Tài liệu Báo cáo khoa học: A second novel dye-linked L-proline dehydrogenase complex is present in the hyperthermophilic archaeon Pyrococcus horikoshii OT-3 pptx

A second novel dye-linked L-proline dehydrogenase
complex is present in the hyperthermophilic archaeon
Pyrococcus horikoshii OT-3
Ryushi Kawakami
1
, Haruhiko Sakuraba
1
, Hideaki Tsuge
2,3
, Shuichiro Goda
1
, Nobuhiko Katunuma
2
and Toshihisa Ohshima
1
1 Department of Biological Science and Technology, Faculty of Engineering, The University of Tokushima, Japan
2 Institute for Health Science, Tokushima Bunri University, Japan
3 The Institutes for Enzyme Research, University of Tokushima, Japan
Dye-linked dehydrogenases (dye-DHs) catalyze the oxi-
dation of various amino acids, organic acids, amines
and alcohols in the presence of artificial electron accep-
tors such as 2, 6-dichloroindophenol (DCIP) and ferri-
cyanide. Although dye-DHs show a high potential for
use as specific elements in biosensors [1], their low
stability has thus far precluded their use in practical
applications and limited our ability to obtain detailed
information about their structures and functions.
Recently, however, much attention has been paid to
the isolation and characterization of enzymes from
hyperthermophilic archaea, as these organisms repre-
sent a source of extremely stable enzymes. Indeed, we

have identified several novel dye-DHs in hyperthermo-
philic archaea, including dye-linked d-proline dehy-
drogenase [2] and dye-linked l-proline dehydrogenase
Keywords
ATP-containing dehydrogenase; dye-linked
L-proline dehydrogenase; hyperthermophilic
archaeon; Pyrococcus horikoshii
Correspondence
T. Ohshima, Department of Biological
Science and Technology, Faculty of
Engineering, The University of Tokushima,
2–1 Minamijosanjima-cho, Tokushima 770–
8506, Japan
Fax: +81 88 656 9071
Tel: +81 88 656 7518
E-mail:
(Received 7 May 2005, revised 3 June
2005, accepted 8 June 2005)
doi:10.1111/j.1742-4658.2005.04810.x
Two distinguishable activity bands for dye-linked l-proline dehydrogenase
(PDH1 and PDH2) were detected when crude extract of the hyperthermo-
philic archaeon Pyrococcus horikoshii OT-3 was run on a polyacrylamide
gel. After purification, PDH1 was found to be composed of two different
subunits with molecular masses of 56 and 43 kDa, whereas PDH2 was
composed of four different subunits with molecular masses of 52, 46, 20
and 8 kDa. The native molecular masses of PDH1 and PDH2 were 440
and 101 kDa, respectively, indicating that PDH1 has an a
4
b
4

structure,
while PDH2 has an abcd structure. PDH2 was found to be similar to the
dye-linked l-proline dehydrogenase complex from Thermococcus profundus,
but PDH1 is a different type of enzyme. After production of the enzyme
in Escherichia coli, high-performance liquid chromatography showed the
PDH1 complex to contain the flavins FMN and FAD as well as ATP.
Gene expression and biochemical analyses of each subunit revealed that
the b subunit bound FAD and exhibited proline dehydrogenase activity,
while the a subunit bound ATP, but unlike the corresponding subunit in
the T. profundus enzyme, it exhibited neither proline dehydrogenase nor
NADH dehydrogenase activity. FMN was not bound to either subunit,
suggesting it is situated at the interface between the a and b subunits.
A comparison of the amino-acid sequences showed that the ADP-binding
motif in the a subunit of PDH1 clearly differs from that in the a subunit
of PDH2. It thus appears that a second novel dye-linked l-proline dehy-
drogenase complex is produced in P. horikoshii.
Abbreviations
dye-DH, dye-linked dehydrogenase; DCIP, 2,6-dichloroindophenol; dye-
L-proDH, dye-linked L-proline dehydrogenase; dye-NADHDH, dye-
linked NADH dehydrogenase.
4044 FEBS Journal 272 (2005) 4044–4054 ª 2005 FEBS
(dye-l-proDH) [3,4], and found these enzymes to be
highly stable and to exhibit a high potential for appli-
cation in amino-acid analyses.
Dye-l-proDH catalyzes the oxidation of l-proline to
D
1
-pyrroline-5-carboxylate in the presence of DCIP.
We first identified this enzyme in the hyperthermo-
philic archaeon Thermococcus profundus DSM9503 [4].

Our functional and structural analyses showed that it
is a novel bifunctional amino-acid dehydrogenase that
exhibits both NADH dehydrogenase and l-proline
dehydrogenase activities [3]. The enzyme is comprised
of four different subunits (a, b, c and d), the genes for
which form an operon [3]. A similar gene cluster also
has been observed in the genome of Pyrococcus
horikoshii OT-3, which has been sequenced completely
[5]. During the course of screening for dye-l-proDH,
we detected two electrophoretically distinguishable
activity bands in the crude extract of P. horikoshii
OT-3, which suggests that in addition to the abcd-type
of dye-l-proDH, another, as yet unknown, dye-
l-proDH is produced by this organism. In the present
study, we identified the gene encoding this other
enzyme, expressed it in Escherichia coli, and examined
the characteristics of its product. We found the enzyme
to be totally different from the abcd-type in both
structure and function; that is, it is comprised of two
different subunits and has no NADH dehydrogenase
activity. In addition, the enzyme complex contained
ATP, FMN and FAD, though abcd-type dye-l-proDH
contains only FAD. Here we describe the molecular
and structural characteristics of this novel, ATP-con-
taining amino-acid dehydrogenase.
Results and Discussion
Distribution of dye-L-proDH in hyperthermophilic
archaea
To identify organisms that produce dye-l-proDH, we
screened enzymes using native-PAGE coupled with

activity staining as described in the ‘Experimental
procedures.’ We observed two separate activity bands
with P. horikoshii and T. peptonophilus; one band
with T. profundus, P. furiosus and P. abyssi; and no
activity bands with T. litoralis (data not shown).
These results suggest that P. horikoshii and T. peptono-
philus each produce two distinguishable forms of dye-
l-proDH. Because its genome has been sequenced
completely [5], we chose to purify the enzymes from
P. horikoshii.
Purification of PDH1 and PDH2 from P. horikoshii
OT-3 and identification of the encoding genes
The steps used to isolate PDH1 and PDH2 from
P. horikoshii OT-3 are summarized in Tables 1 and 2,
respectively. We succeeded in separating the two
enzymes using Butyl Toyopearl column chromatogra-
phy (Fig. 1), and were able to further purify them
using the additional steps listed. Gel filtration
Table 1. Purification of PDH1 from P. horikoshii.
Total protein (mg) Total activity (U) Specific activity (UÆmg
)1
) Yield (%) Fold
Crude extract 4550 65.1 0.0143 100 1.0
Ammonium sulfate
fractionation
3600 58.6 0.0163 90 1.1
DEAE Toyopearl 1040 51.7 0.0497 79 3.5
Butyl Toyopearl 295 7.40 0.0251 11 1.8
Cellulofine HAp 13.3 5.90 0.444 9 31.0
UnoQ 2.63 2.47 0.939 4 65.7

Table 2. Purification of PDH2 from P. horikoshii.
Total protein (mg) Total activity (U) Specific activity (UÆmg
)1
) Yield (%) Fold
Crude extract 4550 65.1 0.0143 100 1.0
Ammonium sulfate
fractionation
3600 58.6 0.0163 90 1.1
DEAE Toyopearl 1040 51.7 0.0497 79 3.5
Butyl Toyopearl 164 14.6 0.0890 22 6.2
Cellulofine HAp 16.2 6.71 0.414 10 30.0
UnoQ 0.536 0.752 1.40 1 97.9
R. Kawakami et al. FAD, FMN and ATP-containing amino-acid dehydrogenase
FEBS Journal 272 (2005) 4044–4054 ª 2005 FEBS 4045
chromatography with Superose 6 showed the molecu-
lar masses of PDH1 and PDH2 to be about 440 and
101 kDa, respectively. Subjecting purified PDH1 to
SDS ⁄ PAGE revealed the enzyme is comprised of two
different subunits (designated a1 and b1) with mole-
cular masses of about 56 and 43 kDa, respectively (data
not shown). After SDS ⁄ PAGE, the stained gel was
scanned, and the relative ratio of the peak areas was
determined to be 1.3 (a1): 1.0 (b1) using NIH image
software ( />Taking into account their molecular masses, the
molecular a1: b1 ratio was calculated to be about
1 : 1, which means the enzyme has a heterooctameric
(a
4
b
4

) structure. The N-terminal amino-acid sequence
of the a1 subunit was determined to be
MRPLDLTEKR, which corresponds to the underlined
amino-acid sequence in ML
MRPLDLTEKR from the
putative protein encoded by the predicted open reading
frame (ORF; PH1363) based on the genome analysis.
This means that ATG, which was situated 7 bp down-
stream from the 5¢-terminus of the predicted ORF, is the
proper initial codon for the a1 gene. The N-terminal
amino-acid sequence of the b1 subunit was determined
to be MLPEKSEIVV, which corresponds to that of the
predicted PH1364 gene product. The a1 and b1 genes
were arranged in tandem (a1 –b1) and were estimated to
encode proteins with molecular masses of 55 316 and
42 685 Da, respectively.
On the other hand, SDS ⁄ PAGE analysis of purified
PDH2 showed four bands (data not shown). The
molecular masses of the a, b, c and d subunits of
PDH2 (designated a2, b2, c2 and d2) were about 52,
46, 20 and 8 kDa, respectively. Although the four dif-
ferent subunits together have a mass of 126 kDa, only
101 kDa has been determined by gel filtration. On the
other hand, SDS ⁄ PAGE after Superose 6 chromato-
graphy showed that all four subunits were present
proportionally in the active fractions. The molecular
ratio of the subunits of PDH2 was determined to be
1 : 1 : 1 : 1 using the same method used for PDH1.
This suggests that the enzyme has a heterotetrameric
(abcd) structure. The N-terminal amino-acid

sequences of the a2, b2, c2 and d2 subunits were
MRINEHPILD, MIGIIGGGII, SEIPNYLKYG and
MKIVCRCNDV, respectively. The N-terminal amino-
acid sequence of the a2 subunit corresponded to the
underlined amino-acid sequence within MEIV
RINEH-
PILD from the putative protein encoded by the predic-
ted ORF (PH1749), except for the first methionine
residue. This means that GTG, which was situated
10 bp down stream from the 5¢-terminus of the predic-
ted ORF, is the proper initial codon for a2. The N-ter-
minal amino-acid sequences of the b2 and c2 subunits
corresponded to the sequences of the predicted
PH1751 and PH1750 gene products, respectively. We
previously suggested the presence of a gene encoding
the d2 subunit (designated PHpdhX) [3], and the
amino-acid sequence of the d2 subunit corresponded
completely to that of the predicted PHpdhX gene prod-
uct. These four genes were arranged in tandem (a2–c2–
d2–b2) and were estimated to encode proteins with
molecular masses of 52 446, 18 974, 10 076 and
42 420 Da, respectively. Although the molecular mass
of d2 subunit was predicted to be 10 kDa, it was deter-
mined to be 8 kDa by SDS ⁄ PAGE. This might be from
the low resolution of the low molecular mass protein at
the used condition (15% gel). We previously reported
that the amino-acid sequences of the a, c, d and b
subunits of T. profundus dye-l-proDH show a high
identity with those deduced from the PH1749, PH1750,
PHpdhX and PH1751 genes of P. horikoshii, respect-

ively [3]. Identification of the genes encoding the four
subunits of PDH2 clearly demonstrates that PDH2 is
similar to the T. profundus dye-l-proDH complex.
In the present study, the genes encoding the PDH1
subunits were found to form an operon (a1–b1), and
similar gene clusters have been observed in the
genomes of P. furiosus (PF1245–PF1246) and P. abyssi
(PAB1842–PAB1843). A gene cluster like that formed
by the PDH2 genes (a2–c2–d2–b2) is also distributed
in these organisms [3]. Together with the results of
activity screening, these observations suggest that both
the a
4
b
4
and abcd dye-l-proDHs are widely distri-
buted within the order Thermococcales in the archaeal
Fig. 1. Elution profile obtained with Butyl Toyopearl chromatogra-
phy. Enzyme solution was applied on a Butyl Toyopearl column and
eluted with a linear gradient of 40–0% ammonium sulfate in buffer
A. The squares and circles show the activity and A
280
, respectively.
FAD, FMN and ATP-containing amino-acid dehydrogenase R. Kawakami et al.
4046 FEBS Journal 272 (2005) 4044–4054 ª 2005 FEBS
domain, though their expression patterns differ from
one another depending up the hyperthermophilic
species examined and cultivation conditions used.
Expression of the PDH1 gene and purification
of the recombinant enzyme

We initially attempted to express the PDH1 gene using
the plasmid pPDH1, but no functional product could
be obtained. We therefore introduced the gene into
pET11a, and were then able to successfully express it
and then isolate it using the steps summarized in
Table 3. E. coli strain BL21 CodonPlus RIL (DE3)
cells transformed with the expression plasmid pEPDH1
exhibited a high level of dye-l-proDH activity, which
was not lost upon incubation at 90 °C for 20 min. The
enzyme was purified to homogeneity from cell extract
using heat treatment and two successive column chro-
matographies:  50 mg of the purified enzyme was
obtained from 2 L of E. coli culture, and the specific
activity of the enzyme was about 2-times that of the
native enzyme. The purified PDH1 showed the same
mobility as the native enzyme on native-PAGE, and
the N-terminal amino-acid sequences of the a1 and b1
subunits of the recombinant enzyme were confirmed to
be identical to those of the native enzyme.
Characteristics of the recombinant PDH1
Recombinant PDH1 showed a high degree of stability
against both temperature and pH (Fig. 2). The enzyme
showed extreme thermostability; no activity was lost
during incubation at 90 °C for 120 min (Fig. 2A).
Using ferricyanide as the artificial electron acceptor,
the optimum temperature for activity was determined
to be about 90 °C (Fig. 2B). By contrast, T. profundus
dye-l-proDH becomes completely inactive within
20 min at 80 °C [4]. Thus, PDH1 is the most thermo-
stable dye-l-proDH described to date. The stability

under various pH conditions was examined while incu-
bating the enzyme at 50 °C for 30 min. The enzyme
lost no activity between pH 5.0 and 10.0 (Fig. 2C),
and the optimum pH was determined to be 7.5
(Fig. 2D).
Table 3. Purification of recombinant PDH1 from E. coli.
Total protein (mg) Total activity (U) Specific activity (UÆmg
)1
) Yield (%) Fold
Crude extract 2280 111 0.0487 100 1.0
Heat treatment 400 279 0.698 251 14.3
Butyl Toyopearl 62 113 1.82 102 37.4
Superdex 200 50 105 2.10 95 43.1
Fig. 2. (A) Thermostability of PDH1. The
enzyme was incubated at 90 °C, and the
residual activity was measured at the indica-
ted times. (B) Effect of temperature on
PDH1 activity. The enzyme activity was
measured at various temperatures between
50 and 95 °C. (C) pH stability of PDH1. The
enzyme was incubated at 50 °C for 30 min
in buffers of various pH, after which residual
activity was measured. (D) Effect of pH on
PDH1 activity. The enzyme activity was
measured at various pHs ranging from 6
and 9. The buffers used were Mes ⁄ NaOH
(pH 6.0–7.0), Hepes ⁄ NaOH (pH 7.0–7.5) and
Tris ⁄ HCl (pH 7.5–9.0).
R. Kawakami et al. FAD, FMN and ATP-containing amino-acid dehydrogenase
FEBS Journal 272 (2005) 4044–4054 ª 2005 FEBS 4047

PDH1 acted exclusively on l-proline; l-hydroxypro-
line, d-proline, cis-4-hydroxyl-d-proline, l-2 pyrroli-
done-5-carboxylate, pyrrole-2-carboxylate and pipecolic
acid were all inert as substrates. Although based on
genome sequencing the enzyme was predicted to have
sarcosine oxidase activity [5], no such activity was detec-
ted. The apparent K
m
values for l-proline and DCIP
were 4 and 0.03 mm, respectively, and the reaction
product of the l-proline dehydrogenation catalyzed
by PDH1 was D
1
-pyrroline-5-carboxylate. These prop-
erties of PDH1 are comparable to those reported for
T. profundus dye-l-proDH [4] with one noteworthy
exception: PDH1 lacks the dye-linked NADH dehy-
drogenase (dye-NADHDH) activity possessed by the
T. profundus enzyme (see below).
Amino-acid sequence alignment and functional
analysis of each subunit
Figures 3 and 4, respectively, show the amino-acid
alignment of the a and b subunits of PDH1, PDH2 and
T. profundus dye-l-proDH. The amino-acid sequence of
the a1 subunit of PDH1 showed 31% and 32% identi-
ties with the a2 subunit of PDH2 and the a subunit of
the T. profundus enzyme, respectively (Fig. 3), while the
b1 subunit showed 56% and 64% sequence identities
with the b2 subunit of PDH2 and the b subunit of the
T. profundus enzyme, respectively (Fig. 4). The b1 sub-

unit of PDH1 contained an ADP-binding motif [6],
which was well conserved in the b subunit of the abcd-
type enzymes (Fig. 4). In addition, expression of the b1
gene in E. coli using a pET15b ⁄ b1 system revealed that,
like the b subunit of the abcd-type enzyme [3], the b1
subunit is capable of catalyzing l-proline dehydrogena-
tion by itself (data not shown). This suggests that the b1
subunit of PDH1 has the same function as that des-
cribed for the b subunit of T. profundus dye-l-proDH;
that is, it catalyzes the first reaction of the incorporation
of electrons from l-proline into the electron transfer sys-
tem [3]. We previously reported that the T. profundus
enzyme exhibits dye-NADHDH activity as well as
l-proline dehydrogenase activity, and functional analy-
sis of each subunit showed that it is the a subunit that
catalyzes the dye-NADHDH reaction [3]. We attempted
to detect dye-NADHDH activity using the PDH1 com-
plex, but found none. In addition, when we expressed
Fig. 3. Amino-acid sequence alignment of
the a subunits of PDH1, PDH2 and T. pro-
fundus dye-
L-proDH: alpha1, a1 subunit of
PDH1; alpha2, a2 subunit of PDH2; and
pdhA, a subunit of T. profundus dye-
L-
proDH. Asterisks(*) represent conserved
residues. ADP-binding motifs [6] are boxed.
FAD, FMN and ATP-containing amino-acid dehydrogenase R. Kawakami et al.
4048 FEBS Journal 272 (2005) 4044–4054 ª 2005 FEBS
the a1 gene in E. coli using a pET11a ⁄ a1 system, the

protein produced also showed no dye-NADHDH activ-
ity. Within the primary structure of the a subunit of the
T. profundus enzyme are two ADP-binding motifs [6]
spanning residues 120–149 and 271–300 (Fig. 3) [3]. The
a1 subunit of PDH1, by contrast, contained only one
ADP binding motif spanning residues 108–136 (Fig. 3),
which suggests the additional ADP-binding motif in the
a subunit might be essential for the dye-NADHDH
activity of the abcd-type enzyme.
Analysis of the prosthetic groups
The absorption spectrum of PDH1 showed pro-
nounced peaks at 370 and 450 nm, suggesting the pres-
ence of flavin compounds. We sought to identify
those compounds using high performance liquid
chromatography (HPLC) and detected both FAD and
FMN within the PDH1 complex (Fig. 5A); moreover,
we determined there to be about 4 mol of FAD and
4 mol of FMN per mol of enzyme complex. Then
using a separate expression system for each subunit
gene, we found that FAD binds to the b1 subunit
(Fig. 5B), which suggests that the ADP-binding motif
in the b1 subunit mediates FAD binding. On the other
hand, FMN was not detected bound to either subunit
(Fig. 5B,C). We also extracted the flavin compounds
from the native enzyme isolated from P. horikoshii
cells and found that levels of FAD and FMN associ-
ated with the native enzyme were similar to those seen
with recombinant PDH1 (Fig. 5D). Taken together,
these findings suggest that the PDH1 complex contains
1 mol of FAD per mol of b1 subunit, but that the a1

and b1 subunits separately produced in E. coli cannot
bind FMN.
While carrying out the procedure to identify the fla-
vin compounds, we observed an unexpected signal that
had about a 4-min retention time on the TSKgel ODS-
80Tm column (Fig. 5A). This signal corresponds to
that of the ATP standard, and when a sample was
injected together with authentic ATP, an enhancement
of the peak was observed. The presence of ATP was
also demonstrated using an Asahipak GS-320HQ col-
umn (data not shown). The ATP content was about
4 mol of ATP per mol of enzyme complex, and
ATP was present in the native enzyme as well as in the
separately produced a1 subunit (Fig. 5C,D), suggesting
that the enzyme complex contains 1 mol of ATP per
mol of a1 subunit. The T. profundus abcd-type enzyme
contains 2 mol of FAD per mol of enzyme complex
[3]. As mentioned above, the a subunit of the enzyme
has two ADP-binding motifs, and in a previous report
we showed it to also contain 1 mol of FAD per mol of
a subunit [3]. In this subunit, any other prosthetic
groups than FAD were not detected [3]. As the a sub-
unit exhibited dye-NADHDH activity, we supposed
that one ADP-binding motif mediates FAD binding
and the other is responsible for the dye-NADHDH
Fig. 4. Amino-acid sequence alignment of
the b subunits of PDH1, PDH2 and T. pro-
fundus dye-
L-proDH: beta1, b1 subunit of
PDH1; beta2, b2 subunit of PDH2; and

pdhB, b subunit of T. profundus dye-
L-
proDH. Asterisks (*) represent conserved
residues. ADP-binding motifs [6] are boxed.
R. Kawakami et al. FAD, FMN and ATP-containing amino-acid dehydrogenase
FEBS Journal 272 (2005) 4044–4054 ª 2005 FEBS 4049
activity. No flavin compounds were bound to the a1
subunit of PDH1, though it did contain one ADP-
binding motif. This suggests the ADP-binding motif in
the a1 subunit mediates the observed ATP binding,
and FMN is likely situated at the interface between
the a1 and b1 subunits within the PDH1 complex. The
dependence of FMN binding on the presence of both
subunits could also be due to a conformational change
of one of the subunits upon heterodimer formation.
There have been several reports of a family of elec-
tron-transfer flavoproteins that utilize both FAD and
FMN as cofactors [7–14]. This family includes cyto-
chrome P450 reductase [8,9], a flavoprotein subunit of
bacterial sulfite reductase [10] and nitric oxide synthase
[11–14]. These enzymes mainly catalyze the transfer
of reducing equivalents from NADPH to a variety of
electron acceptors. In addition, recent studies have
shown that members of this family have similar struc-
tures consisting of two domains, one that binds FMN
and one that binds FAD and NADPH [8–10]. The
FMN domain is homologous to flavodoxins, while the
FAD and NADPH domain is homologous to that
of ferredoxin reductase [8–10,15]. PDH1 shows no
similarity to any of these flavoproteins. The methylo-

trophic bacterium W3A1 reportedly produces an
ADP-containing oxidoreductase, trimethylamine dehy-
drogenase, but not an ATP-containing dehydrogenase,
and the function of ADP in trimethylamine dehydro-
genase remains unknown [16]. Similarly, the catalytic
properties of PDH1 can be interpreted without consid-
ering the function of ATP. It may be that this ATP
has a stabilizing effect on the protein, or that it plays
an unknown regulatory role, as has been suggested for
ADP in trimethylamine dehydrogenase. To the best of
our knowledge, this is the first example of an oxido-
reductase complex that contains an ATP.
In general, dye-DHs play important roles in the
incorporation of electrons from a substrate into
the electron-transfer system. In the present study, the
PDH1 complex was found to contain FAD, FMN and
ATP. That PDH1 is totally different from any known
electron-transfer flavoprotein that utilizes both FAD
and FMN as cofactors, suggests this enzyme may
employ a novel electron transfer pathway from l-pro-
line to the electron-transfer system. Our goal is to bet-
ter understand the relationship between the structure
and function of each subunit of this enzyme and the
physiological functions of the unique prosthetic
groups. An X-ray reflection analysis of PDH1 is now
in progress. These are essential steps in an effort to
achieve the practical application of dye-l-proDHs.
Experimental procedures
Materials
DCIP and l-proline were purchased from Nacalai Tesque

(Kyoto, Japan). [
32
P]ATP[cP] and a ProbeQuant
TM
G-50
microcolumn were from Amersham Bioscience (Tokyo,
Japan). Restriction endonucleases were from New England
Biolabs (Beverly, MA, USA). E. coli strains JM109 and
BL-21 CodonPlus RIL (DE3) were from Stratagene (La
Jolla, CA, USA). The plasmids, pUC18, pUC19, pET11a
and pET15b were from Novagen (Tokyo, Japan). All other
chemicals were of reagent grade.
Microorganism and cell growth
P. horikoshii OT-3 strain was obtained from the Japan
Collection of Microorganisms, (Saitama, Japan), and then
grown at 90 °C for 18 h under anaerobic conditions. The
microorganism was cultured in medium containing 5 g of
tryptone, 1 g of yeast extract, 25 g of NaCl, 1 g of cysteine-
HCl, 1.3 g of (NH
4
)
2
SO
4
, 0.28 g of KH
2
PO
4
, 0.25 g of
MgSO

4
Æ7H
2
O, 0.07 g of CaCl
2
Æ2H
2
O, 0.02 g of FeCl
3
Æ6H
2
O,
Fig. 5. HPLC analyses of the prosthetic groups. Elution profiles of
extracts of recombinant PDH1 (A), the b subunit of PDH1 (B), the a
subunit of PDH1(C), native PDH1 prepared from P. horikoshii cells
(D), and the standard mixture (E).
FAD, FMN and ATP-containing amino-acid dehydrogenase R. Kawakami et al.
4050 FEBS Journal 272 (2005) 4044–4054 ª 2005 FEBS
1.8 mg of MnCl
2
Æ4H
2
O, 4.5 mg of Na
2
B
4
O
7
Æ10H
2

O,
0.22 mg of ZnSO
4
Æ7H
2
O, 0.05 mg of CuCl
2
Æ2H
2
O, 0.03 mg
of Na
2
MoO
4
Æ2H
2
O, 0.03 mg of VOSO
4
Æ2H
2
O, 0.01 mg of
CoSO
4
Æ7H
2
O, and 5 g of elemental sulfur per litre (pH 6.5,
adjusted with 3 m NaOH). After cultivation, the cells were
collected by centrifugation (10 000 g, 15 min), washed twice
with 3% NaCl, suspended in buffer A (10 mm potassium
phosphate, pH 7.2 containing 1 mm EDTA) and stored at

)30 °C.
Enzyme assay and protein determination
Enzyme activity was spectrophotometrically assayed as
described previously using a Shimadzu UV-1200 spectro-
photometer equipped with a thermostat [3]. Protein
concentrations were determined using Bradford method;
bovine serum albumin served as the standard [17].
Purification of dye-L-proDHs from P. horikoshii
OT-3
P. horikoshii cells (wet weight, about 40 g) obtained from
50 L of medium were suspended in 360 mL of buffer A
and disrupted by ultrasonication. After centrifugation
(20 000 g, 20 min) to remove any remaining intact cells and
the cell debris, the supernatant was used as the crude
extract. Ammonium sulfate was added to the extract to
40% saturation, after which it was stirred at 4 °C for 1 h
and centrifuged again (20 000 g, 20 min), and additional
ammonium sulfate was added to the resultant supernatant
containing the enzyme to bring it up to 70% saturation.
Then after 1 h the solution was centrifuged, and the pre-
cipitant, which contained the enzyme, was suspended in
buffer A and dialyzed against the same buffer.
The enzyme-containing solution was then loaded onto a
DEAE Toyopearl column (4.8 · 11 cm; Tosoh, Tokyo,
Japan) equilibrated with buffer A. After washing the col-
umn with the same buffer, the enzyme was eluted with a
1.3-L linear gradient of 0 to 0.2 m NaCl in buffer A. The
active fractions were collected, and the sample was brought
up to 40% saturation with ammonium sulfate and then
loaded onto a Butyl Toyopearl column (3.6 · 10 cm;

Tosoh) equilibrated with buffer A supplemented with 40%
ammonium sulfate. After washing the column with the
same buffer, the enzyme was eluted with a 1-L linear gradi-
ent of 40% to 0% ammonium sulfate in buffer A. Two
peaks containing dye-l-proDH activity appeared in the elu-
tion profile (Fig. 1). The enzymes corresponding to the first
and second peaks were designated PDH1 and PDH2,
respectively. Each enzyme solution was then dialyzed
against 10 mm potassium phosphate, pH 7.2.
The respective PDH1 and PDH2 solutions were sepa-
rately applied to Cellulofine HAp columns (2.6 · 14 cm;
Seikagaku Corp., Tokyo) equilibrated with 10 mm potas-
sium phosphate, pH 7.2. After washing the columns with
100 mm potassium phosphate, pH 7.2, PDH1 was eluted
with 300 mm potassium phosphate, pH 7.2, while PDH2
was eluted with 150 mm potassium phosphate, pH 7.2, and
the respective active fractions were pooled and dialyzed
against buffer A.
PDH1 and PDH2 were then further purified separately
using UnoQ (Bio-Rad, Tokyo, Japan) chromatography.
The respective enzyme solutions were applied to UnoQ col-
umns (0.7 · 3.5 cm) that had been equilibrated with buffer
A and mounted on a fast protein liquid chromatography
(FPLC) system (Bio-Rad). After washing the columns with
buffer A, the enzymes were eluted with a linear gradient of
0 to 0.25 m NaCl in the same buffer, after which the
respective active fractions were pooled and dialyzed against
buffer A. The resultant enzyme solutions were used as the
purified enzyme preparations.
Preparation of the P. horikoshii genomic DNA

and cloning of the enzyme genes
To obtain the genomic DNA containing the PDH1 and
PDH2 genes, P. horikoshii cells were first cultured as des-
cribed above, filtered to remove the sulfur powder and then
centrifuged (10 000 g, 15 min). The cells were then washed
twice with 3% NaCl, and the OT-3 genomic DNA was pre-
pared by the method of Murray and Thompson [18]. To
avoid any nucleotide incorporation errors we did not use
the PCR method to clone the PDH1 and PDH2 genes.
Instead, two oligonucleotide probes (5¢-ATGAGACC
TCTAGATCTAAC-3¢ for the PDH1 gene and 5¢-TATA
TTTAGGTGGAAATTGT-3¢ for the PDH2 gene) were
synthesized based on the DNA sequence in the P. horikoshii
genome database, after which 1.5-pmol samples were labe-
led with [ c -
32
P]ATP (1.85 MBq) using T4 polynucleotide
kinase (10 U), purified on a ProbeQuant
TM
G-50 micro-
column, and used as specific probes for southern and
colony hybridizations.
For preparation of the PDH1 gene, the genomic DNA
was digested with SphI and KpnI; for the PDH2 gene it
was digested with BamHI and SphI, and the resultant
fragments were separated on 0.8% agarose gels. Approxi-
mately 8.0 kbp of fragments digested with SphI and KpnI
and 7.5 kbp with BamHI and SphI were extracted from
the gels and inserted between the SphI and KpnI sites of
plasmid pUC19 and the BamHI and SphI sites of pUC18,

respectively. The E. coli strain JM 109 cells were trans-
formed with these recombinant plasmids and grown on an
LB plate containing 50 lgÆml
)1
ampicillin, 1 mm isopro-
pyl-b-d-thiogalactopyranoside (IPTG) and 200 lgÆml
)1
5-bromo-4-chloro-3-indolyl-b-d-galactoside. The transform-
ants were then subjected to colony hybridization as previ-
ously described [2], which enabled two plasmids, pPDH1
containing the PDH1 gene (insert length; 8.1 kbp) and
pPDH2 containing the PDH2 gene (insert length; 7.4
kbp), to be obtained and used as templates for DNA
R. Kawakami et al. FAD, FMN and ATP-containing amino-acid dehydrogenase
FEBS Journal 272 (2005) 4044–4054 ª 2005 FEBS 4051
sequencing. The sequencing was carried out using the
dideoxynucleotide chain-termination method [19] with an
automated DNA 377 A sequencer (Applied Biosystems,
Tokyo, Japan). The nucleotide sequences were analyzed
using genetyx-sv ⁄ rc9.0 gene analysis software (GEN-
ETYX Corp., Tokyo), and were submitted to the DDBJ
under the accession numbers AB196181 (for the PDH1
gene) and AB196182 (for the PDH2 gene).
Expression of the PDH1 gene and purification
of its product
PDH1 forms a complex comprised of two different subunits
with molecular masses of 56 and 43 kDa, which were desig-
nated a1 and b1, respectively; the genes encoding them
were designated a1 and b1, respectively. Two sets of PCR
primers were prepared to construct the expression plasmid

for the PDH1 gene: 5¢-AGGGATGCATATGAGACCT
CTAGATCTAAC-3¢ and 5¢-AGGCCCCGGGTCACCTC
CTAGCTAGAATTC-3¢ for a1; and 5¢-AGGTGATC
ATATGCTTCTAGAGAAGAGTGAAATA-3¢ and 5¢-AG
AGGATCCTCAGCCCATTTGGAGGGCGG-3¢ for b1.
In each case, the forward primer introduced a unique NdeI
restriction site that overlapped the 5¢-initiation codon, and
the reverse primer introduced a unique SmaIorBamHI
restriction site proximal to the 3¢-end of the termination
codon. PCR was carried out using pPDH1 as the template,
after which the amplified fragments were digested with NdeI
and SmaI for a1 and with NdeI and BamHI for b1. For
ligation to a1, the plasmid pET11a was digested with
BamHI, blunted and then further digested with NdeI. For
ligation to b1, the plasmid pET11a was digested with NdeI
and BamHI. The a1 and b 1 gene fragments were introduced
into pET11a after linearizing it with NdeI and blunted-
BamHI to generate pET11a ⁄ a1 and with NdeI and BamHI
to generate pET11a ⁄ b1, respectively. pET11a ⁄ a1 was then
digested with ClaI, blunted and further digested with SphI.
The resultant fragment containing a1 and the T7 promoter
was introduced into pET11a ⁄ b1 digested with SphI and
BglII (the BglII site had already been blunted) to generate
the expression plasmid pEPDH1, which was then used to
transform E. coli strain BL21 CodonPlus RIL (DE3) cells.
The transformants were grown for 8 h at 37 ° C in SB med-
ium (1.2% tryptone peptone, 2.4% yeast extract, 1.25%
K
2
HPO

4
, 0.38% KH
2
PO
4
and 0.5% glycerol) containing
ampicillin (100 lgÆml
)1
), after which IPTG was added to
1mm, and cultivation was continued for an additional 4 h.
The cells were then collected by centrifugation (10 000 g,
20 min), suspended in 10 mm potassium phosphate, pH 7.0
containing 1 mm DTT and disrupted by sonication. After
centrifugation (20 000 g, 20 min), the supernatant was col-
lected and heated at 90 °C for 10 min, the precipitant was
removed by centrifugation (20 000 g, 20 min), and ammo-
nium sulfate was added to the supernatant to 40% satura-
tion. This enzyme solution was then applied to a Butyl
Toyopearl column (2.6 · 6 cm) equilibrated with 10 mm
potassium phosphate, pH 7.0 containing 0.1 mm DTT (buf-
fer B) supplemented with 40% ammonium sulfate. After
washing the column with the same buffer, the enzyme was
eluted with a 300-mL linear gradient of 40–0% ammonium
sulfate in buffer B. The active fractions were pooled, con-
centrated using an Amicon Ultra-15 (30 000 MWCO),
and applied to a Superdex 200 gel filtration column
(2.6 · 60 cm) on an FPLC system. Buffer B containing 0.2
m NaCl was used as the elution buffer and the flow rate
was 2 mL Æ min
)1

. The active fractions were pooled and dia-
lyzed against buffer B. All buffers used in the purification
were degassed before use.
Subunit gene expression and product purification
Separate expression systems for the a1 and b1 subunits
were constructed to determine the function of each subunit.
For production of the a1 subunit, E. coli strain BL21
CodonPlus RIL (DE3) cells transformed with pET11a ⁄ a1
were grown for 6 h at 37 °C in SB medium in the presence
of ampicillin (100 lgÆml
)1
), after which IPTG was added to
1mm, and cultivation was continued for an additional 3 h.
The cells were then collected by centrifugation (10 000 g,
20 min), suspended in 10 mm potassium phosphate, pH 7.0
containing 1 mm DTT, and disrupted by sonication. After
centrifugation (20 000 g, 20 min), the supernatant was col-
lected and heated at 80 °C for 10 min and centrifuged
(20 000 g, 20 min) again to remove the denatured proteins.
Ammonium sulfate was added to the supernatant to 20%
saturation, after which the enzyme solution was applied to
a Butyl Toyopearl column (2.6 · 6 cm) equilibrated with
buffer B supplemented with 20% ammonium sulfate. After
washing the column with the same buffer, the enzyme was
eluted with a 300-mL linear gradient of 20–0% ammonium
sulfate in buffer B. The active fractions were pooled and
used as the purified enzyme preparation.
When a pET11a ⁄ b1 expression system was used for the
production of the b1 subunit, the enzyme produced was
found mainly in the insoluble fraction as an inclusion body.

To avoid that, we changed the expression system to
pET15b ⁄ b1. The b1 gene fragment, which had been pre-
pared for construction of pET11a ⁄ b1, was introduced into
plasmid pET15b linearized with NdeI and BamHI to gener-
ate pET15b ⁄ b1, which was then used to transform E. coli
strain BL21 CodonPlus RIL (DE3) cells. The transformants
were grown for 6 h at 37 °C in SB medium in the presence
of ampicillin (100 lgÆml
)1
), after which IPTG was added to
1mm, and cultivation was continued for an additional 3 h.
The cells were then collected by centrifugation, suspended
in 10 mm Tris ⁄ HCl, pH 7.5, and disrupted by sonication.
After centrifugation, imidazole and NaCl were added to the
supernatant to 50 mm and 0.5 m, respectively, and the
resultant solution was applied to a HiTrap nickel-charged
chelating column (2.6 · 6 cm; Amersham Biosciences)
FAD, FMN and ATP-containing amino-acid dehydrogenase R. Kawakami et al.
4052 FEBS Journal 272 (2005) 4044–4054 ª 2005 FEBS
equilibrated with 10 mm Tris ⁄ HCl, pH 7.5 containing
50 mm imidazole and 0.5 m NaCl. After washing the col-
umn with the same buffer, the enzyme was eluted with
10 mm Tris ⁄ HCl, pH 7.5 containing 500 mm imidazole and
0.5 m NaCl. The active fractions were pooled, heated at
70 °C for 10 min and centrifuged to remove the precipit-
ants, and then the supernatant was dialyzed against 10 mm
Tris ⁄ HCl, pH 7.5. The resultant enzyme solution was used
as the purified preparation. All buffers used in the purifica-
tion were degassed before use.
Determination of flavin and other prosthetic

groups
Flavin compounds and other prosthetic groups from the
enzyme were extracted with 1% (w ⁄ v) perchloric acid. The
solution was then centrifuged, and after neutralizing the
supernatant with K
2
CO
3
, it was subjected to HPLC using
TSKgel ODS-80Tm (Tosoh) and Asahipak GS-320HQ
(Shodex, Tokyo) columns. An isocratic elution (10 min)
with 10 mm potassium phosphate, pH 6.0 followed by a lin-
ear gradient (30 min) between 0 and 70% methanol in the
same solution was used for the TSKgel ODS-80Tm column.
An isocratic elution with 200 mm sodium phosphate,
pH 5.0 was used for the Asahipak GS-320HQ column. The
flow rate was 1.0 mLÆmin
)1
, and the absorbance at 260 nm
of the effluent from the column was monitored.
Polyacrylamide gel electrophoresis and molecular
mass determination
Native PAGE was carried out with 7.5% polyacrylamide
gel according to the method of Davis [20]. Activity staining
was carried out at 50 °C, as previously described [4].
SDS ⁄ PAGE was carried out using 15% polyacrylamide gel
containing 0.1% SDS according to the method of Leammli
[21]. The subunit molecular mass was determined using
eight marker proteins (New England Biolabs).
The molecular masses of the native enzymes were deter-

mined by gel filtration column chromatography using Supe-
rose 6 HR (Amersham Biosciences) with thyroglobulin
(669 kDa), ferritin (440 kDa), catalase (232 kDa), aldolase
(158 kDa), albumin (67 kDa) and ribonuclease A
(13.7 kDa) serving as molecular standards (Amersham Bio-
science).
Analysis of the N-terminal amino-acid sequences
The N-terminal amino-acid sequences of the enzymes were
analyzed using an automated Edman degradation protein
sequencer. After SDS ⁄ PAGE, the proteins were blotted
onto polyvinylidene difluoride membranes and sequenced
using a PPSQ-10 Protein Sequencer (Shimadzu, Kyoto,
Japan).
Acknowledgements
This study was supported in part by the Pioneering
Research Project in Biotechnology of the Ministry of
Agriculture, Forestry and Fisheries of Japan and by the
National Project on Protein Structural and Functional
Analyses promoted by the Ministry of Education, Sci-
ence, Sports, Culture, and Technology of Japan. R. K.
was supported in part by the Sasakawa Scientific
Research Grant from the Japan Science Society.
References
1 Frew JE & Hill HA (1987) Electrochemical biosensors.
Anal Chem 59, 933A–944A.
2 Satomura T, Kawakami R, Sakuraba H & Ohshima T
(2002) Dye-linked d-proline dehydrogenase from
hyperthermophilic archaeon Pyrobaculum islandicum is a
novel FAD-dependent amino acid dehydrogenase. J Biol
Chem 277, 12861–12867.

3 Kawakami R, Sakuraba H & Ohshima T (2004) Gene
and primary structures of dye-linked l-proline dehydro-
genase from the hyperthermophilic archaeon Thermo-
coccus profundus show the presence of a novel
heterotetrameric amino acid dehydrogenase complex.
Extremophiles 8, 99–108.
4 Sakuraba H, Takamatsu Y, Satomura T, Kawakami R
& Ohshima T (2001) Purification, characterization, and
application of a novel dye-linked l-proline dehydrogen-
ase from a hyperthermophilic archaeon, Thermococcus
profundus. Appl Environ Microbiol 67, 1470–1475.
5 Kawarabayasi Y, Sawada M, Horikawa H, Haikawa Y,
Hino Y, Yamamoto S, Sekine M, Baba S, Kosugi H,
Hosoyama A, et al. (1998) Complete sequence and gene
organization of the genome of a hyperthermophilic
archaebacterium, Pyrococcus horikoshii OT3. DNA Res
5, 55–76.
6 Wierenga RK, Terpstra P & Hol WGD (1986) Predic-
tion of the occurrence of the ADP-binding bab-fold in
proteins, using an amino acid sequence fingerprint.
J Mol Biol 187, 101–107.
7 Murataliev MB, Feyereisen R & Walker FA (2004)
Electron transfer by diflavin reductases. Biochim Biophys
Acta 1698, 1–26.
8 Porter TD (1991) An unusual yet strongly conserved
flavoprotein reductase in bacteria and mammals. Trends
Biochem Sci 16, 154–158.
9 Porter TD & Kasper CB (1986) NADPH-cytochrome
P450 oxidoreductase: flavin mononucleotide and flavin
adenine dinucleotide domains evolved from different

flavoproteins. Biochemistry 25, 1682–1687.
10 Ostrowski J, Barber MJ, Rueger DC, Miller BE, Siegel
LM & Kredich NM (1989) Charactrerization of the
flavoprotein moieties of NADPH-sulfite reductase from
Salmonella typhimurium and Escherichia coli. Physico-
R. Kawakami et al. FAD, FMN and ATP-containing amino-acid dehydrogenase
FEBS Journal 272 (2005) 4044–4054 ª 2005 FEBS 4053
chemical and catalytic properties, amino acid sequence
deduced from DNA sequence of cysJ, and comparison
with NADPH-cytochrome P-450 reductase. J Biol Chem
264, 15796–15808.
11 Alderton WK, Cooper CE & Knowles RG (2001) Nitric
oxide synthases: structure, function and inhibition. Bio-
chem J 357, 593–615.
12 Masters BSS, MacMillan K, Sheta EA, Nishimur JS,
Roman LJ & Martasek P (1996) Neuronal nitric oxide
synthase, a modular enzyme formed by convergent evo-
lution: structure studies of a cystein thiolate-liganded
heme protein that hydroxylates l-arginine to produce
NO as a cellular signal. FASEB J 10, 552–558.
13 Stuehr DJ (1997) Structure-function aspects in the nitric
oxide synthases. Annu Rev Pharmacol Toxicol 37, 339–
359.
14 Stuehr DJ (1999) Mammalian nitric oxide synthetases.
Biochim Biophys Acta 1411, 217–230.
15 Siegel LM, Murphy MJ & Kamin H (1973) Reduced
nicotinamide adenine dinucleotide phosphate–sulfite
reductase of enterobacteria. I. The Escherichia coli
hemoflavoprotein: molecular parameters and prosthetic
groups. J Biol Chem 248, 251–264.

16 Lim LW, Mathews SF & Steenkamp DJ (1988) Identifi-
cation of ADP in the iron–sulfur flavoprotein trimethyl-
amine dehydrogenase. J Biol Chem 263, 3075–3078.
17 Bradford MM (1976) A rapid and sensitive method for
the quantitation of microgram quantities of protein util-
izing the principle of protein-dye binding. Anal Biochem
72, 248–254.
18 Murray MG & Thompson WF (1980) Rapid isolation
of high molecular weight plant DNA. Nucleic Acids Res
8, 4321–4325.
19 Sanger DE, Nicklen S & Coulson AR (1977) DNA
sequencing with chain-terminating inhibitors. Proc Natl
Acad Sci USA 74, 5463–5467.
20 Davis BJ (1964) Disc electrophoresis-2. Method and
application to human serum proteins. Ann NY Acad Sci
121, 404–427.
21 Leammli UK (1970) Cleavage of structural proteins dur-
ing the assembly of the head bacteriophage T4. Nature
227, 680–685.
FAD, FMN and ATP-containing amino-acid dehydrogenase R. Kawakami et al.
4054 FEBS Journal 272 (2005) 4044–4054 ª 2005 FEBS