Identification of Two Mammalian Reductases Involved in the Two-carbon Fatty Acyl Elongation Cascade*

The de novo synthesis of fatty acids occurs in two distinct cellular compartments. Palmitate (16:0) is synthesized from acetyl-CoA and malonyl-CoA in the cytoplasm by the enzymes acetyl-CoA carboxylase 1 and fatty acid synthase. The synthesis of fatty acids longer than 16 carbons takes place in microsomes and utilizes malonyl-CoA as the carbon source. Each two-carbon addition requires four sequential reactions: condensation, reduction, dehydration, and a final reduction to form the elongated fatty acyl-CoA. The initial condensation reaction is the regulated and rate-controlling step in microsomal fatty acyl elongation. We previously reported the cDNA cloning and characterization of a murine long chain fatty acyl elongase (LCE) (1). Overexpression of LCE in cells resulted in the enhanced addition of two-carbon units to C12-C16 fatty acids, and evidence was provided that LCE catalyzed the initial condensation reaction of long chain fatty acid elongation. The remaining three enzymes in the elongation reaction have not been identified in mammals. Here, we report the identification and characterization of two mammalian enzymes that catalyze the 3-ketoacyl-CoA and trans-2,3-enoyl-CoA reduction reactions in long and very long chain fatty acid elongation, respectively.

(FAS) 1 then uses malonyl-CoA, acetyl-CoA, and NADPH to elongate fatty acids in two-carbon increments in the cytosol (2). The principal fatty acid produced by FAS in rodents is palmitic acid, which contains 16 carbons and is designated 16:0 (3).
The mammalian enzymes that elongate palmitic acid (16:0) and very long chain fatty acids (ϾC18) have been localized to the endoplasmic reticulum (ER) and are shown schematically in Fig. 1 (4). Microsomal fatty acid elongation uses malonyl-CoA as the two-carbon donor and consists of four sequential and independent reactions: 1) a condensation between a fatty acyl-CoA and malonyl-CoA to form 3-ketoacyl-CoA; 2) a reduction of the 3-ketoacyl-CoA using NADPH to form 3-hydroxyacyl-CoA; 3) a dehydration of 3-hydroxyacyl-CoA to trans-2,3enoyl-CoA; and 4) a reduction of trans-2,3-enoyl-CoA to saturated acyl-CoA (5). Unlike the multifunctional FAS enzyme, the enzymes that carry out microsomal fatty acid elongation are encoded by separate genes.
Enzymes involved in microsomal fatty acid elongation have been characterized most extensively by genetic deletion studies in Saccharomyces cerevisiae. Three proteins, designated Elo1p, Elo2p, and Elo3p, participate in the initial condensation reaction of microsomal fatty acyl elongation. Elo1p is required for the elongation of C14 to C16 fatty acids (6), and Elo2p and Elo3p are required for the synthesis of very long chain fatty acids (7).
Six mammalian homologues of the yeast ELO genes have been described (Fig. 1). Like their yeast counterparts, these enzymes exhibit some fatty acid chain length substrate specificity. The first mammalian elongase identified was Cig30 (cold-induced glycoprotein of 30 kDa), which is the functional equivalent of yeast Elo2p (8). Ssc1 and Ssc2 (sequence similarity to Cig30 1 and 2) subsequently were identified based on homology to Cig30 (9). Ssc1 is the functional equivalent of Elo3p in yeast (9). Definitive fatty acid substrate specificity has not been assigned to Ssc2, although two of its substrates are arachidonic (20:4) and eicosapentaenoic (20:5) acids (1). ELOVL4 (elongation of very long chain fatty acids-like 4) was identified by linkage and haplotype analysis in families with two forms of autosomal dominant macular dystrophy and is expressed only in tissues with high contents of very long chain fatty acids; therefore, it is likely that ELOVL4 is involved in the elongation of very long chain fatty acids (10). HELO1 was identified based on sequence homology with yeast Elo2p and has a broad range of very long chain fatty acid substrate spec-ificity (11). Recently, we identified a long chain fatty acyl elongase (LCE) that is homologous to the very long chain fatty acyl elongases; however, unlike other family members, the activity of LCE is restricted to long chain fatty acids (C12-C16) (1).
Proteins that participate in the post-condensation reactions of microsomal fatty acid elongation recently were characterized in yeast. Two proteins were identified that participated in the 3-ketoacyl-CoA reductase reaction. The YBR159w gene encoded the protein that was responsible for the majority of the 3-ketoacyl-CoA reductase activity in yeast microsomes (12). Studies using Ybr159p mutants revealed that a small amount of residual 3-ketoacyl-CoA reductase activity was still present, which was subsequently ascribed to 1-acyldihydroxyacetonephosphate reductase (13). No proteins have been identified that carry out the third dehydratase reaction. The final reduction of the trans-2,3-enoyl-CoA is carried out by Tsc13p in yeast (14). In mammals, the enzymes that carry out reactions distal to the condensation reaction have not been identified. In the current studies, we identify and characterize human and mouse reductases that catalyze the second and fourth reactions in microsomal fatty acid elongation.
Cloning of Mammalian 3-Ketoacyl-CoA and trans-2,3-Enoyl-CoA Reductase cDNAs and Construction of Expression Plasmids-cDNAs encoding the putative human and mouse microsomal 3-ketoacyl-CoA reductase (KAR) were identified by a BlastP search of the NCBI data bases using the S. cerevisiae protein Ybr159p. Human and mouse cDNAs (GenBank TM accession numbers NM_016142 and AF064635, respectively) were identified that encode proteins that are ϳ31% identical to the yeast 3-ketoacyl-CoA reductase protein Ybr159p. The trans-2,3-enoyl-CoA reductase (TER) proteins were identified by a BlastP search using the S. cerevisiae protein Tsc13p, which encodes the yeast trans-2,3-enoyl-CoA reductase. Human and mouse cDNAs (GenBank TM accession numbers AAF32373 and AK010984, respectively) were identified that encode proteins that are ϳ34% identical to Tsc13p.
The expression plasmids for human KAR and TER were constructed as follows. cDNAs encoding the full-length protein were obtained by PCR amplification using human adipose tissue first strand cDNA (Clontech, catalogue number 7128-1) as the template and the following  primers: KAR, 5Ј primer, 5Ј-GCCACCATGGGCGGCCGCGAGAGCGC-TCTCCCCGCCGCC-3Ј, and 3Ј primer, 5Ј-TTAGTTCTTCTTGGTTTTC-TTCAG-3Ј; and TER, 5Ј primer, 5Ј-GCCACCATGGGCGGCCGCAAGC-ATTACGAGGTGGAGATT-3Ј, and 3Ј primer, 5Ј-TCAGAGCAGGAAGG-GGATGATGGG-3Ј. The 5Ј primers used to amplify each cDNA contained a Kozak sequence followed by an ATG codon and a NotI restriction enzyme site. The resulting PCR products were ligated into pCMV-Script (Stratagene, La Jolla, CA) and digested with NotI, and three copies of the HA epitope (YPYDVPDYA) were inserted at the 5Ј end of each cDNA. The resulting plasmids were designated pCMV-HA-KAR and pCMV-HA-TER, respectively. The integrity of all PCR products and ligations was confirmed by DNA sequencing. The LCE expression plasmid (pCMV-HA-LCE) was constructed as described previously (1).
Immunofluorescence Microscopy-Chinese hamster ovary K-1 cells (ATCC CCL-61) were set up on glass coverslips in 6-well plates at a density of 1.0ϫ10 5 /well in Dulbecco's modified Eagle's medium/Ham's F-12 medium supplemented with 5% fetal calf serum, 100 units/ml penicillin G sodium, and 100 g/ml streptomycin sulfate (day 0). On day 1, 0.5 g of indicated plasmids was transfected using 3 l of FuGENE 6 (Roche Molecular Biochemicals) in serum-free Dulbecco's modified Eagle's medium/Ham's F-12 medium. On day 3, the cells were washed with PBS and then fixed and permeabilized by incubating in 2 ml of methanol at Ϫ20°C for 10 min. After three washes with PBS, the cells were incubated for 1 h at 4°C in 1% BSA in PBS (buffer A). The cells were then incubated in buffer A at 4°C with a mouse monoclonal HA antibody (HA probe F-7; Santa Cruz Biotechnology, Inc., Santa Cruz, CA) (20 g/ml) and a rabbit anti-calnexin polyclonal antibody (Stress-Gen Biotechnologies Corp., Victoria, Canada) (1:200 dilution) for 1 h. The cells were washed three times with buffer B (0.1% BSA in PBS), and primary antibodies were localized by incubating the cells for 1 h at room temperature in buffer A containing 2 g/ml goat anti-rabbit IgG conjugated to Alexa Fluor 568 and goat anti-mouse IgG conjugated to Alexa Fluor 488 (Molecular Probes, Inc., Eugene, OR). After incubation with the secondary antibody, the cells were washed three times with buffer B, quickly rinsed with PBS and distilled water, and analyzed with a Leica TCS SP confocal microscope (Leica Microsystems Inc., Heidelberg, Germany).
Co-expression of KAR or TER with LCE in HEK-293 Cells-HEK-293 cells (ATCC CRL-1573) were plated at a density of 4 ϫ 10 5 cells/60-mm dish in Dulbecco's modified Eagle's medium supplemented with 10% (v/v) fetal calf serum, 100 units/ml penicillin G sodium, and 100 g/ml streptomycin sulfate (day 0). On days 1 and 2, the cells were transfected with expression plasmids using 6 l of FuGENE 6 according to the manufacturer's instructions. Transfection studies with KAR contained 0.6 g of pCMV-HA-LCE, 0.6 g of pCMV-HA-KAR, 0.8 g of pVAI (15), and 0.02 g of pCMV ␤-galactosidase plasmids/dish. Transfection studies with TER contained 1.0 g of pCMV-HA-LCE, 0.2 g of pCMV-HA-TER, 0.8 g of pVAI (15), and 0.02 g of pCMV ␤-galactosidase plasmids/dish. The total amount of plasmid DNA in each transfection was adjusted to 2 g/dish by adding pCMV-Script. On day 3, the cells were harvested, and cytosolic and membrane proteins were prepared as described previously (1). The microsomal protein was stored at Ϫ80°C after snap freezing in liquid nitrogen. Transfection efficiencies were determined by measuring the ␤-galactosidase activity in the supernatant from the 1.3 ϫ 10 5 ϫ g spin using a ␤-galactosidase assay kit (Stratagene, La Jolla, CA), and the expression of HA-tagged proteins in membranes was confirmed by immunoblot analysis as described (1). The protein concentrations were determined using the method of Lowry et al. (16).
In Vitro Fatty Acid Elongation Assay-Palmitoyl-CoA or BSA-bound fatty acids were used as substrates for all reactions. BSA-bound fatty acids were prepared as 5 or 10 mM solutions as described (1). The elongation assays contained 0.05 mg of microsomal protein in 50 mM potassium phosphate, pH 6. To initiate the elongation reaction, 0.05 mg of microsomal protein from transfected cells was added, and the incubation was continued for the indicated times. The reactions were stopped by adding 0.1 ml of 75% The diagram shows the enzymes and fatty acyl-CoA intermediates that comprise the two-carbon microsomal elongation of fatty acyl-CoAs. The fatty acyl-CoA first undergoes a condensation, which is catalyzed by one of six condensing enzymes discussed in the Introduction. The condensing enzyme utilized is determined by the fatty acyl chain length and degree of desaturation. All of the products of the fatty acyl-CoA condensation reaction then undergo a reduction mediated by the KAR, a dehydration mediated by an unidentified enzyme, and a final reduction mediated by the TER that results in the final elongated fatty acyl-CoA product. The identification and characterization of KAR and TER are described in the current report.
KOH (w/v) and 0.2 ml of ethanol, saponified at 70°C for 1 h, and then acidified by adding 0.4 ml of 5 N HCl with 0.2 ml of ethanol. Fatty acids were collected in three independent extractions using 1 ml of hexane. The extractions were pooled, dried under nitrogen, and separated by TLC using hexane/diethyl ether/acetic acid (30:70:1) as described (1). The TLC plates were exposed to a PhosphorImager screen, the resulting image was analyzed, and the lipids were quantified using a Bio-Imaging Analyzer with BAS1000 MacBAS 2.1 software (Fuji Medical Systems, Stamford, CT).
RNAi-mediated Inhibition of KAR and TER-Double-stranded (ds) RNA oligonucleotides were synthesized by Dharmacon Research (Lafayette, CO) for human KAR, TER, and an irrelevant control gene, vesicular stomatitis virus glycoprotein. The oligonucleotide sequences are listed in Table I. On day 0, HeLa (ATCC CCL-2) or HepG2 (ATTC HB-8065) cells were set up at a density of 4 ϫ 10 5 cells/60-mm dish. HeLa cells were cultured in minimum essential medium supplemented with 10% fetal calf serum, 1ϫ nonessential amino acid mix (Cellgro, Herndon, VA), 1 mM sodium pyruvate, 100 units/ml penicillin G sodium, and 100 g/ml streptomycin sulfate. HepG2 cells were cultured in Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum, 100 units/ml penicillin G sodium, and 100 g/ml streptomycin sulfate, respectively. dsRNAs (0.2 M) were transfected on days 1, 2, and 3 using OligofectAMINE (Invitrogen, Carlsbad, CA) according to the manufacturer's instructions. On day 4, the cells were harvested for membrane protein and total RNA as described (1).
Blot Hybridization of RNA-The multiple tissue Northern blot (Clontech, catalogue number 7780-1) was used to determine KAR and TER expression in human tissues. The blot was hybridized according to the manufacturer's instructions with randomly 32 P-labeled KAR and TER cDNA probes that were made using the full-length human KAR and TER cDNAs as templates (1). For mouse multiple tissue Northern blots, total RNA was extracted from the indicated tissues of C57BL/6J mice using RNA STAT-60 (TEL-TEST, Friendswood, TX), and poly(A ϩ ) RNA was isolated using the MessageMaker mRNA Isolation System (Invitrogen). Three g of poly(A ϩ ) RNA from each tissue was subjected to Northern blot hybridization using 32 P-labeled mouse cDNA probes for mouse KAR and TER (1). Mouse full-length cDNA templates for KAR and TER were PCR-amplified from mouse liver first strand cDNA using the following primers: KAR, 5Ј primer, 5Ј-ATGGAGTGCGCTCCCCCG-GCG-3Ј, and 3Ј primer, 5Ј-TTAGTTCTTCTTCCTTTTCTTCAG-3Ј; TER, 5Ј primer, 5Ј-ATGAAGCACTACGAGGTGGAG-3Ј, and 3Ј primer, 5Ј-TCAGAGCAGGAAGGGAATAAT-3Ј (17). The cyclophilin probe used as a control was described previously (1).
For HepG2 and HeLa cell Northern blots, cells from two 60-mm dishes were used to isolate total RNA as described above. Twenty g of the total RNA was subjected to Northern blot analysis using 32 P-labeled human KAR and TER cDNA probes as described above. All of the Northern blot filters were exposed to a Fuji PhosphorImager and quantified using a Bio-Imaging Analyzer with BAS1000 MacBAS software.
Animal Studies-All of the mice were housed in colony cages with a 12-h light/12-h dark cycle and fed Teklad mouse/rat diet 7002 from Harlan Teklad Premier Laboratory Diets (Madison, WI). Studies using SREBP transgenic mice included five wild-type and five 12-14-weekold TgSREBP-1a, TgSREBP-2 male mice fed a high protein/high carbo-hydrate diet for 2 weeks prior to sacrifice as described (17,19). SCAP liver-specific knockout mice (SCAP f/f ;MX1-Cre) were described previously (20). Four 8 -10-week-old male SCAP f/f ;MX1-Cre and corresponding wild-type mice received four intraperitoneal injections of polyinosinic-polycytidylic acid as described (20). The mice were fed the Teklad chow and sacrificed nonfasted 14 days after the last injection of polyinosinic-polycytidylic acid. Total RNA was extracted from liver as described above, and equal aliquots of RNA from all of the mice were pooled for each treatment group for study by quantitative real time PCR.

RESULTS
To identify and characterize the mammalian reductases involved in microsomal long chain fatty acid elongation, we used the sequences of yeast genes that encode a 3-ketoacyl-CoA reductase (YBR159w) (12) and a trans-2,3-enoyl-CoA reductase (TSC13) (14) to identify potential mammalian orthologues in the human and mouse data bases. The BlastP search of the NCBI human data base revealed two cDNAs (GenBank TM accession numbers NM_016142 and NM_000197), the predicted amino acids of which were 31% identical to the yeast 3-ketoacyl-CoA reductase protein Ybr159p. Similarly, the search of the mouse data base revealed two cDNAs (GenBank TM accession numbers AF064635 and NM_008291), the predicted proteins of which were 32% identical to Ybr159p. The human and mouse cDNAs NM_000197 and NM_008291 encode the hydroxysteroid 17-␤ dehydrogenase 3 protein, which is only expressed in testis (21). Therefore, the proteins encoded by these cDNAs were eliminated as candidates for the microsomal 3-ketoacyl-CoA reductase in liver. The predicted proteins encoded by the human NM_016142 and mouse AF064635 cDNAs were 82% identical and of unknown function.
An alignment of the predicted yeast, mouse, and human 3-ketoacyl-CoA reductase amino acids is shown in Fig. 2A. An in-frame stop codon is present 12 nucleotides prior to the putative initiation methionine in the human KAR cDNA sequence, indicating that the entire coding sequence was represented in the identified cDNA. The translational reading frames of the putative human and mouse KAR cDNAs encode proteins that are 312 amino acids in length. Like in the yeast protein, a conserved dilysine ER retention motif is present at the C-terminal end of human and mouse KAR. Hydropathy analysis using the Kyte and Doolittle algorithm (22) predicts the presence of as many as four putative transmembrane domains (Fig. 2B).
To determine the tissue expression pattern of the putative KAR in mammals, human and mouse multiple tissue Northern blots were performed using species-specific 32 P-labeled cDNA probes (Fig. 2C). A single ϳ2.9-kb mRNA was identified by Northern blot analysis in all of the human tissues represented on the blot. In these, the highest level of expression appeared to be in liver, muscle, and kidney. The expression of KAR also was determined in mouse tissues. The mRNA for mouse KAR was ϳ2.0 kb in size. The difference in mRNA size between the human and mouse transcripts is due to differences in the 3Ј-untranslated sequences. In mouse, high levels of KAR expression were also found in white adipose tissue and brown adipose tissue (Fig. 2C).
The putative human and mouse TER proteins were identified by a BlastP search of the NCBI data base using the yeast TER protein Tsc13p. The identified human (GenBank TM accession number AF222742) and mouse (GenBank TM accession number AK010984) cDNAs encode proteins that are ϳ34% identical to that of Tsc13p (14). The putative human trans-2,3enoyl-CoA reductase cDNA has an in-frame stop codon 21 nucleotides prior to the initiation methionine. The translational reading frames of the human and mouse putative TER cDNAs predict proteins 308 amino acids in length. The overall identity of the human and mouse TER proteins is 95%.
An alignment of the yeast, mouse, and human TER amino acid sequences is shown in Fig. 3A. Unlike the KAR proteins, no consensus ER retention motif is present in the mouse or human TER sequence. Hydropathy analysis using the Kyte and Doolittle algorithm (22) predicts the presence of as many as five transmembrane domains (Fig. 3B). No mitochondrial or peroxisomal targeting sequences were identified in these proteins. The tissue expression pattern of TER was determined using human and mouse multiple tissue Northern blots as described for KAR (Fig. 3C). A single ϳ1.2-kb mRNA was identified by Northern blotting in all of the tissues tested from human and mouse (Fig. 3C). The tissue expression of TER essentially mirrored that of KAR.
To determine whether KAR or TER participated in microsomal fatty acyl elongation, expression vectors utilizing the CMV promoter were assembled that encoded the human KAR or TER proteins with HA epitope tags at their N terminus. HEK-293 cells were transfected with the human KAR or TER expression plasmids, and cytosolic and microsomal proteins were The residue-specific hydropathy index was calculated over a window of 18 amino acids using DNA STAR software, version 5.0. The predicted transmembrane domains are indicated. C, KAR mRNA in human and mouse tissues as measured by blot hybridization. A filter containing 1 g of human poly(A ϩ ) mRNA or 3 g of mouse poly(A ϩ ) mRNA from indicated tissues was hybridized with a 32 P-radiolabeled cDNA probe corresponding to species-specific KAR cDNA. The blots were stripped and rehybridized with a cyclophilin probe. The filters were exposed to Kodak X-Omat Blue XB-1 film with intensifying screens at Ϫ80°C for 5 h. Sk. Muscle, skeletal muscle; Sm. Int., small intestine; PBL, peripheral blood leukocytes; WAT, white adipose tissue; BAT, brown adipose tissue.
FIG. 3. Amino acid sequence alignment, hydropathy plot, and tissue expression of TER. A, amino acid sequence alignment of yeast, mouse, and human TERs. Putative mouse and human TER sequences were identified by a BlastP search using yeast Tsc13p sequences. Amino acids that are 100% conserved are denoted by black boxes. The putative mouse and human TER proteins are 33 and 34% identical to yeast Tsc13p. Mouse and human TER proteins are 95% identical. B, Kyte and Doolittle hydropathy plot of human TER. The residue-specific hydropathy index was calculated over a window of 18 amino acids using DNA STAR software, version 5.0. The predicted transmembrane domains are indicated. C, TER mRNA in human and mouse tissues as measured by blot hybridization. A filter containing 1 g of human poly(A ϩ ) mRNA or 3 g of mouse poly(A ϩ ) mRNA from indicated tissues was hybridized with a 32 P-radiolabeled cDNA probe corresponding to species-specific TER cDNA. The blots were stripped and rehybridized with a cyclophilin probe. The filters were exposed to Kodak X-Omat Blue XB-1 film with intensifying screens at Ϫ80°C for 1 h. Sk. Muscle, skeletal muscle; Sm. Int., small intestine; PBL, peripheral blood leukocytes; WAT, white adipose tissue; BAT, brown adipose tissue. separated by SDS-PAGE to determine the subcellular localization of the proteins. As predicted from the hydropathy profiles, immunoblot analysis using an anti-HA antibody revealed that the expressed KAR and TER proteins were present only in the microsomal fraction (data not shown). To study the subcellular localization of these proteins directly, we performed doublelabel immunofluorescence studies of the HA epitope-tagged KAR, TER, or LCE proteins that were transfected in Chinese hamster ovary K1 cells (Fig. 4). Staining with the anti-HA antibody revealed that KAR and TER co-localized with the ER resident protein calnexin and the condensing enzyme LCE (Fig.  4, C, F, I, and L). Additional stains for a mitochondrial protein, Grp75, showed no significant co-localization with KAR, TER, or LCE, and stains for a cis-compartment Golgi resident protein, GM130, showed no co-localization with KAR and LCE (data not shown). A small degree of co-localization of GM130 and TER was found, the significance of which is not known.
Previously, we showed that the elongation of palmitoyl-CoA (16:0) was increased significantly in microsomes from HEK-293 cells transfected with the condensing enzyme, LCE (1). Elongation activity was determined by measuring the amount of 14 C incorporated from [2-14 C]malonyl-CoA into elongated fatty acid products. LCE overexpression markedly enhanced the initial condensation of palmitoyl-CoA to 3-ketostearoyl-CoA. The LCE-mediated increase in palmitoyl-CoA condensation caused the subsequent reactions to become rate-limiting, leading to the accumulation of elongation intermediates, which could be separated and identified by TLC (1). The accumulation of elongation intermediates provided a tool to study the potential function of the KAR and TER proteins. Working under the hypothesis that KAR functions as a long chain 3-ketoacyl-CoA reductase, the co-expression of KAR with LCE in cells should result in the selective disappearance of 3-ketostearoyl-CoA intermediate in microsomes incubated with palmitoyl-CoA. Similarly, the co-expression of TER with LCE should result in the selective disappearance of the trans-2,3-stearoyl-CoA intermediate if the TER protein functions as a trans-2,3-enoyl-CoA reductase. Fig. 5 (lanes 4 -6 and 10 -12) shows that the overexpression of LCE alone resulted in the accumulation of all of the elongation intermediates in microsomes from HEK-293 cells incu-bated with palmitoyl-CoA as the fatty acid substrate. Co-expression of LCE and human KAR resulted in the selective disappearance of 3-ketostearoyl-CoA, suggesting that the KAR protein enhanced the reduction of 3-ketostearoyl-CoA to 3-hydroxystearoyl-CoA (Fig. 5, lanes 7-9). Similarly, co-expression of LCE and human TER resulted in the disappearance of the trans-2,3-stearoyl-CoA intermediate (Fig. 5, lanes 13-15). This result suggested that TER functions to reduce trans-2,3-stearoyl-CoA to stearoyl-CoA. A duplicate set of experiments was performed using the mouse orthologues of KAR and TER, and similar results were obtained (data not shown).
To determine whether KAR exhibits fatty acid substrate specificity, RNAi was employed to selectively reduce the expression of KAR in cultured cells. Inhibiting KAR expression should result in the accumulation of the 3-ketoacyl-CoA intermediate if the fatty acid tested is a substrate of the enzyme. HeLa cells were transfected with the indicated dsRNAs and microsomal protein, and the total RNA was isolated (Fig. 6). The endogenous mRNA level of KAR was selectively reduced 4-fold in cells transfected with dsRNA oligonucleotides corresponding to KAR, whereas the expression of TER and cyclophilin was unchanged (Fig. 6, lower panels). Microsomes from transfected cells were incubated with long and very long chain fatty acid substrates, and the 14 C-labeled elongation products from the fatty acid elongation reaction were separated by TLC (Fig. 6, upper panel). Microsomes from cells transfected with dsKAR oligonucleotides accumulated the 3-ketoacyl-CoA intermediates for all fatty acids tested in the elongation assay (Fig.  6, lanes 3, 6, 9, and 12). The final elongated fatty acyl-CoA product was reduced by 40 -50% in microsomes from cells transfected with dsKAR oligonucleotides. These results supported the conclusion that KAR functioned as a 3-ketoacyl-CoA reductase and demonstrated that KAR reduced very long chain 3-ketoacyl-CoA substrates as well as long chain fatty acyl-CoAs.
A similar set of RNAi experiments was performed using dsRNAs oligonucleotides corresponding to human TER in HepG2 cells (Fig. 7). HepG2 cells were used for these experiments because the lower endogenous expression of TER apparently facilitated the inhibition of TER by RNAi. The transfection of HepG2 cells with dsTER oligonucleotides resulted in a selective 4-fold reduction in endogenous TER mRNA levels (Fig. 7, lower panels). trans-2,3-Enoyl-CoA intermediates accumulated in microsomes from dsTER oligonucleotide transfected cells for all fatty acid substrates tested in the elongation assay (Fig. 7, lanes 3, 6, 9, and 12). The final fatty acyl-CoA product was reduced by 50 -60% in the elongation assay with all of the fatty acids tested. Together, the data from these overexpression and inhibition studies suggested that TER functioned as a trans-2,3-enoyl-CoA reductase and that TER reduced long and very long chain fatty acid trans-2,3-enoyl-CoA substrates.
All of the previously identified enzymes required for long chain fatty acid biosynthesis are regulated by the sterol regulatory element-binding protein (SREBP) family of transcription factors (23). The SREBP family members are designated SREBP-1a, SREBP-1c, and SREBP-2. The SREBP-1 isoforms preferentially activate genes encoding fatty acid biosynthetic enzymes, whereas SREBP-2 preferentially activates genes specifying cholesterol biosynthetic enzymes. To be active, SREBPs undergo two sequential cleavages that require three proteins: an escort protein designated SCAP and two proteases FIG. 5. Activity of transfected KAR and TER in microsomes from HEK-293 cells co-transfected with LCE. HEK-293 cells were set up on day 0 at a density of 4 ϫ 10 5 cells/60-mm dish and transfected on days 1 and 2 with indicated plasmids. On day 3, the membrane proteins were prepared from the transfected cells, and the fatty acid elongation assay was performed using palmitoyl-CoA as the fatty acid substrate. The reactions were initiated by adding 50 g of membrane protein and incubated for the indicated times at 37°C. The reaction products were saponified and extracted with hexanes, evaporated under nitrogen, and suspended in 40 l of chloroform for spotting onto a silica gel TLC plate. The 14 C-labeled products were separated using hexane/diethylether/acetic acid (30: 70:1) and visualized by exposing the TLC plate to PhosphorImager screen. The identity of the saponified 14 C-labeled products was confirmed by comparing the mobility with authentic standards as described (1). Similar results were obtained in three independent experiments.

FIG. 6. Inhibition of endogenous KAR by RNAi in microsomes from transfected HeLa cells. A, HeLa cells
were set up on day 0 at a density of 4 ϫ 10 5 cells/60-mm dish and transfected with the indicated dsRNA on days 1, 2, and 3 as described under "Experimental Procedures." On day 4, the cells from four 60-mm dishes were harvested, and the membrane proteins were prepared. The fatty acid elongation reaction was performed using BSA-bound palmitic (16:0), ␥-linolenic (18:3n-6), arachidonic (20:4n-6), and eicosapentaenoic acid (20:5n-3) as substrates. The reactions were initiated by adding 50 g of membrane protein and incubated for 30 min at 37°C. The reaction products were saponified, extracted with hexane, and separated by TLC using hexane/diethylether/acetic acid (30:70:1). 14 C-Labeled products were visualized by exposing the TLC plate to a PhosphorImager. B, total RNA was isolated from two 60-mm dishes transfected with the indicated dsRNA, and 20 g of total RNA was subjected to electrophoresis and blot hybridization with the indicated 32 P-labeled probes. The filters were exposed to a PhosphorImager, and the radioactivity of each band was quantified. The fold change relative to that of the nontransfected cells was calculated after correction for loading differences using cyclophilin as the invariant control. Similar results were obtained in three independent experiments.
designated S1P and S2P (24). All three proteins are required for normal SREBP activation inasmuch as the deletion of any one results in the absence of all transcriptionally active forms of SREBPs (24).
To determine whether KAR and TER mRNA levels were regulated in a manner similar to other fatty acid biosynthetic genes, the mRNA levels of FAS, LCE, KAR, and TER were measured in livers from mice that either overexpress the transcriptionally active forms of SREBPs or that lack all SREBP isoforms as a result of inactivating SCAP (17,19,20). Consistent with previous studies, the mRNA levels of FAS and LCE were increased ϳ20-fold in livers from SREBP-1a transgenic mice (TgSREBP-1a) (Table II) (1,17). SREBP-2 overexpression (TgSREBP-2) also increased the expression of FAS and LCE mRNAs, but to a lesser extent than the overexpression of SREBP-1a. Conversely, removing all transcriptionally active forms of SREBPs by deleting SCAP in liver (SCAP Ϫ/Ϫ ) resulted in a 4-fold decrease in FAS expression and a ϳ2-fold reduction in LCE mRNA. The mRNAs for KAR and TER were largely unaffected, either by SREBP overexpression or by the absence of SREBPs. These data suggest that unlike other enzymes required for fatty acid biosynthesis, KAR and TER mRNA levels are not regulated by SREBPs in vivo.

DISCUSSION
In the current studies, we identified two mammalian reductases that participate in the microsomal elongation of long and very long chain fatty acids. BlastP searches of the NCBI data bases identified human and mouse homologues of the S. cerevisiae proteins Ybr159p, a 3-ketoacyl-CoA reductase, and Tsc13p, a trans-2,3-enoyl-CoA reductase. Biochemical studies of the recombinant human and mouse proteins confirmed that they exhibit KAR and trans-2,3-enoyl-CoA reductase activities. The enzymes responsible for microsomal fatty acyl elongation have not been purified previously. Therefore, the genes identified in this study provide an initial molecular characterization of the reductases that carry out the second and fourth steps in microsomal long and very long chain fatty acyl elongation in mammals.
The 3-ketoacyl-CoA reductase, KAR, identified in this study shares sequence similarity with members of the short chain dehydrogenase superfamily (25), which are characterized by a nucleotide co-factor-binding region (Rossmann-fold) and an active site that consists of a triad of catalytically important and highly conserved Ser-Tyr-Lys residues. KAR is expressed in all tissues, with the highest levels of expression occurring in tissues that are directly involved in lipid metabolism. We provide evidence that KAR is a 3-ketoacyl-CoA reductase and that it represents the second enzyme in the microsomal fatty acyl two-carbon elongation cascade. Whether KAR is the only enzyme that can carry out the 3-ketoacyl-CoA reduction in cells could not be addressed in the current studies. In S. cerevisiae, the majority of very long chain fatty acyl 3-ketoacyl-CoA activity is due to Ybr159p; however, the genetic disruption of the YBR159w does not completely abolish all 3-ketoacyl-CoA reductase activity (12). The ybr159⌬ mutants are viable but have a slowed rate of growth. The residual 3-ketoacyl-CoA reductase activity in the ybr159⌬ mutants was attributed to a gene that encodes 1-acyldihydroxyacetone-phosphate reductase (13). A mammalian gene has not been identified; therefore it was not possible to test whether a mammalian orthologue of AYR1 also could mediate the reduction of 3-ketoacyl-CoAs.
The trans-2,3-enoyl-CoA reductase, TER, is ϳ32% identical to the yeast trans-2,3-enoyl-CoA reductase, Tsc13p. Kohlwein et al. (14) identified and characterized the yeast Tcs13 protein as a trans-2,3-enoyl-CoA reductase and reported that it belonged to an evolutionarily conserved family of proteins present FIG. 7. Inhibition of endogenous TER by RNAi in microsomes from transfected HepG2 cells. A, HepG2 cells were set up on day 0 at a density of 4 ϫ 10 5 cells/60-mm dish and transfected with indicated dsRNA on days 1, 2, and 3 as described under "Experimental Procedures." On day 4, cells from four 60-mm dishes were harvested, and the membrane proteins were prepared for the elongation reaction. Fatty acid elongation reaction was performed as described in the legend for Fig. 6. B, total RNA was isolated and subjected to Northern blot analysis as described in the legend for Fig. 6. Similar results were obtained in three independent experiments. in all mammals, yeast, and Arabidopsis thaliana. The human and mouse trans-2,3-enoyl-CoA reductase proteins are also ϳ97% identical to the rat SC2 protein that was originally identified in a screen for cDNAs that encoded synaptic glycoproteins (27). The trans-2,3-enoyl-CoA reductase family members share sequence similarity with steroid 5␣-reductase, an ER enzyme that catalyzes the reduction of testosterone to dihydrotestosterone (27,28). Human TER and steroid 5␣-reductase are ϳ30% identical and 45% similar over the C-terminal ϳ130 amino acids. Neither protein contains classic NADPHbinding sites; however, at least eight amino acid residues at the C-terminal end of steroid 5␣-reductase type 2 are crucial for NADPH binding (29). Four of these eight residues are conserved in the yeast and mammalian TER proteins. Therefore, although the identified TER protein does not contain a classic NAPDH-binding site, the sequence similarity with steroid 5␣reductase suggests that it utilizes NADPH as a co-factor.
The overexpression and inhibition of TER in cultured cells demonstrated that the enzyme is capable of mediating the trans-2,3-enoyl-CoA reduction of both long and very long chain fatty acids. Inhibition of TER by RNAi resulted in the marked accumulation of trans-2,3-enoyl-CoA substrate intermediate for all fatty acids tested (Fig. 7). A small amount of the preceding 3-hydroxyacyl-CoA intermediate also accumulated in microsomes from dsTER oligonucleotide transfected cells. This 3-hydroxyacyl-CoA intermediate is a substrate for the dehydratase enzyme. Although the dehydratase protein has not been identified, Knoll et al. (30) have shown that the dehydratase reaction in microsomal fatty acid elongation is reversible. Therefore, the inhibition of TER could result in the accumulation of the 3-hydroxyacyl-CoA intermediate as a consequence of the reverse reaction. These results do not preclude the possibility that TER may participate in the dehydratase reaction in addition to catalyzing the fourth and final step in the microsomal fatty acyl elongation cascade.
Studies in yeast and mammals have demonstrated that microsomal fatty acyl-CoA condensing enzymes exhibit fatty acyl chain length specificity (1,7,9,10). It has been suggested that the post-condensation enzymes do not exhibit carbon chain length specificity (4). The current studies provide support for this hypothesis. Although all possible fatty acid substrates could not be tested, the data of Fig. 6 show that inhibition of KAR resulted in the accumulation of the 3-ketoacyl-CoA substrate for palmitic (16:0), ␥-linolenic (18:3n-6), arachidonic (20: 4n-6), and eicosapentaenoic (0:5n-3) fatty acids. Similar results were obtained using myristic (14:0), palmitoleic (16:1), docosatetraenoic (22:4n-6), and docosapentaenoic (22:5n-3) fatty acids as substrates in the microsomal elongation assay described in Fig. 6 (data not shown). TER inhibition by RNAi resulted in the accumulation of the trans-2,3-enoyl-CoA intermediates for the same broad range of fatty acid substrates identified as KAR substrates. Despite this accumulation, it is possible that other unidentified 3-ketoacyl-CoA or trans-2,3-enoyl-CoA reductases have greater activities for a given fatty acyl substrate than those characterized in the current studies. The current data demonstrate that the identified KAR and TER do not exhibit the strict fatty acyl chain length substrate specificity displayed by LCE and other characterized condensing enzymes.
All known fatty acid biosynthetic enzymes isolated to date are regulated by the SREBP family of transcription factors (23). The overexpression of SREBPs in liver results in the accumulation of fatty acids that are 18 carbons in length, because of the activation of FAS and LCE (1,23,31). Reducing SREBP levels by eliminating the SCAP protein in liver resulted in a 40-70% reduction in the mRNA levels of all fatty acid biosynthetic genes (20). In contrast to other lipogenic genes, the mRNA levels of KAR and TER were largely unaffected by SREBP expression levels in liver (Table II).
Lipogenesis is hormonally regulated by insulin, and the ability of this hormone to stimulate lipogenesis is mediated by SREBP-1c. The results from the transgenic and knockout mice would suggest that KAR and TER are not regulated by insulin in a manner similar to other lipogenic genes (32,33). The activities of the four microsomal elongation enzymes previously have been measured under conditions of high and low insulin (34). These studies demonstrated that only the initial condensation reaction catalyzed by LCE is regulated by insulin. The mRNA levels of the identified genes responsible for these reactions follow a similar pattern of regulation. The condensing enzyme, LCE, is suppressed in livers of fasted mice (low insulin) and increased more than 20-fold in liver from mice that were fasted and refed a high carbohydrate diet (high insulin) (35). In similar fasting and refeeding studies, the mRNA levels of TER and KAR remain unchanged in mouse liver (data not shown). Together, the in vivo data support the hypothesis that KAR and TER are constitutively expressed and that the initial condensation reaction is the regulated step in microsomal fatty acyl elongation.
In summary, the overexpression and inhibition of the human KAR and TER in cultured cells demonstrate that they function as 3-ketoacyl-CoA and trans-2,3-enoyl-CoA reductases, respectively. The lack of any measurable fatty acid carbon chain length substrate specificity for either KAR or TER suggests that the six known condensing enzymes channel the fatty acyl intermediates to a common series of enzymes that produce the elongated fatty acyl-CoA product (Fig. 1). Whether KAR and TER are essential for the long and very long chain fatty acyl elongation in vivo or whether other proteins also possess 3-ketoacyl-CoA and trans-2,3-enoyl-CoA activities will require analysis in knockout mice. The mice used in these experiments are described under "Experimental Procedures." Total RNA was made from livers of 4 -5 male mice in each group, and equal aliquots were pooled and subjected to quantitative reverse transcription PCR. 36B4 was used as the invariant control. The values represent the relative amount of mRNA in relation to that measured in wild-type mouse liver (WT), which was arbitrarily defined as 1.