Molecular cloning and characterization of (R)-3-hydroxybutyrate dehydrogenase from human heart.

The complete amino acid sequence of human heart (R)-3-hydroxybutyrate dehydrogenase (EC 1.1.1.30) has been deduced from the nucleotide sequence of cDNA clones. This mitochondrial enzyme has an absolute and specific requirement of phosphatidylcholine for enzymic activity (allosteric activator) and is an important prototype of lipid-requiring enzymes. Despite extensive studies, the primary sequence has not been available and is now reported. The mature form of the enzyme consists of 297 amino acids (predicted M(r) of 33,117), does not appear to contain any transmembrane helices, and is homologous with the family of short-chain alcohol dehydrogenases (SC-ADH) (Persson, B., Krook, M., and Jörnvall, H. (1991) Eur. J. Biochem. 200, 537-543) (30% residue identity with human 17 beta-hydroxysteroid dehydrogenase). The first two-thirds of the enzyme includes both putative coenzyme binding and active site conserved residues and exhibits a predicted secondary structure motif (alternating alpha-helices and beta-sheet) characteristic of SC-ADH. Bovine heart peptide sequences (174 residues in nine sequences determined by microsequencing) have extensive homology (89% identical residues) with the deduced human heart sequence. The C-terminal third (Asn-194 to Arg-297) shows little sequence homology with the SC-ADH and likely contains elements that determine the substrate specificity for the enzyme including the phospholipid (phosphatidylcholine) binding site(s). Northern blot analysis identifies a 1.3-kilobase mRNA encoding the enzyme in heart tissue.

The complete amino acid sequence of human heart (R)-3-hydroxybutyrate dehydrogenase (EC 1.1.1.30) has been deduced from the nucleotide sequence of cDNA clones. This mitochondrial enzyme has an absolute and specific requirement of phosphatidylcholine for enzymic activity (allosteric activator) and is an important prototype of lipid-requiring enzymes. Despite extensive studies, the primary sequence has not been available and is now reported. The mature form of the enzyme consists of 297 amino acids (predicted M, of 33,117), does not appear to contain any transmembrane helices, and is homologous with the family of short-chain alcohol dehydrogenases (SC-ADH) (Persson, B., Krook, M., and Jornvall, H. (1991) Eur. J. Biochem. 200,[537][538][539][540][541][542][543] (30% residue identity with human 17@-hydroxysteroid dehydrogenase). The first two-thirds of the enzyme includes both putative coenzyme binding and active site conserved residues and exhibits a predicted secondary structure motif (alternating a-helices and @-sheet) characteristic of SC-ADH. Bovine heart peptide sequences (174 residues in nine sequences determined by microsequencing) have extensive homology (89% identical residues) with the deduced human heart sequence. The C-terminal third  to  shows little sequence homology with the SC-ADH and likely contains elements that determine the substrate specificity for the enzyme including the phospholipid (phosphatidylcholine) binding site(s). Northern blot analysis identifies a 1.3-kilobase mRNA encoding the enzyme in heart tissue.
(R)-3-Hydroxybutyrate dehydrogenase (EC 1.1.1.30) is a * This work was supported in part by grants from the Sara Chait Memorial Foundation, the American Heart Association (to A. R. M.), and the American Heart Association Tennessee Affiliate (to J. 0. M.), by Grant DK 14632 from the National Institutes of Health (to S. F.), and by a National Institutes of Health National Research Service Award (to T. M. D.). This is manuscript 110 from the Brookdale Center for Molecular Biology. 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.  T N 37235. Tel.: 615-322-2132;Fax: 615-343-6833. mitochondrial membrane enzyme with an absolute and specific requirement of phosphatidylcholine (PC)' for enzymic activity (1, 2). It has served as a prototype to study the role of lipid and in turn the nature of lipid-protein interactions in membranes (3-6). The enzyme has been purified from bovine heart (7) and rat liver mitochondria (8, 9) and more recently from rat brain (10). The purified enzyme (apodehydrogenase) is devoid of phospholipid and thereby inactive but can be reactivated by reconstitution with phospholipid vesicles containing PC. The apodehydrogenase inserts spontaneously and unidirectionally into preformed phospholipid vesicles or natural membranes, suggesting that it may be amphipathic (11). Optimal activation of the enzyme is obtained only with membranes containing PC (12,13). However, a phospholipid bilayer is not essential for function since soluble PC (below the critical micellar concentration) will activate the enzyme (12,14). Target inactivation analysis indicates that the enzyme is a tetramer in the mitochondrial inner membrane and after reconstitution into phospholipid bilayers (15). The activation of 3-hydroxybutyrate dehydrogenase by bilayer PC appears to involve site-site interaction (16) and an allosteric mechanism in which PC enhances binding of nucleotide (coenzyme) by more than an order of magnitude (17).

The nucleotide sequence(s) reported in this paper
Chemical derivitization studies have provided insight into structure-function relationships for this novel lipid-requiring enzyme; however, progress has been limited by the lack of the primary sequence. In this study, we report the complete amino acid sequence of human heart (R)-3-hydroxybutyrate dehydrogenase as determined by molecular cloning and partial peptide sequence for the bovine heart form of the enzyme.
The resulting peptide mixture was then reduced by adding 1 pl of 0mercaptoethanol and incubating for 30 min at 37 "C; it was subsequently S-pyridylethylated with 3 pl of 4-vinylpyridine, also for 30 min at 37 "C, under argon. Peptides were then separated on a Vydac 4.6-mm C4 column (The Separations Group, Hesperia, CA) as described (18) and sequenced using a modified (19) Applied Biosystems 477 automated sequenator (Applied Biosystems Inc., Foster City, CA). The N-terminal sequence of the enzyme was determined without modification of the protein. An additional peptide, corresponding to ' The abbreviations used are: PC, phosphatidylcholine; SC-ADH, the family of short-chain alcohol dehydrogenases (27); BDH cDNA, cDNA encoding (R)-3-hydroxybutyrate dehydrogenase; BDH mRNA, mRNA encoding (R)-3-hydroxybutyrate dehydrogenase; GAPDH, glyceraldehyde-3-phosphate dehydrogenase; kb, kilobase(s). the C-terminal of the protein, was prepared by cyanogen bromide cleavage, purified by high pressure liquid chromatography, and sequenced using an Applied Biosystems 475A sequenator equipped with a 900A data system controller as described (20,21).
Cloning-Degenerate synthetic oligonucleotide probes were synthesized based on the amino acid sequence obtained from bovine ( R ) -3-hydroxybutyrate dehydrogenase peptides as described above. The four synthetic oligonucleotides (each labeled according to the peptide from which the sequence was derived) were synthesized as follows (degenerate nucleotides are listed in parentheses; I denotes inosine): 1) antisense BDH KC9,5'-AA(G/A) GC(C/T) TC(C/G) ACI CC(G/ A) AA-3', based on the peptide sequence Phe-Gly-Val-Glu-Ala, and the first 2 base pairs of Phe; 2) antisense BDH KC10, 5'-ATI CC(G/ A) GC(G/A) TT(G/A) TT(C/G) ACI A(G/A)I CCC CA-3', based on the peptide sequence Trp-Gly-Leu-Val-Asn-Asn-Ala-Gly-IIe, and the first 2 base pairs of Ser; 3) antisense BDH KC12, 5'-(G/A)TA (G/ A)TA (G/A)TC CAT IGG (G/A)TG (G/A)TA-3', based on the peptide sequence Tyr-His-Pro-Met-Asp-Tyr-Tyr; 4) sense BDH Nbased on the peptide sequence Ala-Ala-Ser-Val-Asp-Pro-Val, and the first 2 base pairs of Gly.
A mixture of these degenerate oligonucleotides was labeled at the 5' end with [Y-~*P]ATP and hybridized to nitrocellulose filters containing 0.5 X lo6 clones from a human heart cDNA library in hZAPII (Stratagene, La Jolla, CA). Hybridization was at 37 "C, and final washing was at 42 'C with 0.2 X SSC (1 X SSC is 0.15 M NaC1, 0.015 M sodium citrate, pH 7.0). Washed filters were autoradiographed for 1-3 days. Isolated clones were sequenced using the dideoxy chain termination method (22). Sequence analysis was performed using the MacVector software on a MacIntosh IIcx, the Intelligenetics sequence analysis programs on a Sun Systems computer, or the program, Protylze (Scientific and Educational Software) on a PC-AT computer.
cDNA Probes-The probe used for Northern blot analysis was the 450-base pair EcoRIIPstI fragment from the 5' region of the BDH cDNA clone (see Fig.  1). All probes were uniformly labeled with random primers using Klenow and [cY-~*P]~CTP to a specific activity of io9 cpm/pg. RNA Preparation and Northern Blot Analysis-RNA was prepared from rabbit tissues using guanidinium-thiocyanate lysis buffer and centrifugation over a cesium chloride cushion (18,22). RNA was sizeseparated on formaldehyde-agarose gels, and Northern blot transfer was carried out overnight using 10 X SSC. Hybridization was at 42 "C overnight, and washing was at 55 "C with 0.2 X SSC. Films were autoradiographed with a single intensifying screen at -70 "C. To assure that equivalent amounts of RNA were present in each lane, the BDH cDNA probe was eluted, and the Northern blots were then hybridized with cDNA encoding glyceraldehyde-3-phosphate dehydrogenase (GAPDH) from chicken muscle.

RESULTS AND DISCUSSION
cDNA and Amino Acid Sequence Nine clones were isolated and sequenced (sequencing strategy is shown in Fig. 1). The largest clone was a 1.3-kb cDNA, which contained the full coding sequence of the mature form of human heart enzyme. The nucleotide sequence and deduced amino acid sequence of the cDNA encoding human heart ( R ) -3-hydroxybutyrate dehydrogenase (BDH cDNA) are shown in Fig. 2. The corresponding amino acid sequence of the The largest clone contained a 1.3-kb cDNA encoding the full sequence of the mature form of the human heart enzyme. The nucleotide length is shown in the top line, and specific restriction enzyme cleavage sites are shown below. Arrows, length and direction of DNA sequencing; both strands were completely sequenced (see Fig. 2). peptides from the bovine heart enzyme, as determined by microsequencing, is shown below the deduced amino acid sequence for the human enzyme. Two bovine peptide sequences determined by Prasad and Hatefi (23) are also included (peptides D and G). In the regions for which bovine heart sequence has been determined (total of 174 residues), there is extensive homology (89% identical residues) between the enzymes from the two species.
The deduced amino acid sequence of the human enzyme (343 residues) indicates the existence of a leader peptide sequence of a t least 46 residues, which is cleaved to produce the mature form of (R)-3-hydroxybutyrate dehydrogenase (297 residues). The location of the cleavage site is based on alignment with the N-terminal peptide of the bovine enzyme. The putative cleavage site is consistent with but not identical to such cleavage sites in other mitochondrial precursor proteins (24). The calculated size of the preprotein is 38 kDa and of the mature form of the enzyme is 33,117 Da. The size of the leader peptide (-5 kDa) is comparable to the difference in size between precursor and mature forms of rat liver ( R ) -3-hydroxybutyrate dehydrogenase measured by immunoprecipitation of polypeptides from both in vitro and in uiuo translation products (26).

Secondary Structure Predictions
Apparent Lack of Transmembrane Helices-The hydrophilicity profile (28) of the mature form of (R)-3-hydroxybutyrate dehydrogenase (Fig. 3A, leader peptide deleted) does not indicate hydrophobic stretches sufficient to span a phospholipid bilayer. Similar hydrophilicity profile patterns were obtained using algorithms for antigenicity (29) or exposed residues (30) (not shown). Three additional methods (31-33) also failed to detect transmembrane helices. With the Eisenberg et al. (31) method, the maximal average hydrophilicity value for a 21-residue segment was 0.42 for positions 21-41 in the mature human enzyme, whereas a value of 0.68 is considered minimal for a transmembrane helix.
Secondary Structure Motif-The predicted secondary structure (a-helix, /?-sheet, turns, and random coil) of the sequence is illustrated in Fig. 3B. The first 200 amino acids are characterized by alternating segments of a-helix and /?-sheet, consistent with the secondary structure motif for this region of the family of short-chain alcohol dehydrogenases (27)   (residues Gly-16, . In addition, the sequence exhibits significant homology with the other highly conserved residues of the SC-ADH. Of the 77 residues that are conserved (ie. identical) in 10 or more of the 20 dehydrogenases (alignment given in Ref. 27), 37 are identical in human 3-hydroxybutyrate dehydrogenase (denoted by an asterisk in Fig. 2). Alignment of the bovine peptides with the sequence of the human enzyme conforms to this homology; 27 of 59 are conserved in the aligned bovine peptides.
The homology of (R)-3-hydroxybutyrate dehydrogenase (from Lys-10 to  with the SC-ADH extends over about the first two-thirds of the complete sequence and encompasses bozh the putative coenzyme-binding and active site domains of this class of dehydrogenases (27). The coenzymebinding site was assigned to the N-terminal halves of the short-chain dehydrogenases and includes Gly-14, Gly-17, and Gly-19, which are conserved in human and bovine (R)-3hydroxybutyrate dehydrogenase (Fig. 2, residues 16, 20 and 22). Putative active site residues, identified as Ser-139, Tyr-152, and Lys-156 in SC-ADH, are also conserved in human (R)-3-hydroxybutyrate dehydrogenase ( Fig. 2; residues Ser-149, . For the complete sequence of human 3-hydroxybutyrate dehydrogenase (mature form), the highest homology (30% identity) is with human 17P-hydroxysteroid dehydrogenase, a SC-ADH which has 327 amino acids. In the region corresponding to the conserved domains of the SC-ADH (Lys-10 to Gly-193)) the homology was 37% whereas the C-terminal domain  shows only 20% identity. Since the first 193 amino acids of (R)-3-hydroxybutyrate dehydrogenase exhibit homology with a variety of dehydrogenases using different substrates, but which are not known to exhibit a phospholipid requirement, it would seem probable that the C-terminal region (residues 194-297) includes regions that form the substrate-and phosphatidylcholine-binding sites in this lipid-requiring dehydrogenase.

Functional Groups of (R)-3-Hydroxybutyrate Dehydrogenase
Chemical derivatization studies suggest that (R)-3-hydroxybutyrate dehydrogenase has essential arginyl, carboxyl, his-1 2 3 4 5 6 7 BDH FIG. 4. Northern blot analysis of BDH mRNA from rabbit tissues. Total RNA was prepared from rabbit tissues, and the 1.3-kb BDH mRNA was detected by hybridization with a 450-base pair PstI-EcoRI fragment of BDH cDNA (see Fig. 1). A single 1.3-kb mRNA was detected in lane 1, fetal rabbit heart a t 25 days gestation (5 days prenatal); lane 2, 4-week rabbit heart; lane 3, adult rabbit heart (total); lane 4, adult rabbit heart atrium; lane 5, adult rabbit heart ventricle; lane 6, adult rabbit fast twitch skeletal muscle; lane 7, adult rabbit slow twitch skeletal muscle. In some samples (e.g. lanes 3, 5, and 7), a small anount of -3.7-kb mRNA can be detected (arrowhead), but this was not reproducible. The lower part of the figure shows the 1.5-kb GAPDH mRNA detected in each lane of the same Northern blot reprobed with GAPDH cDNA from chicken muscle. tidyl, and thiol residues (6, 37-39). The enzyme can also be derivatized with photoactivatable phospholipid analogues (40,41). In the bovine enzyme, there are two dithiol bridges per monomer and two sulfhydryls (42), the more reactive of which (SHl), although not involved in catalysis (43), is essential for optimal function (37, 44). The reactive sulfhydryl appears to be localized in the nucleotide-binding domain of the enzyme since the same residue was derivatized by arylazido-NAD (23,45). There are also 6 cysteine residues in the sequence of mature human (R)-3-hydroxbutyrate dehydrogenase (residues 17,40,81,163,175, and 242). The sequence of the bovine peptide containing the reactive sulfhydryl was determined by Hatefi and colleagues (23,45) (Fig. 2, peptide G) and is homologous with residues 238-254 of the human enzyme sequence (reactive sulfhydryl located at Cys-242). The seqeuence of the N,N'-dicyclohexylcarbodiimide-reactive peptide (23) (Fig. 2, peptide D ) is homologous with residues 117-128 of the human enzyme sequence (essential carboxyl at Glu-120). Thus, the reactive sulfhydryl and the N,N'-dicyclohexylcarbodiimide-reactive glutamic acid, both of which appear to be in the vicinity of the active center of the enzyme (23), are located in separate regions of the sequence. It is to be noted that the sulfhydryl that reacts with arylazido-NAD is in the C-terminal domain whereas the three conserved glycines (residues 16, 20, and 22), suggested to be in proximity to the nucleotide-binding site (27), are in the N-terminal domain. This may indicate that, in the ternary or quaternary structure of (R)-3-hydroxybutyrate dehydrogenase, the Cterminal domain (that appears to be important for binding to membranes) is folded over the N-terminal domain (which is homologous with the family of the SC-ADH). The proximity of the two domains in the ternary or quaternary structure of (R)-3-hydroxybutyrate dehydrogenase may provide for the catalytic specificity of this novel lipid-requiring dehydrogenase.
Tissue Expression (R)-3-Hydroxybutyrate dehydrogenase is known to be widely distributed in tissues of higher animals (25). Northern blot analysis demonstrates that the 1.3-kb mRNA, encoding (R)-3-hydroxybutyrate dehydrogenase (BDH mRNA), is ex-pressed in all forms of muscle (smooth (not shown), fast and slow twitch, and cardiac muscle (Fig. 4)). To control for the amount of mRNA loaded in each lane, the Northern blots were subsequently probed with cDNA encoding GAPDH. BDH mRNA was expressed in rabbit heart a t stable levels from the fetal stage (5 days prenatal) through the adult stage of development (Fig. 4, lanes 1-3). Expression of the BDH mRNA was equivalent in atrium and ventricle (Fig. 4, lunes 4 and 5, respectively). There was higher expression of BDH mRNA in slow twitch versus fast twitch skeletal muscle (normalized for GAPDH expression) (Fig. 4, lunes 7 and 6, respectively), consistent with the higher content of mitochondria in slow twitch as compared with fast twitch skeletal muscle.