The radioactivity on the blots was detected by autoradiography using Kodak X-Omat AR films.
All other materials and methods were the same as described previously (7,22). The arrowhead shows the 40-kDa band whose partial amino acid sequence was determined.

RESULTS
Assay of Enoyl-CoA Reductase-Both spectrophotometric and radiometric methods were employed to assay the reductase. The enzyme activity was stimulated by the presence of bovine serum albumin in the incubation mixture (Table I). Table I also shows that NADPH cannot be replaced by NADH as a coenzyme. The product of the enzymatic reaction was identified as a saturated fatty acyl-CoA (see below). Although these two assay methods were qualitatively similar, lower activity (ϳ50%) was seen with the radiometric assay (Table I), indicating an isotope effect when 3 H-labeled NADPH was used as the reductant. However, because of high sensitivity and ease of assay, the radiometric method was employed when a large number of samples were to be assayed. The assay method used in other experiments is indicated in the legends of the figures and tables.
Partial Purification of the Peroxisomal Enzyme-The enzyme is apparently bound to the peroxisomal membrane, because after osmotic shock of peroxisomes the enzyme activity remained associated with the membrane fraction (data not shown). A neutral detergent such as Triton X-100 in the presence of high concentrations of KCl or NaCl completely solubilized the enzyme. The enzyme, unlike acyl-DHAP reductase (7), remained soluble after the salt is dialyzed out. This solubilized enzyme was further purified by NADP affinity chromatography (Table II). When analyzed by SDS-PAGE, a 40-kDa protein band was found to be co-purified with the enzyme activity. This band was estimated to correspond to 40 -60% of total proteins in different preparations. A typical SDS-PAGE analysis is shown in Fig. 1. Another major protein (60 kDa) that also binds to the affinity column, seen on the electrophoretogram (Fig. 1), was identified as catalase. Catalase has been shown to contain bound NADPH, although the physiological significance of the bound NADPH is unknown (23). The affinity-purified enzyme rapidly lost its activity upon storage, and attempts to further purify it were not successful.
Partial Amino Acid Sequence of the 40-kDa Protein-The 40-kDa protein band was resistant to Edman degradation, indicating that N-terminal amino acid of the protein is probably blocked. Therefore, after SDS-PAGE the protein band was treated with CNBr, and the resulting peptides were separated by another SDS-PAGE and blotted to a polyvinylidene difluoride membrane. The main bands on the blot were then subjected to Edman degradation. A 3-kDa fragment yielded the longest sequence (27 amino acids) as follows: KSTLALYGKIDFLVNNGGG-QFWSSPEH.
Analysis of another fragment (6 kDa) yielded essentially the same partial amino acid sequence.
Cloning the cDNA of the Reductase-Upon searching the GenBank TM data base, a number of human and mouse EST clones were identified whose DNA sequences coded (80 -90% identity) for the above partial amino acid sequence. One of these mouse EST clones (AA241896) was used to generate a probe for screening a guinea pig liver cDNA library. A nucleotide sequence at the 5Ј-end of the clone, when translated, showed high (90%) amino acid identity with the first 21 residues of the guinea pig reductase partial amino acid sequence we determined. A 0.4-kb fragment containing this stretch of nucleotides was isolated and used as the template to generate 32 P-labeled probe, which was then used to screen the library. A large number of positive clones were isolated in the screen, and the plasmids from two of these (clones 9 and 14) that contained the longest inserts (ϳ1.4 kb) were subjected to nucleotide sequencing. Both of the strands of each insert were sequenced by the "primer-walking" method. The sequences obtained were identical except that the insert from clone 9 was missing 9 base pairs from the 5Ј-end compared with clone 14. The complete sequence (1380 base pairs) of clone 14 is shown in Fig. 2. An open reading frame was identified that started at ATG (ϩ5) within the consensus Kozak (24) translation initiation sequence (GCCATGG) and coded for a polypeptide of 302 amino acids, which contained the same 27 amino acid sequence shown to be present in the peroxisomal 40-kDa protein (Fig. 2). The carboxyl-terminal tripeptide sequence (AKL) is a type 1 peroxisometargeting signal (PTS-1) (25). The 0.5-kb 3Ј-untranslated region contained a noncanonical polyadenylation signal (AGTAAA) near the poly(A) tail (Fig. 2).
Cloning of the Mouse and Human cDNAs-A similar screening of a mouse liver cDNA library using the same probe yielded a number of positive clones from which a homologous mouse cDNA sequence was assembled (Genbank TM accession no. AF232011). By analyzing a number of human EST clones that showed homology to the mouse and guinea pig cDNAs, we have identified a clone (AA232159) containing the full-length (1.9-kb) human reductase cDNA. This human cDNA sequence has also been submitted to the Genbank TM data base (accession no. AF232009).
The three cDNA sequences have 3Ј-untranslated regions of varying chain lengths (0.3 kb for mouse, 0.5 kb for guinea pig, and 0.7 kb for human cDNAs), which do not show close homology with each other.
Translation of the putative open reading frames of these cDNAs shows that these polypeptides are highly homologous (71-75% identical and 90 -93% similar) (Fig. 3), the guinea pig open reading frame being one amino acid shorter than the other two (303 amino acids). The calculated molecular masses of the guinea pig, mouse and human reductases are 32.5, 32.3, and 32.6 kDa, respectively. All of them have the PTS-1, AKL, at the carboxyl end. Genbank TM searches showed that these amino acid sequences are homologous to a large number of eukaryotic and prokaryotic pyridine nucleotide-linked dehydrogenases (see "Discussion"). Fig. 3 also shows the comparison of a putative amino acid sequence of a rat cDNA sequence that is highly homologous to the enoyl-CoA reductase protein sequences in other species, especially mouse (86% identical, 97% similar). This sequence has been submitted to Gen-Bank TM with a tentative identification as either short chain acyl dehydrogenase (AF099742) or peroxisomal 2,4-dienoyl-CoA reductase (AF021854). The main difference between the rat protein and that of the other species is that ARL is the carboxyl-terminal tripeptide sequence for the rat protein instead of AKL for the other three. ARL is not a common mammalian PTS-1, but it is known to be a PTS-1 for S. cerevisiae proteins (25).
Expression and Purification of the Active Recombinant 2-Enoyl-CoA Reductase-The full-length guinea pig liver cDNA was subcloned into a T7lac promoter-containing expression vector (Pet 28a), which was then used to transform E. coli cells as described under "Experimental Procedures." The transformed cells expressed a protein having high 2-enoyl-CoA reductase activity when the T7lac inducer IPTG was added to the growing transformed cells in culture (data not shown). As expected, the recombinant fusion protein, containing six consecutive His residues at the N terminus, binds tightly to a Ni 2ϩ -chelate affinity resin and can be quantitatively eluted by imidazole-containing buffer (Fig. 4A). The recombinant reductase, purified 50-fold from the crude extract, had a specific activity of 1700 nmol/min/mg of protein (spectrophotometric assay). Upon SDS-PAGE, the purified enzyme showed a single 43-kDa band (Fig. 4B). The calculated molecular mass of the recombinant reductase is, however, 36.7 kDa. The purified enzyme is unstable toward storage at 4°C or Ϫ70°C but is somewhat stabilized by thiols such as ␤-mercaptoethanol or dithiothreitol. The half-life of the enzyme in the presence of 1 mM dithiothreitol is about 5 days at 4°C. The purified enzyme completely lost its activity on freezing and thawing.
The human recombinant His-tagged enzyme was also similarly expressed and affinity-purified. The properties of the human reductase are similar to those described for the guinea pig recombinant enzyme (data not shown).
Properties of the Recombinant Reductase-The properties of the guinea pig recombinant reductase are similar to those of the peroxisomal enzyme as evidenced by its specificity toward using NADPH as the coenzyme and inhibition by the thiol-reactive reagent, N-ethylmaleimide (Table III). In a spectrophotometric assay, the recombinant enzyme, like the peroxisomal enzyme (Table I), was found not to use NADH as the coenzyme (data not shown). Unlike liver peroxisomes, the purified recombinant enzyme did not have any 2,4-dienoyl-CoA reductase activity (Table III). This indicates that the peroxisomal 2,4-dienoyl-CoA reductase is an enzyme distinct from the peroxisomal enoyl-CoA reduc-tase whose cDNA we cloned. Both the recombinant and liver peroxisomal enzymes are inhibited by higher concentrations (Ͼ50 M) of decenoyl-CoA (data not shown). The apparent K m values for decenoyl-CoA were 11 and 18 M for the peroxisomal and the recombinant enzymes, respectively. The corresponding K m values for NADPH were 38 and 53 M. The substrate specificity of both the peroxisomal and the recombinant enzymes toward different enoyl-CoAs is similar (except for 6:1) with maximum activity against 10:1 CoA, and the specificity is different from that of guinea pig liver microsomal reductase (Fig. 5). As reported previously (26), it is also seen here that the microsomal reductase has the highest activity with 2-hexenoyl-CoA (Fig. 5). The relatively higher activity of the peroxisomal enzyme with 6:1 CoA compared with the recombinant enzyme is probably due to microsomal contamination (ϳ10%) of the peroxisome preparation.
Identification of the Enzymatic Reaction Product-The product of the reaction between 2-decenoyl-CoA and 4-[ 3 H]NADPH, catalyzed either by peroxisomes or purified recombinant enzyme, was identified as [ 3 H]decanoyl-CoA. This was done by converting the labeled reaction product to methyl ester, which was then cochromatographed with methyl 2-decenoate and methyl decanoate on a silver nitrate-coated TLC plate (see "Experimental Procedures"). All of the radioactivity migrated with the faster moving methyl decanoate (R f ϭ 0.64) and not with the methyl decenoate (R f ϭ 0.56), indicating that the double bond of the 2-decenoyl-CoA is reduced by NADPH to a saturated acyl-CoA. Peroxisomal Localization of the Cloned Reductase-The purification of the enzyme from isolated guinea pig liver peroxisomes and the presence of a PTS-1 in the amino acid sequence of the enzyme indicate that the cloned reductase is a peroxisomal enzyme. This is confirmed by Western blot analysis using a monoclonal antibody raised against the purified enzyme. Although, in addition to peroxisomal activity, enoyl-CoA reductase activity is also present in mitochondria and at a higher concentration in the ER (microsomes), the monoclonal antibody raised against the partially purified enzyme interacted specifically with the peroxisomal fraction (Fig. 6). The weak positive signal seen in mitochondrial and microsomal fractions is probably due to the presence of contaminating peroxisomes in these fractions, as evidenced by the activity of peroxisome membrane-specific DHAP acyltransferase in these fractions (Fig. 6). The specificity of the antibody used for the cloned enzyme is also seen in Fig. 6 by its interaction with the recombinant reductase. As indicated above, because of the presence of additional amino acids, including the His tag at the N terminus of the recombinant enzyme, the molecular mass of the recombinant enzyme is greater (by about 4 kDa) than that of the native enzyme.
Northern Blot Analysis-When a mouse multitissue RNA blot was hybridized with the 32 P-labeled mouse DNA probe as described above, a single radioactive band of ϳ1.3 kb was seen with greatest intensities in liver and kidney and at lower levels in heart, skeletal muscle, and other tissues (Fig. 7). Hybridization of a guinea pig tissue RNA blot with the same 32 P-probe showed a single transcript of 1.5 kb present not only in liver and kidney but also in other tissues in much lower concentrations (data not shown).

DISCUSSION
The induction of the active 2-enoyl-CoA reductase when the cDNAs were expressed in E. coli and the similarity of the properties and specificity of the purified recombinant enzyme with those of the peroxisomal enzyme, but not with the microsomal reductase, show that the cloned cDNAs encode a peroxisomal trans-2-enoyl-CoA reductase. The presence of the PTS-1 in the protein sequences and Western analysis of its subcellular distribution (Fig. 6) corroborate the peroxisomal location of the reductase. These results also confirm the report by Horie et al. (6), which provided good evidence for the presence of an enoyl-CoA reductase in rat liver peroxisomes. The reaction catalyzed by this reductase is shown below (Scheme 1).
As indicated above, the putative amino acid sequences of the reductase (Fig. 3) have high homology to a large number of pyridine nucleotide-linked dehydrogenases. These polypeptide sequences seem to belong to the recently described "short chain acyl dehydrogenase/reductase" family (27). For example, the Nterminal part of the consensus sequence of the reductase has the same NAD(P)H-binding motif, GXXXGXG, at amino acid residues 25-31 (Fig. 3). Similarly, the YX 3-7 K motif shown to be present in the catalytic center of such dehydrogenases is also present in the enoyl-CoA reductase at residues 179 -183 (178 -182 for the guinea pig enzyme). The amino acid sequences of the enoyl-CoA reductase (Fig. 3) have hydrophobic stretches (e.g. from residue 153 to 162 and from 242 to 251), but none of them meet the criteria for a membrane-spanning ␣-helical structure. Interestingly, there are two conserved N-glycosylation motifs (NLT) at positions 131 and 180 (179 for the guinea pig) of the amino acid sequences. Although to date no peroxisomal proteins have been shown to be N-glycosylated (28), recently Elgersma et al. provided evidence that Pex 15P, an S. cerevisiae peroxisomal membrane protein, is O-glycosylated (29).
Among the mammalian dehydrogenases, the enoyl-CoA reductase has high homology to both mitochondrial (31% identity, 51% similarity) and peroxisomal (31% identity, 49% similarity) 2,4-dienoyl-CoA reductase (30,31). The peroxisomal enoyl-CoA reductases also have homology to the bacterial enoyl-ACP reductases that catalyze the same reaction (22% identity and 50% similarity extending throughout the protein sequence of the E. coli enzyme). Surprisingly, the reductase sequences (Fig. 3) have only very limited homology to the enoyl reductase domain of the multifunctional mammalian fatty acid synthase (32,33). This indicates that the enoyl reductases described here belong to a new family of enzymes and may have counterparts in the ER and mitochondria of mammalian cells. Therefore, the availability of these cDNA clones will aid in the discovery of other cellular isoenzymes, and the active recombinant enzyme will facilitate studies into the structure of the enzyme and the molecular details of the catalytic mechanism for the reduction of the double bond.  As discussed above, one rat cDNA recently submitted to GenBank TM (AF021854 and AF099742) from two different laboratories has very high homology to the cloned cDNAs reported here, indicating that it is most probably the rat counterpart of peroxisomal enoyl-CoA reductase. Although AF021854 was tentatively designated as coding for 2,4-dienoyl-CoA reductase, we did not find any such activity in the recombinant guinea pig enzyme. While this manuscript was in preparation, the details of cloning of the AF099742 were described by Zhang and Underwood (34). These authors isolated the clone from fasted rat liver cDNA library prepared by "suppression subtractive hybridization." Although the enzyme it coded was not identified, they also pointed out the homology of the sequence to short chain acyl dehydrogenase/reductase family and the presence of PTS-1 in the putative polypeptide. Consistent with our results, transcript levels were found to be highest in rat liver and kidney, and it has been also shown that the liver transcript level is increased when the rats are fasted for 48 h. Sequence homology, as shown in Fig. 3, strongly suggests that this cDNA encodes the rat peroxisomal 2-enoyl-CoA reductase.
Although the partially purified peroxisomal enzyme seemed to be about 40 -60% pure, it had a much lower specific activity (75 milliunits/mg of protein by radiometric assay) than the purified recombinant enzyme (600 -750 milliunits/mg of protein). This is probably because the enzyme became very labile after the detergent solubilization but a significant fraction of the partially denatured enzyme still was capable of binding to the NADP ϩ affinity matrix. The purified recombinant enzyme had up to 2.0 units/mg of protein activity by the spectrophotometric assay, which is ϳ200 times the specific activity of the liver peroxisomal reductase. The expression of an active recombinant enzyme in E. coli shows that post-translational modifications (e.g. glycosylation) are not necessary for enzyme activity.
The substrate specificity of the enzyme shows that it may be termed a medium chain enoyl-CoA reductase, with specificity similar to that of mitochondrial and microsomal reductases. However, the longer chain enoyl-CoAs (14:1 and 16:1) are also good substrates for both the microsomal and the peroxisomal enzymes (Fig. 7), indicating that this enzyme may have a similar function in both of these organelles. Horie et al. (6) showed that in peroxisomes, octanoyl-CoA is chain-elongated to form dodecanoyl-CoA, and recently, Hayashi and Sato (35) demonstrated that in rat liver peroxisomes dodecanoyl-CoA was chain-elongated to form palmitoyl-CoA, which was reduced to hexadecanol by the NADPH-linked peroxisomal acyl-CoA reductase (36). These results show that fatty acid chain elongation in peroxisomes can be demonstrated in vitro. The physiological importance of the peroxisomal chain elongation system, however, is not clear. Although the specific activity of the peroxisomal enoyl-CoA reductase, the key enzyme for the fatty acid chain elongation system, is similar to that of the ER enzyme, the total activity of cellular ER reductase is much more than the peroxisomal enzyme, because the ER contains about 10 times more protein than peroxisomes per cell (37). Therefore, it seems that the ER is the main cellular site for chain elongation of fatty acids.
There is also a dilemma concerning the peroxisomal and also mitochondrial elongation systems because a highly active fatty acid ␤-oxidation system is also present in the these organelles. It is not clear how these two opposing metabolic pathways could be regulated without compartmentalization. It is possible that the peroxisomal enoyl-CoA reductase may have some functions other than participating in the fatty acid chain elongation system. For example, this reductase may be involved in the reduction of cis-double bonds at even-numbered carbons of unsaturated fatty acids. It has been shown that in mitochondria these fatty acids are mainly metabolized via their conversion to 2,4-dienoyl-CoA followed by reduction (NADPH) to 3-enoyl-CoA, which is then isomerized to trans-2-enoyl-CoA (38). trans-2-Enoyl-CoA is either further metabolized by ␤-oxidation or is reduced by NADPH-linked 2-enoyl-CoA reductase. The presence of 2,4-dienoyl-CoA reductase, ⌬ 3 ,⌬ 2 -enoyl-CoA isomerase (38), and also, as shown here, enoyl-CoA reductase in peroxisomes indicates that a similar conversion of cis-double bonds may occur in peroxisomes. Recent findings show that unsaturated fatty acids having double bond(s) at the odd-numbered carbon atoms are also not oxidized directly in the ␤-oxidation pathway but are converted to medium chain 2-enoyl-CoAs via a reductive pathway (39,40). The key enzyme for such conversion, i.e. ⌬ 3,5 ,⌬ 2,4 -dienoyl-CoA isomerase, has been shown to be present in both liver mitochondria and peroxisomes (41). Because medium chain acyl-CoAs are not the substrates for peroxisomal ␤-oxidation, the peroxisomal enoyl-CoA reductase could catalyze the reduction of the medium chain enoyl-CoAs formed via these reductive pathways to the saturated acyl-CoAs, which may then be converted to the carnitine derivatives and exported to mitochondria for oxidation and energy production. This pathway is probably physiologically important for the complete oxidation of unsaturated fatty acids in liver, especially when the metabolism is switched from utilizing glucose to utilizing fatty acids in conditions such as starvation or an acute diabetic state (42,43). Therefore, the present report of cloning the enoyl-CoA reductase cDNAs will not only facilitate cloning of the cDNAs of other cellular isoforms of this key enzyme of fatty acid metabolism and study of the structure-function relationship of the enzyme but also will be useful for understanding the mechanism of regulation of fatty acid metabolism in different physiological and pathological conditions.