Cloning of a cDNA encoding an aldehyde dehydrogenase and its expression in Escherichia coli. Recognition of retinal as substrate.

The biosynthesis of the hormone retinoic acid from retinol (vitamin A) involves two sequential steps, catalyzed by retinol dehydrogenases and retinal dehydrogenases, respectively. This report describes the cloning of a cDNA encoding a heretofore unknown aldehyde dehydrogenase from a rat testis library and its expression in Escherichia coli. This enzyme has been designated retinal dehydrogenase, type II, RalDH(II). The deduced amino acid sequence of RalDH(II) had the highest identity with mammalian aldehyde dehydrogenases that feature low Km values (μM) for retinal: human ALDH1 (72.2%), rat retinal dehydrogenase, type I (71.5%), bovine retina (72.7%), and mouse AHD-2 (71.5%). RalDH(II) expressed in E. coli recognizes as substrates free retinal, with a Km of ∼0.7 μM, and cellular retinol-binding protein-bound retinal, with a Km of ∼0.2 μM. RalDH(II) also can utilize as substrate retinal generated in situ by microsomal retinol dehydrogenases, from the physiologically most abundant substrate: retinol bound to cellular retinol-binding protein. Rat testis expresses RalDH(II) mRNA most abundantly, followed by (relative to testis): lung (6.7%), brain (6.3%), heart (5.2%), liver (4.4%), and kidney (2.7%). RalDH(II) does not recognize citral, benzaldehyde, acetaldehyde, and propanal efficiently as substrates, but does metabolize octanal and decanal efficiently. These data support a function for RalDH(II) in the pathway of retinoic acid biogenesis.

The hormone RA 1 induces a variety of biological responses by modulating gene expression during development and postnatally, to control differentiation or entry into apoptosis of diverse cell types in numerous organs (1)(2)(3). Vertebrates require RA for normal hematopoiesis, reproduction, bone remodelling, and sustaining epithelia (4,5). Yet, excessive RA causes toxicity, both in the embryo and postnatally (4,6). Presumably, a combination of mechanisms closely controls RA concentrations in vivo including regulation of biosynthesis and catabolism.
RA biosynthesis from retinol (vitamin A) entails two sequential reactions with retinal as an intermediate (7,8). At least three microsomal RoDH isozymes, members of the short-chain dehydrogenase/reductase gene family, catalyze the first and rate-limiting step that generates retinal from the substrate holo-CRBP (9,10). The deduced structures of RoDH isozymes suggest globular proteins with only one transmembrane helix at the N terminus, bounded by four hydrophilic residues, consistent with an enzyme anchored to the endoplasmic reticulum, but exposed to the cytoplasm (11)(12)(13). Such topology would permit access to RoDHs by the cytosolic proteins CRBP and RalDHs. Liver expresses the mRNA of all three known RoDH isozymes, whereas extrahepatic tissues express RoDH(I) and RoDH(II) mRNAs in quantitatively differing patterns. Thus, RoDHs are distributed throughout multiple tissues, consistent with the widespread ability of tissues to synthesize RA (14), and each of the three RoDH isozymes has a characteristic tissue expression pattern, as do the RA receptors (15,16).
Fractionation of rat tissue cytosol by anion-exchange chromatography revealed at least four RalDH isozymes, with the three liver isozymes having K 0.5 values for free retinal ϳ1 M (17). The quantitatively major isozyme in liver, kidney, and testis, RalDH(I), formerly designated P1, has been purified from rat liver (17). In addition to free retinal, RalDH(I) recognizes retinal bound to CRBP as substrate, suggesting that CRBP could facilitate relocation of the retinal product from RoDHs to RalDH(I). Attempts to clone the cDNA encoding RalDH(I) from a rat liver library were complicated by the identification of much more abundant aldehyde dehydrogenase cDNA clones, such as rat phenobarbital-inducible aldehyde dehydrogenase. Therefore, our attention turned to a rat testis cDNA library in an effort to avoid abundant hepatic aldehyde dehydrogenases. During screening for RalDH(I), we identified a cDNA clone, distinct from that of RalDH(I), which encodes a heretofore unknown aldehyde dehydrogenase.
This report describes the cloning of a cDNA encoding this new aldehyde dehydrogenase, the distribution of its mRNA expression in tissues, and the enzymatic characteristics of the enzyme expressed in Escherichia coli. This isozyme has been designated RalDH(II) because it recognizes as substrate, with K m values Ͻ 1 M, unbound retinal and retinal in the presence of CRBP, and also retinal generated in situ by microsomal RoDH from holo-CRBP.

MATERIALS AND METHODS
A cDNA Encoding RalDH(II)-Rat testis RNA served as the template for reverse transcriptase-PCR with the two primers, 5Ј-CTTGG(G/ A)GG(G/A)AA(G/A)AGC-3Ј and 5Ј-AC(T/C)GG(T/C)CCAAA(T/A/G)AT-CTCCTC-3Ј. RNA (3.5 g) was allowed to react with 3 l of random * This work was supported by National Institutes of Health Grant AG13566. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBank TM /EBI Data Bank with accession number(s) U60063 (RalDH(II)).
‡ To whom correspondence should be addressed: 140 Farber Hall, School of Medicine and Biomedical Sciences, SUNY-Buffalo, Buffalo, NY 14214. Tel.: 716-829-2032; Fax: 716-829-2661. 1 The abbreviations used are: RA, all-trans-retinoic acid; CRBP, cellular retinol-binding protein, type I; IPTG, isopropyl-1-thio-␤-D-thiogalactoside; PCR, polymerase chain reaction; RalDH(II), retinal dehydrogenase type II; RalDH(II)/rL, recombinant RalDH(II) expressed from the first ATG with an N terminus histidine tag; RalDH(II)/rS, recombinant RalDH(II) expressed from the second ATG with an N terminus histidine tag; RoDH, retinol dehydrogenase; PCR, polymerase chain reaction; PAGE, polyacrylamide gel electrophoresis; kb, kilobase(s); nt, nucleotide(s). hexamer (Invitrogen), 10 units of rRNase ribonuclease inhibitor and 45 units of avian myeloblastosis virus reverse transcriptase (Promega) in a total volume of 30 l for 2 h at 39°C. The reaction mixture was diluted 1/20 with water, and 10 l of the diluted solution were added to a PCR reaction mixture consisting of (final concentrations): 1 M each primer, 1.5 mM MgCl 2 , 0.2 mM each dNTP, and 2 units of Taq DNA polymerase (Promega) in 50 l of 10 mM Tris-HCl, pH 9.0, 50 mM KCl, 0.1% Triton X-100. PCR used 35 cycles of 2 min at 94°C, 2 min at 55°C, 3 min at 72°C, and 10 min at 72°C after the final cycle. The 0.4-kb products were gel-purified and cloned into pBluescript II SK(ϩ). One clone had a 407-base pair insert with ϳ70% nucleotide sequence identity with the known members of the aldehyde dehydrogenase superfamily. This insert was amplified by PCR with the two original primers, radiolabeled with [␣-32 P]dCTP with the Megaprime DNA Labeling Kit (Amersham), and used to screen a gt11 rat testis cDNA library (Clontech) with a final wash at 45°C with 0.1% SSC and 0.1% SDS. After three rounds of screening, seven of the original 1.6 ϫ 10 5 plaques were positive. The longest insert, ϳ2.2 kb, was excised with EcoRI and cloned into pBluescript II SK(ϩ) to produce pBC/RalDH(II). A series of unidirectional deletions of pBC/RalDH(II) were made and sequenced in both directions by dideoxy chain termination using the fmol TM DNA Sequencing System (Promega).
Northern Blots-The RNA blot was prepared with adult male rat tissue poly(A) ϩ as described previously (11). The probe for RalDH(II), a 796-nucleotide BssHII/EcoRI fragment from a unidirectional deletion mutant that contained nt 1474 through 2240 of pBC/RalDH(II), was labeled with [␣-32 P]dCTP by the method described above. Blot hybridization and washing were done under high stringency conditions as reported (11). Autoradiography was done with Kodak XAR film and one intensifying screen at Ϫ70°C for 24 h.
RNase Protection Assays-The RalDH(II) probe was obtained by in vitro transcription of unidirectional deletion mutants with cDNA from RalDH(II). A plasmid containing nt 1 through 1853 of pBC/RalDH(II) was linearized with ScaI, which cut at nucleotide 1512 to produce a 341-nucleotide fragment consisting of nt 1513 through 1853. This fragment included 291 nt of 3Ј-untranslated cDNA. The 341-nucleotide fragment was used as template for transcription of 32 P-labeled antisense riboprobes by T3 RNA polymerase (Ambion) for 1 h at 37°C in 10 mM dithiothreitol, 0.5 mM each ATP, CTP, and GTP, 5 M UTP, and 50 mCi of [␣-32 P]UTP (800 Ci/mmol). The DNA template was removed through DNase I digestion. The 125-nucleotide-long antisense probe for rat ␤-actin mRNA was transcribed in vitro from p-TRI-␤-actin-125-rat (Ambion) under the same conditions. Probes were purified with a 5% PAGE, 8 M urea gel. The ribonuclease protection assay was done with the RPA II TM Ribonuclease Protection Kit (Ambion). Total RNA (60 g in each sample) was extracted with guanidinium thiocyanate/phenol/ chloroform from adult male Sprague-Dawley rats and was co-precipitated with RNA probes (8 ϫ 10 4 cpm) by 0.5 M ammonium acetate. Each precipitate was resuspended in 20 l of hybridization buffer (80% deionized formamide, 100 mM sodium citrate, 300 mM sodium acetate, 1 mM EDTA, pH 6.4) and then was incubated at 45°C for 18 h. The same amount of probe was hybridized with 10 g of yeast RNA as a control. After digestion with RNase A (1 unit/ml) and RNase T1 (40 units/ml) for 30 min at 37°C, the protected fragments were resolved on 5% polyacrylamide, 8 M urea gels and visualized by autoradiography at Ϫ70°C overnight with one intensifying screen. Quantitative comparisons were made with an LKB UltroScan XL densitometer.
Expression of RalDH(II) in E. coli-Two single-stranded primers were designed to contain nt sequences 27-44 and 1554 -1572 from pBC/RalDH(II): CCGATCATATGCCCGGCGAGGTGAAG and AGGG-ATCCTGGGCCTCTTAGGAGTT, respectively (underlined nucleotides indicate NdeI and BamHI sites, respectively). These primers were used to amplify the coding region of RalDH(II) by PCR (final conditions): 1 M concentration of each primer, 200 ng of DNA template, 0.2 mM each dNTP, 2.5 units of pfu DNA polymerase (Stratagene), 10 mM KCl, 6 mM (NH 4 ) 2 SO 4 , 2 mM MgCl 2 , 0.1% Triton X-100, and 10 g/ml nuclease-free bovine serum albumin in 0.1 ml of 20 mM Tris-HCl, pH 8.2. PCR was done with 35 cycles of 1 min at 94°C, 1 min at 72°C, 1 min at 52°C, and 10 min at 72°C after the final cycle. The 1.5-kb product was gel-purified and cloned into pBluescript II SK(ϩ) by blunt-end ligation. The insert was excised with NdeI and BamHI, gel-purified, and ligated into pET-14b between the unique NdeI and BamHI sites to yield pET/ RalDH(IIL). The insert of pET/RalDH(IIL) was sequenced. pET/ RalDH(IIL) expresses an RalDH(II), RalDH(II)/rL, with an additional 20 amino acids on its N terminus containing a histidine tag: MGSSH-HHHHHSSGLVPRGSH. E. coli strain BL21(DE3) was transformed with pET/RalDH(IIL) by the CaCl 2 method. Bacteria were grown at 37°C until the absorbance at 595 nm reached 0.6. The temperature was lowered to 18°C, 0.4 mM IPTG was added, and incubation was continued for 24 h.
Isolation of RalDH(II)/r from E. coli-Bacteria were pelleted (3500 ϫ g for 10 min), resuspended in 0.25 mM sucrose, 1 mM EDTA, 10 mM Tris-HCl, pH 7.5, and were then disrupted with a French press. The lysate was treated with 200 units/ml DNase I for 15 min at ambient temperature in the presence of 2 mM phenylmethylsulfonyl fluoride, 20 g/ml pepstatin, and 20 g/ml leupeptin, and then was centrifuged at 10,000 ϫ g for 30 min at 4°C. RalDH(II)/r was isolated from the supernatant by nickel affinity chromatography with the His-Bind Buffer Kit (Novagen, Madison, WI). Typically, 2 ml of supernatant (ϳ120 mg of protein) were applied to 1.6 ml of resin. The resin was washed with 16 ml of binding buffer, 9.6 ml of wash buffer and then eluted with 9.6 ml of eluting buffer (1 M imidazole, 0.5 M NaCl, 20 mM Tris-HCl, pH 7.9) as recommended by the manufacturer.
RA Synthesis-Unless stated otherwise, assays were done in duplicate (variation was ϳ10% of the averages) in buffer A (20 mM Hepes, 150 mM KCl, 1 mM EDTA, and 2 mM dithiothreitol, pH 8.5) at 37°C with 400 ng of RalDH(II). Retinal was added in 2 l of dimethyl sulfoxide to a final volume of 0.5 ml. When CRBP-bound retinal was substrate, preincubations of retinal and apo-CRBP were done at 37°C for 20 min to allow CRBP-retinal complex formation. Incubations were initiated by addition of 2 mM cofactor (final concentration) and protein and were done for 10 min. Controls were assays done without protein or without cofactors. Reactions were quenched and RA was quantified by normalphase high performance liquid chromatography as described (18).

RalDH(II) Activity with Aldehydes Other Than
Retinal-Assays were done in 1 ml of buffer A by monitoring for 5 min at 25°C the synthesis of NADH (340 nm). Substrates were added in ethanol (Ͻ10 l), and the reactions were initiated by adding NAD (final concentration, 2 mM). At least six concentrations were used for determining kinetic constants of each substrate, ranging from 0.2-to 20-fold the K m value. Kinetic constants were determined under initial velocity conditions linear with time and protein.
Rat Tissues-Microsomes and RNA were prepared as described previously from the tissues of male rats (ϳ250 g) fed a chow diet (11).
Preparation of CRBP-CRBP was generated in E. coli with the vector pMONCRBP (19), as described (9). The concentration of functional apo-CRBP was determined by saturating an aliquot with retinol, separating free and bound retinol by size-exclusion chromatography, and determining the A 350 /A 280 ratio. Holo-CRBP denotes CRBP saturated with retinol, whereas CRBP-retinal denotes CRBP saturated with retinal.

RESULTS AND DISCUSSION
Nucleotide and Deduced Amino Acid Sequences of RalDH(II)-We selected testis RNA and cDNA, respectively, as templates for synthesizing a probe and for library screening to avoid abundant hepatic aldehyde dehydrogenases. Primers were designed for reverse transcriptase-PCR from two sequences of amino acids in rat phenobarbital-inducible aldehyde dehydrogenase (residues 267-273 and 399 -404), which are highly conserved regions of the mammalian members of the aldehyde dehydrogenase superfamily (20). Reverse transcriptase-PCR with testis RNA produced a 0.4-kb probe (the anticipated size), with a nucleotide sequence only ϳ70% identical with known aldehyde dehydrogenases (Fig. 1, nt 864  through 1270). Library screening with this probe identified an ϳ2.2-kb clone, which was subcloned to yield pBC/RalDH(II).
The single open reading frame in pBC/RalDH(II) has two possible initiator codons; the first starting with adenosine 27 and the second starting with adenosine 63 (Fig. 1). The first one represents the most likely translation initiation site in vivo, according to the Kozak rules (21), and predicts a protein of 511 amino acids. This protein has all 23 of the strictly conserved residues of the aldehyde dehydrogenase superfamily, expressed in phylogenetically diverse organisms (20). It also has all 66 residues, either strictly conserved or "invariantly simi- The protein encoded by pBC/RalDH(II) has the highest amino acid identities and similarities, in parentheses respectively, with chicken (73, 86%), bovine retina (73, 87%), human ALDH1 (72, 87%), rat RalDH(I) (72, 85%), mouse AHD-2 (72, 85%), and rat phenobarbital-inducible (71, 85%) aldehyde dehydrogenases (22)(23)(24)(25)(26)(27). Thus, the overall amino acid sequence similarity, as well as the conservation of specific amino acid residues, denote a heretofore unrecognized aldehyde dehydrogenase. The high amino acid similarity with rat RalDH(I), mouse AHD-2, bovine retina aldehyde dehydrogenase, and human ALDH-1 is especially intriguing, because these enzymes catalyze the conversion of retinal into RA with K m values Ͻ1 M, except for bovine retina aldehyde dehydrogenase, which has a K m value ϳ9 M (17,(27)(28)(29)(30). These enzymes, however, have not been tested for recognition of retinal in the presence of CRBP or retinal generated in situ from holo-CRBP.
RalDH(II) from E. coli-RalDH(II) was expressed in E. coli to study its enzymatic characteristics. Judging from SDS-PAGE analysis, ϳ5% of E. coli protein was RALDH(II)/rL (Fig.  3). Similar results were obtained with the shorter recombinant product, expressed from the second putative initiator (data not shown). RalDH(II)/rL was easily purified via nickel affinity chromatography to a single band on SDS-PAGE. About 1 mg of purified RalDH(II)/rL was obtained from a 100-ml incubation.
RalDH(II)/rL and Retinal-RalDH(II)/rL catalyzed the conversion of 2 M retinal into RA at a rate linear for at least 12.5 min and up to 0.4 g of purified protein and functioned optimally ϳpH 8.5 (not shown). RalDH(II)/rL recognized NAD with a K m of 70 Ϯ 0.6 M and NADP with a K m of 400 Ϯ 7 M (Ϯ S.E., Enzfitter (31)). The V max in the presence of NAD was 5-fold higher than that with NADP. The apparent K m of RalDH(II)/rL for free retinal (i.e. unbound with CRBP) was 0.7 Ϯ 0.3 M, and the V max was 105 Ϯ 4 nmol/min/mg of protein (Ϯ S.D., n ϭ 3) (Fig. 4, top panel). The kinetic constants were re-evaluated in the presence of a 2 molar excess of apo-CRBP at each retinal concentration, which generated an apparent K m of 0.2 Ϯ 0.06 M and a V max of 62 Ϯ 15 nmol/min/mg of protein (n ϭ 3) (Fig.  4, bottom panel). CRBP binds retinal with a K d between 50 and 100 nM (19,32). In the presence of a 2 molar excess of CRBP, the concentration of unbound retinal would range from 0.042 M, at a total retinal concentration of 0.1 M, to 0.1 M at a total retinal concentration of 6 M (calculating from a K d of 100 nM), i.e. it would remain practically constant. In contrast, the CRBP-retinal concentration would range from 0.58 M to 5.9 M. The Michaelis-Menten relationship must therefore occur between CRBP-retinal and the rate of RA production, suggesting that RalDH(II)/rL recognizes retinal in the presence of CRBP as substrate. Because retinal has very poor solubility in aqueous media and there occurs an excess of CRBP relative to retinal in vivo, retinal in cells would likely occur bound to CRBP, in equilibrium with membranes and/or proteins. The ability of RalDH(II)/rL to catalyze the synthesis of RA in the presence of CRBP-retinal shows that it functions under conditions that more closely model physiological conditions than retinal dispersed in the aqueous medium.
Because holo-CRBP is the most abundant form of retinol in vivo, and NADP-dependent microsomal RoDHs are the most active retinol dehydrogenases with respect to holo-CRBP as substrate (33), the ability was determined of RalDH(II)/rL to utilize as substrate retinal generated in situ by microsomal RoDHs from holo-CRBP. Apo-CRBP also was included to ensure complete binding of retinol (K d ϭ 0.1-1 nM (32)) and  (5), heart (6), and brain (7). Lane 1 shows marker DNA; lane 8 shows the digested yeast control; lane 9 shows the probe. The anticipated fragments were observed at 341 nt. The arrow on the right indicates the position of the probe. Lane 9 was exposed for 15 h; the other lanes were exposed for 3 days.  Table II. because this combination of holo-CRBP and apo-CRBP most closely approximates conditions in vivo. Generation of RA from the retinal produced in situ from holo-CRBP by microsomal RoDHs increased with increasing amounts of RalDH(II)/rL titrated into the incubation mixture (Fig. 5). Maximum RA synthesis from holo-CRBP occurred only in the presence of both microsomes and RalDH(II)/rL along with both cofactors, whereas measurable but much less RA production was observed with a combination of microsomes and RalDH(II)/rL, if one cofactor were omitted (Table I). Consistent with previous results, microsomes alone in the presence of both NADP and NAD produced little RA (17,34). These data show that the NAD-supported RalDH(II)/rL can use retinal generated by the microsomal, NADP-dependent RoDHs to biosynthesize RA.
RalDH(II)/rL Substrate Specificities-RalDH(II)/rL does not recognize citral as substrate and inefficiently catalyzes the dehydrogenations of acetaldehyde, benzaldehyde, and propanal (Table II). These are the prototypical substrates used to assay and classify the mammalian aldehyde dehydrogenases. Thus, RalDH(II) differs from many other members of the superfamily not only in primary amino acid sequence, but also in substrate specificity. Medium-chain aldehydes, however, were metabolized efficiently by RalDH(II). Of the eight aldehyde dehydrogenase substrates assayed, octanal and decanal had the most favorable V max /K m values, whereas retinal had the lowest K m . Although the V max /K m for retinal was 3-to 4-fold lower than the values for octanal and decanal, the apparent K m for retinal was an order of magnitude lower. Thus, RalDH(II) can catalyze the dehydrogenation of retinal at much lower concentrations than the other substrates assayed. It should not surprise that medium-chain aldehydes would be accommodated by an active site that recognizes retinal. Octanal and decanal are similar in length to the side chain of retinal and enjoy greater flexibility. Perhaps their greater flexibility and more simple structures (lack of double bonds, lack of methyl groups) and the absence of a conjugated carbonyl account for their greater rates of dehydrogenation, even though they are apparently accommodated less efficiently by the active site.
Characteristics of RalDH(II)/rS-RalDH(II)/rS, the shorter protein expressed in E. coli from the second possible initiator, had enzymatic characteristics similar to those of RalDH(II)/rL. It was NAD-dependent, had a pH optimum ϳ8.5, recognized free retinal with a K m ϳ1 M, and was inefficient in metabolizing benzaldehyde, acetaldehyde, propanal, and hexanal, but metabolized octanal and decanal, efficiently with V max /K m ratios of 175 and 148, respectively. The shorter N terminus, therefore, does not affect activity markedly. This affords the possibility that alternative forms of RalDH(II) are translated in different cell types or under different conditions.
Concluding Summary-The designation of the isozyme reported here as RalDH(II) seems justified by several criteria. Firstly, the amino acid sequence of RalDH(II) has the greatest identity and similarity with RalDH(I) and other aldehyde dehydrogenases that catalyze the conversion of retinal into RA. Secondly, the enzyme recognizes "free" retinal with a low K m and also recognizes retinal in the presence of CRBP with a low K m . Thirdly, RalDH(II) can use retinal generated in situ from holo-CRBP/apo-CRBP by microsomal RoDHs as substrate for RA synthesis. These conditions approximate those that occur in RA-producing tissues. Fourthly, although RalDH recognizes aldehydes other than retinal, their structures are simpler than that of retinal. Their ability to enter into an active site that accommodates retinal would not be unusual, whereas it would be more unusual for the more rigid, branched, and polyunsaturated molecule, retinal, to adapt to a nonspecific enzyme with a low K m and a physiologically effective V max .