cDNA Cloning and Expression of the Human Hepatic Bile Acid-binding Protein A MEMBER OF THE MONOMERIC REDUCTASE GENE FAMILY*

In human liver, we previously identified one isoform of dihydrodiol dehydrogenase activity that expresses high affinity bile acid binding (HBAB) with minimal 3a-hydroxysteroid dehydrogenase (3a-HSD) activity for bile acids. This protein may assist in the rapid intracellular transport of bile acids from the sinusoidal to the canalicular pole of the cell. We now report the cDNA cloning and bacterial expression of this novel, multifunctional protein. A 1252-base pair HBAB cDNA was cloned from a HepG2 hGTl1 library using a rat hepatic bile acid binder cDNA probe. Bacterial expressed recombinant HBAB oxidized racemic trans dihydrodiol benzene (0.455 rmol NADPH/mg/min) with minimal 3a-HSD activity for bile acids (<0.003 rmol NADPH/mg/min). Lithocholic acid and chenodeoxycholic acid dissociation constants as determined by displacement of the fluorescent probe, bis- l-anilino-8 sulfonate, were higher than those previously reported for the native protein (1 pM uersus 10 nM). Significant amino acid sequence homology was found with the human chlordecone confirmed by monitoring reduction in fluorescence of ANS p with concentrations of ANS determined in duplicate in the presence of two different concen- tration of lithocholic acid (0.25 p and chenodeoxycholic acid (1 p ) Apparent Ki then calculated assuming one bile acid-binding site previously

tion into bile. We have previously established that in rat liver the 37-kDa cytosolic enzyme 3a-hydroxysteroid dehydrogenase (rh-3a-HSD)' interacts extensively with bile acids during their intercellular translocation and may be an important determinant in net hepatic bile acid transport (1)(2)(3)(4). Under normal bile acid flux, rapid translocation of bile acids from the sinusoidal to the canalicular poles occurs through the cytosol associated with the rh-3a-HSD and other potential bile acid-binding proteins, such as the dimeric glutathione Stransferases and fatty acid-binding protein. In addition to binding bile acids, this multifunctional oxidoreductase catalyzes the stereospecific reduction of both bile acid precursors and the steroid hormones dihydrotestosterone and progesterone by preferentially utilizing NADP(H) as a cofactor (4-7). rh-3a-HSD is also capable of oxidizing trans-dihydrodiol carcinogens in vitro via its dihydrodiol dehydrogenase activity which has been shown to significantly reduce the mutagenic potential of these procarcinogens in a bacterial mutagenicity test system (8,9). rh-3a-HSD is expressed predominantly in liver and intestine, which is consistent with its role in xenobiotic metabolism and bile acid transport. There is significant amino acid sequence homology with other monomeric NADPH reductases suggesting a subfamily within the large oxidoreductase supergene family (10).
Using human liver, we have previously purified and characterized a cytosolic 36-kDa bile acid-binding protein (HBAB) that exhibits one-to two-orders of magnitude lower bile acid dissociation constants than does rh-3a-HSD (1,11,12). Like rh-3a-HSD, this protein is also a dihydrodiol dehydrogenase, but lacks 3a-HSD activity for bile acids (13). HBAB may function like its rat counterpart in mediating the transcellular cytosolic transport of bile acids. In contrast to the rat, however, in human liver, the other major classes of potential cytosolic bile acid-binding proteins, namely the glutathione S-transferases and fatty acid-binding protein have significantly higher dissociation constants for bile acids (11). Thus, HBAB may be the dominant protein with which bile acids interact during their intracellular transport. In order to fur-' The abbreviations and trivial names used are: rh-3a-HSD, rat hepatic 3a-hydroxysteroid dehydrogenase; HBAB, human bile acid binder; PBS, phosphate-buffered saline; PAGE, polyacrylamide gel electrophoresis; PCR, polymerase chain reaction; chenodeoxycholic acid, 3a,7a-dihydroxyl-5fl-cholanic acid; lithocholic acid, 3a-hydroxyl-5~-cholanic acid bis-ANS, l-anilinonaphthalene-8-sulfonic acid; trans-dihydrodiol benzene, trans-1,2-dihydroxycyclohexa-3,5diene; l-acenaphthenol, 7 hydroxyacenapthene; BSA: bovine serum albumin; 3a-hydroxysteroid dehydrogenase (EC 1.1.1.50); dihydrodiol dehydrogenase (EC 1.3.120). ther characterize the physiological function of this novel protein, we now report the cDNA cloning, deduced amino acid sequence, and bacterial expression of this protein. The knowledge gained from the cDNA cloning will allow for indepth studies to be performed to determine the physiological significance of this multifunctional protein.

MATERIALS AND METHODS
Chemicals-All chemicals purchased from Sigma were of molecular biology grade unless otherwise specified. Lithocholic acid, chenodeoxycholic acid, and 1-acenaphthenol were purchased from the Aldrich, NADP was purchased from United States Biochemical, and bis-ANS was purchased from Molecular Probes. Racemic trans-dihydrodiol benzene was synthesized and the purity confirmed by NMR/IR analysis by Dr. Don Jerina, National Institutes of Health. [m3*P] dCTP (3000 Ci/mmol) and [a-36S]dATP (1400 Ci/mmol) were purchased from Du Pont-New England Nuclear, and [35S]sulfur amino acids (Translabel) were purchased from ICN; restriction enzymes, DNA modification enzymes, and nucleotides were purchased from Pharmacia LKB Biotechnology Inc., Boehringer Mannheim, or Promega and were utilized according to the manufacturer's recommendations. Tissue culture media, Dulbecco's modified Eagle's medium, PBS-trypsin, and fetal calf serum were purchased from GIBCO-BRL.
cDNA Cloning-A Hep GB Xgtll cDNA library was kindly provided by Dr. A. J. Lusis (14) and screened according to the method of Benton and Davis (15) using a 32P-labeled proximal 5'-EcoRI fragment of rh-3a-HSD cDNA (16). Approximately 600,000 plaques were screened on twenty 150-mm plates, and duplicate nitrocellulose filters were prepared as previously described (10,17). Duplicate filters were prehybridized in 7% SDS, 1% BSA (fraction V), 0.5 M sodium phosphate, pH 7.2, and 1 mM EDTA, pH 8, for 1-2 h at 68 "C followed by overnight hybridization at 68 "C with 5 X lo6 cpm of rh-Sa-HSD cDNA probe/filter. Filters were then washed twice in 2 X SSC, 0.1% SDS (1 X standard sodium citrate = 0.15 M sodium chloride, 0.015 M sodium citrate, pH 7) at 22 "C and then in 0.5 X SSC and 0.1% SDS at 55 "C for 30 min, dried, and exposed to Kodak XAR Xomat film with intensifying screens for 24-96 h at -80 "C. Clones were sequentially plaque purified and insert size determined by the polymerase chain reaction using Xgtll-specific primers (18). The largest insert was retrieved by EcoRI digestion, and individual fragments were subcloned into the plasmid vector pSPT 19 (Boehringer Mannheim) and sequenced in both orientations using both vector-and cDNAspecific oligonucleotide primers to generate overlapping DNA sequences in both orientations. DNA sequencing via the chain-terminating technique of Sanger (19) was performed with [a-35S]dATP using the Pharmacia T7 sequencing kit according to the manufacturer's instructions.
Protein Microsequence Analysis of HBAB Protein-Homogeneous HBAB protein (9 pg) was prepared as previously reported (12,13) and digested with trypsin in preparation for gas-phase microsequence analysis. HBAB protein was reduced with 8 M urea and 10 mM dithiothreitol at 50 "C for 15 min, followed by alkylation with 10 mM iodoacetamide for 15 min in the dark. HBAB was then diluted 1:2 with 0.4 M ammonium bicarbonate buffer and incubated with high performance liquid chromatography purified ~-1-tosyl-amido-2-phenylethyl chloromethyl ketone-treated trypsin (Boehringer Mannheim) (50:l weight/weight) overnight at 37 "C. Tryptic fragments were isolated by reverse-phase hydrophobic chromatography using a Vydac C, column, 2.1 X 250 mm, and analyzed on a gas-phase amino acid microsequencer using the same conditions as previously reported (10,20,21).
Cell-free Translation and Immunoprecipitation of HBAB cDNA Clone-Complete HBAB cDNA clone was reconstructed into Promega pGEM 7(+) plasmid at the EcoRI endonuclease restriction site using standard cloning techniques. Plus and minus strand cRNA probes were generated by linearizing 1 pg of pGEM 7-HBAB plasmid by digestion with either ClaI or XbaI restriction enzymes, which served as a template for cRNA production, using appropriate RNA polymerases with rNTP provided in the Gemini Ribroprobe kit. DNA template was digested with Promega RQ1 RNase-free DNase, phenol/ chloroform extracted, and stored at -20 "C in a 100% ethanol, 0.2 M sodium acetate solution (400 pl). Aliquots (50 pl) were centrifuged, resuspended in 2 p1 of 0.1% diethyl pyrocarbonate-treated water, and used to program a reticulocyte lysate expression system using 25 pCi of 35S-labeled sulfur amino acids (Translabel) according to the manufacturer's (Ambion) protocol. Translation products were separated on 12% SDS-PAGE gels and processed for either fluorography using Enhance (Du-Pont New England Nuclear), according to the manufacturer's recommendation or immunoprecipitation according to the method of Doolittle (22) using a previously characterized monospecific rabbit antiserum for either rh-3a-HSD (4) or HBAB (13) followed by fluorography. Fluorograms were vacuum dried and exposed onto Hyperfilm Max (Amersham Corp.) with intensifying screens for 10 days at -80 "C.
Northern Blot Analysis-Poly(A)+-enriched RNA (3 pg) from adult and fetal liver were purchased from (Clontech), transferred unto a nylon membrane after denaturing agarose gel electrophoresis, and hybridized with a random primed 32P-labeled HBAB cDNA probe as previously reported (10).
DNA Sequence Analysis-HBAB-deduced amino acid sequence was compared with the following protein data bases: Genpept version 72 and NBRF protein database using the FASTA program of Pearson and Lipman (23) on a VAX minicomputer using the University of Wisconsin computer program.
Western Blotting of Hep G2 Tissue Culture Lysate-Hep Gz cells were grown to confluence in Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum on 60-mm collagen-coated culture dishes and harvested after trypsinization into 1 ml of PBS using standard procedures. Hep GZ cells were disrupted by three freeze-thaw cycles and the lysates harvested after 15,000 X g centrifugation at 4 "C for 20 min. 2 pl of supernatant was applied onto a 12.5% SDS-PAGE gel and transferred onto nitrocellulose membrane using a semi-dry blotter (Bio-Rad) in preparation for immunoblotting.
Bacterial Expression of HBAB Protein-The prokaryotic expression vector pGEX-GL1, kindly provided by Genlabs Incorporated, is a modification of the pGEX-2T expression vector (Pharmacia) and includes a NcoI restriction site at the cDNA insertion site generating a fusion protein with the glutathione S-transferase of Schistosoma japonicum (24). The NcoI restriction site (underlined) was introduced into the HBAB cDNA at nucleotide position number 1 by incorporating two mutated nucleotides (italicized) that did not alter the amino acid sequence into the 5' PCR product was digested with NcoI and the fragment was then isolated using standard cloning techniques generating a 686-base pair cDNA fragment with NcoI restriction sites at both ends. HBAB cDNA clone was sequentially digested with NcoI and BamHI, whose restriction site at the 3' end of the cDNA was provided for by the polylinker region of the pGem 7(+) plasmid. The HBAB cDNA was ligated into the pGEX-GL1 expression vector in a two-step process by first ligating the NcoI/BamHI HBAB cDNA fragment (686-1252) into the vector, followed by ligation with the NcoI-digested PCR product. Proper orientation of the pGEX-GLI-HBAB construct was confirmed by PCR amplification with primers overlapping the internal NcoI restriction site and restriction mapping. DNA sequencing of the PCR product confirmed no misincorporated nucleotides. HBAB protein was expressed by inducing pGEX-GLI-HBAB transformed Escherichia coli strain HBlOl cells with 1 mM isopropyl-thio-@galactoside for 3 h during mid-log phase of growth. Cells were centrifuged 5000 X g for 10 min at 4 "C, resuspended in 1/50 volume of 20 mM sodium phosphate, pH 7.2, and 150 mM sodium chloride (PBS) and underwent sonication (three bursts for 10 s, 50% power setting) on ice. Triton X-100 was added to a final concentration of 1%, the bacterial lysate was centrifuged at 4 "C and supernatant applied to a 2-ml prepacked glutathione 4B-superose column (Pharmacia) preequilibrated with 20 ml of PBS supplemented with 1% Triton X-100. Column was washed with 30 ml of PBS to remove unbound protein and then washed with 6 ml of 150 mM NaCl, 50 mM Tris-HC1, pH 7.5, and 2.5 mM CaClz (thrombin digestion buffer) in preparation for in situ thrombin digestion of the fusion protein on the column. 2 ml of thrombin digestion buffer containing 10 pg of purified human thrombin (Sigma) was applied and the column was then incubated for 150 min at room temperature. After digestion, the glutathione affinity column was eluted with 50 mM Tris-HCI, pH 7.5, and 150 mM NaCl which released the cleaved HBAB protein. HBAB was purified to homogeneity by application onto a Hi-Trap Blue minicol-umn (Pharmacia), pre-equilibrated with 5 ml of 50 mM Tris-HC1, pH 7.5, and eluted with 2 ml of the same buffer supplemented with 2 M NaCI. HBAB was stored in 50% glycerol a t -20 "C. HBAB used for bile acid binding studies was purified with the exact same protocol except that no Triton X-100 was added after sonification of the cell pellet or used for the initial equilibration of the glutathione affinity column. The lack of Triton X-100 did not affect the purity of the recombinant HBAB protein.
Enzymatic Actiuity-Enzymatic assays for trans-dihydrodiol dehydrogenase and 3a-hydroxysteroid dehydrogenase activities were performed as previously described by Woerner and Oesch (25). Briefly, substrate dissolved in absolute ethanol, methanol, or acetonitrile, which did not exceed 5% of the total volume, and NADP dissolved in water (250 p~) were added to 900 pl of 0.05 M glycine-NaOH, pH 9, and the reaction was initiated by addition of HBAB protein. Enzymatic activity was determined by spectrophotometric monitoring of the conversion of NADP+ to NADPH at 340 nm a t 37 "C. 1-Acenaphthenol (1.47 mM) and racemic trans-dihydrodiol benzene (0.02 to 1 mM), kindly provided by Dr. Don Jerina (National Cancer Institute), were used to monitor trans-dihydrodiol dehydrogenase activity. Lithocholic acid (50 PM) and chenodeoxycholic acid (20 PM) were used to monitor Sa-HSD activity, as previously reported (13). No oxidation of NADP' was detected in the absence of enzyme or substrate. Enzymatic reaction was monitored over a 3-min time period using a Beckman DU-64 spectrophotometer. Data were collected on a Macintosh I1 computer (Apple) and a least square fit of the initial rate was calculated using Cricket Graph Cricket Software.
K,,, and !Irnax were determined by monitoring the rate of NADP+ oxidation while altering either substrate or NADP+, and values were calculated by non-linear least-square fit of the initial reaction rates using SAAM program on a Macintosh Quadra 700. Biuret protein assay was performed according to the manufacturer's recommendation (Bio-Rad) using y-globulin as a protein standard.
Bile Acid Binding Studies-Bile acid binding affinities for lithocholic acid and chenodeoxycholic acid were determined as previously reported using the ANS displacement technique (11,26,27). Initially, bile acid dissociation constants were examined by equilibrium dialysis using a Hoeffer EMD dialysis system with 100-pl chambers. 0.1-5 p~ of recombinant HBAB protein was incubated with 0.005 pCi of [24-"C]lithocholic acid or [24-14C]chenodeoxycholic acid (Du-Pont-New England Nuclear) in 100 p1 of 0.01 M sodium phosphate, pH 7.4, and dialyzed against an equal volume of buffer containing the same nonlabeled bile acid or buffer alone a t 4 "C. Precipitation of the protein within 4-8 h prevented accurate assessment of the bile acid dissociation constants using this technique. Bile acid binding was then assessed by the ANS displacement technique in which bile acids competitively inhibit the binding of the fluorescent probe ANS to the protein, allowing calculation of a Ki for the bile acid. We previously demonstrated that this Ki is comparable to the Kd for bile acid binding reported in previous studies (26, 28). ANS binding to HBAB was monitored by fluorescence emission of the probe (Ex4m/Em4,) with a Perkin and Elmer fluorometer using the same protocol as previously reported (11). Briefly, K d for ANS binding to HBAB was determined by monitoring the increased fluorescence associated with ANS (0.01-50 PM) binding to 2 PM of HBAB in 0.01 M sodium phosphate, pH 7.4, and K d for ANS binding calculated using a double-reciprocal plot. Competitive displacement of ANS by bile acids to HBAB was confirmed by monitoring reduction in the fluorescence of ANS (25 p~) with increasing concentrations of bile acids. K d values for ANS were then determined in duplicate in the presence of two different concentration of lithocholic acid (0.25 and 0.5 p~) and chenodeoxycholic acid (1 and 5 p~) .
Apparent Ki was then calculated assuming one bile acid-binding site as previously reported (11).
Immunoblotting and SDS-PAGE-SDS-PAGE was performed as previously described (13). Gel was electroblotted onto a nitrocellulose membrane (BA85, Schleicher & Schuell) in 20% methanol, 0.192 M glycine, and 20 mM Tris-HC1, pH 8, using a Bio-Rad semi-dry blotter. After blocking the nitrocellulose membrane with 3% BSA in PBS with 0.1% Tween-20, the membrane was incubated with a 1:1200 dilution of a monospecific HBAB rabbit antiserum diluted in 1% BSA in PBS with 0.1% Tween-20 and antibody visualized with goat anti-rabbit IgG antibody coupled to alkaline phosphatase developed according to the manufacturer's recommendation (Promega).
Coomassie Blue staining of the polyacrylamide gel was performed as described (29).

RESULTS AND DISCUSSION
Cloning of HBAB cDNA-Strategy used for the cDNA cloning of HBAB was based on obtaining the primary amino acid sequence from the protein to verify a clone identified by screening a human liver cDNA library using rh-3a-HSD cDNA probe. Hep GP cell lysate was found to contain a 36-kDa protein recognized by monospecific HBAB antiserum, as depicted in Fig. 1. Therefore, a Hep G2 X g t l l cDNA library was screened with rh-3a-HSD cDNA probe, with the assumption that both proteins would share significant sequence homology (14). Five clones identified with the rh-3aHSD cDNA probe were plaque purified, and the largest insert was sequenced in both orientations. This 1252-base pair clone contained an internal EcoRI restriction site, as did the rh-3a-HSD and included a polyadenosine tail at the 3' end. The cDNA encoded for an open reading frame of 323 amino acids with a predicted molecular mass of 37,325 daltons. Concurrent peptide mapping and microsequence analysis of HBAB protein was initiated to confirm the veracity of the cDNA clone identified by rh-3a-HSD. Peptide fragments were isolated, purified, and the sequences for 189 amino acid residues were determined from 16 different peptides. Table I illustrates the location and compares the deduced amino acid sequence with that obtained from tryptic peptide fragments. The deduced cDNA sequence agreed with 184 of 189 residues determined by microsequence analysis. These 5 different amino acids may represent polymorphism of HBAB in Hep Gz cells compared with that of normal human liver, or may represent microheterogeneity. Fig. 2 illustrates the restriction map and sequencing strategy for the cDNA cloning and Fig. 3 shows the nucleotide sequence with its deduced amino acid sequence.
In order to further verify the authenticity of the HBAB cDNA clone, cell-free translation and bacterial expression studies of the HBAB cDNA were undertaken. The HBAB cDNA was reconstructed into the pGem-7 plasmid and used to generate both sense (plus) and antisense (minus) complementary RNA products. Only the sense strand cRNA product was found to produce a 37-kDa protein in a reticulocyte lysate system (data not shown). Monospecific HBAB antibody was capable of immunoprecipitating the sulfur amino acid radiolabeled translation product, whereas preimmune serum and Hep GP cell lysate (2 pl) (lane A ) and normal human hepatic cytosol (2 p1 of 33% (w/v)) (lane B) were subjected to 12% SDS-PAGE, electrotransferred onto nitrocellulose membrane, and immunoblotted with an HBAB monospecific polyclonal rabbit antiserum (13). No protein was identified when rabbit preimmune serum was used (data not shown). Amino acids are listed by their single-letter code. Bold letters underlined represent disagreement between primary amino acid sequence and deduced cDNA sequence. Bold and italic letters underlined represent location of cysteine on cDNA sequence which were oxidized during preparation of HBAB for protein microsequence analysis. Sequences in parentheses are amino acids identified during a single cycle. monospecific rh-Sa-HSD failed to do so, as shown in Fig. 4. These studies confirmed that the protein product of the HBAB cDNA clone was precipitated with antiserum raised against the native protein, and confirmed the Western blot results in Fig. 1.

Peptide
Bacterial Expression of HBAB Immunoreactivity, Enzymatic Activity, and Bile Acid Binding-E. coli was chosen to express HBAB cDNA since similar proteins such as human aldose reductase, human placental 15-hydroxyprostaglandin dehydrogenase, and rh-Sa-HSD have been successfully expressed with enzymatic activity (30)(31)(32). Expression vector pGEX-GL1-HBAB produced a glutathione S-transferase fusion protein from which HBAB, with an additional glycine and serine at the NH2-terminal end was separated by a unique thrombin proteolytic restriction site. Fusion protein was purified by GSH affinity column chromatography to isolate the recombinant protein from other bacterial proteins, followed by thrombin digestion and purification to homogeniety on a Sepharose blue affinity column. Fig. 5 depicts HBAB purification to homogeneity during the three-step protein purification procedure. Purification table for recombinant HBAB is shown in Table I1 starting with 200 ml of bacterial culture.
NO dihydrodiol dehydrogenase activity was detected in the supernatant of the bacterial lysate, indicating that the fusion protein has no detectable catalytic activity. We therefore cannot determine the recovery of the fusion protein from the bacteria pellet by enzyme activity. Monospecific HBAB antiserum was able to identify the fusion protein but with substantially lower affinity than the native protein on Western immunoblotting. By estimating the recovery of the HBAB protein from the glutathione affinity column after thrombin digestion, it was calculated that the fusion protein constitutes at least 8% of the total bacterial lysate proteins. The final purification step led to a significant loss of protein, with no increase in specific activity confirming that the glutathione S-transferase-HBAB fusion protein has little or no enzymatic activity.
Recombinant HBAB protein generated from bacteria was analyzed for both its biochemical activity and its bile acid binding affinities. Table I11 lists the enzymatic features of the recombinant HBAB protein. HBAB efficiently oxidized 1acenaphthenol, the monohydroxy1 substate previously used to monitor dihydrodiol dehydrogenase activity during purification of HBAB from normal human liver. Recombinant enzyme was capable of oxidizing a racemic mixture of transdihydrodiol benzene confirming that the enzyme functions as a dihydrodiol dehydrogenase. Minimal 301-HSD activity was detected when either lithocholic or chenodeoxycholic acid was used as a substrates (<0.003 pmol NADPH/pg/min). Catalytic specificity of the recombinant protein is thus the same as the native HBAB, which had minimal 3a-HSD activity.
Characterization of the bile acid binding to the recombinant protein was attempted initially by equilibrium dialysis, as previously described (11,26). Equilibrium dialysis of HBAB protein (2 p~) with tracer radiolabeled lithocholic acid or chenodeoxycholic acid was performed with various concentrations (0.1-20 p~) of the respective, unlabeled bile acids in a 1OO-wl dual chamber dialysis system. Protein precipitation after 4-6 h at 4 "C prevented accurate determination of the bile acid dissociation constants. Incubation with NADP+, reducing agents, or increasing the p H failed to prevent protein precipitation. Purification of the bacterial expressed protein in the absence of Triton X-100 also had no effect.
Since precipitation of the recombinant protein prevented determination of the bile acid binding affinities by equilibrium dialysis, the ANS displacement technique was used. This technique was used previously to characterize the bile acid binding affinities of native HBAB (11). Binding of ANS to HBAB led to increased fluorescence, which was competitively reduced by increasing the concentration of either lithocholic acid or chenodeoxycholic acid. K d for ANS binding was determined in the absence of bile acid and then compared with an  204 219

T M j O T C C G A C C A G C C T T G G A A A G G T C A C T G A A A A A T C f i C A A T n ; G A T T A T G f i G A C 339
Jmu Val Arq Pro Ala Leu Glu Prrg Ser Leu Lys Asn Leu Gln Leu Asp Tyr Val Asp 112    conditions (13). Further investigation will be required to determine if the tendency to aggregate prevents the recombinant protein from binding bile acids with the same affinity as the native protein. Alternatively, the high affinity bile acid binding of the native protein may be due to post-translational Features of the HBAB Gene-HBAB cDNA was further characterized by Northern blot analysis of human liver RNA and amino acid sequence homology to other monomeric reductases. HBAB cDNA hybridized to an approximately 1.3kilobase mRNA in poly(A)+-selected human hepatic RNA (Fig. 6). Comparable amounts of poly(A)-selected fetal liver RNA demonstrated a 3-fold reduction of HBAB mRNA as compared to adult levels (data not shown). This is similar to the decreased steady state mRNA levels of rh-Sa-HSD in the fetal rat." The HBAB cDNA clone contains the complete 3' end of the gene as evidenced by the presence of both a polyadenylation signal sequence and a polyadenosine tail. The HBAB cDNA polyadenylation signal sequence (ATTAAA) is a variant of the usual signal sequence but is capable of initiating the polyadenylation signal in the SV40 virus and has been identified in other genes as well (33). This sequence has been recently confirmed by sequence analysis of an HBAB genomic clone.' Hybridization to a 1.3-kilobase mRNA size in human liver indicates that only a small portion (50-100 base pairs) of the 5"untranslated region of the HBAB gene is missing. Experiments are underway to identify the complete 5' end by genomic mapping and primer extension studies.

Ser
Homology and Peptide Structure of the HBAB Protein-The deduced HBAB amino acid sequence was directly compared to rh-30-HSD and Genbank data bases to identify other homologous genes and conserved peptide domains. Fig. 7 lists the deduced amino acid sequence for the six most homologous genes (>50% sequence identity) selected by the the FASTA program of Pearson (23). All six of these genes are monomeric oxidoreductases capable of metabolizing a wide variety of endogenous substrates and xenobiotics. The highest sequence homology (83%) was observed with the monomeric NADP(H)-dependent human hepatic chloredecone reductase cDNA gene, which was cloned by immunoscreening of a human hepatic Xgtll expression library (34). The human hepatic chloredecone reductase is a 36-kDa monomeric, oxidoreductase that is capable of reducing the organochlorine pesticide chloredecone to its alcohol metabolite. This biotransformation leads to increased biliary excretion of the pesticide and concomitant reduction of its neurotoxicity since bile is the major excretory route. Significant variation in chloredecone reductase mass and activity have been detected in human hepatic samples suggesting a potential genetic predisposition for this pesticide's toxicity. Bovine lung prostaglandin F synthetase shares 77% sequence identity with HBAB and is capable of reducing PGH, to PGFz, via endoperoxidase activity and of converting PGD, to 9a-11B-PGF2 via 11-keto reductase activity (35, 36). 70% sequence identity and an additional 10% similarity was found with the rh-3a-HSD (10,37). Aldose reductase shares 50% sequence homology and is part of a large subgroup of related genes that    includes the aldehyde reductase and lens crystallin from the common frog. Aldose reductase has been extensively studied and is the best characterized member of this gene family. It is postulated to mediate sorbitol toxicity of diabetes mellitus and is implicated in regulating intracellular osmolarity of kidney cells by generation of sugar polyols in response to osmotic stress (38). Chloredecone reductase, bovine prostaglandin F synthetase, aldose reductase, and rh-3a-HSD all function as reductase i n vivo (4,36,39). The other protein identified with significant sequence homology is the mouse vas deferens protein that is presumed to be an oxidoreductase because of its close homology (82%) to aldose reductase (40). Although the function of the gene product is unknown, it is regulated by androgens and is hypothesized to reduce glucose to sorbitol, an intermediate in the synthesis of fructose or to generate polyols required for osmotic regulation (40). rh-3a-HSD reduces bile acid precursors required for bile acid synthesis and the steroid hormones dihydrotestosterone and progesterone and is capable of oxidizing dihydrodiols i n vitro (3-5, 41, 42). Not shown, aldehyde reductases and epsilon crystalline lens proteins from the common frog also share significant homology (43,44). Sequence homology among these and other oxidoreductases such as the alcohol dehydrogenase gene family suggests evolution from a common ancestral gene by way of gene duplication (45,46). Recent x-ray crystallographic analysis of pig and human aldose reductase should provide important structural information which may be applied to all members of this monomeric reductase gene family (47,48).

DISCUSSION
To our knowledge, we are the first to have cloned and expressed a human hepatic dihydrodiol dehydrogenase that uniquely exhibits a high affinity bile acid binding capacity. The availability of the human and rat cDNA for dihydrodiol dehydrogenase will be beneficial for further characterization of this monomeric reductase gene family. There is significant species variation in the dihydrodiol dehydrogenase family. In rats, only one isoform of dihydrodiol dehydrogenase exists in the liver, whereas in humans, monkeys, and rabbits, multiple isoforms are found. The physiological significance of this diversity is unknown but the diversity is analogous to the cytochrome P-450 supergene family in which overlapping substrate specificities exist in multiple, similar enzymes.
We previously reported resolution of six isoforms of dihydrodiol dehydrogenase activity in the 35-30-kDa gel filtration fraction of human hepatic cytosol by monitored oxidation of the substrate 1-acenaphthenol (13). Only one isoform of dihydrodiol dehydrogenase demonstrated high affinity bile acid binding which suggests that the bile acid-binding site may be distinct from the substrate-binding site. Hara and colleagues (49) reported the purification of five forms of human hepatic dihydrodiol dehydrogenase that were initially separated by anion-exchange chromatography. A more recent report by this group concluded that their D2 isoform was comparable to the HBAB protein since bile acids were capable of completely inhibiting the dihydrodiol dehydrogenase activity with Ki values of 5-10 nM (50). Although their Ki values are comparable to the Ki value for bile acid displacement of ANS binding in the native protein, we were unable to reproduce these results using the recombinant protein. This disagreement may be due to expression in bacteria of a recombinant protein that lacks the high affinity bile acid binding of the native protein. Penning and Sharp (51) reported the identification of dihydrodiol dehydrogenase activity using benzene dihydrodiol as a substrate in a 35-kDa gel filtration fraction of human hepatic cytosol and demonstrated recognition of a 35-kDa protein on Western blotting with rat hepatic 3a-HSDl dihydrodiol dehydrogenase antiserum. This protein was not further purified.
Direct comparison of amino acid sequences of the rat and human bile acid-binding proteins is an ideal means to define subtle differences which may confer either different substrate specificities for reductase activity or bile acid binding affinities. The rh-3a-HSD binds mono-and dihydroxyl bile acids and their glycine or taurine conjugates with K d values of 1-10 PM whereas the K d for comparable bile acids by HBAB is 10-20 nM (1,11,26). The location of the bile acid-binding site on the human bile acid binder is unknown. The only

cDNA Cloning and Expression of H B A B
Cys-Ala-Ser-Lys (located at position 205-209 of HBAB) has been identified within a group of evolutionary diverse oxidoreductases and has been implicated by Krook et al. (56) to play a key role in catalytic activity. Bovine prostaglandin f synthetase, chloredecone reductase, and HBAB all have a conservative substitution of a phenylalanine for tyrosine at position 205. The replacement of tyrosine with a phenylalanine in some of these oxidoreductases suggests that the hydrophobic domain of the phenyl ring is the key feature for catalytic activity. Site-specific mutation of this tyrosine to an alanine in a related enzyme, human placenta 15-hydroxyprostaglandin dehydrogenase, leads to complete loss of enzymatic activity confirming the significance of this residue (31). Review of the 1.65-angstrom x-ray crystal structure of the human aldose reductase provides an explanation for the conservation of tyrosine or phenylalanine residue within the diverse dehydrogenase gene family. Tyrosine 209 in human aldose reductase hydrogen binds to the nicotinamide group of the cofactor thereby positioning the cofactor in the proper orientation for transfer of the critical 4 R hydrogen. Conformational analysis of the human and porcine aldose reductase thus provides a paradigm for other members of this gene family. According to both models, aldose reductase is composed of a parallel eight chain a/@ barrel protein, with each @ portion of the chain constituting part of the elliptical catalytic pocket (47,48).
The physiologic role of the human HBAB in bile acid transport or metabolism is unknown. Unlike the rat bile acid binder, HBAB is not required for bile acid synthesis. HBAB may serve to retain bile acid within the hepatic cytosol as in the rat so that they may be rapidly excreted into bile by the canalicular bile acid transporter. Other potential cytosolic bile acid-binding proteins in human liver, such as the glutathione S-transferases and fatty acid-binding protein, have much higher bile acid dissociation constants which eliminate these proteins as significant physiologic bile acid binders.
In addition to their potential role in mediating intracellular bile acid transport, the high affinity bile acid binding suggests another role in monitoring the intrahepatic bile acid concentration. Bile acid synthesis is tightly regulated predominantly by expression of the rate-limiting enzyme in bile acid synthesis, cholesterol 7a-hydroxylase (57-61). Multiple factors modulate expression of this labile enzyme. In bile fistula rats, depletion of the bile acid pool leads to induction of cholesterol 7a-hydroxylase gene transcription, increased steady state mRNA levels, and increase in both protein mass and enzymatic activity within 48-72 h. Replacement of the bile acid pool with hydrophobic bile acids and those with 7a-hydroxyl group were capable of down-regulating both protein levels and gene expression. Whether bile acids themselves or alterations of another oxysterol in response to bile acids are responsible for this alteration is unknown. These studies suggest that active monitoring of the intrahepatic bile acid concentration is in part responsible for regulation of cholesterol 7a-hydroxylase gene. The high affinity binding of HBAB renders it a potential candidate for monitoring the intracellular bile acid concentration. An unoccupied HBAB protein would occur only at very low intracellular concentrations which may serve as a signal for the induction of bile acid synthesizing capacity in combination with other factors. HBAB lacks all previously recognized features of a DNAbinding protein, thereby rendering it unlikely to directly interact with cis acting DNA regulatory elements in the cholesterol 7a-hydroxylase gene. Future experiments will determine the role of this high affinity bile acid-binding protein in bile acid transport and metabolism.
In conclusion, we have cloned and expressed a novel human hepatic dihydrodiol dehydrogenase that has high bile acid binding affinities. The availability of the rh-3a-HSD cDNA, a highly homologous protein with different catalytic specificity, and bile acid binding affinities will be beneficial for identifying protein domains that have subtle differences.