Molecular Cloning and Expression of a Mouse Thiamin Pyrophosphokinase cDNA*

Thiamin pyrophosphokinase (EC 2.7.6.2) catalyzes the pyrophosphorylation of thiamin with adenosine 5′-triphosphate to form thiamin pyrophosphate. A mouse thiamin pyrophosphokinase cDNA clone (mTPK1) was isolated using a combination of mouse expressed sequence tag database analysis, a two-step polymerase chain reaction procedure, and functional complementation screening with aSaccharomyces cerevisiae thiamin pyrophosphokinase-deficient mutant (thi80). The predicted protein contained 243 amino acid residues with a calculated molecular weight of 27,068. When the intact mTPK1 open reading frame was expressed as a glutathione S-transferase fusion protein in Escherichia coli lacking thiamin pyrophosphokinase, marked enzyme activity was detected in the bacterial cells. The corresponding 2.5-kilobase pair mRNA was expressed in a tissue-dependent manner and was found at relatively high levels in the kidney and liver, indicating that the mode of expression of mTPK1 genes differs with cell type. The expression ofmTPK1 genes in cultured mouse neuroblastoma and normal liver cells was unaffected by the thiamin concentration in the medium (10 μm versus 3.0 nm). This is the first report on identification of the primary sequence for mammalian thiamin pyrophosphokinase.

and conformations of many of these TDP-dependent enzymes from various cells have been well studied, and the machinery and configuration of TDP as a transient intermediate carrier of the aldehyde group in the enzymes are characterized (6). In diseases such as beriberi and Wernicke-Korsakoff syndrome, which are caused by thiamin deficiency, decreases in the activities of one or more of the TDP-dependent enzymes are thought to account for the impairment of cell functions (7). Moreover, TDP is the precursor for thiamin triphosphate that is suspected to have a specific role in neuronal activity (8,9).
An important aspect of thiamin metabolism in mammals is the regulation of the intracellular levels of the four thiamin forms in a narrow range. The main factors involved in this regulation are membrane-associated thiamin transport systems and the cellular enzyme, thiamin pyrophosphokinase (TPK; EC 2.7.6.2). A number of biochemical analyses suggest that several types of thiamin transport systems exist in mammals, and in several cells the driving force for thiamin uptake appears to be its phosphorylation to TDP (3)(4)(5). Recently, the mammalian thiamin transport protein, THTR-1, was identified as the gene mutated in thiamin-responsive megaloblastic anemia (Online Mendelian Inheritance in Man number: 249270) (10 -12). However, although the enzyme has been purified from several mammalian sources (13)(14)(15), the nucleotide sequences of TPK have been determined only for the microorganisms (16,17) Saccharomyces cerevisiae (THI80), Schizosaccharomyces pombe (tnr3), and Paracoccus denitrificans (pTPK1). The isolation of the cDNA or genomic DNA for TPK from mammalian cells is indispensable for the elucidation of thiamin metabolism and its regulation in higher animals. In this study, we isolated a new mouse cDNA, whose translation product is homologous to the TPKs of the microorganisms, by the combination of expressed sequence tag (EST) database analysis, a two-step polymerase chain reaction (PCR) procedure (18), and functional complementation screening using the yeast TPK-deficient mutant (thi80) (19). The expression experiment in Escherichia coli showed that the cloned cDNA, termed mTPK1, produced the TPK activity. We also investigated the steadystate mRNA levels in various mouse tissues and two cultured cells grown under thiamin-sufficient or -deficient conditions.

EXPERIMENTAL PROCEDURES
Organisms and Cultures-The E. coli strains DH5␣ and BL21 were used to amplify plasmids and express the recombinant protein, respectively. The S. cerevisiae thi80 mutant strain used in this study was T48 -2D (␣ thi80 -1 ura3-52 his3-⌬200 leu2-⌬1 trp1-⌬63) (16). The media and the growth conditions for the yeast and bacterial cells were as described previously (16). Glucose in the yeast minimal medium was replaced with 2% galactose for inducing the transcription from the GAL1 promoter in the yeast expression vector pYES2 (Invitrogen). The mouse neuroblastoma cell line, Neuro 2a (ATCC CCL-131), and the mouse normal liver cell line, NMuLi (ATCC CRL-1638), were purchased from Dainippon Pharmaceutical Co. Ltd.; Suita, Japan. The mouse cells were grown in the special Dulbecco's modified Eagle's medium without thiamin (Life Technologies, Inc.) and supplemented with glucose (6 mg/ml), 5% fetal bovine serum (Equitech-Bio; Ingram, TX), and 2 mM glutamine at 37°C in 5% CO 2 . Under this condition the only source of thiamin was the fetal bovine serum, and the final thiamin concentration was 3.0 nM, which was determined using the method previously described (20); this medium will be referred to as thiamin-deficient medium. Also, thiamin-sufficient medium contained an additional 10 M thiamin hydrochloride (Nacalai Tesque; Kyoto, Japan) than the thiamin-deficient medium.
Cloning of mTPK1-General methods for DNA and RNA manipulation and yeast or bacterial transformation were as described previously (21,22). The deduced amino acid sequence of the S. pombe TPK gene (tnr3) (17) was used to conduct a text-based search of the mouse EST database at the National Center for Biotechnology Information. The sequence information of the obtained mouse clone AA981202 allowed the synthesis of three specific primers, F1 (5Ј-TGACCAAATCATGGC-CTCTGTG), R1 (5Ј-TTCATGCTGAGCAACGAACCAG), and R2 (5Ј-TTAGCTCTTGATGGCCATGGT). PCR using the F1 and R1 primers produced a predicted 409-bp cDNA fragment from the SuperScript mouse 15.5-day embryo cDNA library made with pCMV-SPORT 2 as a cloning vector (Life Technologies, Inc.). Amplification of the 5Ј-region was carried out following the two-step PCR procedure (18) using an Advantage®-HF PCR kit (CLONTECH), which is capable of proofreading activity. The first PCR was performed using only the R1 primer to amplify a single-strand cDNA from the mouse embryo library as a template in a 25-l reaction volume for 40 cycles. Each cycle consisted of denaturation at 94°C for 15 s, annealing at 63°C for 30 s, and extension at 68°C for 3 min. The first PCR product was used as a template in the second run, where pCMV-F (5Ј-TACGCCTGCAGGTAC-CGGT) and R2 were used as primers to amplify the 5Ј-region cDNA. The PCR was carried out for 15 s at 94°C, 30 s at 60°C, and 3 min at 68°C for 40 cycles. The second PCR product was directly ligated with pYES2 at the PvuII site and was subsequently transformed into the E. coli strain DH5␣. Thus, the constructed library was used for the functional complementation analysis using the yeast thi80 mutant. The fulllength cDNA of mTPK1 was obtained from mouse embryo poly(A) ϩ RNA (Ambion) using a SMART RACE cDNA amplification kit (CLON-TECH) according to the manufacturer's instructions. The two specific oligonucleotides used were R4 (5Ј-ACAGTGACAAGGCCGGACCCATC-GTAGG) for 5Ј-rapid amplification of cDNA ends (RACE) and F3 (5Ј-C-TCTCTCATCTACCTCCTCCAACCCGGG) for 3Ј-RACE.
DNA Sequencing-DNA sequencing of plasmid DNAs was performed using a 377 DNA sequencer (Applied Biosystems) with synthesized nucleotides. The sequence of the full-length cDNA was confirmed by sequencing two clones that were generated using independent PCRs. Nucleotides and deduced amino acid sequences were analyzed with programs from the GENETYX software package (Software Development; Tokyo, Japan).

Expression of Recombinant Protein in E. coli-
The intact open reading frame (ORF) of mTPK1 was amplified from pYES2-MT by the primers F2 (5Ј-cgcggatccATGGAGCATGCCTTTACCC), containing a BamHI site, and R3 (5Ј-cattggctcgagGCTCTTGATGGCCATGGT), containing a XhoI site. A bacterial expression vector coding for a glutathione S-transferase (GST)-mTPK1 fusion protein, pGEX-MT was made by inserting the mTPK1 ORF into a BamHI/XhoI site in pGEX-4T-1 (Amersham Pharmacia Biotech). The E. coli strain BL21 bearing pGEX-4T-1 or pGEX-MT was grown overnight in 2 ml of LB medium containing 50 g/ml ampicillin, pelleted, grown for 2 h in 20 ml of fresh medium, and induced for 2 h with isopropyl-␤-D-thiogalactopyranoside at 0.1 mM. The cells were then pelleted, resuspended in 2 ml of phos-phate-buffered saline buffer (pH 7.4) containing 1 mM phenylmethylsulfonylfluoride and 20 g/ml pepstatin A, and sonicated on ice for 10 min with an ultrasonic oscillator (Kubota; Tokyo). Cell debris was removed by centrifugation at 20,000 ϫ g for 20 min, and the supernatant was served as a crude extract.
Blot Analysis-For examination of expression of the mTPK1 gene in mouse tissues, a multiple tissue Northern blot containing 2 g of poly(A) ϩ RNA per lane (CLONTECH) was hybridized with the mTPK1 ORF (see above) and the human ␤-actin cDNA as an internal control that had been labeled with [␣-32 P]dCTP (3,000 Ci/mmol) (ICN Biomedicals) following the random primer labeling procedure. The washed membrane was exposed to X-Omat AR film (Eastman Kodak Co.) for 2 weeks (mTPK1) or 2 days (␤-actin) at Ϫ80°C. The content of mTPK1 mRNA in cultured cells was investigated using reverse transcriptase-PCR. Total RNAs isolated with the RNeasy kit (Qiagen) were treated with deoxyribonuclease (Takara; Otsu, Japan) to eliminate residual genomic DNA. The first-strand cDNA was synthesized from 5 g of total RNA in a total volume of 50 l with the SuperScript preamplification system (Life Technologies, Inc.) using oligo(dT) as a primer. The synthesized cDNA (2 l) was subsequently subjected to PCR with F1 and R1 primers in a total volume of 25 l. After 15 cycles, 5 l of the PCR product was fractionated on a 1.2% agarose gel, blotted onto a Hybond-Nϩ nylon membrane (Amersham Pharmacia Biotech), and hybridized with the mTPK1 probe. The membrane was exposed to the film for 1 day at Ϫ80°C. The expression of glyceraldehyde-3-phosphate dehydrogenase in cultured cells was investigated by Northern blot analysis using 10 g of total RNA as described previously (16). The membrane hybridized with the probe for glyceraldehyde-3-phosphate dehydrogenase (CLONTECH) was exposed for 1 day at Ϫ80°C.
Enzyme Assays-Thiamin-repressible acid phosphatase activity of the thi80 mutant on plates was detected using a staining method based on the diazo-coupling reaction (23). TPK activity was measured using the procedure described previously (16). Briefly, the 1.5 ml of assay mixture containing 0.02 M Tris-HCl (pH 7.5), 0.1 mM thiamin, 1 mM ATP, 1 mM MgCl 2 , and the enzyme source was incubated for 30 min at 37°C. The reaction was stopped with 0.3 ml of 30% trichloroacetic acid. TDP in the reaction fluid was extracted with ethyl ether and determined using high-performance liquid chromatography after conversion to the corresponding thiochrome (24). The protein was determined using a protein assay kit (Bio-Rad) with bovine serum albumin as the standard.

RESULTS AND DISCUSSION
The deduced amino acid sequence of the S. pombe TPK gene, tnr3, was used to search a mouse EST database, and the entry FIG. 1. Cloning strategy of mTPK1 cDNA. For amplification of the 5Ј-region containing the intact ORF, R1 primer was used for the first PCR, and R2 and pCMV-F were used for the second PCR. The products of the second PCR were ligated with yeast expression vector and employed for functional complementation screening with the S. cerevisiae TPK-deficient mutant. The full-length cDNA was obtained following the 5Ј-and 3Ј-RACE methods using R4 and F3 primers, respectively.

FIG. 2. Functional complementation of the yeast thi80 mutant.
A yeast thi80 mutant strain, T48 -2D, bearing pYES2 or the pYES2 derivative was incubated for 3 days on agar medium containing 10 Ϫ6 M thiamin with 2% glucose or galactose as a carbon source and tested for acid phosphatase activity. The plasmid, pYES2-THI80, was constructed by ligating pYES2 with an EcoRI/SphI fragment containing the intact THI80 ORF from pTrc99A-THI80 (16).
AA981202 was identified as a potential homologue. The sequence of the AA981202 EST clone was 411 bp in length, and the predicted translation product had significant sequence similarity with the carboxyl-terminal moiety of the tnr3 protein.
The F1 and R1 primers based on the sequence of AA981202 amplified a predicted 409-bp cDNA from a mouse embryo cDNA library, suggesting that the corresponding cDNA clone was present in this library. A two-step PCR was then performed to isolate the 5Ј-region of the cDNA, as described in Fig.  1. Because the R2 primer used in the second PCR involved the deduced stop codon of AA981202, the second PCR product was expected to contain the intact ORF and was inserted into the yeast expression vector pYES2 downstream of the yeast GAL1 promoter. Thus, the constructed library was employed to isolate the mouse TPK cDNA using functional complementation screening with the S. cerevisiae TPK-deficient thi80 strain. The expression of enzymes involved in yeast thiamin metabolism, such as thiamin-repressible acid phosphatase coded by the PHO3 gene (25), are repressed by thiamin in the medium via the thiamin regulatory system, in which TDP serves as a corepressor (19). However, the expression of PHO3 in the thi80 mutant is not repressed even under high concentrations of thiamin in the medium (Fig. 2), which are caused by an insufficient increase in the intracellular concentration of TDP because of the low levels of TPK activity (16). A thi80 mutant, T48 -2D, was transformed with the pYES2 library and plated on the agar medium containing 10 Ϫ6 M thiamin and 2% galactose without uracil. After incubation at 30°C for 3 days, a total of 800 uracil-positive transformants were examined for the thiamin-repressible acid phosphatase activity using staining. About 100 transformants appeared as white colonies, and 8 colonies showed all the same plasmid, designated pYES2-MT, based on their restriction maps. As shown in Fig. 2, T48 -2D with pYES2-MT, as well as with the yeast THI80 gene, showed a decrease in the phosphatase activity when grown in galactose medium, suggesting that the intracellular concentration of TDP in the thi80 mutant was increased by the product of the mouse cDNA inserted in pYES2-MT.
We identified an 835-bp mouse cDNA in pYES2-MT that contained a single ORF whose ATG translation start codon conformed to Kozak's rules (26). The predicted protein encoded by the ORF contained 243 amino acid residues with a calculated molecular weight of 27,068. This numerical value was similar in size to the determined molecular mass of the purified TPK polypeptide from human red cells (28,000 Da) (13). To ascertain whether this mouse ORF actually encodes TPK, we attempted to express the ORF as a GST fusion protein in E. coli, in which TPK does not exist and TDP is synthesized from thiamin monophosphate by thiamin-monophosphate kinase (EC 2.7.4.16) (27). Fig. 3A shows the patterns of SDS-polyacrylamide gel electrophoresis of the crude extracts of bacterial cells expressing GST or the GST fusion protein. A 53-kDa band, which was almost consistent with the expected size, appeared in the crude extract of cells inducing the fusion protein. After the samples were dialyzed sufficiently, the TPK activities in the crude extracts were determined. As shown in Fig. 3B, marked TPK activity was detected in the cells expressing the GST fusion protein, whereas no enzyme activity was detected in the cells expressing GST alone. When thiamin monophosphate, for which the TPK is completely inactive (1), was used as substrate in place of thiamin, TDP formation was not observed in the cells. Moreover, potassium ions, which are required for thiamin-monophosphate kinase activity of E. coli (27), were not included in the mixture used for TPK activity. It was therefore concluded that this mouse ORF encodes a mouse TPK, and the cDNA was termed mTPK1. Until now almost all purified and recombinant TPKs have had dimeric structures (13)(14)(15)(16). From the comparison of the elution volume of TPK activity with those of standard proteins using gel filtration analysis, the molecular weight of the GST-mTPK1 fusion protein was estimated to be 110,000 (data not shown). This finding suggests that the native mouse TPK also exists in the dimer form of identical subunits. We compared the putative amino acid sequence of the mTPK1 protein with those of TPKs of other organisms such as S. pombe, S. cerevisiae, and P. denitrificans (Fig. 4). The amino acid sequence of mTPK1 showed 39% identity with the carboxyl-terminal half of tnr3, which appears as a fusion protein (569 amino acids) with unknown function at the amino-terminal half (17). The mTPK1 protein was also similar to THI80 and pTPK1 over their full lengths (31% of amino acid identity). Amino acid sequence homology could not be detected between the conserved regions of TPKs and the TDP-dependent enzymes, suggesting that the structural feature in the interaction of amino acids with thiamin differs from that with TDP. It was also not possible to assign the ATP-binding site to any region of TPKs although many types of consensus sequences for peptide segments involved in nucleotide-binding sites are proposed (28). The determination of functional domains in TPK polypeptides may facilitate the comprehension of the transfer mechanism of pyrophosphate from ATP to thiamin.
To determine the expression pattern of the mTPK1 gene, we performed Northern blot analysis on various mouse tissues with the coding region as the probe. As shown in Fig. 5A, a 2.6-kilobase pair single band was detected predominantly in the kidney, followed by the liver, when the membrane was exposed to x-ray film for 2 weeks. The faint bands of the same length were detected in the heart, brain, and testis when analyzed using the bioimaging analyzer system (BAS2000, Fuji Photo Film; Tokyo) (data not shown). The length of these detected bands was about three times longer than that of mTPK1 cDNA inserted in pYES2-MT, suggesting that our cloned cDNA was not the full-length cDNA for the mTPK1 gene. Therefore, the full-length cDNA was isolated using the 5Ј-RACE and 3Ј-RACE methods. The length of isolated transcript from mouse embryo mRNA was 2,563 bp, which was almost consistent with that of the detected bands. The full-length mTPK1 cDNA contained a 5Ј-noncoding region of 56 bp and a long 3Ј-noncoding region of 1.7 kilobase pairs, in which four putative polyadenylation sites, AATAAA (29), were found (nucleotides 1000, 1796, 1800, and 2410. These findings indicated that the mTPK1 gene was expressed in a tissue-specific manner. An unexpected finding was that a very low level of mTPK1 expression was observed in the brain, in which the glucose and energy metabolism are aggressive. However, the distinctive mode of expression of the mTPK1 gene may be present in some areas of the central nervous system. It was previously demonstrated that the expression of the TPK gene (THI80) in S. cerevisiae is partially repressed by thiamin in the medium (16), in which intracellular TDP acts as a corepressor. In mammals, the relationship between the expression of TPK and the provision of thiamin are indistinct. Sanioto et al. (30) reported that TPK activity was reduced in liver and heart muscle of thiamin-deficient rats, whereas Trebukhina et al. (31) reported that the activity was increased transiently in the livers of mice deprived of thiamin. To examine whether the expression of the mTPK1 gene is controlled by thiamin or its phosphorylated derivatives, the effect of thiamin content in the medium on the steady-state mRNA level in mouse neuroblastoma and normal liver cell lines was investigated. Fig. 5B shows the results of a reverse transcriptase-PCR experiment for mTPK1 in two cell lines growing for 2 weeks in thiamin-sufficient (10 M) or -deficient (3.0 nM) medium. As expected from the results of the multiple tissue blot, the relative message level in neuroblastoma cells was about 20% of that in the liver cells by quantitation of radioactivity using the bioimaging analyzer system. However, the mRNA levels of both nerve and liver cells were unaffected by the thiamin concentration of the medium, suggesting that thiamin or a thiamin derivative does not participate in the regulation of mTPK1 in mouse cells. This finding was in contrast to TDP-dependent enzymes, such as transketolase and pyruvate dehydrogenase, whose mRNA levels in human cultured cells are decreased by thiamin deficiency (32).
In conclusion, this is the first report on identification of the primary sequence for mammalian TPK. It was not possible to isolate mammalian TPK cDNA using a conventional functional complementation strategy, possibly because of the very low levels of expression of TPK. The successful cloning revealed in this study appeared to be caused by selective enrichment of the desired clone using the two-step PCR procedure. This study suggests that, using the mTPK1 cDNA, progress will be made in understanding mammalian thiamin metabolism at the molecular level. Recently, we have isolated a human cDNA whose nucleotide sequence was nearly identical to mTPK1, and investigation to clarify the function of this human clone is in progress.
FIG. 5. Expression pattern of mTPK1. A, detection of mTPK1 mRNA in mouse tissues. A Northern blot of multiple mouse tissues was hybridized with 32 P-labeled probes. RNA size markers are indicated on the right. B, detection of mTPK1 transcript in mouse cell lines. The mTPK1 gene expression in mouse cells grown either in thiamin-sufficient (S) or -deficient (D) medium was monitored using reverse transcriptase-PCR assay. The glyceraldehyde-3-phosphate dehydrogenase (G3PDH) mRNA level was determined as an internal control by Northern blot analysis.