Xenopus Cytosolic Thyroid Hormone-binding Protein (xCTBP) Is Aldehyde Dehydrogenase Catalyzing the Formation of Retinoic Acid*

Amino acid sequencing of an internal peptide fragment derived from purified Xenopus cytosolic thyroid hormone-binding protein (xCTBP) demonstrates high similarity to the corresponding sequence of mammalian aldehyde dehydrogenase 1 (ALDH1) (Yamauchi, K., and Tata, J. R. (1994) Eur. J. Biochem. 225, 1105–1112). Here we show that xCTBP was co-purified with ALDH and 3,3′,5-triiodo-l-thyronine (T3) binding activities. By photoaffinity labeling with [125I]T3, a T3-binding site in the xCTBP was estimated to reside in amino acid residues 93–114, which is distinct from the active site of the enzyme but present in the NAD+ binding domain. The amino acid sequences deduced from the two isolated xALDH1 cDNAs (xALDH1-I and xALDH1-II) were 94.6% identical to each other and very similar to those of mammalian ALDH1 enzymes. The two recombinant xALDH1 proteins exhibit both T3 binding activity and ALDH activity converting retinal to retinoic acid (RA), which are similar to those of xCTBP. The mRNAs were present abundantly in kidney and intestine of adult femaleXenopus. Interestingly, their T3 binding activities were inhibited by NAD+ and NADH but not by NADP+ and NADPH, whereas NAD+ was required for their ALDH activities. Our results demonstrate that xCTBP is identical to ALDH1 and suggest that this protein might modulate RA synthesis and intracellular level of free T3.

A major characteristic of 3,3Ј,5-triiodo-L-thyronine (T 3 ), 1 the active form of thyroid hormone at the cellular level, is the multiplicity of physiological processes. These include such diverse functions as postembryonic and fetal development and postnatal growth in mammals, amphibian metamorphosis, maturation of central nervous system, energy metabolism in homeotherms, and environmental adaptation in poikilotherms (1)(2)(3)(4). It is now generally accepted that, at the molecular level, most of these actions of thyroid hormone are initiated by the interaction between thyroid hormone receptor and T 3 . Thyroid hormone receptor is a member of a multigene family of nuclear receptors that act as transcription factors in combination with transcriptional co-activators and co-repressors and chromatinmodifying factors (5)(6)(7)(8)(9). It is, however, not clear as to how T 3 enters the cell and reaches the nucleus and what determines the dynamics of cytoplasm-to-nucleus transfer of the hormone.
A key component of this intracellular process is most likely to be the cytosolic thyroid hormone-binding protein (CTBP). Recently, CTBPs have been detected in mammalian and amphibian cells (10 -15), an interesting feature of which is that these exhibit different biochemical properties. We have earlier described a CTBP in adult Xenopus liver (xCTBP) (11) which is a 59-kDa protein with a higher affinity for T 3 than L-thyroxine (T 4 ), T 3 binding being neither Ca 2ϩ -nor NADPH-dependent, as is the case for Rana and rat CTBPs (12)(13)(14)(15).
The physiological actions of retinoic acid (RA) and other retinoids are also considered to be exerted through nuclear retinoic acid receptors (7,16). There is also good evidence that a large fraction of RA and retinoids is present in the cell bound to cytoplasmic proteins, identified as cytosolic retinoic acid (CRABP) and retinol-(CRBP) binding proteins (16,17). It has been suggested that these binding proteins may not only determine the intracellular concentration of free ligands but may also act as their transporters into the nucleus. A similar suggestion, based on indirect evidence, has also been made for a mammalian CTBP that has been identified as a monomer of pyruvate kinase subtype M2 (18). Although there are many similarities between CTBP, CRABP, and CRBP, on the one hand, and retinoic acid receptors and thyroid hormone receptors, on the other, a major difference is the multiple types of CTBPs, unlike CRABP and CRBP (17). We have previously reported three types of xCTBPs (19), each with a distinct pattern of expression, which raises the possibility of a tissuespecific role for CTBPs. It therefore became important to characterize CTBPs in greater detail.
A unique feature of xCTBP found in adult liver is that it has a region similar to those of mammalian class 1 aldehyde dehydrogenase (ALDH1) (aldehyde:NAD ϩ oxidoreductase, EC 1.2.1.3) (11), which is one of the enzymes catalyzing the oxidation of various aliphatic and aromatic aldehydes to the corresponding acids. An important and rather specific activity of ALDH1 is to act as an enzyme catalyzing the synthesis of RA * 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) AB016717 and AB016718.
Here we report studies carried out to unambiguously identify xCTBP as ALDH1. Toward this aim, we have cloned two cDNAs encoding ALDH1 from a Xenopus hepatic cDNA library with human ALDH1 cDNA as the probe (25), determined their nucleotide sequences, and examined both the enzyme and T 3 binding activities using the recombinant proteins expressed in Escherichia coli. Deduced amino acid sequences and studies of their activities clearly showed that the two translated products are ALDH1 with T 3 binding activity. The corresponding mRNAs were expressed predominantly in the kidney and intestine in adult Xenopus. T 3 binding and ALDH activities of these proteins seem to be expressed alternatively depending on NAD ϩ binding.

EXPERIMENTAL PROCEDURES
General-Enzymes and chemicals were obtained from the following sources: restriction enzymes from Life Technologies, Inc., New England Biolabs, Boehringer Mannheim, Takara Shuzo, and Toyobo; a multiprime DNA labeling kit was from Amersham Pharmacia Biotech; Moloney murine leukemia virus-reverse transcriptase, exonucleases III and VII, guanidinium isothiocyanate, and cesium chloride were from Life Technologies, Inc.; Taq DNA polymerase was from Biotech International; a DNA sequencing kit was from Toyobo; ethyl-3-aminobenzoate methanesulfonic acid salt was from Aldrich; [␣-32 P]dCTP (110 TBq/ mmol) was from ICN Biomedicals Inc.; [ 125 I]T 3 (122 MBq/g; carrierfree) was from NEN Life Science Products; and unlabeled T 3 , D-T 3 , T 4 , 3,3Ј,5-triiodo-L-thyroacetic acid and all-trans-retinal were from Sigma. Acetaldehyde was obtained from Merck and AG 1-X8 resin from Bio-Rad. Other reagents of molecular biology grade were purchased from Wako Pure Chemicals and ICN Biomedicals. Adult Xenopus hepatic cDNA library in ZAP II vector was kindly provided by Dr. A. Kawahara, Hiroshima University, Japan. Human ALDH1 cDNA was a gift from Dr. A. Yoshida, Beckman Research Institute of the City of Hope, CA. Protein was determined by the dye binding method with bovine ␥-globulin as a standard (26).
Preparation of Cytosol-Adult female Xenopus laevis were anesthetized by immersing in 0.2% ethyl-3-aminobenzoate methanesulfonic acid salt. Animals were first perfused with ice-cold Barth-X amphibian Ringer (27) containing 0.2 mg/ml heparin and then with ice-cold Barth-X amphibian Ringer alone. Dissected tissue was minced with scissors in Barth-X amphibian Ringer, followed by several washings in the same solution. The minced tissue was homogenized in 4.5 volumes of 0.25 M sucrose, 10 mM Tris-HCl, 1 mM EDTA, 1 mM MgCl 2 , 1 mM dithiothreitol (DTT), 1 mM benzamidine hydrochloride, and 1 mM phenylmethylsulfonyl fluoride, pH 7.5, as described previously (11). After successive differential centrifugations at 3,000 ϫ g for 10 min, 12,000 ϫ g for 20 min, and 100,500 ϫ g for 60 min, a clear supernatant was obtained and stored in 10% glycerol at Ϫ85°C until its use as a cytosol.
Purification of Native xCTBP from Liver Cytosol-xCTBP was purified as described with some modifications (11). All the following procedures were carried out at 4°C, unless otherwise noted. In brief, solid ammonium sulfate at a final concentration of 1.4 M was added to 30 ml of cytosol, and the precipitate was removed by centrifugation at 12,000 ϫ g for 15 min. More solid ammonium sulfate, at a final concentration of 2.5 M, was added to the supernatant. The precipitate obtained was collected by centrifugation in the same way and dissolved in 2-3 ml of 20 mM sodium phosphate, 0.5 mM DTT, pH 7.5. It was applied to a Cellulofine GCL-1000 column (3.0 ϫ 76.5 cm, Seikagaku Co.), which had been equilibrated with the same buffer, and eluted at a flow rate of 0.3 ml/min. The eluates with T 3 binding activity were applied to CM cation-exchange and DEAE anion-exchange Sepharose columns (5 ml packed Fast Flow columns, Amersham Pharmacia Biotech), which had been tandemly connected and equilibrated with 20 mM sodium phosphate, 0.5 mM DTT, pH 7.5. Flow-through fractions were collected, and the pH of the combined eluate was adjusted to 5.3 with 1 M acetic acid. The eluate was subjected to chromatography on Mono S cation-exchange column (5 ϫ 50 mm; Amersham Pharmacia Biotech), which had been equilibrated with 20 mM sodium acetate, 0.5 mM DTT, pH 5.3, and the proteins were eluted with a 60-min gradient of this buffer to 20 mM sodium phosphate, 0.5 mM DTT, pH 7.5, at a flow rate of 0.5 ml/min in a fast protein liquid chromatography apparatus (Amersham Pharmacia Biotech). The proteins in the peak fractions exhibiting T 3 binding activity were further fractionated by chromatography on a hydroxyapatite column (model Taps, Tonen) with a 20-min gradient of 20 mM sodium phosphate, 0.5 mM DTT, pH 7.5, to 300 mM sodium phosphate, 0.5 mM DTT, pH 7.5, at a flow rate of 0.5 ml/min in a fast protein liquid chromatography apparatus. Finally, the proteins with T 3 binding activity were further isolated by chromatography on a hydrophobic interaction phenyl 5PW column (7.5 ϫ 75 mm, Tosoh, Tokyo, Japan), equilibrated with 0.5 M ammonium sulfate in 5 mM sodium phosphate and 0.5 mM DTT, pH 7.0. The proteins were resolved with a 20-min linear gradient of the buffer to 60% ethylene glycol in 5 mM sodium phosphate, 0.5 mM DTT, pH 7.0, at a flow rate of 0.5 ml/min, using a high performance liquid chromatography (HPLC) apparatus (Jusco 851-GI system, Japan Spectroscopic Co.).
Screening of cDNA Library and Sequence Analysis-An hepatic cDNA library was screened with 32 P-radiolabeled human ALDH1 cDNA (25). The entire sequences of the two cDNAs were determined for both strands by the method of Sanger et al. (28). The computer program, Clustal W (1.60) in DNA Data Bank of Japan was used on multiple sequence alignment and the construction of unrooted tree by the Neighbor-joining method (29).
Northern Blot Analysis-Total RNA was prepared from them by the acid guanidinium isothiocyanate/phenol/chloroform method (30). Total RNA (15 g) was electrophoresed on a 1% agarose gel containing 2.6 M formaldehyde, and the separated RNAs were transferred onto a nylon filter. Hybridization and washing were performed under high stringency conditions as described (31). The probe for Xenopus ALDH1 (xALDH1) cDNAs, a 0.3-kbp fragment that contained nt 1708 -2014 of xALDH1-I cDNA, was amplified by polymerase chain reaction (PCR) and labeled with [␣-32 P]dCTP. To check the amount of total RNA loaded, 28 S ribosomal RNA hybridization signals on the same filter were estimated as a loading control. Xenopus 28 S ribosomal cDNA was amplified at the nt positions 1-346 by PCR, after the reaction with Moloney murine leukemia virus-reverse transcriptase in the presence of (dT) 12-18 at 37°C for 1 h. Autoradiography was done with Kodak XAR5 film with intensifying screen at Ϫ85°C for 1-7 days.
Expression of Recombinant xALDHs in E. coli-The coding sequences of xALDH1-I and -II cDNAs, with a NdeI site engineered into the start codon and a BglII site downstream from the stop codon, was prepared by PCR, subcloned into pET15b expression vector (Novagen, Madison, WI), and designated pET15b/xALDH1-I and pET15b/xALDH1-II. These plasmids were transformed into E. coli BL21. Bacteria were grown at 37°C until the absorbance at 600 nm reached 0.5. The temperature was lowered to 24°C, 0.2 mM isopropyl-1-thio-␤-D-galactopyranoside was added, and incubation was continued for 24 h unless otherwise noted.
Purification of Recombinant xALDHs from E. coli-Bacteria were pelleted by centrifuging (1200 ϫ g), resuspending in 0.3 M NaCl, 50 mM Tris, pH 8.0, 10 mM imidazole, 1 mg/ml lysozyme, 1 mM benzamidine hydrochloride, 1 mM phenylmethylsulfonyl fluoride, and 50 mM 2-mercaptoethanol, and then keeping on ice for 30 min. The cells were disrupted by sonication for 10 s three times on ice at the range 5 (UR200P type, Tomy) and subsequently by three cycles of freezing and thawing, and the lysate was centrifuged at 105,000 ϫ g for 40 min at 4°C. Recombinant proteins with a histidine tag were isolated from the other proteins in the supernatant by a nickel affinity chromatography (1 ml of the resin) (ProBond Resin, Invitrogen, CA), with 0.3 M NaCl, 50 mM Tris, pH 8.0, 250 mM imidazole, after washing the column with six times column volume of 0.3 M NaCl, 50 mM Tris, pH 8.0, 80 mM imidazole. The purified proteins were stored in 1 mM EDTA, 1 mM DTT, and 10% glycerol at Ϫ85°C until use.  (11), and these radioactivities were measured in a ␥ counter (Auto Well Gamma System ARC-2000, Aloka, Japan). The amount of [ 125 I]T 3 bound nonspecifically was derived from the radioactivity in the sample incubated with 5 M unlabeled T 3 and subtracted from amount of the total bound T 3 to give the values for specific binding. The values for the dissociation constant (K d ) and maximum binding capacity were calculated from Scatchard plot (32).
Photoaffinity Labeling-Photoaffinity labeling with underivatized [ 125 I]T 3 was carried out as described previously (11). The proteins were resolved by SDS-polyacrylamide gel electrophoresis (PAGE) (33) and visualized using Coomassie Brilliant Blue R-250 staining or silver staining. Phosphorylase b, bovine serum albumin, ovalbumin, and carbonic anhydrase were used as molecular weight standards. The labeled proteins were detected by autoradiography exposed to x-ray XAR5 film (Kodak) at Ϫ85°C for 2-5 days.
ALDH Activity-Assay was performed in duplicate or triplicate (values within 10% of the mean) in 100 l of 50 mM Tris, pH 8.0, 3.3 mM pyrazole, 100 mM KCl, 1 mM DTT, 0.33 mM NAD ϩ , and various concentrations of substrates by monitoring for 1-2 min at 24°C the formation of NADH (⑀ at 340 nm ϭ 6220) with aldehydes other than retinal. The reactions were initiated by adding the enzyme. At least six concentrations were used for determining kinetic constants of acetaldehyde, propionaldehyde, and retinal, ranging 1-32 mM, 0.2-16 mM and 1-32 M, respectively. For RA synthesis, the formation of NADH and retinoic acid (⑀ at 340 nm ϭ 6220 ϩ 39,200 Ϫ 22,800) were monitored under dim light (34). In some cases, the formation of RA was monitored at 340 nm by HPLC (35). There are few differences in the values of kinetic parameters obtained from the two methods. Kinetic constants were determined under initial velocity conditions linear with time and protein.
Western Blot Analysis-Cytosolic proteins from adult liver and two recombinant xALDH1 proteins were separated by electrophoresis on a SDS-10% polyacrylamide gel, transferred onto a nitrocellulose membrane, and immunoblotted for 1 h at room temperature with a primary polyclonal antibody to a peptide of xCTBP, which is identical to amino acid residues 239 -261 in xALDH1-II (see Fig. 2B). Binding was detected by using the chemiluminescence kit (Boehringer Mannheim) according to the manufacturer's directions.
Protein Sequencing-The photoaffinity labeled xCTBP was digested with lysyl endopeptidase, and the peptides were fractionated by HPLC as described previously (36). The peptide with radioactivity was hydrolyzed and analyzed by gas phase sequence analyzer (Applied Biosystem, 470A).

Characteristics of xCTBP from Liver Cytosol-
The elution profile of the final step of the chromatography of xCTBP, which was purified on the basis of [ 125 I]T 3 -binding on a phenyl 5PW column, resembled that described earlier (11). The major protein peak contained a single species of protein with 59 kDa that was specifically photoaffinity labeled with [ 125 I]T 3 and corresponded to the peak of [ 125 I]T 3 -binding activity (Fig. 1). To determine whether or not xCTBP has ALDH activity, each fraction was assayed with acetaldehyde as a substrate. The major protein peak also coincided with the peak of ALDH activity. These results indicate that the 59-kDa protein has both activities. The information about the sequence involved in the [ 125 I]T 3 -binding site was obtained by the identification of the peptide with radioactivity after digestion of the affinity labeled xCTBP with lysyl endopeptidase. Protein sequencing revealed that the peptide (K)LADLVERDRLILSTM-, which corresponded well to 91-106 residues of human ALDH1 enzyme (25) and to 92-107 residues of xALDH1-I and -II enzymes Relative efficiencies of ALDH activity of xCTBP for two aldehyde substrates, acetaldehyde and retinal, are shown in Table I. xCTBP catalyzed the conversion of acetaldehyde into acetic acid or retinal into RA at a linear rate for at least 1-2 min at 1.85 g of the purified protein. The K m value for acetaldehyde (380 Ϯ 20 M) was 220 times higher than that for retinal (1.7 Ϯ 0.5 M), although the V max for acetaldehyde (0.54 Ϯ 0.01 mol/min/mg) was 7 times higher than that for retinal (0.073 Ϯ 0.002 mol/min/mg). xCTBP exhibited allosteric characteristics for retinal with the Hill coefficient of 2.5 Ϯ 0.2 but not for acetaldehyde as the substrate (Fig. 5A).
Cloning and Analysis of xALDH1 cDNAs-We selected adult Xenopus liver cDNA library for cloning ALDH cDNA because T 3 binding activity is higher in liver cytosol than in other tissues of adult Xenopus (36). Human ALDH1 cDNA (25) was used as a screening probe from the fact that amino acid sequence of xCTBP showed high identity to that of human ALDH1 in the two regions selected (see Fig. 2B). Two cDNAs with 2.4 kbp, xALDH1-I and xALDH1-II, which were quite similar to but slightly different from each other, were thus isolated. xALDH1-I was distinguished from xALDH1-II in one of the two internal PstI restriction sites ( Fig. 2A). When amino acid sequences were deduced from the two cDNAs and aligned with human ALDH1 sequence (25), it is likely that xALDH1-I cDNA starts at the third nt of the putative ATG codon and ends at nt position 2301 and that xALDH1-II cDNA starts at 21 nt upstream from possible start codon and ends at nt position 2341 (Fig. 2B). For xALDH1-II cDNA, the flanking sequence of the possible start site conformed partially to the Kozak criteria (37). A putative polyadenylation signal was present beginning at nt position 2276 for xALDH1-I cDNA and 2317 for xALDH1-II cDNA, in their 3Ј-untranslated regions. Deduced amino acid sequences of xALDH1-I and -II consisted of both 502 residues including the start site Met and whose molecular weights were calculated to be 55,020 and 55,215, respectively, which agreed well with 59 kDa for xCTBP estimated by SDS-PAGE. Estimated pI values of xALDH1-I (7.08) and -II proteins (7.44) also agreed well with the measured pI value of xCTBP, 7.0 Ϯ 0.1 (11). The amino acid compositions of the xALDH1-I and -II were highly similar to that of the purified xCTBP (11). The amino acid sequence of xALDH1-I showed 94.6% identity with that of xALDH1-II through 502 residues. The two se- quences exhibited the highest identity (74 -80%) with ALDH1 sequences from various species, as well as the subclass of ALDH1 reported as a retinal dehydrogenase type II, RalDH(II) (38,39), and the second (71%), third (67-68%), and fourth (65-66%) highest identities with human ALDH6, mammalian ALDH2, and human ALDH5, respectively. The cladogram derived from multiple alignment of amino acid sequences of several classes of ALDHs clearly suggests that the two Xenopus proteins belong to the class of ALDH1 (Fig. 3).
As regards functional significance of the above comparisons, these Xenopus proteins had all 23 of the strictly conserved residues of the aldehyde dehydrogenase superfamily, expressed in phylogenetically diverse organisms (40). By analogy to the human ALDH1 sequence, Cys 304 (41) and Glu 270 (42) might play catalytically essential roles in Xenopus proteins, whereas human ALDH2 enzyme also require the Gln 489 (43). All three residues were conserved in the two xALDH1 sequences. Three of the amino acids interacted with NAD ϩ in the NAD ϩ binding domain of rat ALDH3 (44) and bovine ALDH2 (45), corresponding to Trp 170 , Asn 171 , and Glu 197 in the Xenopus sequences, and were conserved in xALDH1 as well as mammalian ALDH1 sequences.
To identify whether or not the xCTBP purified from liver cytosol is xALDH1, the amino acid sequences of the two regions in xCTBP determined by direct protein sequencing were compared with those deduced from the two xALDH1 cDNAs. The sequences (K)LADLVERDRLILSTM and (M)DIDKVAFTG-STEVGKLIKEAAG were identified to the amino acid positions 92-107 and 239 -261 of xALDH1-II, but both were distinct from the corresponding sequences of xALDH1-I at amino acid positions 92 and 256. Thus xCTBP is more likely to be xALDH1-II than xALDH1-I.
Characteristics of Recombinant xALDH1 Expressed in E. coli-The xALDH1-I and -II proteins expressed in E. coli contain an additional 20 residues of a histidine tag. We purified them to almost single band by a nickel affinity chromatography (Fig.  4A). Approximately 7 mg of the purified proteins were obtained from a 250-ml culture. In SDS-PAGE, the apparent molecular weights of the recombinant xALDH1-I and -II proteins with a histidine tag were estimated to be 60 ϫ 10 3 , which was a bit bigger than that of xCTBP in liver cytosol, 59 ϫ 10 3 (Fig. 4, B and C). The two purified recombinant proteins were specifically photoaffinity labeled with [ 125 I]T 3 (Fig. 4B), like the xCTBP purified from liver cytosol (see inset in Fig. 1). The photoaffinity labeling of xALDH1-II was more strongly inhibited by 5 M unlabeled T 3 than that of xALDH1-I. Polyclonal antibody to the peptide of xCTBP recognized both recombinant proteins as well as xCTBP in adult liver cytosol (Fig. 4C).
ALDH activity was found in the two recombinant proteins expressed from the xALDH1-I and -II cDNAs with all the substrates examined. The K m and V max values for acetaldehyde, propionaldehyde, retinal, and NAD ϩ are summarized in Table I. Both proteins showed the lowest K m values for retinal (6.9 Ϯ 0.5 M in xALDH1-I and 4.2 Ϯ 0.2 M in xALDH1-II) (Fig. 5, B and C); the K m values for propionaldehyde (0.45 Ϯ 0.13 mM in xALDH1-I and 0.32 Ϯ 0.14 mM in xALDH1-II) and acetaldehyde (3.2 Ϯ 0.4 mM in xALDH1-I and 1.7 Ϯ 0.2 mM in xALDH1-II) were 2 and 3 orders of magnitude higher than those for retinal. V max values of the two proteins for acetaldehyde (0.40 Ϯ 0.12 in xALDH1-I and 0.14 Ϯ 0.02 mol/min/mg in xALDH1-II) were similar to those for propionaldehyde (0.31 Ϯ 0.07 in xALDH1-I and 0.12 Ϯ 0.01 mol/min/mg in xALDH1-II) and were 1 order of magnitude higher than those for retinal (0.062 Ϯ 0.005 in xALDH1-I and 0.045 Ϯ 0.003 mol/min/mg in xALDH1-II). The V max /K m values indicate that substrate preference of the two xALDH1 proteins is the following order: retinal Ͼ propionaldehyde Ͼ acetaldehyde. The kinetics for NAD ϩ were characterized by K m of 38 Ϯ 3 and 9.2 Ϯ 1.8 M, and V max of 0.44 Ϯ 0.10 and 0.18 Ϯ 0.03 mol/min/mg, for xALDH1-I and -II, respectively, at the concentration of NAD ϩ ranging from 6 to 120 M. Compared with the kinetics of the purified xCTBP, the K m of Xenopus recombinant proteins exhibited values 3-8 times higher, although their V max value was very similar. The order of substrate specificity of the Xenopus enzymes was in agreement with that of mammalian ALDH1 enzymes (38,46,47) but quite distinct from other classes of mammalian ALDHs (48 -51). Positive cooperativity showing allosteric kinetics could only be detected when the ALDH activities of the xALDH1-I and -II proteins were examined with various concentration of retinal (Hill coefficients, n ϭ 1.8 Ϯ 0.4 and 2.7 Ϯ 0.6, respectively) (Fig. 5, B and C) but were undetectable with acetaldehyde and propionaldehyde as substrates.
Relationship between T 3 Binding and ALDH Activities on xCTBP/xALDH1-To confirm whether or not the two recombinant xALDH1 proteins are dual-functional proteins, T 3 bind- ing activity was also examined. The two recombinant proteins bound specifically T 3 (insets in Fig. 6). Their binding specificities were T 3 Ͼ D-T 3 Ͼ T 4 Ͼ 3,3Ј,5-triiodo-L-thyroacetic acid, which was very similar to that of the xCTBP in adult liver cytosol (11). [ 125 I]T 3 binding to xALDH1-II was more strongly inhibited by 320 nM unlabeled T 3 than that to xALDH1-I, which was in good agreement with the results of the photoaffinity labeling (Fig. 4B). The results of Scatchard analysis of the xALDH1-I and -II proteins shown in Fig. 6 and their K d values for T 3 binding are compared with those of the xCTBP from liver cytosol determined previously (11) (Table II). The K d values, 142 Ϯ 0 and 48.0 Ϯ 7.2 nM for the recombinant xALDH1-I and -II proteins, were 16 and 5 times higher than that for the xCTBP purified from liver cytosol (9 nM), respectively.
As the recombinant xALDH1 proteins required 10 Ϫ5 to 10 Ϫ4 M NAD ϩ to display ALDH activity (Table I), the effects of NAD ϩ on T 3 binding by the two recombinant proteins were examined. NAD ϩ can inhibit the T 3 binding to the two xALDH1 proteins in a dose-dependent manner. The concentrations of NAD ϩ necessary to inhibit 50% of specific T 3 binding were about 100 and 43 M for xALDH1-I and xALDH1-II, respectively (not shown). These findings suggest that xCTBP/xALDH1 expresses alternatively T 3 binding and ALDH activities dependent on NAD ϩ binding. Therefore, we next examined the effect of four coenzymes, including NAD ϩ , on the T 3 binding to the recombinant xALDH1 proteins. NAD ϩ and NADH both inhibited T 3 binding activity by 47 and 27% for xALDH1-I and 23 and 18% for xALDH1-II, respectively, at the concentration of 0.2 mM, but restriction sites. B, first lines, nucleotide sequences of xALDH1-I cDNA; second lines, amino acid sequence deduced from xALDH1-I cDNA sequence; third lines, amino acid sequence deduced from xALDH1-II cDNA sequence; fourth lines, nucleotide sequences of xALDH1-II cDNA. The AATAAA polyadenylation signal close to the 3Ј end of the cDNAs is written in bold letters. The two underlined peptides show the regions corresponding to the sequences determined by a direct protein sequencing. The N-terminal region was determined using the peptide derived from the liver xCTBP after digestion with lysyl endopeptidase, and the C-terminal one was determined using the peptide after treatment with CNBr (11). Residues essential for catalytic activity (41-43), Trp 170 , Asn 171 Glu 197 , Glu 270 , Cys 304 , and Gln 489 , are underlined. Amino acid residues of xALDH1-I are represented by dots where they are identical to the xALDH1-II sequence.
Expression of xCTBP/xALDH1 in Different Tissues-In all tissues examined, a single band of mRNA was detected in a size between 28 S and 18 S ribosomal RNAs. The accumulation of xCTBP/xALDH1 transcripts was particularly strong in kidney and intestine, with smaller amounts found in liver and stomach. These transcripts were expressed at very low levels in heart and skeletal muscles (Fig. 8). DISCUSSION From the following five lines of evidence, we conclude that xCTBP is xALDH1. First, the xCTBP purified from liver cytosol displayed ALDH activity ( Fig. 1 and Table I), and its enzymatic properties agreed with those of mammalian ALDH1. Second, the recombinant proteins expressed from two xALDH1 cDNAs in E. coli also showed both ALDH and T 3 binding activities (Figs. 4 -7), all of which were similar to those of the liver xCTBP (Tables I and II). Third, in the two internal regions, the primary sequence of xCTBP coincided exactly with the corresponding sequences in xALDH1-II and in xALDH1-I except for two residues. Fourth, the apparent molecular weights estimated by SDS-PAGE and pI values and amino acid compositions of xCTBP, described previously (11), were consistent with the molecular weights, pI values, and amino acid compositions calculated from the sequence data of the xALDH1 cDNAs. Fifth, polyclonal antibody to a peptide of xCTBP reacted to both xCTBP and the two recombinant xALDH1 proteins. Since retinal is a preferred substrate for xALDH1, our conclusion raises the possibility that xCTBP/xALDH1 can modulate actions of RA and T 3 via their nuclear receptors by regulating RA synthesis and intracellular levels of free T 3 .
As regards specifically the Xenopus proteins, the K m and K d values of the recombinant xALDH1 proteins were several times higher than those of xCTBP from liver cytosol. The lower affinities of the recombinant xALDH1 proteins for the substrates and T 3 might be due to the presence of the histidine tag at their N termini, which may be a contributory factor for the variation.  A comparison of xALDH1-I with xALDH1-II reveals some differences between their molecular, enzymatic, and T 3 binding properties. However, it is not possible to ascertain from these results that the two proteins have distinct functions. Sequencing data of both the xCTBP (11) and the two cDNAs strongly indicate that the xCTBP we purified from liver cytosol is xALDH1-II, rather than xALDH1-I. Even if xALDH1-I was contained in the fractions with T 3 binding activity of the final purification step, by phenyl 5PW column chromatography, the ratio of xALDH1-I to xALDH1-II might be low.
It is well known that ALDH isozymes form a superfamily (40). The determination of the nucleotide sequences of the isolated cDNAs allowed us to compare their amino acid sequences with those of the known ALDHs from various species and to construct the cladogram shown in Fig. 3. First, this tree clearly illustrates that translated products of the two Xenopus cDNAs and the mammalian and chicken ALDH1 proteins group together. This relationship among the ALDH1 sequences reflects the evolutional distances among vertebrates. The second interesting point about this tree is the position of the two rodent RalDHs(II) and the rat retinal dehydrogenase type I, RalDH(I). The RalDH(II) proteins were outside of the vertebrate ALDH1 group with the bootstrap value of 91.4%, whereas RalDH(I) protein fell within the vertebrate ALDH1 group, suggesting that RalDH(II) is a sister group of ALDH1 and that it is hard to distinguish RalDH(I) from ALDH1 on the basis of the data of primary sequences alone. We could not exclude the possibility that a Xenopus homolog of the rodent RalDH(II), which has not yet been found so far, exists and also plays a role in the synthesis of RA from retinal. It is worth noticing that the ability to convert to RA is found in many ALDH1 proteins, including RalDH(II) as a subtype, from various vertebrate species (indicated by asterisks in Fig. 3). In view of the involvement of RA in many developmental processes, it would be valuable to survey the presence of subtypes of ALDH1 and to examine the expression patterns of a group of xALDH1 proteins during early embryogenesis and limb formation in Xenopus.
Sequence comparison of ALDHs depicted in the cladogram (Fig. 3) shows that the residues Glu 270 , Cys 304 , and Gln 489 , which are thought to be involved in the catalytic role of ALDH enzymes (41)(42)(43), were highly conserved in Xenopus sequences. The region participating in T 3 binding by xCTBP/xALDH1 is located at amino acid positions 92-107. The corresponding region is present away from the catalytic domain but in the NAD ϩ binding domain in bovine ALDH2 (45). Although this region does not seem to interact directly with NAD ϩ , the structural basis for the effects of NAD ϩ on T 3 binding will probably require a structure for ALDH1.
Both recombinant xALDH1 enzymes can catalyze the dehydrogenation of acetaldehyde, propionaldehyde, and retinal (Table I). The substrate specificity estimated as a V max /K m is the highest for retinal, this value being 7-12-and 50 -90-fold higher than those for propionaldehyde and acetaldehyde, respectively, whereas for xCTBP, the V max /K m for retinal is 30fold higher than that for acetaldehyde. Interestingly, allosteric kinetics showing positive cooperativity was observed for the recombinant xALDH1 enzymes and the xCTBP with the Hill coefficient of 1.8 -2.7 and 2.5, respectively, when retinal was used as the substrate (Table I). A similar observation was made for a rat RalDH(I), with Hill coefficient of 1.4 -1.8, by Napoli's group (52)(53)(54)(55). Napoli's group also reported that retinal associated with cytosolic retinoid-binding proteins (CRABP and CRBP) could be important for its recognition as a substrate by the enzyme, since CRABP stimulated the production of RA and CRBP suppressed it (53). This interesting finding suggests that it would be useful in future studies to determine a similar effect of retinoid-binding proteins on the enzyme activities of xCTBP/xALDH1.
The two xALDH1 enzymes reported here are quite similar to each other as regards their the primary sequences, substrate specificities, and T 3 binding properties (Tables I and II, and Fig. 2). Probably the presence of two genes is most likely due to the tetraploidy of the Xenopus genome (56). For this reason we used the 3Ј non-coding region of the xALDH1-I cDNA as a probe for Northern analysis, since it would recognize both xALDH1 transcripts but not those of the other members of ALDH superfamily. The relatively high amounts of xALDH1 mRNAs, migrating as a single band with a mobility intermediate to those of ribosomal 18 S and 28 S RNAs in adult Xenopus kidney and intestine (Fig. 8), are compatible with the finding for rat kidney ALDH1 (57). The level in liver was not so high, although T 3 binding activity was found almost exclusively in liver (19). This discrepancy suggests the possibility of posttranscriptional or -translational regulation, or some other form of modulation of T 3 binding activity of xCTBP. Dual functional properties of ALDH1 have also been reported for other species, such as for elephant shrew (23) and giant octopus crystallins as major component of lens proteins (see Fig. 3) and human 56-kDa androgen-binding protein in genital skin fibroblast (24). Interestingly, the formation of RA from retinal via rabbit ALDH (probably ALDH1) was stimulated by diethylstilbestrol, dehydroisoandrosterone, estrone, and cortisone but inhibited by progesterone, deoxycorticosterone, testosterone, and androsterone (34), thus suggesting that compounds with steroid hormonal activities could bind to rabbit ALDH. Although very recent study indicated that human liver ALDH1 and ALDH2 both could bind T 3 and 3,3Ј,5-triiodo-L-thyroacetic acid, their K d values were micromolar ranges (58). It would therefore be interesting to determine whether or not the same or other hydrophobic signaling molecules bind to xCTBP/xALDH1. Among other studies on CTBPs, there are those reporting an enzyme-linked CTBP, which is a monomeric form of pyruvate kinase subtype M2 in human cell lines and whose conversion to the tetramer is regulated by an intermediate of the glycolytic pathway, fructose 1,6-bisphosphate (10,18). Recently, Shi et al. (59) isolated the cDNA for a Xenopus homolog of the human pyruvate kinase subtype M2 and described high levels of its transcript in Xenopus tadpole tail just before metamorphic climax, in hindlimb during the progression of metamorphosis, whereas relatively low levels were detected in the intestine during metamorphosis. However, it is uncertain whether or not the monomer of pyruvate kinase subtype M2 functions as a CTBP in Xenopus as in the human cell lines. Our earlier studies have shown that xALDH1 is the predominant CTBP in adult Xenopus (11,19). It is worth pointing out that there is no similarity of primary sequence between the monomer of the pyruvate kinase subtype M2 and xCTBP/xALDH1. Another type of NADPH-activated CTBPs has been reported in rat kidney (13), liver (14), and brain (15), comprising molecular species with different molecular weight values. For example, kidney CTBP is a monomer of 58 kDa, liver CTBP is a homodimer with 76 kDa, and rat brain CTBP is a 58-kDa protein. T 3 binding activity of all of these CTBPs is strongly activated by NADPH and slightly by NADH, but not by NADP ϩ , whereas NAD ϩ has no effect on the T 3 binding activity of the rat CTBPs (13,14). A very recent study indicated that human kidney CTBP is a 38-kDa protein, which is homologous to kangaroo crystallin (60). On the other hand, T 3 binding activity of xCTBP is inhibited by NAD ϩ and NADH, and neither NADPH nor NADP ϩ is effective. If the total concentration of cytoplasmic NAD ϩ plus NADH, which ranges 10 Ϫ4 M in mammalian cells (61), changes in similar ranges in Xenopus cells, it would be one of the important factors modulating T 3 binding to xCTBP/ xALDH1. These observations and our present results show that xCTBP might be a novel type of CTBP.