Human Pyridoxal Kinase cDNA CLONING, EXPRESSION, AND MODULATION BY LIGANDS OF THE BENZODIAZEPINE RECEPTOR*

Peptide fragments of a porcine benzodiazepine-bind- ing protein were used to isolate the cDNA of a related human protein. The cDNA encodes a polypeptide of 312 amino acid residues that is homologous to a bacterial pyridoxal kinase. Transient expression of the cDNA in human embryonic kidney cells confirmed that it encodes human pyridoxal kinase. The recombinant en- zyme displayed a K m value of 3.3 (cid:109) M for pyridoxal and was inhibited competitively by 4-deoxypyridoxine ( K i (cid:53) 2.8 (cid:109) M ). Benzodiazepine receptor ligands that bound to the purified porcine protein also exerted a potent inhib- itory effect on human pyridoxal kinase activity. Transcripts of the pyridoxal kinase gene were detectable in all human tissues examined, and were particularly abundant in the testes. The gene is localized on chromo- some 21q22.3 and represents a candidate gene for at least one genetic disorder that has been mapped to this region (autoimmune polyglandular disease type 1).

Pyridoxal-5-phosphate (PLP) 1 is an essential cofactor for numerous enzymic reactions of intermediary metabolism (1). Mammals cannot synthesize PLP de novo and require dietary precursors such as pyridoxal, pyridoxamine, and pyridoxine (classified collectively as vitamin B 6 ). Synthesis of PLP from these inactive precursors requires a phosphorylation reaction that is catalyzed by pyridoxal kinase (EC 2.7.1.35). This enzyme activity has also been detected in bacteria where it functions in a related PLP salvage pathway (2).
Dietary precursors of PLP are phosphorylated by pyridoxal kinase in the liver and released to the bloodstream in association with albumin (3)(4)(5). However, circulating PLP must be dephosphorylated by membrane-associated phosphatases before gaining entry to target cells. After diffusing through cell membranes, it is converted back to the active cofactor by intracellular pyridoxal kinase. In consequence, there is a requirement for ubiquitous expression of the kinase in mammalian tissues.
Although the major pathways of vitamin B 6 metabolism have been established for many years, there remains little understanding of how PLP homeostasis is controlled in mammals. Recently, valuable insights to this problem have been provided by genetic manipulation of an enzyme involved with PLP dephosphorylation (6). However, similar studies on the role of pyridoxal kinase have not been possible owing to the absence of any identified pyridoxal kinase genes from eukaryotic sources.
Previously, a cytosolic protein of unknown function was purified from porcine brain by benzodiazepine-affinity chromatography (7). Here, peptide fragments of this protein were used to isolate a homologous human cDNA. This cDNA encodes a protein that exhibits both the physical and enzymic properties of pyridoxal kinase. The activity of the recombinant enzyme was shown to be modulated by ligands of mammalian benzodiazepine receptors. In addition, the cDNA sequence was used to determine the tissue-expression pattern and chromosomal location of the human pyridoxal kinase gene.
Isolation of a Human Pyridoxal Kinase cDNA (PKH)-Purified P36 protein (7) was cleaved with cyanogen bromide, and three peptides were resolved by reverse-phase high performance liquid chromatography using standard methodology (8). Each of the purified peptides was coupled to D-phenylenediisothiocyanate glass, and subjected to solidphase Edman degradation as described previously (9). The three P36 peptide sequences were searched against the human cDNA data base (HCD) (10), using the TBLASTN algorithm (11). All three peptides displayed significant homology with the same translated cDNA sequence (see Fig. 1). This cDNA assembly (1210 bp) contained an open reading frame of 936 bp from the first in-frame methionine residue (see Fig. 1). The open reading frame was amplified from human testes cDNA (CLONTECH) using the primers, 5Ј-ccggccctcgaggatCCAGGCCCG-GCATGGAGGAGGAGT and 5Ј-ccggccatgcatCAGGGACAAACACG-GAGACACCAA. Amplification at 95°C for 45 s, 60°C for 1 min, and 72°C for 2 min was performed for 35 cycles using the XL PCR system (Perkin-Elmer). Reaction products were purified from agarose gels and either sequenced directly or ligated into pCDM8 (Invitrogen) for expression studies. The cloned cDNA was sequenced over its entire length to ensure that no mutations had been introduced during amplification.
Expression of the PKH cDNA-Human embryonic kidney cells (HEK-293) were maintained in Dulbecco's modified Eagle's medium, supplemented with calf serum (10%), penicillin (50 units/ml), and streptomycin (50 g/ml). Cells were plated at ϳ20% confluence 24 h prior to transfection. Transfection of the PKH cDNA, cloned in pCDM8, was performed with a calcium phosphate-DNA precipitate in HEPES buffer (8). Following incubation with the precipitate for 24 h, the cells were washed and cultured for a further 48 h before harvesting. Transfection efficiency was assessed by using pCMV␤ (CLONTECH) for expression of ␤-galactosidase activity in the transfected cells (12). Prior to harvesting, the adherent cells were washed five times with ice-cold phosphatebuffered saline (10 mM sodium phosphate, 140 mM sodium chloride, 5 mM potassium chloride, pH 7.4). The cells were then scraped in phosphate-buffered saline, collected by centrifugation (1000 ϫ g, 5 min, * 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.
Pyridoxal Kinase Assay-The pyridoxal kinase activity of cell extracts was measured by a fluorometric assay (13). Briefly, cell extracts (5-10 g of protein) were incubated at 37°C in 200 l of a substrate solution containing 0.1 M potassium phosphate (pH 6.4), 1 mM ATP, 0.1 mM zinc chloride, and 1-20 M pyridoxal. After 60 min, 40 l of 6 mM hydoxylamine was added, and the fluorescence was measured using a CytoFluor 2350 fluorometer (Millipore). Excitation and emission wavelengths were 360 and 460 nm, respectively. A unit of enzyme activity is defined as the amount of protein that catalyzes the formation of 1 nmol of PLP/min at 37°C. Preparatory studies demonstrated that the rate of PLP production was constant for at least 60 min under these assay conditions.
Northern Blot Analysis-The open reading frame of the PKH cDNA ( Fig. 1) was labeled with 32 P using the NEblot random priming system (New England Biolabs). This probe was hybridized with human RNA blots (CLONTECH) according to manufacturer instructions. The blots were washed at 60°C in 0.1 ϫ SSC, 0.1% SDS prior to exposure. Blots were stripped of probe by boiling in 0.5% SDS and re-hybridized with a 32 P-labeled fragment of the human glyceraldehyde-3-phosphate dehydrogenase cDNA (nucleotides 789 -1140) (14).

RESULTS
Previously, a cytosolic protein (P36) of unknown function was purified from porcine brain by benzodiazepine-affinity chromatography (7). A purified preparation of this protein was cleaved with cyanogen bromide, and three peptide fragments were resolved by high performance liquid chromatography. The sequences of these peptides (see Fig. 1) were used to search the HCD data base of translated human cDNAs. The cDNA sequences in HCD are assembled from overlapping expressed sequence tags (ESTs) that were derived from a random selection of human cDNA clones (10). All three peptides displayed homology to the same translated cDNA assembly (86 -94% identity). This assembly (1210 bp) contains a large open reading frame encoding a polypeptide of 312 amino acid residues (Fig. 1). The sequence of the open reading frame was confirmed after obtaining an independent cDNA clone from human testes mRNA.
The sequence of the open reading frame was used to search data bases of DNA and protein sequences. This search identified many anonymous fragments of human DNA that display exact identity with the cDNA sequence (see below). In addition, several genes of non-human origin were found to exhibit significant homology (Table I). Of these, the most similar sequences are hypothetical proteins of unknown function, derived from yeast or bacteria. However, it was also possible to detect significant homology with the pdxK gene of Escherichia coli (27% amino acid identity with 5 gaps, Fig. 2A). This gene encodes a pyridoxine/pyridoxal/pyridoxamine kinase that functions in a salvage pathway of vitamin B 6 metabolism (15). An enzyme with similar properties is known to exist in eukaryotic cells (16,17) though the genes that encode it have not yet been identified. Preparations of pyridoxal kinase have been purified from several mammalian sources (17)(18)(19) and have led to the isolation of two peptide fragments from the ovine enzyme (20,21). Significantly, both of these peptides are homologous to the translated cDNA sequence (Fig. 2B). These sequence comparisons indicate that the cloned cDNA encodes a human homologue of pyridoxal kinase (termed PKH).
To examine the PKH protein for enzymic activity, the cDNA was expressed transiently in human embryonic kidney cells. These cells normally contain only low levels of pyridoxal kinase activity (Ͻ0.01 units/mg of protein at 20 M pyridoxal). However, this activity was increased by more than 100-fold following transfection with the PKH cDNA (1.2-1.5 units/mg of protein). The recombinant enzyme displayed a K m value of 3.3 M for pyridoxal (Fig. 3A). It was inactive in the presence of 5 mM EDTA or in the absence of ATP. Enzymic activity was inhibited by 4-deoxypyridoxine (K i ϭ 2.8 M). This inhibition was competitive with respect to pyridoxal (Fig. 3A).
Previously, the porcine homologue of PKH was purified to homogeneity by virtue of its specific interaction with an immobilized benzodiazepine, 1012S. This interaction could be blocked by low concentrations (10 -100 M) of ␤-carbolines or free 1012S but was insensitive to all other benzodiazepinereceptor ligands that were examined (7). In the presence of a high concentration of pyridoxal (20 M), both 1012S and ethyl-␤-carboline-3-carboxylate were potent inhibitors of the human enzyme (IC 50 ϭ 2 and 5 M, respectively; Fig. 3B). Inhibition was not competitive with respect to either pyridoxal or ATP but displayed a complex pattern that could not be explained adequately by simple noncompetitive or uncompetitive mechanisms (data not shown). Other ligands of benzodiazepine receptors, such as flunitrazepam and PK-11195, had little effect on recombinant PKH activity (Fig. 3B).
Expression of the PKH gene in human tissues was assessed by probing mRNA blots with the PKH cDNA at high stringency (Fig. 4). A major transcript of 1.5 kb was detectable in all tissues examined, and was particularly abundant in the testes. Two larger transcripts (4.4 and 7.9 kb) were also detectable in most tissues. In brain, the 7.9-kb transcript was the major hybridizing species. However, in all other tissues examined, the larger transcripts were of relatively low abundance. The GenBank TM data base of DNA sequences contains several accessions that are exact matches to fragments of the PKH cDNA. These anonymous sequences are either human ESTs, sequence-tagged sites, products of exon-trapping, or large fragments of genomic DNA. Several of the accessions are known to be derived from chromosome 21q22 (e.g. Z47290) and thereby provide the location of the PKH gene. Furthermore, efforts by Stanford Human Genome Center to sequence a potential disease locus at chromosome 21q22.3 have revealed a BAC clone from this region (BAC B159G9) that contains almost the entire PKH gene on 10 distinct exons. 2

DISCUSSION
This study describes the first example of a eukaryotic gene that encodes pyridoxal kinase activity. Identification of the PKH gene arose from analysis of an unusual benzodiazepinebinding protein (P36) that was isolated previously from porcine brain (7). The PKH protein is the human homologue of P36 and displays significant sequence similarity to a pyridoxal kinase of E. coli. Regions of the PKH protein are also homologous to peptide fragments of purified ovine pyridoxal kinase. Together, these data indicate that PKH and P36 are human and porcine pyridoxal kinases, respectively. Previously, P36 was detected in the cytosolic fraction of several porcine tissues and was demonstrated to consist of two subunits (each ϳ36,000 Da) (7). These properties are consistent with the reported properties of purified ovine pyridoxal kinase (18).
To provide conclusive identification, the PKH cDNA was expressed transiently in human embryonic kidney cells. The pyridoxal kinase activity of these cells was increased by more than 100-fold following transfection with the PKH cDNA. The recombinant enzyme displayed a K m value for pyridoxal (3.3 M) that is within the range of values reported for mammalian preparations of pyridoxal kinase (3-50 M) (16,17,22). Competitive inhibition by 4-deoxypyridoxine and a requirement for ATP and divalent cations are properties of PKH that are also consistent with mammalian pyridoxal kinases (16,17).
Previous studies of the porcine P36 protein, demonstrated that it could bind to an immobilized benzodiazepine-receptor ligand, 1012S. This binding could be blocked by low concentrations of ␤-carboline derivatives but not by other structural classes of benzodiazepine-receptor ligands (7). It was therefore predicted that 1012S and ␤-carbolines, but not flunitrazapam or PK-11195, would bind to recombinant PKH. Consistent with this prediction, 1012S and ethyl-␤-carboline-3-carboxylate were found to be potent inhibitors of PKH activity while flunitrazapam and PK-11195 had no effect. The site through which this inhibition is mediated appears to be distinct from the binding sites for pyridoxal or ATP and may represent a new target to probe the structural requirements for PKH catalysis. It also provides a convenient means for purification of pyridoxal kinase.
Consistent with its ubiquitous role in vitamin B 6 metabolism, pyridoxal kinase activity has been detected in a wide variety of mammalian tissues (16), and transcripts of the PKH  (15) is aligned with the deduced amino acid sequence of the human cDNA (PKH). Conserved residues are indicated by asterisks above the pdxK sequence. B, peptide fragments (PKS_PEP1 and 2) that were derived from purified ovine pyridoxal kinase (20,21) are aligned with a fragment of the PKH peptide sequence (residues 226 -269). Conserved residues are indicated by vertical bars.
gene were detected in all tissues examined (Fig. 4). The authors are unaware of any previous studies in which the pyridoxal kinase activity of testes has been examined. The very high level of hybridizing mRNA in this tissue indicate that it may be a rich source of the enzyme. For the majority of tissues examined, the most abundant species of hybridizing transcript is approximately 1.5 kb in length. However, larger transcripts (4.4 and 7.9 kb) were also detectable in most tissues and were particularly abundant in the brain. These mRNAs may reflect the use of alternative polyadenylation sites or the existence of highly homologous genes. Human homologues of the PKH cDNA were not detected in current EST data bases. However, it may be relevant that an antiserum raised against porcine P36 crossreacted with a larger protein (M r ϭ 73,000) that is located in brain membranes (7). Future screening of human genomic DNA libraries with the PKH cDNA should help to resolve the question of potential gene homologues.
The PKH gene is located on chromosome 21q22.3. Consistent with this finding, individuals with chromosome 21 trisomy (Down's syndrome) have been reported to display increased pyridoxal kinase activity (23). Although vitamin B 6 metabolism is known to be altered in Down's syndrome (24), its relevance to symptoms of the disorder is unknown at present. A genetic disease of unknown etiology, autoimmune polyglandular disease type 1, has also been mapped to chromosome 21q22.3 (25,26). Although there is no obvious connection between vitamin B 6 metabolism and this unusual disorder, the PKH gene is clearly a candidate for mutational analysis in affected families.
Identification of the PKH gene now permits detailed analyses of pyridoxal kinase function at both the molecular and physiological levels. Site-directed mutagenesis will complement studies of protein-modification (20,21) that have begun to define the structural requirements for catalytic activity. The future use of transgenic animals to explore the physiological consequences of disrupting the function of pyridoxal kinase is also likely to provide valuable insights to its role in the control of vitamin B 6 metabolism.