Characterization of mouse short-chain aldehyde reductase (SCALD), an enzyme regulated by sterol regulatory element-binding proteins.

Sterol regulatory element-binding proteins (SREBPs) enhance transcription of genes encoding all of the proteins required for the cellular synthesis and uptake of cholesterol and unsaturated fatty acids. Here, we use suppression subtractive hybridization to identify a previously unrecognized SREBP-enhanced gene in mice. The gene encodes a membrane-bound enzyme that we designate SCALD, for short-chain aldehyde reductase. We expressed SCALD in bacteria, purified it extensively, and studied its catalytic properties in detergent solution. The enzyme specifically uses NADPH to reduce a variety of short-chain aldehydes, including nonanal and 4-hydroxy-2-nonenal. The enzyme also reduces retinaldehydes, showing equal activity for all-trans-retinal and 9-cis-retinal. Northern blot analysis indicates that SCALD is expressed most abundantly in mouse liver and testis. In the liver of mice, SCALD is suppressed by fasting and induced by refeeding, consistent with regulation by SREBPs. In testis, SCALD expression is restricted to pachytene spermatocytes, as revealed by visualization of mRNA and protein. SCALD is also expressed in four layers of the retina, including the outer segment of rods and cones, as revealed by immunohistochemistry. SCALD appears to be the mouse ortholog of the human protein that has been designated variously as prostate short-chain dehydrogenase/reductase 1, retinal reductase 1, and retinol dehydrogenase 11. In view of its ability to reduce short-chain aldehydes in addition to retinals, we propose that SCALD may be induced by SREBP in liver and other tissues to prevent toxicity from fatty aldehydes that are generated from oxidation of unsaturated fatty acids that are synthesized as a result of SREBP activity.

Sterol regulatory element-binding proteins (SREBPs) 1 are transcription factors that activate at least 20 target genes involved in cholesterol and unsaturated fatty acids biosynthesis (1)(2)(3). The SREBPs differ from other transcription factors because they are synthesized as membrane-bound proteins, the active fragments of which must be released by proteolysis to enter the nucleus and activate transcription. The proteolytic release is controlled by the lipid content of the cells; release is rapid when cells are depleted of cholesterol, and it is blocked when cholesterol overaccumulates (4,5).
Mammalian cells express three isoforms of SREBP. SREBP-1a and SREBP-1c are produced from a single gene through the use of alternate transcription start sites, and SREBP-2 is encoded by a separate gene (6). In vivo, SREBP-1c, and SREBP-2 are the predominant isoforms, and their physiological roles have been studied in transgenic mice overexpressing the cleaved nuclear form of SREBPs (nSREBPs) in liver (1). In these mice, nSREBP-1c preferentially enhances transcription of genes required for fatty acid and triglyceride synthesis: acetyl-CoA carboxylase, fatty-acid synthetase, longchain fatty-acyl elongase, and stearoyl-CoA desaturase. nSREBP-2 preferentially increases the level of mRNAs encoding multiple enzymes in the cholesterol biosynthetic pathway, including 3-hydroxy-3-methylglutaryl coenzyme A, 3-hydroxy-3-methylglutaryl coenzyme A reductase, farnesyl-diphosphate synthase, squalene synthase, lanosterol demethylase, and others. As a result of these changes, there is a massive increase in the content of unsaturated fatty acids and cholesterol in livers of transgenic mice overexpressing nSREBPs (1).
In the current study, we used the technique of suppression subtractive hybridization (7) to identify additional genes whose mRNAs are regulated by SREBPs (8). Specifically, we compared the expression levels of mRNAs from two sources of brown adipose tissue, one from wild-type mice and the other from transgenic mice overexpressing nSREBP-1c in adipose tissue (9). Among the most differentially expressed genes, we found a nucleotide sequence encoding a protein, hereafter referred to as short-chain aldehyde reductase (SCALD), that is regulated by all three SREBP isoforms.
SCALD is the mouse ortholog of a human protein that has already been given three names: prostate short-chain dehydrogenase/reductase 1 (PSDR1) (10), retinal reductase 1 (RalR1) (11), and retinol dehydrogenase 11 (RDH11) (12). The human mRNA was discovered as a gene that is expressed at very high levels in human prostate (10) and at lower levels in a variety of other human organs (12). The mRNA was induced 3-fold when prostate adenocarcinoma cells were treated with androgen. The cDNA sequence predicted that the protein was an enzyme of the family of short-chain aldehyde reductases, hence the original name PSDR1. The protein was overexpressed in insect tissue culture cells by cDNA transfection, and this led to an increase in the activity of a membrane-bound enzyme that reduced all-trans-retinal to the corresponding alcohol using NADPH as an electron donor (11). The enzyme also reduced 9-cis-retinal and 13-cis-retinal. This led to the renaming of the enzyme as RalR1. A subsequent study using histochemistry demonstrated that PSDR1 is expressed in human and bovine retinal pigment epithelium, and the enzyme was renamed RDH11 (12). The previous publications on the human enzyme did not examine the ability of the enzyme to reduce any aldehydes other than retinals.
To determine the substrate specificity of mouse SCALD, we produced the enzyme in bacteria, purified it to near homogeneity, and reconstituted it in active form. Here, we show that the enzyme has activity against short-chain aldehydes such as nonanal in addition to retinaldehydes. Polyunsaturated fatty acids can give rise to a variety of aldehydes, including nonanal, when they undergo oxidation (13). The promiscuous specificity of SCALD stands in contrast to another aldehyde dehydrogenase (designated photoreceptor retinol dehydrogenase (prRDH) (14)) that we show is specific for retinaldehydes and does not act upon nonanal. Moreover, we show that SCALD is up-regulated by SREBPs, which also up-regulate the expression of the enzymes that synthesize and elongate unsaturated fatty acids (1). Unlike prRDH, which is expressed in rod outer segments but not in eight other tissues (14), SCALD is expressed in a wide variety of organs, including testis, liver, and retina. Considered together, these findings suggest that SCALD plays a role in the metabolism of short-chain aldehydes in addition to retinaldehydes, and therefore we propose the less restrictive name SCALD.
General Methods-Lipoprotein-deficient newborn calf serum was prepared as described (15). Northern blot hybridization of total RNA was carried out as described (9). Filters were exposed to X-Omat blue film (Kodak) with intensifying screens at Ϫ80°C for ϳ16 h. Quantitative real-time PCR was carried out as described (16), using SCALDspecific primers (5Ј-GCGGCTCGTGAGATCCAA-3Ј and 5Ј-GGCAAAG-GCTCGAATAGACTTG-3Ј). Poly(A) ϩ RNA was isolated from total RNA using oligo(dT)-cellulose columns (Amersham Biosciences). Immunoblot analysis was carried out as described (17) with the use of the Super Signal CL-HRP substrate (Pierce). Filters were exposed to X-Omat blue film for 1-5 min at room temperature.
Subtractive Hybridization and Selection of Differentially Expressed Genes-Suppression subtractive hybridization was performed using the PCR-Select cDNA subtraction kit (Clontech) according to the manufacturer's protocol. Briefly, cDNA was prepared from 2 g of poly(A) ϩ RNA obtained from the brown adipose tissue of wild-type and aP2-SREBP-1c transgenic mice (9). Both the wild-type and transgenic cDNA pools were digested with RsaI. The digested aP2-SREBP-1c "tester" cDNA pool was divided into two groups: one pool was ligated to the supplied adaptor set 1, and the other to adaptor set 2R. After separate melting and hybridization of these pools with wild-type "driver" cDNA, the two tester pools were annealed and subjected to adaptor-specific PCR to yield "forward"-subtracted cDNA. The subtraction was also performed with wild-type cDNA as tester and aP2-SREBP-1c cDNA as driver to produce a "reverse"-subtracted cDNA pool. The forward-subtracted cDNA was subcloned into pT-Adv TA cloning vector (Clontech), and transformed cells were plated onto ampicillin-containing Luria-Ber-tani-agar plates, generating a subtracted cDNA library. Recombinant colonies were grown in 100-l Luria-Bertani cultures in 96-well plates at 37°C for 2 h with shaking. The cloned inserts were amplified using the PCR-Select differential screening kit (Clontech). The cDNAs were arrayed on duplicate filters and hybridized with 32 P-labeled probes derived from the forward-and reverse-subtracted cDNA pools after sequential digestion with RsaI, SmaI, and EagI to remove the adapter sequences. Approximately 350 clones that hybridized specifically with the forward-subtracted probes with a signal intensity of Ͼ2-fold relative to the reverse-subtracted probes were each sequenced in their entirety, and 26 clones with novel sequences were identified and further characterized by Northern blots to confirm their differential expression. Four of the twenty-six clones showed a Ͼ10-fold higher expression in brown adipose tissue from aP2-SREBP-1c mice versus wild-type mice. All four of these clones corresponded to SCALD.
cDNA Cloning of SCALD-A cDNA library ligated to the Marathon cDNA adaptor was prepared from the mouse aP2-SREBP-1c brown adipose tissue poly(A) ϩ RNA using the Marathon cDNA amplification kit (Clontech). The 5Ј-RACE fragment of SCALD was obtained by PCR using the 5Ј-primer AP1 (5Ј-CCATCCTAATACGACTCACTATAGGGC-3Ј) provided by the manufacturer and the 3Ј-primer based on the 5Ј sequence of the SCALD fragment (5Ј-GTCCAGACCACTGAAGAGCTT-TCTC-3Ј). The 3Ј-RACE fragment was amplified with the 5Ј primer based on the 3Ј sequence of the SCALD fragment (5Ј-GAGAAG-GAAGCTCTTCAGTGGTCTGGAC-3Ј) and the 3Ј-primer AP1. 5Ј-and 3Ј-RACE fragments were sequenced, and the full-length SCALD cDNA was amplified from the cDNA library using primers 5Ј-GTCTAGGAT-CAGAAGCCATAAGTCC-3Ј and 5Ј-GTCAGCCCGAGGAAATGGCAAA-CTCCC-3Ј based on 5Ј-and 3Ј-RACE fragment sequences, respectively. The 2.7-kb PCR product was subcloned into the pT-Adv TA cloning vector (Clontech). The resulting plasmid was designated pT-Adv-SCALD and sequenced.
Expression Plasmids-pSCALD-Flag and pSCALD-His encode epitope-tagged versions of mouse SCALD and were generated by PCR with pT-Adv-SCALD as a template using the 5Ј-primer (5Ј-CTGAGAT-GTTCGGATTCCTGCTTCTGCTCTCTC-3Ј) for both plasmids and the 3Ј-primer (5Ј-TCACTTGTCGTCGTCGTCCTTGTAGTCCCAATCCAC-TGGGAGGCCCAGCAGGT-3Ј) for the FLAG-tagged version and the 3Ј-primer (5Ј-TTAATGATGATGATGATGATGCCAATCCACTGGGAG-GCCCAGCAGGT-3Ј) for the His-tagged version. The PCR product was cloned into the pTarget vector (Promega) by TA cloning. A similar construct without a COOH-terminal tag was generated by PCR and designated pSCALD. pHis-SCALD encodes a histidine-tagged version of mouse SCALD and was generated by PCR using the 5Ј-primer (5Ј-GGAATTCCATATGTTCGGATTCCTGCTTCTGCTC-3Ј) and 3Ј-primer (5Ј-CCTTAAGCATATGTTACCAATCCACTGGGAGGCCCAG-3Ј) and pT-Adv-SCALD as a template. The resulting PCR product was digested with NdeI and ligated into a NdeI-digested and dephosphorylated pET-15b vector (Novagen) encoding six consecutive histidines, followed by the cloning site. All expression plasmids were verified by DNA sequencing of the entire coding regions and cloning sites.
The prRDH-His expression plasmid encodes a His-tagged version of the bovine photoreceptor retinol dehydrogenase, as described previously by Rattner et al. (14).
Purification of Detergent-solubilized Recombinant SCALD-The bacterial strain BL21(DE3) was transformed with pHis-SCALD and plated on Luria-Bertani-agar containing 100 g/ml carbenicillin and 1% (w/v) glucose to repress basal expression of SCALD. Expression and purification of SCALD was carried out according to the pET System Manual (Novagen). Briefly, after isolation of inclusion bodies by centrifugation, SCALD was solubilized using solubilization buffer (50 mM CAPS at pH 11, 1 mM dithiothreitol, and 1% (v/v) Nonidet P-40) and centrifuged at 10 5 ϫ g at 4°C for 30 min to pellet the remaining insoluble material. Bovine serum albumin (50 mg/ml solution) was added to the soluble fraction to achieve a final concentration of 1 mg/ml, after which the fraction was dialyzed sequentially against buffer A (20 mM potassium phosphate at pH 8, 1 mM dithiothreitol, and 1% Nonidet P-40) for 3 h at 4°C and buffer B (100 mM potassium phosphate at pH 7.5, 1 mM dithiothreitol, 1% Nonidet P-40, and 10% (v/v) glycerol) overnight at 4°C. The protein was further purified by fast protein liquid chromatography using a cation exchange column (HiTrap TM SP-Sepharose HP from Amersham Biosciences) and buffer B containing increasing concentrations of NaCl (0 to 1 M). Fractions containing SCALD were pooled, and the protein content was estimated by SDS-PAGE using bovine serum albumin as standard.
Affinity Purification of Rabbit Polyclonal SCALD Antibody-A rabbit polyclonal antibody against mouse SCALD was prepared by immunizing rabbits with a synthetic peptide (amino acids 301 to 316 of SCALD) coupled with keyhole limpet hemocyanin. For immunoblotting, whole serum was used at a 1:1000 dilution. For immunohistochemistry, the antibody was affinity-purified on a column in which the above synthetic peptide was coupled to agarose beads using a SulfoLink kit (Pierce).
Cell Culture, Transfection, and Fractionation-On day 0, human embryonic kidney 293 (HEK-293) cells were plated at a density of 2 ϫ 10 6 cells/100-mm dish in medium A (Dulbecco's modified Eagle's medium containing 100 units/ml penicillin and 100 g/ml streptomycin sulfate) supplemented with 10% (v/v) fetal calf serum. On day 1, cells were cotransfected with 0.5 g of pVA1 (18) plus 2 g of one of the following vectors: pTarget, pSCALD, pSCALD-Flag, pSCALD-His, or prRDH-His using the MBS transfection kit (Stratagene) according to the manufacturer's instructions. Briefly, the cells were switched to medium A supplemented with 6% (v/v) MBS solution. DNA suspensions (2.5 g/dish) were incubated with the cells for 3 h at 35°C under 3% CO 2 . Monolayers were then switched back to medium A supplemented with 10% fetal calf serum. Cells were harvested on day 2 for preparation of cell fractions.
M19 cells are a mutant line of CHO-K1 cells that lack nuclear SREBPs because of a deletion in the Site-2 protease gene (19). N-BP1a and N-BP2 cells are lines of transfected M19 cells that express the cleaved nuclear forms of human SREBP-1a and SREBP-2, respectively, under control of a ponasterone-inducible nuclear receptor system (20). On day 0, N-BP1a and N-BP2 cells were plated at a density of 7.5 ϫ 10 5 /100-mm dish in medium B (a 1:1 mixture of Ham's F-12 medium and Dulbecco's modified Eagle's medium containing 100 units/ml of penicillin and 100 g/ml streptomycin sulfate) supplemented with 5% fetal calf serum, 5 g/ml cholesterol, 1 mM sodium mevalonate, 20 M sodium oleate, and 0.05% (v/v) ethanol. On day 2, the monolayers were switched to medium B supplemented with 5% lipoprotein-deficient serum in the absence or presence of 1 M ponasterone A in the case of N-BP1a cells and in the absence or presence of 10 M ponasterone A for N-BP2 cells (20). Cells were harvested on day 3 for preparation of membrane fractions.
Murine embryonic fibroblasts (MEF-1 cells) were plated at a density of 5 ϫ 10 5 /100-mm dish in medium A supplemented with 10% fetal calf serum. On day 1, the monolayers were switched to medium A supplemented with 10% lipoprotein-deficient serum containing 0.2% ethanol in the absence or presence of sterols (10 g/ml cholesterol and 1 g/ml 25-hydroxycholesterol), 50 M compactin, and 50 M sodium mevalonate as indicated. Cells were harvested on day 2 for preparation of membrane fractions.
For SDS-PAGE, cell fractionation was carried out as described (21), and the membrane fractions were either resuspended in SDS buffer and subjected to electrophoresis or treated with 100 mM Na 2 CO 3 for 15 min at room temperature or 0.25% Nonidet P-40 for 30 min at 4°C, pelleted at 10 5 ϫ g to separate soluble and insoluble material, and then subjected to electrophoresis (21). For assays of enzymatic activity, a microsomal fraction (10 5 ϫ g pellet) was prepared as described (22).
Animal Tissues-Tissues were collected immediately after sacrifice of the animals. For Northern blots, tissues were quickly frozen in liquid nitrogen for subsequent extraction of RNA. For immunoblotting, all operations were done at 4°C. Tissues were homogenized with a Polytron into ϳ5 volumes of sucrose buffer (5 mM Hepes at pH 7.6, 250 mM sucrose, 0.3% (v/v) Triton X-100, and a mixture of protease inhibitors (1 mM phenylmethylsulfonyl fluoride, 10 g/ml aprotinin, 10 g/ml leu- Overbars denote the two signature sequences for the superfamily of shortchain dehydrogenase/reductases (37). GenBank TM accession numbers for mouse and human SCALD are AF474027 and AAF89632, respectively. Human SCALD is also called PSDR1 (10), RDH11 (12), and RalRI (11). B, Kyte-Doolittle hydropathy plot of mouse SCALD. The residuespecific hydropathy index (38) was calculated over a window of 18 residues using DNA STAR software Version 5.0. C, membrane localization of SCALD by immunoblot analysis of cell fractions. On day 0, cultured HEK-293 cells were set up and transfected on day 1 with either empty vector (Mock) or pSCALD-Flag as described under "Experimental Procedures." On day 2, the cells were harvested, and whole cell (Whole), supernatant cytosolic (S), and pelleted membrane (P) fractions were isolated (lanes 1-4). Aliquots of the membrane fraction were treated with Na 2 CO 3 or Nonidet P-40 as described under "Experimental Procedures" and fractionated again into supernatant (S) and pellet (P) fractions (lanes 5-8). Aliquots of protein (30 g) from the different fractions were subjected to SDS-PAGE and immunoblotted with 5 g/ml of mouse monoclonal anti-Flag antibody.
RNA in Situ Hybridization-Probe template corresponding to SCALD cDNA nucleotides 3 to 413 (where ϩ1 corresponds to A of initiator methionine) was amplified by PCR using primers (5Ј-GTTCG-GATTCCTGCTTCTGCTC-3Ј and 5Ј-CCGTCTGCAGTCTTCGAGTAG-TAGG-3Ј) and cloned into pCRII-TOPO (Invitrogen). 35 S-Labeled sense and antisense probes were generated by Sp6 and T7 RNA polymerases, respectively, from linearized cDNA templates by in vitro transcription using the Maxiscript kit (Ambion, Inc., Austin, TX). Radioisotopic in situ hybridization was performed as described (25). Autoradiographic exposure ranged from 21 to 28 days at 4°C.
Immunohistochemistry-Serial sections of testis and retina were incubated with either 0.8 g/ml of affinity-purified anti-SCALD antibody or 10 g/ml of preimmune serum according to immunoperoxidase methods described previously (26,27) for testis and immunofluorescence methods (28) for retina.
Assay of Enzyme Activity-All reactions were carried out in buffer D (100 mM potassium phosphate at pH 7.2, 200 mM NaCl, 1 mM dithiothreitol, 2 mg/ml bovine serum albumin, 1% glycerol, and 150 M NADPH). Reactions with detergent-solubilized recombinant SCALD were carried out in a final volume of 1 ml of buffer D containing 0.3% (w/v) Zwittergent 3-16 and 1 g of enzyme. Reactions with transfected microsomal SCALD and prRDH were carried out in the absence of detergent in a final volume of 0.5 ml of buffer D containing either 200 g of microsomal protein (for reactions with steroids and carbonyl compounds) or 10 g of microsomal protein (for reactions with retinaldehydes). All reactions were initiated by a 1-l addition of ethanol containing the indicated concentration of substrate.  A and B) and protein (C and D). A, Northern blot analysis of mouse tissue RNAs. Total RNA was isolated from white (WAT) and brown (BAT) adipose tissues of wild-type (WT) and transgenic mice expressing the nuclear form of SREBP-1c (9). Total liver RNA was isolated from wild-type and transgenic mice expressing the nuclear forms of SREBP-1a (39) and SREBP-2 (40). Aliquots of total RNA (20 g) were subjected to electrophoresis and blot hybridization with a mouse SCALD-specific 32 P-labeled probe (nucleotides 190 -783, where ϩ1 corresponds to A of initiator methionine) and a control 32 P-labeled probe directed against mouse ␤-actin (39). B, relative amounts of SCALD mRNA in livers from wild-type (WT) and SCAP f/f ; MX1-Cre (SCAP Ϫ/Ϫ ) mice (34) were subjected to fasting (F) and refeeding (RF). Total RNA from livers of mice (four mice/group) was pooled and subjected to real-time PCR quantification as described (16). Each value represents the amount of mRNA relative to that in the nonfasted wild-type mice in the same experiment (control), which is arbitrarily defined as 1. C, immunoblot analysis of N-BP cells. On day 0, N-BP1a and N-BP2 cells were set up as described under "Experimental Procedures." On day 2, both cell lines were switched to medium B supplemented with 5% lipoprotein-deficient serum in the absence or presence of ponasterone A as indicated. After incubation for 16 h at 37°C, the cells were harvested and fractionated, and aliquots of the membrane fraction (30 g) were subjected to SDS-PAGE and immunoblotted with a 1:1000 dilution of rabbit anti-SCALD antiserum. D, immunoblot analysis of MEF-1 cells. On day 0, MEF-1 cells were set up as described under "Experimental Procedures." On day 1, the cells were switched to medium A supplemented with 10% lipoprotein-deficient serum in the absence or presence of sterols (10 g/ml of cholesterol plus 1 g/ml of 25-hydroxycholesterol) and/or compactin (50 M compactin plus 50 M sodium mevalonate) as indicated. After incubation for 16 h at 37°C, cells were harvested, and the membrane fractions were subjected to immunoblot analysis as described in C.
With steroids and carbonyl compounds as substrates, SCALD activity in detergent-solubilized recombinant protein was measured by monitoring the rate of NADPH oxidation spectrophotometrically at 340 nm during a 5-min incubation at room temperature (ϳ22°C) using an Ultrospec 3000 spectrophotometer (Pharmacia Biotech). Enzyme activity was calculated based on an absorption coefficient of 6220 M Ϫ1 cm Ϫ1 at 340 nm for NADPH.
With steroids and carbonyl compounds as substrates, SCALD and prRDH activity in microsomes from transfected HEK-293 cells was measured by monitoring the production of 3 H-labeled products. For these experiments, (4S)-[4-3 H]NADPH was prepared from D-[1-3 H]glucose (20 Ci/mmol) as described (29) and used instead of unlabeled NADPH in buffer D. After incubation at 37°C for 1 h, the 3 H-labeled reaction products were extracted with 6 volumes of chloroform:methanol (2:1), and one-sixth aliquots of the reactions were quantified by scintillation counting. Blank values obtained in the absence of substrates were subtracted from measured values.
With retinaldehydes as substrates, SCALD and prRDH activity in detergent-solubilized recombinant protein and in microsomes from transfected HEK-293 cells was measured by monitoring the reduction of retinaldehydes by reverse-phase HPLC. Reactions were incubated for 15 min at room temperature (when using detergent-solubilized SCALD) or 37°C (when using microsomal SCALD and prRDH) and terminated by the addition of an equal volume of cold ethanol. Extraction of retinoids was performed twice with 4 volumes of hexane, and the pooled organic phases were dried under nitrogen at room temperature. Retinoids were resuspended into 100 l of methanol:water (95:5) and subjected to HPLC analysis. The stationary phase was a supelcosil LC-18 column from Supelco, and the mobile phase was methanol:water (95:5). The flow rate was 2 ml/min, injection volume was 40 l, and the detection wavelengths were 325 nm for retinol detection and 373 nm for retinal detection. Retinoids were quantified by comparing their peak areas to calibration curve of standards made using the indicated wavelength. All procedures involving retinoids were performed under dim light.
Data Analysis-Kinetic constants (apparent K m and V max ) were assessed with the ENZFIT computer program (Elsevier Biosoft), using the Michaelis-Menten equation for curve fitting.

RESULTS
We used the technique of suppression subtractive hybridization to identify novel target genes that are activated by SREBPs in brown adipose tissue of transgenic mice that express dominant-positive nSREBP-1c. Forward-and reversesubtracted cDNA libraries were derived from brown adipose tissue mRNAs obtained from wild-type and transgenic mice.
After PCR amplification, recombinant cDNAs were arrayed on filters and hybridized using the forward-and reverse-subtracted cDNA pools as probes. 350 clones hybridized with the forward-subtracted probe to a significantly higher degree than with the reverse-subtracted probe. All of these clones were sequenced. Among them, some were already known to be induced by SREBPs. These included stearoyl-CoA desaturase (6 clones), fatty-acid desaturase (4 clones), isopentenyl-diphosphate:dimethylallyl-diphosphate isomerase (5 clones), and glycerol-3-phosphate acyltransferase (1 clone). We also isolated four clones derived from different fragments of a single mRNA that was not known previously to be regulated by SREBPs. We cloned this cDNA in its entirety and named the encoded protein SCALD. In February 2002, we deposited this sequence in Gen-Bank (accession no. AF474027). Thereafter, Moore et al. (30) independently cloned the same sequence from the mouse. Fig. 1A presents the amino acid sequence of mouse SCALD. It is 85.4% identical to human PSDR1 (10), also named RalRI (11) and RDH11 (12); it is 100% identical to the sequence reported by Moore et al. (30). SCALD belongs to the family of short-chain dehydrogenase/reductases (31,32). Members of this family are NAD(P)(H)-dependent oxidoreductases. The various members exhibit only 15-30% identity, but all possess two highly conserved motifs that are the signature of this family. Those motifs are overlined in Fig. 1A. The first motif, GXXXGXG, is in the coenzyme-binding domain, and the second motif YXXXK is involved in catalytic activity. The hydropathy plot of SCALD (Fig. 1B) shows a protein of 316 amino acids with a hydrophobic NH 2 -terminal segment of 20 amino acids that likely functions as a signal sequence for insertion into the endoplasmic reticulum. If this sequence is not cleaved, it could also serve as a membrane anchor. As shown in Fig. 1C, when a FLAG-tagged version of SCALD was expressed in HEK-293 cells through transfection, the protein was bound to membranes and could not be dissociated by treatment with sodium carbonate, but it was solubilized with Nonidet P-40. This indicates that SCALD remains bound to the membrane through its uncleaved hydrophobic NH 2 -terminal sequence. This finding is in agreement with that of Kedishvili et al. (11). Fig. 2A shows that SCALD was overexpressed in white and

FIG. 3. Reduction of nonanal by detergent-solubilized recombinant SCALD.
Enzyme reactions were monitored spectrophotometrically for 5 min at room temperature as described under "Experimental Procedures." A, enzyme concentration curve. Increasing amounts of detergent-solubilized SCALD were added to the standard reaction at pH 7.5 in the presence of 100 M nonanal. B, pH curve. Each reaction was carried out at the indicated pH in the presence of 1 g of SCALD and 100 M nonanal. C, NADPH concentration curve. The same conditions as in B were used with the indicated concentration of NADPH at pH 7.2. D, nonanal concentration curve. The same conditions as in B were used with the indicated concentration of nonanal at pH 7.2. B-D, a blank value (ϳ0.4 nmol/min/tube) based on the rate of disappearance of NADPH in parallel reactions lacking SCALD was subtracted from each measured value. Inset in A, SDS-PAGE (12% gel) of detergent-solubilized SCALD (2 g), followed by staining with GelCode Blue reagent (Pierce). brown adipose tissue of the aP2-SREBP-1c mice as determined by Northern blotting of the mRNA (lanes 1-4). The SCALD mRNA was also increased in livers of previously described transgenic mice that overexpress the nuclear forms of SREBP-1a or SREBP-2 in liver ( lanes 5-7). In livers of wildtype mice, the amounts of nSREBPs are known to decline with fasting and rise with the refeeding of a low-fat diet (33). Fig. 2B shows that the amount of SCALD mRNA declined with fasting and rose with refeeding, consistent with regulation by nSREBPs. This conclusion was supported by the finding that SCALD mRNA is markedly reduced in livers of gene-targeted mice with a disruption in the hepatic gene encoding SREBP cleavage-activating protein (SCAP). These livers lack nSREBPs as a result of failure to process the SREBP precursors to the mature nuclear form (34). In these livers the SCALD mRNA declined only slightly with fasting, and it failed to show the normal rise with refeeding, again supporting the conclusion that hepatic SCALD is regulated by SREBPs.
To test this regulation in another way, we used N-BP cells (20), which are lines of Chinese hamster ovary cells that have been engineered to express individual isoforms of nSREBPs in response to induction by the steroid ponasterone (Fig. 2C). We blotted these cells with an antibody directed against a peptide derived from SCALD (see "Experimental Procedures"). Immunodetectable SCALD rose when ponasterone induced the production of nSREBP-1a (lane 2) or nSREBP-2 (lane 4). The SCALD protein was induced in mouse embryonic fibroblasts (MEF-1 cells) that were treated with the 3-hydroxy-3-methylglutaryl coenzyme A reductase inhibitor compactin (Fig. 2D,  lane 3), and it was reduced when sterols were added (lane 2). All of these findings are consistent with regulation of SCALD by SREBPs.
To characterize the enzymatic activity of purified SCALD, we expressed the protein in bacteria and purified it from inclusion bodies using fast protein liquid chromatography. Although the protein contained a His tag, we were unable to use heavy metal chromatography because this procedure irreversibly inacti-vated the enzyme. Fig. 3A (inset) shows that the purified protein gave a single band on a Coomassie-stained SDS-polyacrylamide gel. This band stained with the antibody to SCALD (not shown). Enzyme activity was measured by the substrate-dependent oxidation of NADPH as determined by a decrease in absorbance at 340 nm. After screening several aldehydes as substrates, we selected the short-chain saturated aldehyde, nonanal, for detailed study. In the presence of nonanal, oxidation of NADPH was linear with enzyme protein up to 1 g in a 1-ml volume (Fig. 3A). The reaction was optimal between pH 6.5 and 7.5 (Fig. 3B). The apparent K m values for NADPH and nonanal were 20 and 30 M, respectively (Fig. 3, C and D). The V max was 5 nmol/min/tube. This value corresponds to a K cat of 167 min Ϫ1 . The enzyme did not oxidize NADH in the presence of nonanal, suggesting specificity for NADPH. We observed no reduction of NADP or NAD in the presence of nonanol, suggesting that the enzyme highly favors aldehyde reduction as opposed to alcohol oxidation (data not shown).
We tested a variety of aldehydes and keto-containing compounds for their ability to support the oxidation of NADPH in the presence of SCALD (Fig. 4). For this purpose, we used two enzyme preparations: 1) the detergent-solubilized protein described above, which was produced in bacteria and purified by fast protein liquid chromatography; and 2) crude microsomes from HEK-293 cells transfected with a cDNA encoding SCALD. The latter was designed to mimic the enzyme in its native state embedded in membranes of the endoplasmic reticulum.
For the purified enzyme in detergent solution, we used the spectrophotometric NADPH oxidation assay described above. For the microsomal assay, we measured the incorporation of 3 H from [ 3 H]NADPH into the indicated substrate as measured by organic extraction (see "Experimental Procedures"). To compare the relative activities for each substrate, we normalized the data to the activity observed with nonanal, which was set at 1. All substrates were tested at a concentration of 100 M, which is a near-saturating level for nonanal. The results in Fig.  4 are arranged in a linear order from lowest to highest activity FIG. 4. Substrate specificity of SCALD: comparative activity of detergent-solubilized enzyme (A) and microsomal enzyme (B). Assays in A were carried out with 1 g of detergentsolubilized recombinant SCALD using the spectrophotometric assay as described under "Experimental Procedures." Assays in B were carried out with 200 g of microsomal protein from pSCAP-transfected HEK-293 cells using the [ 3 H]NADPH assay as described under "Experimental Procedures." In both assays, the indicated substrates were added at a final concentration of 100 M. The results (means of triplicate assays) are expressed as relative activities compared with the activity of nonanal, which was set at a value of 1.0. The activity for nonanal in A and B were 4.5 nmol/min/tube and 9821 cpm/h/ tube, respectively. for the assay with purified SCALD. In general, we observed a close correlation between the activities of the purified and microsomal enzyme preparations (Fig. 4), although there were some exceptions. Neither enzyme preparation showed significant reductive activity toward the keto-containing steroids progesterone and dehydroisoandrosterone, whereas both preparations were active against a variety of short-and medium-chain aldehydes of 8 -10 carbons containing zero to two double bonds.
We could not use the spectrophotometric assay to measure reduction of retinaldehydes because these substrates absorb light at 340 nm. However, we could use the [ 3 H]NADPH reduction assay to measure the reduction of retinaldehydes in the SCALD-expressing microsomes. At the standard substrate concentration of 100 M, the reduction of all-trans-retinal and 9-cis-retinal was markedly slower than the reduction of nonanal (Fig. 4B).
To measure retinaldehyde reduction with the purified recombinant enzyme, we use HPLC to separate the retinol products from the retinaldehyde substrates (Fig. 5). For comparative purposes in the same experiment, we measured the reduction of nonanal using the spectrophotometric assay. The results showed that the K m values for all-trans-retinal (Fig. 5B) and 9-cis-retinal (Fig. 5C) were in the same range as the K m for nonanal (Fig. 5A). The V max for the retinaldehydes was lower than the V max for nonanal.
To test the significance of the broad substrate specificity of SCALD, we compared the activity of this enzyme with the activity of prRDH, an enzyme that is specifically expressed in retinal rod and cone outer segments and that appears to be specialized for the reduction of all-trans-retinal (14). We obtained an expression plasmid encoding prRDH as a generous FIG. 5. Reduction of nonanal, all-trans-retinal, and 9-cis-retinal by detergent-solubilized SCALD. All assays were carried out with 1 g of detergent-solubilized SCALD. A, reactions were incubated at room temperature for 5 min and assayed spectrophotometrically as described in Fig. 3. B and C, after incubation at room temperature for 15 min, the reactions were terminated by the addition of an equal volume of cold ethanol, after which the retinoids were extracted twice with four volumes of hexane and reconstituted into 100 l of methanol: water (95:5). Forty-l aliquots were analyzed by reverse-phase HPLC as described under "Experimental Procedures." gift from the laboratory of Jeremy Nathans. HEK-293 cells were transfected with a cDNA encoding His-tagged prRDH or a cDNA encoding His-tagged SCALD. Microsomes were subjected to immunoblotting with an anti-His antibody, and the expression levels of the two proteins were comparable (Fig. 6A). The microsomes were assayed for their ability to reduce nonanal using the [ 3 H]NADPH assay and their ability to reduce all-trans-retinal using the HPLC assay. The apparent K m for the SCALD-mediated reduction of all-trans-retinal was much lower with the microsomal preparation (1.3 M) than it was for the purified enzyme in detergents (57 M, see Fig. 5). This difference was reproducible with several enzyme preparations. The microsomal preparation of prRDH had a similarly low K m for all-trans-retinal (Fig. 6C). However, whereas SCALD was active on nonanal, prRDH showed no measurable activity (Fig. 6B).
Human PSDR1 was reported to have greatest expression in the prostate (10), whereas mouse PSDR1, which is identical to mouse SCALD, was reported to be expressed most highly in testis and liver and much less in prostate (30). The Northern blots of Fig. 7A confirm the highest levels of SCALD expression in testis with relatively high levels in liver, adrenal, and ovary. Expression of SCALD was barely detectable in mouse prostate. A similar relative distribution of SCALD (testis Ͼ liver Ͼ Ͼ Ͼ prostate) was seen in rat tissues (Fig. 7B). Immunoblots of mouse tissues with the anti-SCALD antibody revealed similar levels of protein expression in liver and testis despite the dif-ferences in mRNA levels. The apparent molecular mass of the protein was 31 kDa, which is consistent with the size predicted from the mRNA sequence. This antibody was not potent enough to detect SCALD protein in any of the other mouse tissues that were examined (Fig. 7C).
In situ hybridization experiments revealed that SCALD expression in testis was confined to the seminiferous tubules (Fig.  8A). Only a narrow layer of the tubules was visualized, suggesting that SCALD is expressed in a particular stage of sperm cell development. No hybridization signal was obtained with the control sense probe (Fig. 8B). Higher magnification suggested that SCALD mRNA is concentrated in the peripheral layer of the seminiferous tubule (Fig. 8C), which contains pachytene spermatocytes as indicated by bright-field examination of the same hematoxylin-stained section (Fig. 8D). Immunochemical staining confirmed the expression of SCALD protein in pachytene spermatocytes (Fig. 8, E and F). The protein was concentrated in a perinuclear body consistent with localization to the Golgi complex. Remarkably, we were unable to detect SCALD mRNA or protein in either less mature or more mature sperm, suggesting that the expression of SCALD is highly stage-specific in sperm differentiation.
Immunohistochemical staining of the retina with an affinitypurified anti-SCALD antibody revealed specific immunoreactivity within distinct layers of the retina (Fig. 9, top). Punctate cytoplasmic staining was evident in the outer segment of the layer of rods and cones. In the outer plexiform layer, the stain- FIG. 7. Expression of SCALD mRNA  (A and B) and protein (C) in tissues from mice and rats. Total RNA was isolated from the indicated tissues of adult male (except for ovary) mice (A) and male rats of the indicated age (B). Aliquots of total RNA (20 g) were subjected to electrophoresis and Northern blot hybridization with a mouse SCALD-specific 32 Plabeled probe (see Fig. 2A) and a control 32 P-labeled probe directed against rat cyclophilin as described (41). C, immunoblot analysis of SCALD in mouse tissues. Homogenates of the indicated male adult mouse tissues (10 3 ϫ g supernatant fraction) were prepared as described under "Experimental Procedures." Aliquots (30 g) were subjected to SDS-PAGE and immunoblotted with a 1:1000 dilution of rabbit anti-SCALD antiserum.
ing pattern was restricted to the innermost region and assumed a fibrillar morphology. Punctate staining was evident in a subset of cells in the innermost region of the inner nuclear layer, where amacrine cells predominate. Neurons in the layer of ganglion cells were also positively stained. DISCUSSION The data in this paper extend our knowledge of the mammalian short-chain dehydrogenase/reductase that is known variously as PSDR1, RalR1, RDH11, and SCALD. For purposes of this discussion, we will use the name SCALD. The new findings are: 1) transcription of the SCALD gene is regulated by SREBPs. As a result, in tissue culture cells the gene is transcribed actively when the cells are deprived of sterols and is repressed when sterols are abundant. In liver, SCALD mRNA declines with fasting and reappears with refeeding; 2) in contrast to the reported enrichment of the corresponding human enzyme in prostate, SCALD is expressed at very low levels, if at all, in the prostate of mice and rats, and it is enriched at the mRNA level in testis, liver, adrenal, and ovary of rodents. At the protein level, the enzyme was present at equal levels in mouse testis and liver and at much lower levels in other tissues. Immunohistochemical studies of the mouse retina showed that SCALD is expressed in four layers of the retina in contrast to the more restricted expression of RDH11 (exclusively in bovine and human retinal pigment epithelium) (12) and prRDH (exclusively in the outer segments of macaque rods and cones) (14); 3) in mouse testis, SCALD shows a highly restricted pattern of expression in pachytene spermatocytes, but not in progenitor spermatogonia or in mature sperm; 4) in contrast to the previously described enzyme designated prRDH, which is expressed in retinal rods and reduces alltrans-retinal, purified recombinant SCALD shows a broad substrate specificity, reducing short-chain aliphatic aldehydes such as nonanal as actively as it reduces retinals. Considered together, these data indicate that mouse SCALD and its human ortholog (PSDR1/RalR1/RDH11) have functions that may be much broader than simply reducing retinaldehydes.
The reason for the tight regulation of SCALD transcription by SREBPs is unknown, but the following hypothesis must be considered. SREBPs are known to enhance the synthesis of fatty acids, their elongation to stearate, and their subsequent unsaturation to yield oleate (1). In addition, through their induction of the low density lipoprotein receptor mRNA, SREBPs enhance the cellular uptake of lipoproteins that contain polyunsaturated fatty acids, especially linoleate and arachidonate. Unsaturated fatty acids are subject to oxidation by molecular oxygen (13). Oxidation yields a wide variety of C 2-12 fatty aldehydes, including 4-hydroxy-2-nonenal and nonanal. Such aldehydes are toxic to cells because they form Schiff bases with lysine residues in cellular proteins. Hence, as a protective maneuver SREBPs may induce SCALD so as to reduce these toxic aldehydes to nontoxic alcohols. If this hypothesis is correct, it would predict that fatty aldehydes or their protein adducts should accumulate when SREBP activity is stimulated in livers of knock-out mice that lack the SCALD gene. We are currently in the process of preparing and studying SCALD knock-out mice for this purpose.
The reason for the expression of SCALD specifically in pachytene spermatocytes remains to be elucidated. Recently, Wang et al. (35) reported the specific expression in mouse and rat spermatogenic cells of an aberrantly spliced mRNA that encodes a truncated form of SREBP-2 that terminates prior to the membrane-spanning sequence. We have shown previously that such a truncated SREBP-2 (encoded by a mutant gene in sterol-resistant Chinese hamster ovary cells) goes directly to the nucleus and activates transcription without a requirement for proteolytic processing, and hence it is not down-regulated FIG. 8. Visualization of SCALD mRNA and protein in adult mouse testis. A, low-power view of a dark field-illuminated testis that was hybridized with an antisense SCALD probe as described under "Experimental Procedures." Note heterogeneity of expression within the seminiferous tubules. B, hybridization of same field with the sense probe (control). C, high-power view of boxed area in A (arrows). Note expression in the peripheral layers corresponding to the pachytene spermatocytes. D, bright-field illumination of identical field as in C. E, immunohistochemical localization with affinity-purified anti-SCALD antibody, illustrating labeling of Golgi complex in pachytene spermatocytes (arrows). F, incubation of same field with preimmune serum (control). A-F, each bar represents 100 m.
FIG. 9. Immunofluorescence localization of SCALD in the mouse retina. Tissue sections were double-labeled with propidium iodide to stain the nucleus (red) and affinity-purified anti-SCALD antibody (green) in the top panel or preimmune serum in the bottom panel as described under "Experimental Procedures." S denotes the sclera; numbers refer to the following layers of the retina: 1, pigment epithelium; 2, rods and cones; 3, external limiting membrane; 4, outer nuclear layer; 5, outer plexiform layer; 6, inner nuclear layer; 7, inner plexiform layer; 8, ganglion cell layer. Bar, 40 m. by sterols (36). Further studies are necessary to determine whether the specific expression of SCALD in pachytene spermatocytes is triggered by the endogenous truncated form of SREBP-2.