Spectroscopic characterization of a higher plant heme
oxygenase isoform-1 from Glycine max (soybean)
)
coordination structure of the heme complex and
catabolism of heme
Tomohiko Gohya
1
, Xuhong Zhang
2
, Tadashi Yoshida
2
and Catharina T. Migita
1
1 Department of Biological Chemistry, Faculty of Agriculture, Yamaguchi University, Japan
2 Department of Biochemistry, Yamagata University School of Medicine, Japan
Heme oxygenase (HO, EC 1.14.99.3) catalyzes the con-
version of heme to biliverdin IX
a
, CO and free iron
through successive reduction and oxygenation reac-
tions in the presence of molecular oxygen and elec-
trons supplied by NADPH. Studies on the structure
and function of HO have been conducted mostly in
mammalian enzymes, as HO was first identified in
mammals [1–3]. During the last decade, the HO genes
Keywords
ferredoxin; heme catabolism; heme
complex; higher-plant heme oxygenase;
spectroscopic characterization
Correspondence
C. T. Migita, Department of Biological

in the presence of
NADPH ⁄ ferredoxin reductase ⁄ ferredoxin, or ascorbate. During the heme
conversion, an intermediate with an absorption maximum different from
that of typical verdoheme–heme oxygenase or CO–verdoheme–heme oxyge-
nase complexes was observed and was extracted as a bis-imidazole com-
plex; it was identified as verdoheme. A myoglobin mutant, H64L, with
high CO affinity trapped CO produced during the heme degradation. Thus,
the mechanism of heme degradation by GmHO-1 appears to be similar to
that of known heme oxygenases, despite the low sequence homology. The
heme conversion by GmHO-1 is as fast as that by SynHO-1 in the presence
of NADPH ⁄ ferredoxin reductase ⁄ ferredoxin, thereby suggesting that the
latter is the physiologic electron-donating system.
Abbreviations
AtHO-1, heme oxygenase isoform 1 of Arabidopsis thaliana; BVR, biliverdin reductase; CPR, cytochrome P450 reductase; Fd, plant
ferredoxin; FNR, ferredoxin:NADP
+
reductase; GmHO-1, heme oxygenase isoform 1 of Glycine max; heme, iron protoporphyrin IX, either
ferrous or ferric forms; hemin, ferric protoporphyrin IX; HO, heme oxygenase; hydroxyheme, iron meso-hydroxyl protoporphyrin IX; KPB,
potassium phosphate buffer; rHO-1, heme oxygenase isoform 1 of Rattus norvegicus; SynHO-1, heme oxygenase isoform 1 of
Synechocystis sp. PCC 6803.
5384 FEBS Journal 273 (2006) 5384–5399 ª 2006 The Authors Journal compilation ª 2006 FEBS
have been identified in a wide range of biological
species, especially in pathogenic bacteria, and some of
them have been expressed and characterized [4–6].
Higher-plant HOs, however, have not been investi-
gated on a molecular basis by applying multiple spect-
roscopic methods to the purified protein, and are now
the least studied HOs.
HO in plants is one of the plastid enzymes partici-
pating in phytochromobilin synthesis. This enzyme

HOs are highly homologous to each other; for example,
soybean (Glycine max) HO isoform-1 (GmHO-1) has
71.7% homology to A. thaliana HO-1 (AtHO-1), and
HOs from other plant species have similar levels of
homology. On the contrary, the homology in amino
acid sequences between plant HOs and HOs from other
biological species is quite low, e.g. 21% to cyanobacte-
rial HO-1 (Synechocystis sp. PCC 6803) (SynHO-1),
22% to rat HO-1 (rHO-1), 23% to corynebacterial
HmuO, or 21% to neisserial HemO. Comparison of the
sequence alignment reveals that the catalytically pivotal
residues, Gly139 and Asp140, of human HO-1 (and
also rHO-1) are replaced by Ala and His, respectively,
Fig. 1. Amino acid sequence of GmHO-1 as compared with the sequences of Arabidopsis, Synechocystis and rat HO-1s. The lightly shaded
letters indicate residues with sequence identity, and heavily shaded histidine residues are proximal heme ligands. Bars below the alignments
show a-helical parts (AH) in the crystal structures of heme–SynHO-1 and heme–rHO-1 and those presumed for heme–GmHO-1.
T. Gohya et al. Heme catabolism by soybean heme oxygenase-1
FEBS Journal 273 (2006) 5384–5399 ª 2006 The Authors Journal compilation ª 2006 FEBS 5385
in GmHO-1 (Fig. 1), although the residues comprising
the distal F-helix part are relatively well conserved in
the whole sequence of GmHO-1. In mammalian HO-1,
Gly139 directly contacts with heme, and Asp140 is
known to be a key residue for enzymatic activity [3,13].
In addition, the proximal residue for the substrate
heme binding, His25 in rHO-1 (also in human HO-1),
is replaced by Lys, and only one His in the correspond-
ing distal A-helix part occupies a position 13 residues
away from the N-terminus. Moreover, the Arg183 resi-
due of mammalian HO-1, which participates in a-meso-
specific heme decomposition, is not conserved (Leu in

the heme complexes of SynHO-1 and rHO-1. We
found that, in spite of the low homology of the amino
acid sequence with those of known HOs, the heme–
GmHO-1 complex has similar spectroscopic character-
istics to those of the heme complexes of cyanobacterial,
mammalian or bacterial HOs [15,18,19]. GmHO-1 con-
verts combined heme into biliverdin IX
a
, retaining
a-regiospecificity, and releasing CO and free iron, in
the presence of oxygen and NADPH ⁄ FNR ⁄ Fd, with-
out requiring additional reducing agents, albeit the
coordination structure of the verdoheme intermediate is
apparently different from that of the known verdo-
heme–HO complexes.
Results
Expression and purification of GmHO-1
By culturing the cells at two temperatures, first at
37 °C and then at 25 °C, we avoided the accumulation
of inclusion bodies of GmHO-1. The harvested cells
were brown in color, unlike the cells expressing rHO-1
or SynHO-1, which were greenish due to the accumu-
lated biliverdin; nevertheless, the E. coli cells expressed
active GmHO-1, as will be described later. It has been
reported that the E. coli cells expressing the HY1 gene
encoding AtHO-1 have a yellowish-brown tinge [9]. We
purified the GmHO-1 from the soluble fraction by
ammonium sulfate fractionation and subsequent col-
umn chromatography on Sephadex G-75 and DE-52.
The ammonium sulfate fraction and active G-75 frac-

¼ 5.95, g
y
¼ 5.68 and g
z
¼ 2.00 (Fig. 3A).
Here, anisotropy of the g
xy
component is apparently
larger than that of heme–rHO-1, as shown in the partly
expanded spectra (a-1 in Fig. 3), indicating that in-plane
anisotropy of heme is relatively large. In addition, small
amounts of low-spin species are also observed in the
neutral solution, as distinctly seen in the partly expan-
ded spectrum (a-2 in Fig. 3). The EPR spectrum of the
15
NO-bound GmHO-1 (nitrosylheme GmHO-1) is
characteristic of six-coordinate heme proteins with the
Heme catabolism by soybean heme oxygenase-1 T. Gohya et al.
5386 FEBS Journal 273 (2006) 5384–5399 ª 2006 The Authors Journal compilation ª 2006 FEBS
nitrogenous proximal ligand, indicating hyperfine
splitting due to a
14
N nucleus (nuclear spin 1, giving the
triplet splitting) in addition to a
15
N nucleus (nuclear
spin 1 ⁄ 2, giving the doublet splitting) at the g
2
compo-
nent (Fig. 3C). This strongly suggests coordination of a

GmHO-1
k
max
(nm)
SynHO-1
k
max
(nm)
rHO-1
k
max
(nm)
Soret Visible Soret Visible Soret Visible
Ferric 405 500, 630 402 498, 631 404 500, 631
e (m
M
)1
Æcm
)1
) 127 128 140
Ferrous 428 557 427 555 431 554
Oxy 415 541, 578 410 537, 574 410 540, 575
CO-bound 420 539, 569 427 536, 566 419 535, 568
Alkaline 414 539, 577 427 537, 575 414 540, 575
pK
a
8.2 8.9 7.6
Fig. 2. Absorption spectra of the various forms of heme–GmHO-1.
Spectra are of the ferric (red), ferrous (blue), ferrous–CO (black)
and ferrous–oxy (green) forms. Inset: titration plots of GmHO-1

Accordingly, this species was determined to be the
alkaline form of heme–GmHO-1, which coordinates a
hydroxyl anion at the distal site of heme. Another
low-spin species, denoted as g* in Fig. 3A, seems to be
a denatured form of heme–GmHO-1, because similar
low-spin species were sometimes observed for other
heme–HO complexes (data not shown).
Single species of the alkaline form were observed for
heme–GmHO-1, the same as for rHO-1, but distinct
from SynHO-1, which exhibits two kinds of alkaline
forms (Table 2). The anisotropy in g-values, g
ak
1
–g
ak
3
,
of the alkaline form of heme–GmHO-1 is somewhat
smaller than that of the other two HO complexes. This
might indicate that the axial ligand field is relatively
strong in GmHO-1, due to steric control imposed by
the distal helix, in accord with the observation that
the in-plane anisotropy is large in the ferric state of
heme.
Determination of the proximal ligand
Based on the EPR results for nitrosylheme–GmHO-1,
candidates for the proximal ligand of heme were
searched for in the amino acid sequence of GmHO-1,
around the position corresponding to the proximal His
of other HOs (Fig. 1). His30 was found to be only

GmHO-1, the apparent equilibrium constant for bind-
ing of nitrogenous ligands, imidazole and azide, to
heme–GmHO-1 was evaluated. As imidazole was
added to the solution of heme–GmHO-1, the Soret
Table 2. EPR parameters of the low-spin and
15
NO-bound forms of
heme–HO-1 complexes.
Protein
GmHO-1 SynHO-1 rHO-1
Alkaline Neutral Alkaline-1 Alkaline-2 Alkaline
Low spin
g
1
2.63 2.86 2.78 2.68 2.67
g
2
2.21 2.29 2.14 2.20 2.21
g
3
1.82 1.59 1.74 1.80 1.79
15
NO
a (
15
N), G 27 31 26
a (
14
N), G 7.6 7.1 7.5
g

between the Soret band maxima of the azide-free
form at 405 nm and of the azide-bound form at
421 nm, so the relative amounts of the nonbound and
azide-bound forms were estimated directly from the
values of absorbance at these maxima (data not
shown). The estimated association constants for imi-
dazole and azide binding to heme–GmHO-1 are sum-
marized and compared with those for heme–SynHO-1
and heme–rHO-1 in Table 3.
Heme degradation by GmHO-1
Heme degradation by HOs can be monitored through
the changes in optical absorption spectra, because
ferric heme, oxyheme and verdoheme intermediate
complexes of HO and free biliverdin, products of the
HO reaction, all exhibit characteristic absorption
bands (Scheme 1). As shown in Fig. 5A, addition of
ascorbate to the solution of heme–GmHO-1 initiates
the reaction, as revealed by gradual diminution of the
Soret band. After several minutes, a broad band
appears at around 660–675 nm; this increases in inten-
sity with time, indicating the formation of biliverdin.
In this case, neither bands of the oxy form (at 541 and
578 nm) nor bands of verdoheme and CO–verdoheme
(at  690 and at  640 nm, respectively) are observed,
implying that the first step of heme conversion is rate-
limiting. The apparent initial rate of heme degradation
by GmHO-1 is about three times higher than that of
degradation by SynHO-1, but nearly four times lower
than that of degradation by rHO-1 in the presence of
1200 equivalents of ascorbate (Table 4).

) 35 (1) 210 (6) 1400 (40)
K
azide
(· 10
2
M
)1
) 43 (1) 17 (0.4) 200 (4.7)
Scheme 1. Pathways of heme degradation by heme oxygenase (HO), elucidated for the mammalian HO-1. Most HOs other than those of
mammalian origin are also known to cleave heme at the a-meso position selectively to produce biliverdin IX
a
.
T. Gohya et al. Heme catabolism by soybean heme oxygenase-1
FEBS Journal 273 (2006) 5384–5399 ª 2006 The Authors Journal compilation ª 2006 FEBS 5389
NADPH ⁄ FNR ⁄ Fd of GmHO-1 was also compared
with that of SynHO-1 under similar conditions, as well
as that of rHO-1 in the presence of NADPH ⁄ cyto-
chrome P450 reductase (CPR; NADPH:cytochrome
P450 reductase; EC 1.6.2.4) (Table 4). This result sug-
gests that the GmHO-1 reaction occurs as fast as the
SynHO-1 reaction, supported by the NADPH ⁄ FNR ⁄ Fd
reducing system, and as fast as the rHO-1 reaction sup-
ported by the NADPH ⁄ CPR reducing system under
similar conditions.
Bilirubin assay of the heme oxygenase activity
of GmHO-1
In mammalian HO reactions, the end-product, biliver-
din IX
a
, is further reduced to bilirubin by NADPH:

0.25 l
M; and NADPH of indicated equivalent.
Protein
GmHO-1
v (l
MÆmin
)1
)
SynHO-1
v (lMÆmin
)1
)
rHO-1
v (lMÆmin
)1
)
Ascorbate (1200 eq.) 0.17 0.051 0.63
NADPH (4 eq.) 0.26 0.35 0.52
NADPH (8 eq.) 0.41 0.39 –
Table 5. Bilirubin yields (%) in the heme conversion by HO coupled
with the biliverdin conversion by BVR.
Proteins
GmHO-1 SynHO-1 rHO-1
NADPH (4 eq.) +
NADPH (8 eq.) ⁄ BVR
a
28 28 44
NADPH (20 eq.) +
desferrioxamine (1.1 mg) + BVR
b

proving liberation of CO in the heme conversion by
GmHO-1 in the presence of either ascorbate (Fig. 6A)
or NADPH ⁄ FNR ⁄ Fd (Fig. 6B).
Next, to determine whether verdoheme is produced
in the GmHO-1 reaction, this reaction was performed
under a CO atmosphere. The high partial pressure of
CO should enhance coordination of CO to the verdo-
heme if it is produced. As shown in Fig. 6C, when
ascorbate was added to the heme–GmHO-1 solution
presaturated with CO in a sealed cuvette, the Soret
band maximum shifted from 405 to 420 nm, and peaks
Fig. 6. Detection of CO produced during the GmHO-1 reaction by the H64L mutant of myoglobin (A, B) and detection of CO–verdoheme pro-
duced in the GmHO-1 reaction under CO (C, D). (A) Difference spectra of optical absorption spectra obtained for the reaction of heme–
GmHO-1 (5 l
M) in the presence of H64L (4 lM) minus those for the reaction of H64L (4 lM) alone, after addition of ascorbate (6 mM)at
appropriate times. (B) Difference spectra obtained for the reactions described in (A), except that NADPH (10 l
M), FNR (0.22 lM) and Fd
(1 l
M) were used in place of ascorbate. (C) Spectra obtained for the reaction of heme–GmHO-1 (5 lM) with ascorbate (6 mM). (D) Spectra
obtained for the reaction of heme–GmHO-1 (5 l
M) with FNR (0.22 lM), Fd (1 lM), and NADPH (40 lM). All solutions were in 0.1 M KPB
(pH 7.0).
T. Gohya et al. Heme catabolism by soybean heme oxygenase-1
FEBS Journal 273 (2006) 5384–5399 ª 2006 The Authors Journal compilation ª 2006 FEBS 5391
concomitantly appeared at 539 and 571 nm, indicating
formation of the CO-combined heme–GmHO-1 com-
plex (Table 1). Injection of oxygen gas into this cuvette
decreased the Soret band so that it almost vanished
after 30 min, and instead, an absorption peak
appeared at 637 nm, suggesting the formation of

to nearly one-third and a relatively strong broad
band appeared in the visible region (k
max
¼
660 nm) (Fig. 7). This 660 nm band is very similar to
that observed in heme degradation by GmHO-1 in the
presence of NADPH ⁄ FNR ⁄ Fd (Figs 5B and 6D), sug-
gesting that the same intermediate, namely the 660 nm
species, is accumulated.
To isolate and identify the 660 nm species, the heme–
GmHO-1 reaction was carried out with six equivalents
of H
2
O
2
under anaerobic conditions, to avoid the degra-
dation or successive conversion of the intermediate by
oxygen. The green pigment was extracted from the reac-
tion product with acetone containing imidazole. The
spectrum of the extract is shown in Fig. 8B, exhibiting
peaks at 404, 534, 636 and 684 nm; the latter two differ
from the 660 nm band of the protein complex (Fig. 8A).
When the solution of the extract was exposed to CO, the
684 nm band gradually shifted to 636 nm. The reported
band maxima of bis-imidazole-coordinated verdoheme
are 400, 536 and 685 nm [23], so the spectra shown in
Fig. 8B are considered to be a mixture of a CO-coordi-
nated monoimidazole complex and a bis-imidazole com-
plex of verdoheme. Using the same methods, extracts
of the H

(6 eq. of the heme) under anaerobic condi-
tions. (B–D) Acetone extracts containing excess imidazole from the
reaction mixtures of the heme–HO-1 complexes with H
2
O
2
(6 eq.)
in anaerobic conditions.
Heme catabolism by soybean heme oxygenase-1 T. Gohya et al.
5392 FEBS Journal 273 (2006) 5384–5399 ª 2006 The Authors Journal compilation ª 2006 FEBS
species produced in the course of heme degradation by
GmHO-1 is verdoheme.
Discussion
Homology and heme binding
Estimation of the secondary structure of GmHO-1 sug-
gests that, in spite of low homology in the amino acid
sequence, GmHO-1 protein should consist of eight
a-helices common to other HOs whose crystal struc-
tures are known [3,25–29] (Fig. 1). A recent modeling
study on pea HO-1 also suggested a similar structure
[12]. In GmHO-1, however, a critical residue for HO
activity, the proximal His, which fixes heme to the
heme pocket of the enzyme and participates in the
activation of heme, is not at the position in which
it is present in SynHO-1 (His17) or rHO-1 (His25).
Instead, there is only one His (His30) in the predic-
ted proximal A-helix region (Fig. 1). Experiments on
heme binding to GmHO-1 have demonstrated 1 : 1
stoichiometry, and the result of EPR investigations of
nitrosylheme–GmHO-1 indicate a nitrogenous proxi-

molecule interacts with respective Asp residues
(Asp140 of rHO-1 and Asp131 of Syn HO-1) in the
distal helix through hydrogen-bonding networks via
water molecules in crystals [25,27]. Unfortunately, the
resolution of the reported crystal structures of heme–
SynHO-1 and rHO-1 is not sufficiently high to allow
accurate quantitative comparison of the length of the
hydrogen-bonding network, so the reason for such a
large difference in pK
a
values is unclear. In GmHO-1,
the corresponding residue to the Asp is His150, which
is also competent as a partner of the indirect hydrogen
bonding with the heme-bound water. This difference in
the hydrogen-bonding counterpart would also affect
the pK
a
value.
The EPR parameters of the nitrosylheme–HO com-
plexes also give useful information on the heme pocket
structure. As shown in Table 2, the hyperfine splitting
constants of the
15
N nucleus of the distal NO and of
the
14
N nucleus of the proximal His of the GmHO-1
complex are closer to those of the rHO-1 complex than
to those of the SynHO-1 complex. Thus, Fe–N(O)
r-bonding in heme–GmHO-1 might be comparably

heme, and conversely, the steric effect of the distal
residues might reduce the accessibility of azide. The
amino acid residues comprising the presumed distal
T. Gohya et al. Heme catabolism by soybean heme oxygenase-1
FEBS Journal 273 (2006) 5384–5399 ª 2006 The Authors Journal compilation ª 2006 FEBS 5393
helix of GmHO-1 include several hydrophobic amino
acids, and the distal side of the heme pocket of Syn-
HO-1 has been reported to be less polar in total than
that of rHO-1 [27]. Thus, the less polar characteristics
of the heme pocket of GmHO-1 might be associated
with the smaller values of K
azide
for GmHO-1 and
SynHO-1, and with the ferric character of the heme.
The K
imidazole
values of the three kinds of heme–
HO-1 complexes are very different, and that of heme–
rHO-1 is strikingly large. The K
imidazole
might reflect
the magnitude of the vacancy in the distal side or the
steric hindrance at the opening of heme pockets, due
to the relatively large size of imidazole molecules,
although flexibility of the distal pocket would also
affect it. Comparison of the opening side of heme
pockets in crystal structures shows that Ile137 of
SynHO-1 droops over the heme distal site, whereas the
corresponding Val146 of rHO-1 is located relatively
high above the distal side of heme (as is also true in

in the
presence of either ascorbate or NADPH ⁄ FNR ⁄ Fd. The
oxyheme is apparently observed in the time-dependent
spectra in Fig. 5B, and CO excision from heme has
been verified. Furthermore, the intermediate with
k
max
¼ 660 nm has been confirmed to be verdoheme
(Figs 6 and 8). Hydrogen peroxide also drives the con-
version of heme–GmHO-1 to verdoheme–GmHO-1, as
is the case for other heme–HO complexes (Fig. 7) [24].
Consequently, GmHO-1 has been established to be an
HO that site-specifically oxygenates and cleaves heme
into biliverdin IX
a
, CO and free iron in a manner sim-
ilar to mammalian HO enzymes (Scheme 1).
The initial heme degradation rate of GmHO-1, indi-
cated in Table 4, is comparable to that of SynHO-1 in
the presence of NADPH, FNR, and Fd, and also to
that of rHO-1 in the presence of NADPH and CPR.
Here, NADPH ⁄ CPR has been established to be the
physiologic electron-donating system for mammalian
HO. The yield of ferric biliverdin in the heme–GmHO-
1 reaction in the presence of NADPH ⁄ FNR ⁄ Fd is
inferred to be also comparable to that in the heme–
rHO-1 reaction, because chelating of the ferric iron
enhances the bilirubin yield of the GmHO-1 and
SynHO-1 reactions to a similar level as that with the
rHO-1 reaction (Table 5). The lower yield of bilirubin

bate-supported GmHO-1 reaction (Fig. 5A). Even for
mammalian HO-1, the reducing ability of ascorbate
is known to be thermodynamically insufficient to
reduce ferric heme–HO-1 (E¢
Asc
¼ + 80 mV vs.
Heme catabolism by soybean heme oxygenase-1 T. Gohya et al.
5394 FEBS Journal 273 (2006) 5384–5399 ª 2006 The Authors Journal compilation ª 2006 FEBS
E
heme–human HO-1
¼ ) 65 mV, at pH 7.4 and 25 °C)
[34]. We consider that this high resistance to the
reduction of heme is disadvantageous to the plant
HOs if they utilize ascorbate as the reducing agent
for HO reactions.
Low CO affinity of heme–GmHO-1
It has been reported that heme catabolism by mamma-
lian HOs under a CO atmosphere stops at the CO–
verdoheme stage, due to the high affinity of CO
for the verdoheme [35]. In contrast to this, in the
NADPH ⁄ FNR ⁄ Fd-assisted heme–GmHO-1 reaction
performed under a CO atmosphere, CO-bound verdo-
heme was not detected (Fig. 6D), and in the ascorbate-
supported reaction, CO–verdoheme was observed only
momentarily (Fig. 6C). Thus, the affinity of verdoh-
eme–GmHO-1 for CO appears to be very low. The
optical absorption spectrum of verdoheme–GmHO-1 is
unique, having a broad absorption band with a maxi-
mum at 660 nm (Fig. 7), not at the 686 nm of verdoh-
eme–rHO-1 [23]. Therefore, the electronic state of the

EagI, NcoI and HindIII. First, a 50-bp double-stranded
synthetic oligonucleotide with unique sites for the afore-
mentioned restriction enzymes was ligated between the NdeI
and HindIII sites of a T7-promotor-based bacterial expres-
sion vector pMW172, to make a plasmid tentatively
referred to as pMW-A. Ten oligonucleotides and their
complements, 44–91 nucleotides in length, were synthesized
to construct a 681-bp synthetic gene coding for the mature-
type GmHO-1 from the ATG initiation codon to the TAA
stop codon. Each nucleotide was phosphorylated with T4
polynucleotide kinase, and then annealed with its comple-
ment to make a double-stranded DNA, e.g. Oligo I to
Oligo X. Oligo I was designed so that the 5¢-end could be
ligated to the NdeI site, whereas its 3¢ cohesive end was
complementary to the 5¢-end of Oligo II. The 3¢-end of
Oligo II could be ligated to the BstEII site. Similarly, the
5¢-ends of Oligos III, V, VII and IX were designed to ligate
to the BstEII, BglII, EagI and NcoI sites, respectively, and
their 3¢-ends had sequences for ligation to the 5¢-ends of
Oligo IV, VI, VIII and X. The 3¢-end of Oligo X had a
sequence designed to ligate to the HindIII site. To complete
the GmHO-1 expression vector pMWGmHO-1, double-
stranded Oligo I to Oligo X were ligated step by step into
the restriction enzyme sites of pMW-A.
To construct an expression plasmid for the H30G mutant
of GmHO-1, PCR was used according to the method of
Nelson and Long [36].
The nucleotide sequence was determined with an Applied
Biosystems (Foster City, CA, USA) 373A DNA sequencer.
Expression of GmHO-1 and purification

phosphate buffer (KPB) (pH 7.4) and applied to a Sepha-
dex G-75 column (3.6 · 50 cm), pre-equilibrated with the
same buffer. The protein fractions eluted in the KPB, with
an intense 26 kDa band on SDS ⁄ PAGE, were collected
and directly loaded onto a column of DE-52 (2.6 · 28 cm).
The column was washed with 50 mL of 20 mm KPB
(pH 7.4) ⁄ 50 mm KCl, and the protein was eluted with
400 mL of 20 mm KPB (pH 7.4), using a linear gradient of
50–250 mm KCl. Only fractions with a single band at
26 kDa on SDS ⁄ PAGE were collected.
Other proteins
The following proteins were expressed in E. coli and
purified to apparent homogeneity on SDS ⁄ PAGE accord-
ing to the methods described in each reference: SynHO-1
[15], a truncated form of rat HO-1 [37], H64L mutant of
myoglobin [38], maize ferredoxin type III [39], maize
FNR [40], a truncated form of human CPR [41], and rat
BVR [42].
Heme binding and preparation of heme–GmHO-1
To determine the stoichiometry of heme binding to
GmHO-1, 24 lL of a 200 lm protein solution (0.1 m KPB,
pH 7.0) was titrated with each 2 lL of a 400 lm hemin
solution, prepared by diluting 2 mm alkaline solution
(NaOH, 10 mm) in 0.1 m KPB (pH 7.0). Absorbance at
405 nm was recorded for each addition of the hemin solu-
tion to construct titration curves vs. the volume of added
hemin solution. Separately, the absorbance at 405 nm of
each free hemin solution was measured, and these were
plotted together with that of the corresponding hemin–
protein solutions. The equivalent was determined at an

each addition. Optical absorption spectra were recorded
after each addition of azide, and the absorbance values at
405 and 419 nm were monitored to estimate the mole frac-
tions of azide-free and azide-bound forms of the complex,
respectively. Then both mole fractions were plotted against
the concentration of azide of each solution. The intersection
point corresponding to the mole fraction of 0.5 of the
respective forms gives the amount of azide necessary to
attain the equilibrium, thereby giving the equilibrium con-
stant for azide binding, K
azide
. Imidazole binding was also
performed in a similar way with the use of 5 m or 10 m imi-
dazole stock solutions. K
imidazole
was estimated on the basis
of the mole fractions of imidazole-free and imidazole-bound
forms, which were estimated from the values of absorbance
at 402 and 420 nm, respectively, in the difference spectra of
imidazole-free minus imidazole bound forms.
EPR spectra were recorded on a Bruker (Karlsruhe,
Germany) E500 spectrophotometer, operating at 9.35–
9.55 GHz, and at 6–8 K for ferric heme complexes or 20–
20 K for nitrosylheme complexes, with an Oxford liquid
helium cryostat (ESR900) (Oxford, UK). The
15
NO-bound
form of heme–GmHO-1 was prepared by adding dithionite
to the solution of heme–GmHO-1 containing Na
15

Æcm
)1
for heme–SynHO-1, and A
404
, e ¼
140 mm
)1
Æcm
)1
for heme–rHO-1, where the molar
Heme catabolism by soybean heme oxygenase-1 T. Gohya et al.
5396 FEBS Journal 273 (2006) 5384–5399 ª 2006 The Authors Journal compilation ª 2006 FEBS
absorption coefficients were all determined by the pyridine
hemochrome method, using A
557
, e ¼ 34.4 mm
)1
Æcm
)1
).
Heme conversion by GmHO-1 in a CO atmosphere was
carried out in the aforementioned anaerobic UV cell. The
reaction was initiated by the addition of O
2
(1 mL) with a
syringe.
Bilirubin assay
Heme degradation by GmHO-1 or SynHO-1 (each 5 lm in
0.1 m KPB, pH 7.0) was conducted in the presence of
NADPH ⁄ FNR ⁄ Fd (20 lm, 0.22 lm,1lm, respectively),

2
O
2
solution (480 lm) was injected into 2 mL of
a nitrogen-saturated solution of heme–GmHO-1 ( 80 lm in
0.1 m KPB, pH 7.0) in a capped UV cell filled with nitrogen
gas. Immediately, the color of the solution turned to green,
indicating formation of the 660 nm species. Then, a concen-
trated imidazole ⁄ acetone solution was added to the cell (1 : 1
v ⁄ v), and the mixture was incubated for 30 min on ice to
complete the protein precipitation. The supernatant of the
solution was transferred to an eppendorf tube, followed by
spinning down precipitates, and used for the optical absorp-
tion measurements.
Other procedures
The secondary structure of GmHO-1 was obtained by
use of nps@(Network Protein Sequence Analysis). HPLC
analysis of reaction products was performed as previously
described [32]. Homology calculations were executed on
genetyx-mac network version 14.0.3.
Acknowledgements
This work was supported in part by grants-in-aid from
the Ministry of Education, Science, Culture and Sports,
Japan (18570140 for CTM and 16580108 for TY).
The bacterial expression vector pMW172 was a gift
from Dr K. Nagai, MRC Laboratory of Molecular
Biology, Cambridge, UK. The expression plasmid for
the myoglobin mutant H64L was a gift from Professor
J. S. Olson, Rice University. E. coli expression plas-
mids for maize FNR and maize Fd III were gifts from

8 Davis SJ, Bhoo SH, Durski AM, Walker JM & Viersta
RD (2001) The heme-oxygenase family required for
phytochrome chromophore biosynthesis is necessary for
proper photomorphogenesis in higher plants. Plant
Physiol 126, 656–669.
9 Muramoto T, Tsurui N, Terry MJ, Yokota A & Kohchi T
(2002) Expression and biochemical properties of a
T. Gohya et al. Heme catabolism by soybean heme oxygenase-1
FEBS Journal 273 (2006) 5384–5399 ª 2006 The Authors Journal compilation ª 2006 FEBS 5397
ferredoxin-dependent heme oxygenase required for
phytochrome chromophore synthesis. Plant Physiol 140,
1958–1966.
10 Emborg TJ, Walker JM, Noh B & Vierstra RD (2006)
Multiple heme oxygenase family members contribute to
the biosynthesis of the phytochrome chromophore in
Arabidopsis. Plant Physiol 140, 856–868.
11 Noriega GO, Balestrasse KB, Batlle A & Tomaro ML
(2004) Heme oxygenase exerts a protective role against
oxidative stress in soybean leaves. Biochem Biophys Res
Commun 323, 1003–1008.
12 Linley PJ, Landsberger M, Kohchi T, Cooper JB &
Terry MJ (2006) The molecular basis of heme oxygenase
deficiency in the pcd1 mutant of pea. FEBS J 273,
2594–2606.
13 Fujii H, Zhang X, Tomita T, Ikeda-Saito M & Yoshida
T (2001) Role for highly conserved carboxylate, aspar-
tate 140, in oxygen activation and heme degradation by
heme oxygenase-1. J Am Chem Soc 123, 6475–6484.
14 Zhou H, Migita CT, Sato M, Sun D, Zhang X,
Ikeda-Saito M, Fujii H, Yoshida T (2000) Participation

Yoshida T (2000) Histidine 20, the crucial proximal
axial heme ligand of bacterial heme oxygenase Hmu O
from Corynebacterium diphtheriae. J Biol Chem 275,
17494–17500.
22 Rohlfs RJ, Mathews AJ, Carver TE, Olson JS, Springer
BA, Egeberg KD, Sligar SG (1990) The effects of amino
acid substitution at position E7 (residue 64) on the
kinetics of ligand binding to sperm whale myoglobin.
J Biol Chem 265, 3168–3176.
23 Takahashi S, Mansfield Matera KM, Fujii H, Zhou H,
Ishikawa K, Yoshida T, Ikeda-Saito M & Rousseau DL
(1997) Resonance Raman spectroscopic characterization
of
a-hydroxyheme and verdoheme complexes of heme
oxygenase. Biochemistry 36, 1402–1410.
24 Wilks A & Ortiz de Montellano PR (1993) Rat liver
heme oxygenase. J Biol Chem 268, 22357–22362.
25 Sugishima M, Omata Y, Kakuta Y, Sakamoto H,
Noguchi M & Fukuyama K (2000) Crystal structure of
rat heme oxygenase-1 in complex with heme. FEBS Lett
471, 61–66.
26 Schuller DJ, Zhu W, Stojiljkovic I, Wilks A & Poulos
TL (2001) Crystal structure of heme oxygenase from the
Gram-negative pathogen Neisseria meningitidis and a
comparison with mammalian heme oxygenase-1.
Biochemistry 40, 11552–11558.
27 Sugishima M, Migita CT, Zhang X, Yoshida T &
Fukuyama K (2004) Crystal structure of heme oxyge-
nase-1 from cyanobacterium Synechocystis sp. PCC
6803 in complex with heme. Eur J Biochem 271, 4517–

acid biosynthetic enzyme. J Biol Chem 281,
15662–15670.
Heme catabolism by soybean heme oxygenase-1 T. Gohya et al.
5398 FEBS Journal 273 (2006) 5384–5399 ª 2006 The Authors Journal compilation ª 2006 FEBS
34 Liu Y, Moe
¨
nne-Loccoz P, Hildebrand DP, Wilks A,
Loehr TM, Mauk AG & Ortiz de Montellano PR
(1999) Replacement of the histidine iron ligand by a
cysteine or tyrosine converts heme oxygenase to an
oxidase. Biochemistry 38, 3733–3743.
35 Yoshida T, Noguchi M & Kikuchi G (1982) The step of
carbon monoxide liberation in the sequence of heme
degradation catalyzed by the reconstituted microsomal
heme oxygenase system. J Biol Chem 257, 9345–9348.
36 Nelson RM & Long GL (1989) A general method of
site-specific mutagenesis using a modification of the
Thermus aquaticus polymerase chain reaction. Anal Bio-
chem 190, 147–151.
37 Ishikawa K, Sato M, Ito M & Yoshida T (1992) Impor-
tance of histidine residue 25 of rat heme oxygenase for
its catalytic activity. Biochem Biophys Res Commun 182 ,
981–986.
38 Springer BA & Sligar SG (1987) High-level expression
of sperm whale myoglobin in Escherichia coli. Proc Natl
Acad Sci USA 84, 8961–8965.
39 Hase T, Mizutani S & Mukohata Y (1991) Expression
of maize ferredoxin cDNA in Escherichia coli. Plant
Physiol 97, 1495–1401.
40 Onda Y, Matsumura T, Kimara-Ariga Y, Sakakibara T,