cDNA cloning and expression of a human aldehyde dehydrogenase (ALDH) active with 9-cis-retinal and identification of a rat ortholog, ALDH12.

This report describes the isolation of a heretofore uncharacterized aldehyde dehydrogenase (ALDH) with retinal dehydrogenase activity from rat kidney and the cloning and expression of a cDNA that encodes its human ortholog, the previously unknown ALDH12. The human ALDH12 cDNA predicts a 487-residue protein with the 23 invariant amino acids, four conserved regions, cofactor binding motif (G(209)XGX(3)G), and active site cysteine residue (Cys(287)) that typify members of the ALDH superfamily. ALDH12 seems at least as efficient (V(m)/K(m)) in converting 9-cis-retinal into the retinoid X receptor ligand 9-cis-retinoic acid as two previously identified ALDHs with 9-cis-retinal dehydrogenase activity, rat retinal dehydrogenase (RALDH) 1 and RALDH2. ALDH12, however, has approximately 40-fold higher activity with 9-cis- retinal than with all-trans-retinal, whereas RALDH1 and RALDH2 have equivalent and approximately 4-fold less efficiencies for 9-cis-retinal versus all-trans-retinal, respectively. Therefore, ALDH12 is the first known ALDH to show a preference for 9-cis-retinal relative to all-trans-retinal. Evidence consistent with the possibility that ALDH12 could function in a pathway of 9-cis-retinoic acid biosynthesis in vivo includes biosynthesis of 9-cis-retinoic acid from 9-cis-retinol in cells co-transfected with cDNAs encoding ALDH12 and the 9-cis-retinol/androgen dehydrogenase, cis-retinoid/androgen dehydrogenase type 1. Intense ALDH12 mRNA expression in adult and fetal liver and kidney, two organs that reportedly have relatively high concentrations of 9-cis-retinol, reinforces this notion.

The ligand-activated transcription factors retinoic acid receptor and retinoid X receptor control expression of diverse genes essential for embryonic development and throughout the life spans of vertebrates (1)(2)(3)(4). tRA 1 seems to function as the major ligand of retinoic acid receptor in vivo but does not bind with retinoid X receptor. 9cRA, in contrast, can serve as a high-affinity ligand for both retinoic acid receptor and retinoid X receptor in vitro and has been identified as a retinoid X receptor ligand in cells treated with tRA (5,6). tRA derives from the major parent retinoid in animals, all-trans-retinol. Sequential reactions generate tRA from all-trans-retinol: reversible dehydrogenation of all-trans-retinol into all-trans-retinal, and irreversible dehydrogenation of all-trans-retinal into tRA (7). Dietary all-trans-retinyl esters and all-trans-␤-carotene provide all-trans-retinol and serve as the major sources of vitamin A activity for mammals. Diet also contains substantial 9-cis-␤-carotene, which can undergo metabolism into 9-cis-retinol (8,9). Several short-chain alcohol dehydrogenases convert all-trans-and/or 9-cis-retinol into all-trans-and 9-cis-retinal, respectively (10), and RALDH isozymes convert both all-transand 9-cis-retinal into tRA and 9cRA, respectively (11).
Understanding how tissues maintain steady-state concentrations of retinoic acid requires identification of the enzymes that catalyze the two reactions of retinoic acid biosynthesis and when and where each functions. Because rat kidney contains multiple RALDH activities (16), we used protein purification to obtain the amino acid sequence for a heretofore uncharacterized ALDH. The rat amino acid sequence was identical with the deduced amino acid sequence of a partial human genomic clone. We generated the full-length human cDNA and expressed and characterized the enzyme as an ALDH, i.e., ALDH12. ALDH12 has considerable activity with 9-cis-retinal, has much lower activity with all-trans-retinal, and can contribute to a pathway of 9cRA biosynthesis in cells co-transfected with a cDNA that encodes CRAD1, a 9-cis-retinol dehydrogenase (25).

EXPERIMENTAL PROCEDURES
Identification and Sequencing of a Rat Kidney ALDH-Kidney cytosol (270 mg of protein) from male Harlan Sprague-Dawley rats (16) was applied to a Mono-Q column (1.5 ϫ 8 cm) in 20 mM Tris-HCl, pH 8 (buffer A). The column was washed with buffer A until protein elution ceased and then was eluted with buffer A containing 300 mM NaCl. The protein recovered was applied to Affi-Gel blue (0.75 ϫ 15 cm) in 100 mM phosphate buffer, pH 7.5 (buffer B). The column was washed with buffer B and eluted with 2 mM NAD ϩ in buffer B. Each of the two fractions was applied to a Mono-Q HR 5/5 column (Amersham Pharmacia Biotech) in 20 mM Hepes, pH 7.5 (buffer C) and eluted at 0.5 ml/min with a NaCl gradient from 0 to 250 mM in buffer C. Active fractions were analyzed by SDS-polyacrylamide gel electrophoresis. A protein band of ϳ54 kDa from the NAD ϩ eluted fraction was excised and sent to Bill Lane of the Harvard Microchemistry Facility for sequence analysis.
cDNA Cloning of a Human Kidney ALDH-We identified a human genomic clone in GenBank TM (accession number AL021939) that had a partial coding sequence of an ALDH with nine of the unique peptide sequences identified in the rat protein band. To clone the cDNA, 5Ј-RACE (5Ј-AGCTGGACCTGACGGTTGCCGGAATGC-3Ј) and 3Ј-RACE (5Ј-CCAGAGGTGCCCCTGATCTCCTTCACC-3Ј) were done with Human Kidney Marathon-Ready cDNA (CLONTECH). The program was as follows: 1 cycle at 94°C for 3 min, 5 cycles at 94°C for 30 s and 72°C for 3 min, 5 cycles at 94°C for 30 s and 70°C for 3 min, and 25 cycles at 94°C for 30 s and 68°C for 3 min. A primer (5Ј-GGAGGTGATTGAAA-GAGCCAACAACGT-3Ј) was used for nested 3Ј-RACE on the original 3Ј-RACE product to further amplify the sequence. The complete cDNA was cloned by end-to-end polymerase chain reaction with the same template and a program of 1 cycle at 94°C for 2 min; 35 cycles of 94°C for 45 s, 55°C for 45 s, and 72°C for 3 min; and 1 cycle of 72°C for 10 min.
mRNA Blots-A 334-base pair probe, corresponding to nucleotides 1361 to 1694 of the cDNA, was amplified by polymerase chain reaction, labeled with 32 P, and purified using a Micro Bio-Spin 30 column (Bio-Rad). The probe was hybridized to the Human Multiple Tissue Expression Array (CLONTECH). Prehybridizations and hybridizations were done in 10 ml of ExpressHyb solution at 65°C for 30 min and overnight, respectively. The blot was washed five times in 2ϫ SSC (300 mM NaCl and 30 mM sodium citrate) with 1% SDS at 65°C for 20 min and twice with 0.1ϫ SSC/0.5% SDS at 55°C for 20 min. The array was exposed to Kodak X-OMAT LS film for 36 h. This blot is normalized for RNA loading of each dot to account for average tissue differences in mRNA levels of eight housekeeping genes. Thus, the data show relative abundance of target transcripts in different tissues. The same probe was hybridized to the Human Multiple Tissue Northern Blot (CLONTECH). Prehybridizations and hybridizations were done in 10 ml of ExpressHyb solution at 68°C for 30 min and 1 h, respectively. The blot was washed four times with 2ϫ SSC/0.05% SDS for 30 min at room temperature and twice in 0.1ϫ SSC/0.1% SDS for 30 min at 68°C and exposed to Kodak X-OMAT LS film overnight. The blot was reprobed with a ␤-actin probe (CLONTECH) using the same protocol.
Enzyme Assays-The ADLH12 coding region was polymerase chain reaction-amplified with 5Ј-GGAAGCTTATGGCTGGAACAAACGCAC-T-3Ј and 5Ј-CGGGATCCTAGTGGCTCCACCATTAGCA (underlining indicates restriction sites) and cloned into the HindIII/BamHI sites of pcDNA3 (Invitrogen) to give pcDNA3/ALDH12. CHO cells were transfected with pcDNA3/ALDH12 (8 g/100-mm plate) and LipofectAMINE (Life Technologies, Inc.). Cells were harvested 24 h later and sonicated in 20 mM Hepes, 150 mM KCl, 1 mM EDTA, 10% sucrose, and 2 mM dithiothreitol (buffer D), pH 7.5. A supernatant (800 ϫ g spin for 10 min) was used for enzyme assays. Assays were done in triplicate in buffer E (20 mM Hepes, 150 mM KCl, 1 mM EDTA, and 2 mM dithiothreitol), pH 8.5, under initial velocity conditions. Assays with nonretinoids were done in 1 ml of buffer E by monitoring NAD(P)H synthesis (340 nm) at ambient temperature. Substrates were added in 4 l of ethanol. Reactions were initiated by adding NAD(P) ϩ at a final concentration of 1 mM. Assays for retinoids were done in 0.5 ml of buffer E with 2 mM NAD ϩ at 37°C. Retinoids were added in 2 l of Me 2 SO. Retinoic acid was analyzed by high pressure liquid chromatography as described previously (18,23). Lysates from CHO cells transfected with pcDNA3 were used as controls. Kinetic data were fit by nonlinear regression analyses using Graph Pad Prism.
Whole Cell Assays-CHO cells were transfected with combinations of pcDNA3, pcDNA3/CRAD1 (25), and pcDNA3/ALDH12. Twenty-four h after transfection, the medium was replaced with fresh medium (5 ml), and 9-cis-retinol was added in 5 l of Me 2 SO to a final concentration of 1 M. One or 2 h later, the cells and their medium were analyzed for 9cRA by high pressure liquid chromatography (18,23).

RESULTS AND DISCUSSION
Identification of a Novel Rat ALDH-Two ALDH isozymes in rat kidney and liver that catalyze retinal metabolism have been cloned and characterized: (a) RALDH1, and (b) RALDH2 (16 -19). To characterize another, rat kidney cytosol was applied to Mono-Q chromatography to separate the P1 fraction, which contains RALDH1, from the P2 fraction, which contains additional ALDH isozymes (16). Mono-Q does not retain P1 at pH 8 but does retain P2. P2 was eluted from the column with NaCl and separated into two fractions by Affi-Gel blue affinity chromatography. The P2 subfractions were applied separately to anion exchange chromatography. The one not retained by the Affi-Gel blue column behaved as P2c, as reported previously (16). The one that eluted with NAD ϩ behaved as P2a (Fig. 1A). P2c was not studied further because its low pI implies that it represents RALDH2, which has a pI of 5.1 (24). Fractions with FIG. 1. A, anion-exchange chromatography of RALDH fractions. RALDH fraction P2 from rat kidney cytosol was separated into two subfractions by Affi-Gel blue chromatography: one eluted in the flowthrough, and a second eluted with NAD ϩ . Each was analyzed by anionexchange chromatography as described under "Experimental Procedures." The Affi-Gel blue binding fraction eluted in fractions 2-5 from the anion-exchange column, as had the previously reported RALDH fraction P2a (16). Fractions 3 and 4 of P2a contained 62% and 27% of the total recovered activity, respectively. The Affi-Gel blue flow-through fraction eluted at a high salt concentration similar to RALDH fraction P2c, likely representing RALDH2 (16). B, SDS-polyacrylamide gel electrophoresis analysis of P2a isolated from rat kidney cytosol. Fractions 3 and 4 from P2a were analyzed on a 10% SDS-polyacrylamide gel electrophoresis gel stained with Coomassie Blue. The arrowhead denotes the ϳ54-kDa band in fraction 3 (the peak fraction) that was analyzed for peptide sequences. Lane M shows molecular mass markers. The numbers on the right show the migration of the molecular mass markers (kDa).
cDNA Clone Encoding a Human Homolog of the Rat ALDH-The cDNA sequence of AL021939 was generated by 5Ј-and 3Ј-RACE and polymerase chain reaction with primers to both ends with a human kidney cDNA template. The cDNA produced encodes a deduced protein of 487 amino acids with the 23 invariant amino acids and the four conserved regions that typify members of the ALDH superfamily (Fig. 2). Also included are a cofactor binding motif (G 209 XGX 3 G) and a cysteine residue (Cys 287 ) in the appropriate locus to serve as an active site nucleophile (12,13). ALDH12 shows the closest nucleotide/ amino acid similarity with the hydroxymuconic semialdehyde dehydrogenase of Pseudomonas putida but has no more than 50% similarity with other human ALDHs (Table I). Human ALDH genes 1-11 have been cloned and characterized, as have two named but unnumbered human ALDHs (succinic semialdehyde dehydrogenase and methylmalonal semialdehyde dehydrogenase); the new human ALDH therefore represents ALDH12.
mRNA Expression of the Human ALDH-Adult human liver and kidney expressed a 2.5-kilobase ALDH12 mRNA intensely. Expression in other tissues was below detection limits by Northern blot analysis (Fig. 3). Dot-blot analysis confirmed intense mRNA expression in adult liver and kidney ( Fig. 4; Table II). Brain and spinal cord had a detectable but much lower level of expression. Other tissues with detectable expression included mammary gland, thymus, adrenal, testis, prostate, and parts of the gastrointestinal tract. Tissues with no detectable expression included ovary, adult heart, spleen, thyroid gland, bone marrow, tumor cells, and parts of the gastrointestinal tract. Expression was also detected in fetal liver, kidney, brain, heart, thymus, and lung. One of the eight negative controls (Escherichia coli DNA) had a strong signal, suggesting a complementary sequence in E. coli.
Enzymatic Properties-ALDH12 had no detectable activity with 2 mM benzaldehyde at pH 5 or pH 6, but activity increased from pH 7 to pH 9. Activities at the physiological pH of 7.4 and the pH used for assays (pH 8 -8.5) were 62% and ϳ88% of pH 9 activity, respectively (Fig. 5). At substrate concentrations of 2 mM, ALDH12 was most active with benzaldehyde and NAD ϩ (Table III). NADP ϩ provided lesser but substantial activity. Under these conditions, ALDH12 catalyzed the highest reaction rate with decanal among the straight-chain, aliphatic monofunctional aldehydes tested. As chain length decreased, so did activity. The addition of a second functional group to aliphatic monofunctional aldehydes had an inconsistent effect on activity. ALDH12 metabolized the four-carbon bifunctional succinic semialdehyde at a greater rate than longer-chain aliphatic monofunctional aldehydes (e.g., hexanal and octanal) but metabolized the five-carbon bifunctional aldehyde glutaraldehyde at a lower rate than shorter-chain aliphatic monofunctional aldehydes (e.g., octanal and propanal). No activity was detected with eight additional xenobiotic or naturally occurring aldehydes. Substrates were not converted into detectable products by supernatants of mock-transfected cells, with three exceptions: hexanal, decanal, and glutaraldehyde

Relative mRNA expression in human tissues of ALDH12
The numbers in parentheses denote expression relative to liver.    had activities in mock supernatants that were 8.5%, 74%, and 23%, respectively, of their total activities in pcDNA3/ALDH12transfected cells.
Identification of an Alternative Transcript-We identified an alternative ALDH12 transcript lacking nucleotides 910 FIG. 4. mRNA expression of ALDH12 in human tissues. Data were obtained with a 32 P-labeled ALDH12 probe and a commercial Dot-blot with RNA samples predominantly from human tissues. Table  II shows the identities of individual Dot-blots and relative signal intensities. Assays were done at ambient temperature for 5 min with 33 g of protein and 1 mM NAD ϩ , except for assays for 9-cis-retinal, which were done at 37°C for 10 min with 4 g of protein and 2 mM NAD ϩ . Data represent the means Ϯ S.D. of triplicates from a representative experiment: top, succinic semialdehyde (q) and acetaldehyde (E); bottom, 9-cis-retinal (q) and benzaldehyde (E). through 1071 during cloning. The missing fragment consists of a single exon, indicated by the partial genomic structure of AL021939. The shorter transcript would encode a protein with 54 fewer amino acids, lacking 2 of the 23 invariant residues: Gly 284 and Cys 287 . The latter serves as the catalytically active nucleophile (13). Indeed, the shorter form had no activity (data not shown). Northern blot analysis did not reveal two transcripts, probably because they were not separated well by gel electrophoresis.
Concluding Summary-This report identifies and characterizes a new human ALDH, ALDH12, and demonstrates expression of a rat ortholog. Evidence consistent with the possibility that ALDH12 might function in a pathway of 9-cis-retinal metabolism includes the ϳ40-fold higher rate of catalysis with 9-cis-versus all-trans-retinal and intense expression in liver and kidney, two organs that reportedly have higher concentrations of 9-cis-retinol than other tissues (5,26). Experiments with cells co-transfected with the microsomal 9-cis-retinol/androgen dehydrogenase CRAD1 and the cytosolic ALDH12 show that ALDH12 accesses 9-cis-retinal generated in microsomes from 9-cis-retinol and can therefore function in a pathway of 9cRA biosynthesis in vivo. These experiments also suggest that the expression level of ALDH12, rather than CRAD1, determined the rate of 9cRA formation.
Two other ALDHs convert 9-cis-retinal into 9cRA, rat RALDH1 and RALDH2 (11, 27, 28). RALDH1 has equivalent efficiency (V m /K 0.5 ) for all-trans-and 9-cis-retinal, whereas RALDH2 shows ϳ4-fold greater efficiency for all-trans-versus 9-cis-retinal. ALDH12, therefore, is the first example of an ALDH with greater efficiency for 9-cis-retinal. Comparative efficiencies of RALDH1, RALDH2, and ALDH12 for 9-cis-retinal can be assessed from published data (17,23,27). The K 0.5 values for RALDH1 purified from rat liver and purified recombinant RALDH2 with 9-cis-retinal are ϳ5 and 0.5 M, respectively (11,27). The V m values of RALDH1 and RALDH2, respectively, with 9-cis-retinal are ϳ2.7-fold and ϳ0.2-fold of V m values with all-trans-retinal. The highest V m values of RALDH1 and RALDH2 with all-trans-retinal, obtained with purified recombinant proteins under similar conditions, were 52 and 105 nmol/min/mg protein, respectively (17,23). These data indicate efficiencies of ϳ27 and 38, respectively, for RALDH1 and RALDH2 with 9-cis-retinal, compared with 0.8 ((0.065 ϫ 40 nmol/min/mg)/3.15 M) for ALDH12. The lack of ALDH12 purity also must be considered because activity was measured in the 800 ϫ g supernatant of transfected cells. Fifty-fold represents a conservative estimate for a purification factor, placing ALDH12 near or greater than RALDH1 and RALDH2 in efficiency for catalyzing 9-cis-retinal metabolism.
ALDH12 seems too restricted in substrate recognition to serve as a general xenobiotic clearing enzyme. That is, despite its seemingly disparate substrate recognition (Table III), ALDH12 showed remarkable specificity in some respects. High efficiency for benzaldehyde was eliminated by adding substituents to the phenyl ring-note the lack of detectable activity with coniferyl aldehyde and trans-4-stillbenecarboxyaldehyde-or by separating the phenyl ring from the carboxyaldehyde (no detectable activity with trans-cinnamaldehyde). Despite activity with all-trans-retinal, albeit low, ALDH12 had no detectable activity with citral, a simplified analog of retinal. Appreciable activity with short-chain aliphatic aldehydes was eliminated with the addition of functional groups (pyruvaldehdye, betaine, and malonaldehyde), except for glutaraldehyde. Taken together, these data suggest an active site structured or adaptable to specific classes of substrate(s), with recognition of short-and medium-chain aliphatic aldehydes possible because of their size and flexibility, not necessarily because they are preferred substrates. The situation might be similar to that of RALDH2, which has a protruding "arm" that wraps around all-trans-retinal to form an organized active site (29,30); yet RALDH2 also recognizes acetaldehyde and medium-chain acylaldehydes, resulting from its initially unorganized binding pocket (23).
A function for the short transcript that translates an enzymatically inactive ALDH12 remains uncertain. It could serve as a control element for ALDH12 action (substrate sequestration?), perhaps produced under certain conditions, or it might function as a structural protein. Uses of ALDHs as crystallin and an androgen-binding protein indicate that such a transcript could translate a protein with a function unrelated to aldehyde metabolism (14,15,31).

TABLE IV
Kinetic constants of ALDH12 with select substrates Data were obtained at ambient temperature except for 9-cis-retinal, which was assayed at 37°C. K m and V m values represent the averages of two independent determinations obtained with six to eight substrate concentrations in duplicate or triplicate. The numbers in parentheses are the independent values (Ϯ SEM for the K m values) determined by nonlinear regression analysis. Because data were generated from different transfections, the V m /K m ratios were normalized, setting the average V m of benzaldehyde in each transfection as 100.