Genomic organization, tissue distribution and deletion mutation
of human pyridoxine 5¢-phosphate oxidase
Jeong Han Kang
1
, Mi-Lim Hong
1
, Dae Won Kim
2
, Jinseu Park
2
, Tae-Cheon Kang
3
, Moo Ho Won
3
,
Nam-In Baek
4
, Byung Jo Moon
1
, Soo Young Choi
2
and Oh-Shin Kwon
1
1
Department of Biochemistry, College of Natural Sciences, Kyungpook National University, Taegu, Korea;
2
Department of Genetic
Engineering, Division of Life Sciences, and
3
Department of Anatomy, College of Medicine, Hallym University, Chunchon, Korea;
4

were studied by creating sequential truncation mutants. Our
results showed that deletion of the N-terminal 56 residues
affects neither the binding of coenzyme nor catalytic activity.
Keywords: deletion mutation; genomic organization; PNP
oxidase; polyadenylation; tissue distribution.
Pyridoxal 5¢-phosphate (PLP), the metabolically active form
of vitamin B
6
, is a required coenzyme for numerous
enzymes involved in amino acid metabolism [1]. The
functions of PLP include coenzymatic participation in
reactions leading to the formation of several neurotrans-
mitters [2]. Moreover, it appears that PLP modulates
steroid–receptor interactions and is involved in the regula-
tion of immune function [3]. The enzymes that are
conventionally involved in vitamin B
6
metabolism are an
ATP-dependent pyridoxal kinase (PDXK; EC 2.7.1.35)
[4,5], FMN-dependent pyridoxine 5¢-phosphate oxidase
(PNPO, EC 1.4.3.5) [6,7] and pyridoxal phosphatase
(PDXP, EC 3.1.3) [8,9].
PNPO catalyzes the conversion of pyridoxine 5¢-phos-
phate (PNP) and pyridoxamine-5¢-phosphate (PMP) to
PLP, with O
2
as an electron acceptor. Kinetic studies
published by Choi et al. [10], have established that the
oxidase can function via either a binary or ternary complex
mechanism, depending upon the nature of the substrate.

Enzymes: ATP-dependent pyridoxal kinase (EC 2.7.1.35); FMN-
dependent pyridoxine 5¢-phosphate oxidase (PNPO, EC 1.4.3.5);
pyridoxal phosphatase (EC 3.1.3).
(Received 23 February 2004, revised 16 April 2004,
accepted 20 April 2004)
Eur. J. Biochem. 271, 2452–2461 (2004) Ó FEBS 2004 doi:10.1111/j.1432-1033.2004.04175.x
stand the structure and regulatory mechanisms of this
enzyme. A high degree of sequence homology exists between
PNPO from different sources suggesting that all members of
this enzyme group share a common three-dimensional fold
and catalytic mechanism. Recently, the E. coli [20–22] and
human enzymes [23] have been cloned and crystallized.
In contrast with the abundant data on the mechanism of
catalysis very little is known about the genomic structure
and expression of PNPO. Here we present a characteriza-
tion of the genomic organization, the structure of the
mRNA isoforms produced by alternative polyadenylation,
and the tissue distribution of the transcript. To our
knowledge this study describes the first detailed investiga-
tion of the transcription of human PNPO. In addition, the
minimum size necessary for enzymic function was deter-
mined by deletion mutagenesis.
Materials and methods
Materials
A Marathon-Ready
TM
cDNA library from human brain, a
multiple tissue Northern blot (MTN
TM
Blot) and a dot blot

cDNA library (human whole brain, Clontech) as a template.
PCR was carried out in GeneAmp PCR system 2400
(PerkinElmer Life Sciences) for 30 cycles of denaturation
94 °C for 1 min, annealing 55 °C for 1 min and extension
72 °C for 2 min. The PCR product was cloned into the
pGEM-T vector (Promega) and sequenced (GenBank
TM
/
EBI accession number AF468030).
To facilitate expression vector construction, a BamHI
recognition site was introduced at both ends of the ORF by
PCRwithprimersshowninTable1.ThePCRmixturewas
analysed on a 0.8% agarose gel, and the product band was
extracted from the gel, purified, and ligated into the pGEM
vector. Then a BamHI digested fragment was subcloned
into pET28a expression vector (pET28a/PNPOx) and used
to transform BL21(DE3) competent cells.
For the construction of deletion mutants, convenient
restriction sites and PCR-based strategies were used
(Table 1). Each PNPO deletion mutant was subcloned into
pET28a. These constructs encode the following residues
of human PNPO: D1–56, residues 57–262; D1–72, residues
73–262; D238–262, residues 1–237. The structures of these
plasmids were verified by restriction and sequence analysis
to ensure that the reading frame was maintained.
In silico
analysis
The full-length ORF sequence of PNPO (GenBank
TM
/EBI

TM
Hybridization solution (Clontech). The
filters were then hybridized at 65 °Cfor16hwith
32
P-labelled specific cDNA probes containing either the
complete ORF or the 3¢-UTR of PNPO as required. The
3¢-UTR of 1 kb had been cloned using the PNPO-specific
primers (sense, 5¢-TACACAGGGTGGTCCACAAGC
Table 1. PCR primers used in the expression constructions for wild-type and deletion mutants. PNPO deletion mutants were constructed using PCR
amplication of the relevant portions of PNPO cDNA followed by restriction digestion and subsequent subcloning into pGEM and pET28a vector.
Primer Primer sequence Restriction enzyme
Wild Forward 5¢-TAAGGATCCCCCATGACGTGC-3¢ BamHI
Reverse 5¢-CAGGATCCAGAGTTAAGGTGCAAG-3¢ BamHI
D1–56 Forward 5¢-CCGAATTCGACCCAGTGAAACAGTTT-3¢ EcoRI
Reverse 5¢-GGAAGCTTAGTTAAGGTGCAAGTCTCTC-3¢ HindIII
D1–72
a
Forward 5¢-CGGATCCGAGGAGGCTGTTCAGTGT BamHI
D238–262
a
Reverse 5¢-AGGATCCCTAGGGTAGGCCCCGCCG-3¢ BamHI
a
The reverse and forward primer of wild-type were used for constructions of D1–72 and D238–262, respectively.
Ó FEBS 2004 Human pyridoxine 5¢-phosphate oxidase (Eur. J. Biochem. 271) 2453
CAGG-3¢;antisense,5¢-GGGGCGGTAACGGCTGG
ACAGAGAA-3¢). To obtain the full-length ORF, we
performed PCR amplifications using the specific primers for
human PDXP [9] and human PDXK (sense, 5¢-CAG
GCCCCATATGGAGGAGGAGTGCCGG-3¢;antisense,
5¢-GGGGATCCTCACAGCACCGTGGC-3¢) [27]. After

and purification of recombinant
human PNPO
The PNPO cDNA was cloned between the BamHI of
pET28a expression vector (Novagen Inc.) after PCR
amplification. Transformants of E. coli BL21(DE3) har-
bouring pET28a/PNPO were cultured at 37 °CinLuria–
Bertani medium with 50 lgÆmL
)1
kanamycin. When that
culture had grown to an A
600
of 0.5, isopropyl thio-b-
D
-galactoside was added to a final concentration of 1 m
M
.
After inducing the expression of the PNPO protein for 3 h
at 37 °C, cells were harvested by centrifugation (10 000 g at
4 °C for 10 min), and the pellet was suspended in lysis
buffer (20 m
M
Tris/HCl pH 7.4, 1 m
M
EDTA, 200 m
M
NaCl, 10 m
M
2-mercaptoethanol, 0.5 m
M
phenyl-

)1
at 25 °C. The
value of K
m
and k
cat
were determined from double
reciprocal plots of initial velocity and substrate concentra-
tion. The concentration of enzyme was determined by the
Bradford method.
Results and discussion
Genomic organization of human PNPO
Using PNPO cDNA as a query sequence, a
BLAST
analysis
(available through the NCBI web site) mapped the PNPO
gene to human chromosome 17q21.32. The gene spans over
7743 bp, and the coding region of the gene was divided into
seven discrete exons as shown in Fig. 1A. All exon/intron
boundaries were found to contain the canonical 5¢ donor GT
and 3¢ acceptor AG sequences (Table 2). The ORF encodes
a 261 amino acid protein with a molecular mass of 30 kDa.
A computer calculation reveals that the isoelectric point for
the protein is 6.61.
SCANPROSITE
software analysis by
EXPASY
showed that the deduced human protein has the following
putative post-translational modification sites: a sulfation
site, nine phosphorylation sites, three N-myristoylation sites

Sp1 sites (data not shown). The absence of a TATA-box is
indeed a noticeable feature of many housekeeping genes [29].
The mouse gene encodes a protein of 261 amino acids of
m 30 114 Da, and it is located on chromosome 11 which has
a very similar genomic organization to that of humans
(Fig. 1B). The longest cDNA contains 1991 bp consisting
of a 786 bp ORF, a 118 bp 5¢-untranslated region and a
1087 bp 3¢-noncoding region. As in humans, the mouse
PNPO gene is encoded by seven exons and the intron/exon
junctions also follow the GT/AG rule. The 3¢-end of the
sequence contains a poly(A) stretch, preceded by a putative
polyadenylation signal AATAAA. The mouse PNPO gene
has CpG islands extending from position )511 to 276 and
from )82 to +227 with a CG content of 61%. The deduced
protein with a predicted pI of 8.35 has a putative sulfate
site, eight phosphorylation sites, two N-myristoylation sites
and one RGD cell attachment sequence. Human and mouse
PNPO share 90% identity at the amino acid level.
Table 2. The intron/exon junctions of the human PNPO gene. The nucleotide sequences at exon (uppercase letters) and intron (lowercase letters)
junction are shown. Exon and intron sizes are indicated in bp.
Exon (bp) 5¢-splice donor Intron (bp) 3¢-Splicing acceptor Exon
I (243)
CGAGAG/gtgccg 1 (1492) tcctag/GCATTT II
II (125)
CACCAG/gtgggc 2 (1185) tcctag/AGATGG III
III (100)
GAGCTG/gtgggt 3 (843) ttctag/GACTCT IV
IV (54)
CGTCAG/gtgagt 4 (248) gagcag/GTGCGT V
V (129)

of a highly related gene that cross-hybridizes with the PNPO
probe. There are several possible mechanisms by which
multiple transcripts could be generated from the same gene:
(1) use of alternative polyadenylation sites; (2) use of
alternate transcription start sites; and (3) differential splicing
of pre-mRNA. In the Western blot analysis as shown in
Fig. 3, no protein with a molecular mass higher than
30 kDa could be detected with mAbs against sheep PNPO.
This line of evidence may rule out the existence of an
alternative splicing product.
To further elucidate the presence of isoform message, this
filter was reprobed with the DNA probes specific for the
3¢-UTR between the two potential poly(A) signals. The
results showed that only the 3.4 kb band was detected
(Fig. 2A, right), which supports the hypothesis that the two
mRNA species are generated by alternate usage of poly-
adenylation sequences. The putative schematic structure of
the mRNA isoforms is shown in Fig. 2B.
Two putative polyadenylation signals ) one an ATT
AAA motif 1472 bp downstream of the termination codon
and the other an AATAAA motif 27 bp upstream of the
end of the gene ) were found within the genomic primary
sequence. It is known that the most common polyadeny-
lation signal is AATAAA, and that ATTAAA is  80% as
efficient as the terminal sequence [30]. Thus, both polyade-
nylation sites of PNPO worked, implying some read-
through of the first site by an unknown mechanism. A
search of the human EST database with the human PNPO
sequence also supported this hypothesis. Alternate usage of
polyadenylation signals is frequently seen in testis tissue.

Fig. 3. Western blot analysis of human PNPO. SDS/PAGE (A) and
immunoblot with mAb (B) for human tissue and cell homogenates.
LaneM,Molecularmassstandards;lane1,brain;lane2,liver;lane3,
lung; lane 4, prostate; lane 5, human breast cancer (MCF-7); lane 6,
human uterine carcinoma (HL3T1); lane 7, stomach tissue.
2456 J. H. Kang et al. (Eur. J. Biochem. 271) Ó FEBS 2004
mRNA expression levels in selected tissue for each enzyme
are shown in Table 3. Consistent with their ubiquitous role
in vitamin B
6
metabolism, all three transcripts have been
detected in a wide variety of tissue. Analysis of the array
revealed that human PDXK was expressed in essentially all
organs with the highest levels observed in descending order
testes, kidneys and placenta. A relatively high level of
PDXK transcript was expressed in foetal organs. In
contrast, human PDXP mRNAs appear to be strikingly
abundant in the brain indicating a more specific role [9].
These results imply that the three enzymes are differentially
expressed and regulated in a tissue specific manner.
The regulation of PLP could be controlled by several
factors. The synthesis of PLP requires the joint action of
PDXK and PNPO, and the PLP availability is dependent
on the degree of protein binding of the synthesized
coenzyme and transport of the precursors [33,34], and
phosphatase action [35]. PNPO does play a kinetic role in
regulating in vivo PLP formation [2,36], whereas PDXK
plays an additional trapping role whereby pyridoxal is
diffusible across the cell membrane [33]. Tissue with high
oxidase activities, however, produce PLP not only for

PNP and PMP, respectively, from Lineweaver–Burk
(double-reciprocal) plots (Table 4).
In order to delineate the region of human PNPO that is
essential for catalysis, we expressed the sequential trunca-
tion mutants in E. coli and determined the effect of each
deletion on activity. In this work, the role of both the N- and
C-terminal regions of human PNPO were studied by the
truncation mutants: D1–56, D1–72 and D238–262 (Fig. 5B).
V
max
values of 0.10 and 0.05 lmolÆmin
)1
Æmg
)1
for the
recombinant wild-type enzyme were obtained for PNP and
PMP, respectively, whereas the deletion of the noncon-
served 56-amino acid at N-terminal domain (D1–56) caused
about a twofold increase in catalytic activity (Table 4). The
K
m
value of the mutant, however, is about threefold higher
Table 3. Comparison of mRNA expression levels of vitamin B
6
regula-
ting enzymes. A dot blot array containing human poly(A)
+
RNAs
from various tissues were hybridized with probes as described in Fig. 4.
Expression levels of selected tissues for PNPO, PDXK, and PDXP are

Oesophagus 7.5 13.4 2.8
Stomach 17.2 43.5 26.1
Duodenum 9.3 27.2 18.9
Jejunum 13.3 43.1 26.1
Ileum 5.8 23.0 21.2
Ilocecum 6.1 60.6 35.9
Appendix 0.7 20.5 21.2
Colon, ascending 1.2 5.7 21.9
Colon, transverse 7.9 8.0 21.6
Colon, desending 2.8 9.4 6.5
Rectum 2.3 16.4 15.8
Kidney 85.6 58.0 31.4
Skeletal muscle 34.1 15.4 20.7
Spleen 13.5 32.0 8.8
Thymus 6.4 29.9 18.6
Peripheral blood leukocyte 0.4 26.3 13.3
Lymph node 4.6 51.2 17.8
Bone morrow 11.2 42.0 22.8
Trachea 4.8 23.0 5.8
Lung 4.6 24.8 7.6
Placenta 30.2 61.4 8.1
Bladder 6.8 12.4 9.2
Uterus 2.9 19.4 9.5
Prostate 16.5 37.0 18.2
Testis 8.4 100.0 49.3
Ovary 4.3 20.1 28.9
Liver 100.0 56.0 61.8
Table 3. (Continued).
PNPO PDXK PDXP
Pancreas 2.5 59.0 33.3

(l
M
)
V
max
(lmolÆmin
)1
Æmg
)1
)
k
cat
/K
m
(
M
)1
ÆS
)1
)
Wild-type PNP 2.1±0.2 0.10±0.06 5.2 · 10
4
PMP 6.2±0.3 0.05±0.01 8.2 · 10
3
PLP 3.8
D1–56 PNP 6.2±0.2 0.21±0.02 3.1 · 10
4
PMP 20.8±0.4 0.08±0.01 3.6 · 10
3
PLP 23.0

disturbance in the folding process during expression caused
by a missing structural unit. The presence of the first helical
sequence might be solely structural, as it does not have a
direct interaction with either PLP or FMN [23]. In addition,
a deletion of 25 residues at the C terminus (D238–262)
resulted in essentially inactive enzymes, indicating that this
region is required for function.
Conclusions
In this report, we have described the genomic organization
of PNPO, tissue distribution and deletion mutagenesis.
(1) The human PNPO gene is composed of seven exons
and six introns spanning  7.7 kb of the genomic DNA.
The 5¢-flanking region has the characteristic features of
housekeeping genes. Due to alternate usage of polyadeny-
lation sites, two species of mRNA existed in all examined
tissue. Nevertheless, no protein isoforms were detected.
Fig. 5. Deletion analysis of recombinant PNPO. (A) Expression and purification of recombinant human PNPO. SDS/PAGE analysis (12%
acrylamide)ofcrudecellextractsofE. coli BL21(DE3) containing the expression vector without and with the coding sequences for the wild-type or
mutants. Lane M, Low molecular mass standards (Bio-Rad); lane 1, crude extracts from cultured cells harbouring pET28a; lane 2, cells containing
pET28a/PNPO in the presence of 1 m
M
isopropyl thio-b-
D
-galactoside; lane 3, purified recombinant PNPO from Ni
2+
resin; lanes 4–6, purified
deletion mutants: D1–56, D1–72 and D238–262, respectively. (B) Left, schematic structure of wild-type PNPO and the N- and C-terminal deletion
mutants used in this study. Numbers refer to the amino acid position along the primary sequence of PNPO. Right, the effect of N- and C-terminal
deletion on PNPO activity was expressed as a percentage of enzymatic activity in wild-type enzyme. Solid black and crosshatched bars are for
substrate PNP and PMP, respectively. The results shown are the means ± SD from triplicate assays.

doxine) 5¢-phosphate oxidase. In Vitamin B
6
Metabolism and Role
in Growth (Tryfiates, G.P., ed.), pp. 1–26. Food and Nutrition
Press, Westport, CT.
3. Robson, L.C. & Schwartz, M.R. (1975) Vitamine B
6
deficiency
and the lymphoid system I. Effects on cellular immunity and in
vitro incorporation of
3
H-uridine by small lymphocytes. Cell.
Immunol. 16, 135–144.
4. McCormick, D.B., Gregory, M.E. & Snell, E.E. (1961) Pyridoxal
phosphokinase I: assay, distribution, purification and properties.
J. Biol. Chem. 236, 2076–2084.
5. Hanna, M.C., Turner, A.J. & Kirkness, E.F. (1997) Human
pyridoxal kinase. cDNA cloning, expression, and modulation by
ligands of the benzodiazepine receptor. J. Biol. Chem. 272, 10756–
10760.
6. Kwok, F. & Churchich, J.E. (1992) Pyrdoxine-5¢-P oxidase. In
Chemistry and Biochemistry of Flavoenzymes (Muller, F., ed.),
Vol. 3, pp. 1–20. CRC Press, London.
7. Ngo,E.O.,LePage,G.R.,Thanassi,J.W.,Meisler,N.&Netter,
L.M. (1998) Absence of PNP oxidase (PNPO) activity in neo-
plastic cells: isolation, characterization, and expression of PNPO
cDNA. Biochemistry 37, 7741–7748.
8. Fonda, M.L. (1992) Purification and characterization of vitamine
B
6

17. Notheis, C., Drewke, C. & Leistner, E. (1995) Purification and
characterization of the pyridoxol-5¢-phosphate: oxygen oxido-
reductase (deaminating) from Escherichia coli. Biochim. Biophys.
Acta 1247, 265–271.
18. Zhao, G. & Winkler, M. (1995) Kinetic limitation and cellular
amount of pyridoxine (pyridoxamine) 5¢-phosphate oxidase of
Escherichia coli K-12. J. Bacterol. 177, 883–891.
19. Di Salvo, M.L., Safo, M.K., Musayev, F.N., Bossa, F. & Schirch, V.
(2003) Structure and mechanism of Escherichia coli pyridoxine
5¢-phosphate oxidase. Biochim. Biophys. Acta 1647, 76–82.
20.Safo,M.K.,Mathews,I.,Musayev,F.N.,DiSalvo,M.L.,
Thiel, D.J., Abraham, D.J. & Schirch, V. (2000) X-ray structure of
Escherichia coli pyridoxine 5¢-phosphate oxidase complexed with
FMN at 1.8 A
˚
resolution. Structure 8, 751–762.
21. Safo, M.K., Musayev, F.N., De Salvo, M.L. & Schirch, V. (2001)
X-ray structure of Escherichia coli pyridoxine 5¢-phosphate oxi-
dase complexed with pyridoxal 5¢-phosphate at 2.0A
˚
resolution.
J. Mol. Biol. 310, 817–826.
22. Di Salvo, M.L., Ko, T.P., Musayev, F.N., Raboni, S., Schirch, V.
& Safo, M.K. (2002) Active site structure and stereospecificity of
Escherichia coli pyridoxine-5¢-phosphate oxidase. J. Mol. Biol.
315, 385–397.
23. Musayev, F.N., Di Salvo, M.L., Ko, T.P., Schirch, V. & Safo, M.K.
(2003) Structure and properties of recombinant human pyridoxine
5¢-phosphate oxidase. Protein Sci. 12, 1455–1463.
24. Quandt, K., Frech, K., Karas, H., Wingender, E. & Werner, T.

6
.
In Metabolism of Vitamins and Trace Elements (Florkin,M.&
Stotz, E.H., eds), Vol. 21, pp. 47–71. Elsevier Scientific Publishing
Co, Amsterdam.
34. Anderson, B.B., Newmark, P.A. & Rawlins, M. (1974) Plasma
binding of vitamine B6 compound. Nature 250, 502–504.
35. Lumeng, L. & Li, T.K. (1975) Characterization of the pyridoxal
5¢-phosphate and pyridoxamine 5¢-phosphate hydrolase activity in
rat liver. Identity with alkaline phosphatase. J. Biol. Chem. 250,
8126–8131.
36. Kwon, O.S., Kwok, F. & Churchich, J.E. (1991) Catalytic
and regulatory properties of native and chymotrypsin-treated
pyridoxine-5-phosphate oxidase. J. Biol. Chem. 266, 22136–
22140.
Ó FEBS 2004 Human pyridoxine 5¢-phosphate oxidase (Eur. J. Biochem. 271) 2461