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
Chemistry, Faculty of Agriculture,
Yamaguchi University, 1677-1 Yoshida,
Yamaguchi 753-8515, Japan
Fax ⁄ Tel: +81 83 9335863
E-mail:
T. Yoshida, Department of Biochemistry,
Yamagata University School of Medicine,
Iidanishi 2-2-2, Yamagata 990-9585, Japan
Fax: +81 23 6285225
Tel: +81 23 6285222
E-mail:
(Received 9 August 2006, revised 1 October
2006, accepted 9 October 2006)
doi:10.1111/j.1742-4658.2006.05531.x
Heme oxygenase converts heme into biliverdin, CO, and free iron. In
plants, as well as in cyanobacteria, heme oxygenase plays a particular role
in the biosynthesis of photoreceptive pigments, such as phytochromobilins
and phycobilins, supplying biliverdin IX
a
as a direct synthetic resource. In
this study, a higher plant heme oxygenase, GmHO-1, of Glycine max (soy-
bean), was prepared to evaluate the molecular features of its heme com-
plex, the enzymatic activity, and the mechanism of heme conversion. The
similarity in the amino acid sequence between GmHO-1 and heme oxygen-
ases from other biological species is low, and GmHO-1 binds heme with
1 : 1 stoichiometry at His30; this position does not correspond to the prox-
imal histidine of other heme oxygenases in their sequence alignments. The
heme bound to GmHO-1, in the ferric high-spin state, exhibits an acid–
base transition and is converted to biliverdin IX
a
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
catalyzes the cleavage of heme into biliverdin IX
a
,
which is then reduced and isomerized to form (3E)-
phytochromobilin, a chromophore of the photorecep-
tor protein of the phytochrome family, which plays
critical roles in mediating photomorphogenesis, by
sensing far-red and red light [7]. HO genes of higher
plants have been identified in a few moss plants,
several angiosperms (tobacco, tomato, pea, soybean,
rice plant, sorghum, Arabidopsis thaliana), and a gym-
nosperm (loblolly pine) [8]. So far, the HY1 gene and
the HO3 and HO4 genes of Arabidopsis have been
expressed in Escherichia coli, and Cd-induced expres-
sion of HO-1 in soybean leaves has also been reported
[8–11]. In the latter study, it was suggested that plant
HOs also play a role in protection against oxidative
cell damage [11]. More recently, the PsHO1 gene of
pea was expressed, and the HO activity of the protein
product was examined [12]. These studies have shown
that the obtained proteins bind heme to generate a
1 : 1 complex, and CO and biliverdin IX
a
are gener-
ated through heme catabolism, thereby confirming HO
activity. However, characterization of the heme com-
plexes on a molecular basis and determination of the
kinetics of heme catabolism have not been performed
yet.
The amino acid sequences reported for higher-plant
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 (AH) 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
GmHO-1) [14]. Thus, our first concern in examining
plant HOs is to determine whether the plant HO lack-
ing the residues critical for HO activity catalyzes heme
degradation in a similar fashion to the mammalian
enzymes.
The next concern is to establish the mechanism of
electron transfer from NADPH to GmHO-1. Muramoto
et al. reported that the AtHO-1 reaction required addi-
tional reductant besides ferredoxin reductase (FNR;
ferredoxin:NADP
+
reductase; EC 1.27.1.2) ⁄ ferredoxin
(Fd) and NADPH [9]. On the other hand, we
have clarified that cyanobacterial SynHO-1 (and also
HO-2) shows full activity when coupled with
NADPH ⁄ FNR ⁄ Fd, without a secondary reductant,
which had been suggested to be necessary for the HO
activity of cyanobacterial proteins [15–17]. Then, we
wanted to determine whether the NADPH ⁄ FNR ⁄ Fd
reducing system works fully in the heme conversion
into biliverdin by GmHO-1 as in the SynHO-1
reaction.
To investigate these phenomena, we purified the
recombinant mature form of GmHO-1 protein, exclu-
ding the plastid transit peptides, based on the reported
amino acid sequence [8] by constructing a bacterial
expression system. Spectroscopic analyses of the
molecular features of the heme–GmHO-1 complex and
of the mechanism of heme degradation were per-
formed, and the results were compared with those for
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-
tions were tinged with yellow. We do not know the
nature of this yellow substance(s) at present. The final
preparation after chromatography on a DE-52 column
was clear and colorless, and gave a single band of
26 kDa with about 97% purity on SDS ⁄ PAGE, the size
expected from the deduced GmHO-1 amino acid
sequence (26.1 kDa). About 100 mg of protein was
obtained from 1 L of culture.
Spectroscopic features of the heme–GmHO-1
complex
The optical absorption spectra of the heme–GmHO-1
complex in the ferric, ferrous, CO-bound and
O
2
-bound forms are typical of heme proteins and
similar to those of the SynHO-1 or rHO-1 complexes
(Fig. 2). The stoichiometry of the heme binding was
confirmed to be 1 : 1 by the titration plots shown in
the inset. The optical absorption data for heme–
GmHO-1, together with those of SynHO-1 and rHO-1,
are summarized in Table 1. The absorption maxima of
the O
2
-bound and CO-bound forms of heme–GmHO-1
are slightly red-shifted compared with those of heme–
SynHO-1 and heme–rHO-1.
The EPR spectrum of heme–GmHO-1 at pH 7.0
shows the heme mostly in the rhombic ferric high-spin
state, with g
x
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
histidinyl residue to the proximal site of the heme. The
spectral features of
15
NO-heme–GmHO-1 are somewhat
different from those of the nitrosylheme complexes of
SynHO-1 (Fig. 3D) and rHO-1 (Fig. 3E), whereas those
of the latter two are very alike. The EPR parameters
of the nitrosylheme–HO complexes as well as those of
the low-spin heme–HO complexes are listed in Table 2.
Acid–base transition of heme–GmHO-1
The features of the optical absorption spectrum of
the ferric heme–GmHO-1 complex reversibly change,
depending on pH, between acidic (pH 6.5) and alka-
line (pH 10.5) conditions. The absorption maxima
of the alkaline form are listed in Table 1. The
pK
a
value of this acid–base transition was estimated
to be 8.2 by the method described in Experimental
procedures.
Such an acid–base transition of heme–GmHO-1 was
also observed by EPR. The EPR spectrum of heme–
GmHO-1 at pH 8.7 exhibited a small high-spin signal
at g
xy
6 and prominent peaks of the low-spin heme
Table 1. Optical absorption data of the heme–HO-1 complexes.
Types of
heme
Protein
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
(4.8 nmol) with hemin (0.4 m
M), monitored by the increase in
absorbance at 405 nm. The pink off-line dots indicate the results
for the titration without protein.
Fig. 3. EPR spectra of the ferric heme–GmHO-1 (A, B) and nitro-
sylheme–HO (C–E) complexes. EPR conditions were: microwave
frequency, 9.35 GHz; microwave power, 1 mW for (A) and (B) and
0.2 mW for (C)–(E); field modulation frequencies, 100 kHz; field
modulation amplitude, 10 G for (A) and (B) and 2 G for (C)–(E); sig-
nal acquisition temperature, 8 K for (A) and (B) and 25 K for (C)–(E).
(A) The ferric heme complex of GmHO-1 at pH 7.0 (0.1
M, KPB).
a-1, Expansion of the g
xy
region of (A) (solid line) and of the corres-
ponding part of the spectrum of ferric heme–rHO-1 (dotted line).
a-2, Expansion of the higher-field region of (A). (B) The ferric
heme–GmHO-1 complex at pH 8.7 (50 m
M, Tris ⁄ HCl). (C–E) The
15
N-nitrosylheme complexes of GmHO-1, SynHO-1, and rHO-1,
respectively.
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 5387
at the lower magnetic fields (Fig. 3B). The g-values of
this low-spin species were the same as those of the
low-spin species observed at neutral pH (denoted as
g
ak
in Fig. 3A, expanded lower-field spectrum).
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
nitrogenous ligand capable of coordinating to the fer-
ric heme, so the GmHO-1 mutant H30G was prepared,
and the optical absorption and EPR spectra of its
heme complex were determined. The H30G mutant
also accommodated heme with 1 : 1 stoichiometry, like
other HO mutants that lack the proximal histidine,
but the optical spectrum of the heme–H30G complex
exhibited an asymmetrically broadened Soret band
with the blue-shifted maximum at around 390 nm,
compared with the Soret-band features of the wild-type
complex, and no other characteristic bands at the vis-
ible region (data not shown). Such features of the spec-
trum are commonly seen in the heme complexes of
HO mutants lacking the proximal His [20,21]. The
heme–H30G complex did not decompose the bound
heme enzymatically in the presence of either ascorbate
or NADPH ⁄ FNR ⁄ Fd (data not shown). EPR meas-
urements on the heme–H30G complex provided critical
evidence for the lack of coordination of a protein resi-
due at the proximal site, yielding the deformed spec-
trum of high-spin heme (Fig. 4A), which means that
the heme is in the multiple configuration in the heme
pocket, and the spectrum of nitrosylheme is typical of
the
15
NO coordination without the sixth ligand
(Fig. 4B). Accordingly, it has been established that the
proximal ligand of heme–GmHO-1 is His30.
Exogenous ligand binding
To investigate the nature of the heme pocket in
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
1
2.09 2.08 2.08
g
2
2.01 2.00 2.01
g
3
1.96 1.96 1.97
Fig. 4. EPR spectra of (A) ferric heme and (B)
15
N-nitrosylheme
complexes of the H30G mutant of GmHO-1. EPR conditions were
the same as those described in Fig. 3.
Heme catabolism by soybean heme oxygenase-1 T. Gohya et al.
5388 FEBS Journal 273 (2006) 5384–5399 ª 2006 The Authors Journal compilation ª 2006 FEBS
band maximum of the complex shifted from 405 to
410 nm and decreased in intensity, without showing a
distinct isosbestic point. At the same time, the heights
of the absorption peaks in the visible region (500 and
630 nm) also decreased, and then new peaks appeared
at 531 and 568 nm with increasing intensity, indica-
ting formation of imidazole-bound heme–GmHO-1
(data not shown). The association constant was esti-
mated on the basis of the changes in absorbance at
402 nm (imidazole-free form) and at 422 nm (imidaz-
ole-bound form) in the difference spectra of imidaz-
ole-free minus imidazole-titrated forms, as described
in Experimental procedures. In the same way, azide
binding to heme–GmHO-1 was also examined. In this
case, a clear isosbestic point was observed at 412 nm,
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).
To establish the physiologic electron-donating system
in the higher-plant HO reaction, the heme–GmHO-1
reaction was carried out in the presence of NADPH
coupled with FNR ⁄ Fd (Fig. 5B). After addition of
NADPH, the Soret band maximum of heme–GmHO-1
immediately shifted from 405 to 415 nm, and at the
same time, distinct absorption bands of oxyheme
appeared at 540 and 579 nm. Then, a broad band
appeared at around 660 nm, and was maximal 9–12 min
after initiation of the reaction. The spectral features of
the final reaction mixture were analogous, but not iden-
tical, to those of the ascorbate reaction, which was
mainly due to free biliverdin, probably because of the
overlapping of absorption bands of the intermediate
complexes. Product analysis by HPLC revealed that
only the a-isomer of biliverdin IX was produced in both
the NADPH ⁄ FNR ⁄ Fd-supported and ascorbate-assis-
ted GmHO-1 reactions (data not shown). Catalase did
not affect the heme–GmHO-1 reaction in the presence
of either ascorbate or NADPH ⁄ FNR ⁄ Fd. The apparent
initial heme degradation rate in the presence of
Table 3. Equilibrium constants for imidazole and azide ion binding
to the heme–HO-1 complexes. Numbers in parentheses indicate
relative values normalized to the values of GmHO-1.
Protein
GmHO-1 SynHO-1 rHO-1
K
imidazole
(M
)1
) 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:
biliverdin reductase (BVR; EC 1.3.1.24). To estimate
the yield of free biliverdin produced by the GmHO-1
reaction, the bilirubin assay was carried out by use of
rat BVR. The overall yields of bilirubin after the heme
conversion followed by biliverdin reduction under dif-
ferent conditions were estimated and compared with
the bilirubin yields for SynHO-1 or rHO-1 reactions
under comparable conditions (Table 5). Bilirubin yields
were somewhat low for GmHO-1 and SynHO-1, but
were comparable ( 50%) for the three HOs when des-
ferrioxamine, a chelating agent of Fe
3+
, was applied to
assist the extraction of Fe
3+
from the ferric biliverdin–
HO complexes. Interestingly, when an excess amount
of ascorbate was used as a reducing agent, the bilirubin
yield in the GmHO-1 reaction was as high as 47%.
Detection of CO liberation and identification
of reaction intermediates
Plausible verdoheme and CO–verdoheme intermediates
were not detected in the course of heme degradation
Table 4. Apparent rates of initial heme degradation (v) by GmHO-1,
SynHO-1 and rHO-1 in the presence of ascorbate, NADPH ⁄ FNR ⁄ Fd
(for GmHO-1 and SynHO-1), and NADPH ⁄ CPR (for rHO-1). The con-
centration of reactants in 0.1
M KPB (pH 7.0) are: sodium ascor-
bate, 6 m
M; heme–HO, 5 lM; Fd, 1 lM; FNR, 0.22 lM; CPR,
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
50 42 47
Ascorbate (2400 eq.) 47 – –
a
Additional NADPH and BVR were supplied 45 min after the first
addition of NADPH (4 eq.).
b
Desferrioxamine and BVR were sup-
plied 20 min and 30 min after addition of NADPH, respectively.
Fig. 5. Heme conversion by GmHO-1. (A) Spectra were recorded at
the indicated time after addition of ascorbate (6 m
M) to the solution
of heme–GmHO-1 (5 l
M in 0.1 M KPB, pH 7.0). (B) Spectra were
recorded at the indicated times after addition of NADPH (40 l
M)to
the solution of heme–GmHO-1 (5 l
M; 0.1 M KPB, pH 7.0), FNR
(0.22 l
M), and Fd (1 lM).
Heme catabolism by soybean heme oxygenase-1 T. Gohya et al.
5390 FEBS Journal 273 (2006) 5384–5399 ª 2006 The Authors Journal compilation ª 2006 FEBS
by GmHO-1 by optical absorption spectrometry
(Fig. 5). To ascertain the liberation of CO, this reac-
tion was performed in the presence of H64L, a myo-
globin mutant with 40 times greater affinity for CO
than the wild type [22]. As shown in Fig. 6A,B, the
difference spectra, the spectra obtained in the presence
minus those obtained in the absence of H64L, show an
absorption peak at 423 nm, which is known to be spe-
cific for the CO-bound form of myoglobin, thereby
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
CO–verdoheme. The 637 nm band disappeared gradu-
ally and was replaced by a new broad band with a
maximum at approximately 675 nm, identical to the
absorption band of free biliverdin. The heme–GmHO-1
reaction under a CO atmosphere was also carried out
with the NADPH ⁄ FNR ⁄ Fd reducing system (Fig. 6D).
In this case, however, the 637 nm band was not
observed, but a broad band with a maximum at around
657 nm appeared. Therefore, in the presence of ascor-
bate (Fig. 6C), verdoheme is probably produced, but in
the presence of NADPH ⁄ FNR ⁄ Fd, formation of the
verdoheme intermediate is still unclear because the
657 nm band is different from that of the well-known
verdoheme or CO–verdoheme bands at 688 and 637 nm,
respectively [23].
Heme degradation by HO is driven by hydrogen
peroxide, which substitutes for molecular oxygen and
probvides electrons to convert heme into verdoheme
[24]. Therefore, heme–GmHO-1 was reacted with
H
2
O
2
to ascertain whether verdoheme was actually
produced. Soon after addition of H
2
O
2
to the heme–
GmHO-1 solution, the Soret band intensity diminished
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
2
O
2
reaction intermediates of heme–SynHO-1
and heme–rHO-1 were obtained. The optical absorption
spectra of the acetone extracts (Fig. 8C,D) showed
similar features, indicating a common chromophore.
In conclusion, it has been confirmed that the 660 nm
Fig. 7. Heme degradation by GmHO-1 in the presence of H
2
O
2
.
Spectra were recorded at the indicated times after addition of H
2
O
2
(40 lM in N
2
-saturated 0.1 M KPB) to the heme–GmHO-1 solution
(5 l
M in N
2
-saturated 0.1 M KPB).
Fig. 8. Optical absorption spectra of the intermediates of heme–
HO reactions. (A) Obtained from the reaction mixture of heme–
GmHO-1 with H
2
O
2
(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-
mal ligand of the heme, whereas the heme–H30G
complex shows neither a proximal protein ligand
(Fig. 4) nor HO activity. These findings have firmly
established that His30 of GmHO-1 is the proximal
heme ligand.
Coordination structure of heme–GmHO-1
The optical absorption data for the GmHO-1 complex
in ferric, ferrous, oxy and CO-bound forms of heme
show that the coordination structure of the heme is
generally like that of SynHO-1 or rHO-1 (Fig. 2 and
Table 1). The EPR spectrum of the ferric resting form
of heme–GmHO-1, however, shows that the rhombic
anisotropy is relatively large and close to that in the
ferric a-hydroxyheme complex of rHO-1 [30]. This
means that the ligand field on the heme plane is relat-
ively anisotropic in GmHO-1. For this reason, it is
possible that the surrounding helices exert a greater
anisotropic ligand field effect on the heme plane than
the corresponding residues in SynHO-1 and rHO-1,
the amino acid sequences of which are considerably
different from that of GmHO-1.
The observed acid–base transition strongly suggests
that the sixth, distal ligand of heme is a water mole-
cule, and the heme-bound water is supposed to be con-
nected to dissociable distal residue(s) through direct
or indirect hydrogen bonding. The estimated pK
a
value
of 8.2 is between the values of heme–rHO-1 (7.6) and
heme–SynHO-1 (8.9); in the latter two, the distal water
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
strong to that in heme–rHO-1 [15]. The strength of the
Fe–N(His) bonding in the GmHO-1 and in rHO-1
complexes also appears to be the same, thereby imply-
ing that the imidazole part of His30 is neutral and
probably forms a hydrogen bond with Gln34, such
that His25 of rHO-1 is stabilized by the hydrogen
bonding with Glu29.
Characterization of the heme pocket of GmHO-1
by exogenous ligand binding
Azide, like imidazole, is a ligand with both r-donor
and p-donor characteristics, but is a relatively stron-
ger p-donor for stabilization of the higher oxidation
states of metal ions. In agreement with this, the bind-
ing constants for the binding of azide to ferric heme–
HOs are one to two orders larger than those for
imidazole (Table 3). Comparison of the K
azide
values
shows that heme–GmHO-1 and heme–SynHO-1 have
smaller values than heme–rHO-1, suggesting weak
relevance of the similarity in the amino acid sequences
of distal helices. This difference does not necessarily
mean that the ferric character of the former two is
less than that of the latter, because polarity of the
heme milieu as well as the steric conditions of the
distal heme pocket could also affect K
azide
. Polar resi-
dues might either stabilize the azide coordinated to
heme or facilitate the access of anionic azide to the
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
human HO-1) [3,25,27]. The difference in K
imidazole
between heme–SynHO-1 and heme–rHO-1 may be
attributable to this structural difference. As shown in
Fig. 1, the presumed A-helix of GmHO-1 is 3–4 turns
longer than the A-helices of SynHO-1 and rHO-1. Fur-
thermore, the assumed F-helix, corresponding to the
distal helices of SynHO-1 and rHO-1, is also relatively
long, suggesting a unique structure of the opening of
the heme pocket of GmHO-1. Such a structure might
inhibit the approach of imidazole to the heme in
GmHO-1, explaining the smallest K
imidazole
value.
Mechanism of heme degradation by GmHO-1
Although some of the details are still unknown, the
degradation of heme by HO follows the mechanism
shown in Scheme 1. Heme catabolism by mammalian
HO proceeds sequentially by way of oxyheme,
hydroxyheme, verdoheme and ferric biliverdin, and the
presence of each intermediate is monitored by respect-
ive specific absorption bands. Heme catabolism by
HOs of mammals, pathogenic bacteria, cyanobacteria
and probably insects is considered to have a similar
mechanism, because the characteristic absorption
bands of verdoheme and CO–verdoheme are observed
or CO produced is detected [1,5,15,16,31,32].
In heme degradation by GmHO-1, it has been con-
firmed that the final product is biliverdin IX
a
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
in the GmHO-1 and SynHO-1 reactions in the pres-
ence of NADPH ⁄ FNR ⁄ Fd probably occurs due to
reduction of ferric biliverdin to yield ferrous ion being
less efficient with the limited number of electrons sup-
plied from NADPH. Accordingly, it can be concluded
that GmHO-1 is fully active when electrons are sup-
plied successively from NADPH by way of FNR and
Fd to the combined heme in the presence of molecular
oxygen. This finding is inconsistent with previous
reports that cyanobacterial or plant HO reactions
require an additional reducing agent such as ascorbate
[9,12,17].
We propose that the NADPH ⁄ FNR ⁄ Fd reducing
system could be the native electron-donating partner
in the GmHO-1 reaction in plastids, as well as in
the SynHO-1 reaction in the cytosol. Although
ascorbate, which is abundant in plant tissues (repor-
ted to be present at 1–20 mm [33]), probably contri-
butes to some extent to the efficient reduction of the
ferric ion and extraction of biliverdin from the ferric
biliverdin–enzyme complex, it is unlikely that ascor-
bate initiates the heme–GmHO-1 reaction as a physi-
ologic electron donor. Kinetic experiments show that
the first step of heme degradation by GmHO-1 and
SynHO-1 is obviously slower than that by rHO-1 in
the presence of ascorbate (Table 4). This means that
the first reduction of the ferric heme is harder to
achieve in the GmHO-1 and SynHO-1 reactions than
in the rHO-1 reaction (Scheme 1), and that the first
step of heme conversion is rate-limiting in the ascor-
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
verdoheme and accordingly the axial coordination
structure of verdoheme in GmHO-1 seem to be unu-
sual. A resonance Raman study on verdoheme–rHO-1
suggested that verdoheme is in the six-coordinate state
with the proposed ligand of hydroxide or water [23],
so that the sixth ligand of verdoheme–GmHO-1 might
be different from that of known HO complexes of
verdoheme. The stronger distal coordination should
result in weaker affinity of CO for verdoheme. In the
ascorbate-supported reaction, such distal coordination
onto verdoheme is not discernible, and instead, in
a CO atmosphere, transient CO coordination is
observed. This is probably because the verdoheme is
accumulated less, due to the limitation on verdoheme
generation imposed by the first step of heme conver-
sion. Furthermore, a large excess of ascorbate might
be advantageous for verdoheme degradation, by redu-
cing verdoheme accumulation. To address coordina-
tion of the distal ligand onto verdoheme–GmHO-1,
site-directed mutations of possible distal residues are
being studied.
Experimental procedures
Construction of expression vector pMWGm HO-1
GmHO-1 expression vector pMWGmHO-1 was constructed
in the same way as pMWSynHO-1 or pMWSynHO2,
described previously [14,15]. According to the reported
amino acid sequence of GmHO-1 [8], we designed a nucleo-
tide sequence encoding a mature-type enzyme from Ser26
to Ser250 without the N-terminal transit region and with
unique sites for the restriction enzymes NdeI, BstEII, BglII,
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
A 5 mL inoculum in LB medium (+ 50 lgÆmL
)1
ampicil-
lin ⁄ 0.1% glucose) was prepared from a plate of trans-
formed E. coli BL21 (DE3) cells carrying pMWGmHO-1.
Five-hundred-milliliter cultures were inoculated with
200 lL of the inoculum and grown in LB medium
(+ 200 lgÆmL
)1
ampicillin) at 37 °C until the A
600 nm
reached 0.8–1.0. The cells were grown for an additional
24 h at 25 °C, harvested by centrifugation at 5000 g for
10 min using a Kubota RA-3 rotor (Tokyo, Japan), and
stored at ) 80 °C prior to use. The typical yield of cells
from a 500 mL culture was 1.7 g.
The E. coli cells (10 g), resuspended in 90 mL of 50 mm
Tris ⁄ HCl buffer (pH 7.4, + 2 mm EDTA), were lysed
[2 mg lysozymeÆ(g cells)
)1
] with stirring at 4 °C for 30 min.
After sonication (Branson 450 Sonifire, Danbury, CT,
USA) and centrifugation at 100 000 g for 1 h using a
Hitachi RP50T rotor (Tokyo, Japan), the resulting superna-
tant was covered with a 60–90% ammonium sulfate frac-
tion and centrifuged at 12 000 g for 15 min using a Kubota
RA-3 rotor. The subsequent precipitates, containing the
GmHO-1 protein, were dissolved in 20 mm potassium
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 5395
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
inflection point of the titration curve. Reconstitution of
heme–GmHO-1 was carried out by adding a small excess of
hemin to the protein solution, and then removing the excess
hemin with a Sephadex G-25 column pre-equilibrated with
0.1 m KPB (pH 7.0).
Optical absorption and EPR spectroscopy
All of the optical absorption spectra were recorded on a
Shimadzu (Tokyo, Japan) UV-2200 spectrophotometer at
25 °C. Protein solutions were in 0.1 m KPB (pH 7.0), unless
otherwise specified. The ferrous form of heme–GmHO-1
was obtained by the addition of an appropriate amount of
anaerobic dithionite solution (1 m) to heme–GmHO-1
(5 lm) in a UV cell (Aldrich, St Louis, MO, USA) with a
screwtop open-cap tube furnished with a rubber septum,
under anaerobic condition. The CO-bound form was
prepared by exposing the ferrous form to gaseous CO
introduced into the anaerobic cuvette. The oxy form of
heme–GmHO-1 was obtained by reduction of the ferric
form with NADPH (1 eq.) ⁄ FNR ⁄ Fd and exposing the
reduced complex to small amount of oxygen gas.
The pH titration was performed using 1 m Tris ⁄ HCl
buffer for pH 6.5–8.3 and 1 m NaOH solution for
pH 8.3–11.4; aliquots were added to heme–GmHO-1 (8 lm)
in 18 mm KPB (pH 6.00), and the pK
a
value was estimated
according to the published method [14].
Azide binding was carried out by addition of azide solu-
tions (0.1 or 1 m) to 2 mL of heme–GmHO-1 solution
(5 lm), followed by incubation for 10–15 min on ice after
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
NO
2
in an
EPR tube made of extra-pure synthetic quartz.
Heme conversion and kinetic measurements
Protein solutions were in 0.1 m KPB (pH 7.0), unless other-
wise specified. Heme conversion reactions in the presence of
sodium ascorbate were started by adding appropriate
amounts of sodium ascorbate solution (1 m) to the heme–
GmHO-1 solution (5 lm). When NADPH ⁄ FNR ⁄ Fd was
used as a reducing system, the reaction was started by addi-
tion of a given amount of NADPH (5–15 mm in 0.1 m
KPB, pH 7.0) to the solution of heme–GmHO-1 (5 lm), Fd
(1 lm), and FNR (0.22 lm), or to the solution of rHO-1
(5 lm) and CPR (0.22 lm). Optical absorption spectra of
the reaction mixtures were recorded between 240 and
900 nm at appropriate time intervals until the HO reaction
was completed. The initial rate of heme degradation was
estimated from the decreasing rate of the Soret band
(A
405
, e ¼ 127 mm
)1
Æcm
)1
for heme–GmHO-1, A
402
, e ¼
128 mm
)1
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),
and after the reaction was complete (approximately 45 min
after initiation of the reactions), BVR (1 lm) and additional
NADPH (40 lm) were supplied. In the rHO-1 (5 lm) reac-
tion, CPR (0.2 lm ) was used in place of FNR ⁄ Fd. In the
ascorbate-supported reactions, the heme–enzyme solution
containing BVR (1 lm) and NADPH (40 lm) was reacted
with ascorbate (12 mm ). The end of each reaction was con-
firmed by the complete disappearance of the Soret band,
and the quantity of bilirubin was determined from the A
468
with e
468
¼ 55 mm
)1
Æcm
)1
[43].
Detection of CO
To each solution of heme–GmHO-1 (5 lm) and FNR ⁄ Fd
(0.22 and 1 lm, respectively) or heme–GmHO-1 (5 lm)
alone, H64L (4 lm) was added, and the heme conversion
reaction was initiated by the addition of NADPH (10 lm)
or ascorbate (6 mm), respectively. Optical absorption spec-
tra of the reaction mixtures were recorded at appropriate
time intervals. As a control experiment, the reaction of
H64L with NADPH ⁄ FNR ⁄ Fd or ascorbate was carried
out, and the difference spectra, in the presence minus in the
absence of heme–GmHO-1, were obtained.
Extraction of the 660 nm intermediate
Anaerobic H
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
Professor T. Hase, Osaka University. The cDNA of
human CPR was a gift from Dr F. Gonzalez, NIH.
References
1 Tenhunen R, Marver HS & Schmid R (1969) Micro-
somal heme oxygenase. J Biol Chem 244, 6388–6394.
2 Yoshida T & Kikuchi G (1978) Features of the reaction
of heme degradation catalyzed by the reconstitution
microsomal heme oxygenase system. J Biol Chem 253,
4230–4236.
3 Schuller DJ, Wilks A, Ortiz de Montellano PR &
Poulos TL (1999) Crystal structure of human heme
oxygenase-1. Nat Struct Biol 6, 860–867.
4 Schmitt MP (1997) Utilization of host iron sources by
Corynebacterium diphtheriae: identification of a gene
whose product is homologous to eukaryotic
heme oxygenases and is required for acquisition of
iron from heme and hemoglobin. J Bacteriol 179,
838–845.
5 Zhu W, Wilks A & Stojiljkovic I (2000) Degradation of
heme in Gram-negative bacteria: the product of the
hemO gene of neisseriae is a heme oxygenase. J Bacteriol
182, 6783–6790.
6 Ratliff M, Zhu W, Deshmukh R, Wilks A & Stojiljko-
vic I (2001) Homologues of neisserial heme oxygenase
in Gram-negative bacteria: degradation of heme by the
product of pigA gene of Pseudomonas aeruginosa.
J Bacteriol 183, 6394–6403.
7 Smith H (2000) Phytochromes and light signal percep-
tion by plants ) an emerging synthesis. Nature 407,
585–590.
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
of carboxylate amino acid side chain in regiospecific
oxidation of heme by heme oxygenase. J Am Chem Soc
122, 8311–8312.
15 Migita CT, Zhang X & Yoshida T (2003) Expression
and characterization of cyanobacterium heme oxyge-
nase, a key enzyme in the phycobilin synthesis. Eur J
Biochem 270, 687–698.
16 Zhang X, Migita CT, Sato M, Sasahara M & Yoshida T
(2005) Protein expressed by the ho2 gene of the cyanobac-
terium Synechocystis sp. PCC 6803 is a true heme oxyge-
nase. FEBS J 272, 1012–1022.
17 Rhie G & Beale SI (1995) Phycobilin biosynthesis:
reductant requirements and product identification for
heme oxygenase from Cyanidium caldarium. Arch Bio-
chem Biophys 320, 182–194.
18 Takahashi S, Wang J, Rousseau DL, Ishikawa K,
Yoshida T, Host JR, Ikeda-Saito M (1994) Heme–heme
oxygenase complex. J Biol Chem 269, 1010–1014.
19 Chu GC, Tomita T, So
¨
nnichsen FD, Yoshida T &
Ikeda-Saito M (1999) The heme complex of Hmu O, a
bacterial heme degradation enzyme from Corynebacter-
ium diphtheriae. J Biol Chem 274, 24490–24496.
20 Ito-Maki M, Ishikawa K, Mansfield Matera K, Sato M,
Ikeda-Saito M, Yoshida T (1995) Demonstration that
histidine 25, but not 132, is the axial heme ligand in rat
heme oxygenase-1. Arch Biochem Biophys 317, 253–258.
21 Chu GC, Katakura K, Tomita T, Zhang X, Sun D,
Sato M, Sasahara M, Kayama T, Ikeda-Saito M &
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–
4525.
28 Sugishima M, Hagiwara Y, Zhang X, Yoshida T,
Migita CT & Fukuyama K (2005) Crystal structure of
dimeric heme oxygenase-2 from Synechocystis sp. PCC
6803 in complex with heme. Biochemistry 44, 4257–
4266.
29 Hirotsu S, Chu GC, Unno M, Lee D, Yoshida T, Park
S, Shiro Y & Ikeda-Saito M (2004) The crystal struc-
tures of the ferric and ferrous forms of the heme com-
plex of HmuO, a heme oxygenase of Corynebacterium
diphtheriae. J Biol Chem 279, 11937–11947.
30 Mansfield Matera K, Takahashi H, Fujii H, Zhou H,
Ishikawa K, Yoshimura T, Rousseau DL, Yoshida T &
Ikeda-Saito M (1996) Oxygen and one reducing equiva-
lent are both required for the conversion of a-hydroxy-
heme to verdoheme in heme oxygenase. J Biol Chem
271, 6618–6624.
31 Wilks A & Schmitt MP (1998) Expression and
characterization of a heme oxygenase (Hmu O) from
Corynebacterium diphtheriae. J Biol Chem 273,
837–841.
32 Zhang X, Sato M, Sasahara M, Migita CT & Yoshida
T (2004) Unique features of recombinant heme oxyge-
nase of Drosophila melanogaster compared with those of
other heme oxygenases studied. Eur J Biochem 271,
1714–1724.
33 Conklin PL, Gatzek S, Wheeler GL, Dowdle J,
Raymond MJ, Rolinski S, Isupov M, Littlechild JA
& Smirnoff N (2006) Arabidopsis thaliana VTC4
encodes L-galactose-1-P phosphatase, a plant ascorbic
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,
Sugiyama T & Hase T (2000) Differential interaction of
maize root ferredoxin:NADP
+
oxidoreductase with
photosynthetic and non-photosynthetic ferredoxin
isoproteins. Plant Physiol 123, 1037–1045.
41 Yamano S, Aoyama T, McBride OW, Hardwick JP,
Gelboin HV & Gonzalez F (1989) Human NADPH-
P450 oxidoreductase: complementary DNA cloning,
sequence and vaccinia virus-mediated expression and
localization of the CYPOR gene to chromosome 7.
Mol Pharmacol 36, 83–88.
42 Kikuchi A, Park SY, Miyatake H, Sun D, Sato M,
Yoshida T & Shiro Y& (2001) Crystal structure of rat
biliverdin reductase. Nat Struct Biol 8, 221–225.
43 Yoshida T & Kikuchi G (1978) Purification and proper-
ties of heme oxygenase from pig spleen microsomes.
J Biol Chem 253, 4224–4229.
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 5399