Human Pyridoxal Phosphatase

Pyridoxal phosphatase catalyzes the dephosphorylation of pyridoxal 5′-phosphate (PLP) and pyridoxine 5′-phosphate. A human brain cDNA clone was identified to the PLP phosphatase on the basis of peptide sequences obtained previously. The cDNA predicts a 296-amino acid protein with a calculated Mr of 31698. The open reading frame is encoded by two exons located on human chromosome 22q12.3, and the exon-intron junction contains the GT/AG consensus splice site. In addition, a full-length mouse PLP phosphatase cDNA of 1978 bp was also isolated. Mouse enzyme encodes a protein of 292 amino acids with Mr of 31512, and it is localized on chromosome 15.E1. Human and mouse PLP phosphatase share 93% identity in protein sequence. A BLAST search revealed the existence of putative proteins in organism ranging from bacteria to mammals. Catalytically active human PLP phosphatase was expressed in Escherichia coli, and characteristics of the recombinant enzyme were similar to those of erythrocyte enzyme. The recombinant enzyme displayed Km and kcat values for pyridoxal of 2.5 μm and 1.52 s–1, respectively. Human PLP phosphatase mRNA is differentially expressed in a tissue-specific manner. A single mRNA transcript of 2.1 kb was detected in all human tissues examined and was highly abundant in the brain. Obtaining the molecular properties for the human PLP phosphatase may provide new direction for investigating metabolic pathway involving vitamin B6.


EXPERIMENTAL PROCEDURES
Materials-We purchased a human brain cDNA library as well as a dot blot array containing poly(A) ϩ RNAs from human tissues from Clontech (Palo Alto, CA). Restriction endonuclease and other cloning reagents were purchased from New England Biolabs, Inc. (Beverly, MA) or Promega. Double-stranded DNA probes were radiolabeled with [␣-32 P]dCTP (3000 Ci/mmol) from Amersham Biosciences using a commercial random priming kit (Amersham Biosciences). All other reagentgrade chemicals were obtained from standard suppliers.
Identification and Cloning of Human Brain PLP Phosphatase (hPLPP) cDNA-BLAST searches, conducted with PLP phosphatase peptides (AQGVLFDCDGVL and AVLVGYDEHFSFAK) as query sequences, revealed a full match with the predicted amino acid sequence of a cDNA (GenBank TM /EBI accession number NM020315). This clone was used to design PCR primers for the cloning of human PLP phosphatase cDNA.
A multiple protein sequence alignment of the human PLP phosphatase clone, along with the sequences most closely related to it, was performed using the ClustalW (version 1.74) program (23), and the resulting aligned sequence were shaded using Boxshade. Identification of the chromosomal localization of the human PLP phosphatase was performed by a BLASTN search of the human genome sequences of the National Center for Biotechnology Information (NCBI).
Expression in E. coli and Purification of Human PLP Phosphatase-The PLP phosphatase cDNA was cloned between the BamHI of the bacterial expression vector pQE30 (Qiagen) after PCR amplification. Transformants of E. coli M15/pRER4 with the resulting pQE30-hPLPP construct were grown at 37°C in 200 ml of LB medium with 100 g/ml ampicillin and 25 g/ml kanamycin. The plasmid pREP4 constitutively expresses the Lac repressor protein encoded by the lacI gene to reduce the basal level of expression (Qiagen). When that culture had grown to an A 600 of 0.6, isopropyl ␤-D-thiogalactopyranoside was added to a final concentration of 1 mM. After inducing the expression of the PLP phosphatase protein for 16 h at 25°C cells were harvested, washed, and resuspended in 20 ml of 20 mM Tris-HCl buffer, pH 7.4, containing 200 mM NaCl and 20 mM imidazole.
The cell suspension was sonicated, and the lysate was cleared by centrifugation at 10,000 ϫ g and 4°C for 30 min. The supernatant was then poured into the column loaded with the nickel-nitrilotriacetic acid agarose (Qiagen), washed with Tris buffer containing 40 mM imidazole, and eluted protein was eluted with 200 mM imidazole. The purity of the eluted protein was evaluated by 12% SDS-PAGE using Coomassie blue staining to visualize the protein.
Enzyme Assay and Kinetic Characterization-The enzymatic activity of PLP phosphatase was measured at pH 7.4 in 40 mM triethanolamine-HCl. The rate of production of pyridoxal from PLP was measured by following the decrease in absorbance at 390 nm for at least 3 min. The initial velocity data were fitted by a least-squares method to the Lineweaver-Burk transformation of the following equation, where [S] represents the concentration of the varied substrate PLP, and K m represents the Michaelis constant.
A molecular weight of 64,000 for PLP phosphatase with a specific activity of 1.4 mol⅐min Ϫ1 mg Ϫ1 was used in the calculations of molar enzyme concentrations. One unit of specific activity is defined as the amount of protein that catalyzes the formation of pyridoxal/min from PLP. PLP is known to have an extinction coefficient of 4900 cm Ϫ1 M Ϫ1 at pH 7. Protein concentrations were determined by the Bradford assay using the reagent and micro procedure from Pierce Chemical Co. Bovine serum albumin was used as a standard.
In addition, hydrolysis of all other substrates was measured in 40 mm triethanolamine-HCl, pH 7.4, containing 4 mM MgCl 2 . The release of free phosphate from potential substrates was measured colorimetrically as the molybdate complex with malachite green (16,24).
Multiple Tissue Expression Array-A Northern filter containing 8 human tissue-specific poly(A) ϩ RNAs and a dot blot array containing human poly(A) ϩ RNAs from various adult tissues, fetal tissues, and cancer cell lines were prehybridized at 65°C for 1 h in ExpressHyb TM hybridization solution (Clontech). The filters were then hybridized at 65°C for 16 h with 32 P-labeled cDNA probe containing the complete ORF. After washing as recommended by the manufacturer, blots were exposed to x-ray films at Ϫ70°C with an intensifying screen for the appropriate time period. For scanning densitometry, the blot was scanned and BioLab Image software was used to quantify the signals. Blots were reprobed with ␤-actin as a loading control.

RESULTS
Identification and Cloning of the PLP Phosphatase cDNA from Human and Mouse-Using two tryptic peptide sequences from the erythrocytes PLP phosphatase previously reported by Gao and Fonda (21), we conducted NCBI BLAST searches of the human EST data base. A cDNA clone encoding both peptide sequences, designated a hypothetical protein dj37E 16.5 from lung carcinoma (GenBank TM /EBI accession number NM020315) was identified. When the tryptic peptide sequences were aligned with the deduced amino acid sequence there was a perfect match, confirming that the cDNA in the EST GenBank TM data base codes for PLP phosphatase.
We used the sequence information for designing specific primers for PCR amplification utilizing a human brain cDNA library. Sequencing of this clone revealed that the brain cDNA is identical to the hypothetical protein from lung carcinoma. The ORF encodes a 296-amino acid protein with a molecular mass of 31698 Da. A computer calculation reveals that the isoelectric point for the protein is 6.12. ScanProsite software analysis by ExPASy showed that the deduced human protein has the following putative post-translational modification sites: 2 N-glycosylation sites, 7 phosphorylation sites, and 4 N-myristoylation sites.
In addition, we cloned a complete mouse homologue by 5Ј-RACE and 3Ј-RACE-PCR using an EST sequence (GenBank TM / EBI accession number AK043228). As shown in Fig. 1, the longest cDNA contains 1978 bp consisting of a 879-bp ORF, a 13-bp 5Ј-untranslated region, and a 1086-bp 3Ј-noncoding region. The 3Ј-end of the sequence contains a poly(A) stretch, preceded by a putative polyadenylation signal AATAAA. The open reading frame encodes a 292-amino acid protein with a predicted molecular mass of 31512 Da and pI of 5.53. The deduced protein has a putative N-glycosylation site, 5 phosphorylation sites, and 4 amidation sites.
Using BLAST searches of the EST data bases, we found putative PLP phosphatases in various species. Multiple alignment of the protein sequence of PLP phosphatase with other homologous proteins or putative proteins was performed using the ClustalW program as shown in Fig. 2. All of the PLP phosphatases are highly homologous at the amino acid sequence level. The mouse PLP phosphatase sequence shows a high degree of similarity (94% identity and 98% similarity) with the human enzyme. The sequence identities of human PLP phosphatase to the hypothetical proteins from Danio rerio, Drosophila melanogaster, Caenorhabditis elegans, Arabidopsis thaliana, Saccharomyces cerevisiae, and E. Coli are 36, 28, 26, 25, 31, and 27%, respectively, and similarities share 77, 65, 68, 60, 70, and 66%, respectively. This result suggests that PLP phosphatase is highly homologous and widely expressed throughout evolution. The tryptic sequences from human erythrocytes enzyme that are perfectly matched with the deduced amino sequence are indicated by asterisks in Fig. 2.
Genomic Organization and Chromosomal Localization-Using PLPP cDNA as a query sequence, a BLAST analysis of the genomic sequence available through NCBI Web site mapped the PLP phosphatase gene to human chromosome 22q12.3. Mouse PLP phosphatase gene is located on chromosome 15.E1 with a very similar genomic organization to that observed in human. The ORF of PLP phosphatase is encoded by two exons FIG. 1. Nucleotide and deduced amino acid sequence of a full-length cDNA encoding mouse PLP phosphatase (GenBank TM accession number AY366300). The predicted amino acid sequence of PLP phosphatase is shown below the nucleotide sequence. The termination codon is marked by an asterisk, and the polyadenylation signal is double underlined. Putative glycosylation sites (g), phosphorylation sites (p), and N-myristoylation sites (m) in mouse PLP phosphatase are indicated. A potential exon/intron junction is also indicated by a slash. The numbers on the right refer to nucleotide and amino acid positions. as shown in Fig. 3. The intron/exon junctions followed the GT/AG rule (25) and were flanked by conserved sequences. The lariate branch-point in the consensus sequence is located at 23 nucleotides upstream of the 3Ј splice site. The possible exon/ intron junction site in mouse cDNA is also indicated by a slash in Fig. 1.
Functional Expression and Catalytic Properties of the Human Recombinant Enzyme-After E. coli M15/pQE30-hPLPP was induced with isopropyl ␤-D-thiogalactopyranoside, the His 6 -tagged recombinant PLP phosphatase yielded a major band with an apparent molecular mass of 32 kDa on an SDS-PAGE gel corresponding to about 15% of the total applied protein (Fig. 4). Most of the expressed protein was found in the soluble extract after lysis of the cells. As shown in Table I, the crude extraction from 200 ml of cells contained ϳ14.5 mg of total protein and 0.14 units/mg of PLP phosphatase activity. Purification by binding to Ni 2ϩ -nitrilotriacetic acid-agarose resin and elution with buffer containing 200 mM imidazole resulted in a near homogeneous preparation. The purified recombinant protein had a specific activity of 1.4 units/mg and appeared as a single band with a molecular mass of 32 kDa agreeing with the value calculated from the deduced sequence (Fig. 4, lane 6).
Activity of the human PLP phosphatase toward PLP, PNP, and PMP was determined (Table II). Recombinant human enzyme followed Michaelis-Menten kinetics with respect to PLP, and the Michaelis constant and k cat value for this reactant was 2.5 M and 1.52 s Ϫ1 , respectively. K m for PNP and PMP were 43 and 81 M, respectively, and k cat values were 1.25 and 0.45 s Ϫ1 , respectively, whereas activity with p-nitrophenyl phosphate was very low. Thus, the specificity constant of the recombinant PLP phosphatase was highest with PLP followed by PNP. The recombinant enzyme requires a divalent ion for activity, and Mg ϩ2 is the most effective in catalyzing the dephosphorylation of PLP.
To investigate the state of aggregation of recombinant human PLP phosphatase, the purified enzyme was examined by FPLC gel filtration at pH 7.4 in 50 mM triethanolamine-HCl buffer. The elution profile of the enzyme exhibits a symmetrical peak characterized by a molecular mass of about 60 kDa when the concentration of the recombinant protein in the fraction is around 10 M (data not shown). These data clearly indicate that the recombinant enzyme is a dimer, agreeing with previous data for the erythrocyte enzyme.
Tissue-specific Expression of Human PLP Phosphatase-Dot blot array and Northern blot analyses were used to determine the expression pattern for PLP phosphatase in human tissues. As evident by multiple tissue expression (MTE TM ) array analysis, PLP phosphatase transcripts were detected in most tissues, indicating ubiquitous expression. However, the level of expression was markedly variable. PLP phosphatase was most highly represented in all the regions of central nerve system except the spinal cord (Fig. 5). The other major sites of expression were found in liver and testis. In fetus, expression levels of PLP phosphatase transcript showed a rather even distribution in all organs except brain. Like adult, fetal brain expressed PLP phosphatase transcript with a very high level. To determine the size of mRNA transcripts, Northern blots analyses were performed. A single 2.1-kb transcript was detected with variable intensity in all adult tissues examined: heart, brain, placenta, lung, liver, skeletal muscle, kidney, and pancreas (data not shown).

DISCUSSION
In this report, we describe the first identification and characterization of cDNAs encoding a PLP phosphatase from human and mouse. A human cDNA clone of PLP phosphatase was identified by NCBI BLAST search on the basis of the amino acid sequences of two peptides from the purified human erythrocytes enzyme. Several lines of evidence confirm that the cDNA is indeed encoding the PLP phosphatase. First, the amino acid sequences of the two peptides of the purified protein are contained in the amino acid sequence deduced from the ORF (Fig. 2). Second, the molecular mass of the protein of 32 kDa calculated from the deduced cDNA sequence is identical to the values from SDS-polyacrylamide gel electrophoresis for the purified erythrocyte protein and the recombinant protein.
Third, the kinetic parameters of the recombinant protein are similar to those of purified enzyme. The K m values for the substrate PLP (2.5 M) are practically indistinguishable from those of the erythrocyte (1.5 M) (16).
Human PLP phosphatase cDNA encodes a 296-amino acid protein with a predicted molecular mass of 31698 Da. The hydropathy profile of PLP phosphatase indicated a rather even distribution of hydrophobic and hydrophilic regions throughout the molecule (data not shown). Therefore, no evidence for integral membrane domains in PLP phosphatase was found. The isolated cDNA contains a short 5Ј-untranslated segment similar to lung carcinoma cDNA. No larger product has been generated in repeated RACE experiments using different primer sets. Furthermore, the 2.1-kb message determined by Northern analysis agrees well with the size of the full-length cDNA, suggesting that the 5Ј-end of the cDNA is at or near the transcription start site. Therefore, the cloned human brain cDNA may be complete. By contrast, analysis the EST clone for mouse PLP phosphatase shows that it was incomplete. The mouse EST sequence (GenBank TM /EBI accession number AK043228) contains several unidentified nucleotides and a stop codon in ORF, but no polyadenylate tail. Thus, we cloned a full-length cDNA of mouse PLP phosphatase by 5Ј-and 3Ј-RACE-PCRs, and the sequence has been deposited in GenBank TM /EBI as accession number AY366300.
Comparison of the identified cDNA sequences with the genome sequences revealed the genomic structure and chromosomal localization. The PLP phosphatase gene is divided into two exons, and the splice site sequences are remarkably well conserved as shown in Fig. 3. In mammals two ends of the intron are conserved by a donor-site consensus sequence AG/GURAGU and an acceptor-site consensus sequence (Y)nNCAG/G, where R, Y, and N indicate purine, pyrimidine, and any nucleotide, respectively (25,26). A branch-point consensus sequence, YNCUGAC, is usually about 15-40 nucleotides upstream of the 3Ј splice site, and the adenosine in the sequence forms the 2Ј-5Ј linkage in the lariat structure. In addition, the genomic sequences were examined for the pres- FIG. 5. Tissue-specific expression of the human PLP phosphatase mRNA. The human multiple tissue expression (MTE TM ) array was hybridized with a 32 Plabeled human PLP phosphatase-specific cDNA probe. Tissue sources for the RNA are indicated below the blot. ence of CpG islands using the CpG plot program in European Bioinformatics Institute (EBI), as defined by Gardiner-Garden and Frommer (27). The hPLPP gene contains a CpG island with a CG obs /CG exp ratio in excess of 0.8 and a GC content of 74% spanning a region from 343-bp upstream to 614-bp downstream of the start codon, whereas the mPLPP gene has a CpG island extending from Ϫ282 bp to ϩ643 bp with a CG content of 71%.
We also reported here the functional expression of human PLP phosphatase. The expressed protein was purified by a one-step affinity chromatographic method, and consequently endogenous PLP phosphatase produced by E. coli was eliminated. Enzymological analysis of the recombinant human protein expressed from this cDNA clone showed a high specificity for the hydrolysis of PLP. Characteristics of the recombinant enzyme were similar to those from the native enzyme purified from human erythrocytes, except kinetic values for PNP (16). Noteworthy is the relatively high V max and K m with PNP as a substrate. As shown in Table II, the K m values of recombinant enzyme for PNP and PLP were 43 and 2.5 M, respectively, whereas the V max values for PNP were surprisingly high as 1.2 units/mg, which is almost same to the value for PLP (1.4 units/mg). Judging from the kinetic results, the catalytic efficiency (k cat /K m ) for PNP was 20-fold lower than that of PLP. On the other hand, the kinetic values for the native enzyme were somewhat different (16). The enzyme had relatively low K m and V max values with PNP. K m values with respect to PNP and PLP were 5.2 and 1.5 M, respectively, whereas the V max were 3.2 units/mg and 0.7 units/mg, respectively. In consequence, catalytic efficiency value for PNP was 17-fold low than that of PLP, which is practically indistinguishable from that of the recombinant enzyme.
An interesting aspect of our work is the finding that PLP phosphatase is differentially expressed in a tissue-specific manner. Based on the levels of mRNA expression, the widespread distribution of PLP phosphatase in human tissues is consistent with its essential role in cellular metabolism. In addition to its housekeeping role in PLP metabolism, however, PLP phosphatase may play a more specific role especially in brain, where its level is substantially higher. In this connection, it is should be noted that pyridoxal kinase is expressed in essentially all mammalian organs (4), whereas PNP oxidase is selectively present with the highest activity found in liver and kidney (28).
In conclusion, the results obtained in this study will provide valuable information on the action mechanism of human PLP phosphatase and the physiological role of this enzyme. The availability of the cDNAs encoding PLP phosphatase, and the recombinant enzyme will be helpful to expand our view of metabolic pathways that could involve PLP phosphatase and to evaluate its functional role in human cellular homeostasis.