Alteration of the substrate specificity of the angular dioxygenase carbazole 1,9a-dioxygenase

Article abstract of DOI:10.1271/bbb.80512

Carbazole 1,9a-dioxygenase (CARDO) consists of terminal oxygenase (CARDO-O) and electron transport components. CARDO can catalyze specific oxygenation for various substrates: angular dioxygenation for carbazole and dibenzo-p-dioxin, lateral dioxygenation for anthracene, and monooxygenation for methylene carbon of fluorene and sulfide sulfur of dibenzothiophene. To elucidate the molecular mechanism determining its unique substrate specificity, 17 CARDO-O site-directed mutants at amino acid residues I262, F275, Q282, and F329, which form the substrate-interacting wall around the iron active site by CARDO-O crystal structure, were generated and characterized. F329 replacement dramatically reduced oxygenation activity. However, several mutants produced different products from the wild-type enzyme to a large extent: I262V and Q282Y (1-hydroxycarbazole), F275W (4-hydroxyfluorene), F275A (unidentified cis-dihydrodiol of fluoranthene), and I262A and I262W (monohydroxydibenzothiophenes). These results suggest the possibility that the respective substrates bind to the active sites of CARDO-O mutants in a different orientation from that of the wild-type enzyme.

Full text of DOI:10.1271/bbb.80512

Biosci. Biotechnol. Biochem., 72 (12), 3237–3248, 2008
Alteration of the Substrate Specificity
of the Angular Dioxygenase Carbazole 1,9a-Dioxygenase
1;2;
1
1;3
1;*,**
1;***
*
Hiromasa UCHIMURA,
Yusuke USAMI,1 Li LI,
Tadafumi HORISAKI,
Takashi UMEDA, Haruko NOGUCHI,
3;****
4
3;4
Tohru TERADA,
1;*****
Shugo NAKAMURA, Kentaro SHIMIZU,
Tetsuo TAKEMURA,2 Hiroshi HABE,
Hisakazu YAMANE,1 and Hideaki NOJIRI
Kazuo FURIHATA, Toshio OMORI,
1;******
5
y
1;3;
1Biotechnology Research Center, The University of Tokyo, 1-1-1 Yayoi, Bunkyo-ku, Tokyo 113-8657, Japan
2Department of Chemistry, Faculty of Science, Tokyo University of Science,
1-3 Kagurazaka, Shinjuku-ku, Tokyo 162-8601, Japan
3Professional Program for Agricultural Bioinformatics, The University of Tokyo,
1-1-1 Yayoi, Bunkyo-ku, Tokyo 113-8657, Japan
4Department of Biotechnology, The University of Tokyo, 1-1-1 Yayoi, Bunkyo-ku, Tokyo 113-8657, Japan
5Department of Applied Biological Chemistry, The University of Tokyo,
1-1-1 Yayoi, Bunkyo-ku, Tokyo 113-8657, Japan
Received July 28, 2008; Accepted August 27, 2008; Online Publication, December 7, 2008
[doi:10.1271/bbb.80512]
Carbazole 1,9a-dioxygenase (CARDO) consists of
terminal oxygenase (CARDO-O) and electron transport
components. CARDO can catalyze specific oxygenation
for various substrates: angular dioxygenation for car-
bazole and dibenzo-p-dioxin, lateral dioxygenation for
anthracene, and monooxygenation for methylene car-
bon of fluorene and sulfide sulfur of dibenzothiophene.
To elucidate the molecular mechanism determining its
unique substrate specificity, 17 CARDO-O site-directed
mutants at amino acid residues I262, F275, Q282, and
F329, which form the substrate-interacting wall around
the iron active site by CARDO-O crystal structure, were
generated and characterized. F329 replacement dra-
matically reduced oxygenation activity. However, sev-
eral mutants produced different products from the
wild-type enzyme to a large extent: I262V and Q282Y
(1-hydroxycarbazole), F275W (4-hydroxyfluorene),
F275A (unidentified cis-dihydrodiol of fluoranthene),
and I262A and I262W (monohydroxydibenzothio-
phenes). These results suggest the possibility that
the respective substrates bind to the active sites of
CARDO-O mutants in a different orientation from that
of the wild-type enzyme.
Key words: angular dioxygenation; carbazole 1,9a-
dioxygenase; Rieske non-heme iron oxy-
genase; site-directed mutagenesis; substrate
specificity
Metabolically versatile bacteria play an important role
in decomposing released aromatic compounds regard-
less of their origin. Although sufficient induction time is
occasionally needed to detect degradation, most natural
and synthetic aromatic compounds are mineralized or
partially converted to other compounds. Aerobic degra-
dation of various aromatics, including toxic solvents,
y
To whom correspondence should be addressed. Fax: +81-3-5841-8030; E-mail: anojiri@mail.ecc.u-tokyo.ac.jp
These two authors contributed equally to this work.
*
**
Present address: Laboratory of Proteomic Sciences, 21st Century COE Program, Kyoto Pharmaceutical University, Yamashina-ku, Kyoto 607-
8412, Japan
***
****
Present address: School of Environmental Science and Engineering, Shandong University, Jinan 250100, People’s Republic of China
Present address: Computational Science Research Program, RIKEN, 2-1 Hirosawa, Wako, Saitama 351-0198, Japan
Present address: National Institute of Advanced Industrial Science and Technology (AIST), Central 5-2, 1-1-1 Higashi, Tsukuba, Ibaraki 305-
8565, Japan
*****
******
Present address: Department of Industrial Chemistry, Shibaura Institute of Technology, 3-9-14 Shibaura, Minato-ku, Tokyo 108-8548, Japan
Abbreviations: AD, angular dioxygenation; Ap, ampicillin; BDO, biphenyl 2,3-dioxygenase; BDO-O, terminal oxygenase component of BDO;
CARDO, carbazole 1,9a-dioxygenase; CARDO-F, ferredoxin component of CARDO; CARDO-O, terminal oxygenase component of CARDO;
CARDO-R, ferredoxin reductase component of CARDO; GC-MS, gas chromatography-mass spectrometry; HPLC, high-performance liquid
chromatography; LD, lateral dioxygenation; MSTFA, N-methyl-N-trimethylsilyltrifluoroacetamide; NDO, naphthalene 1,2-dioxygenase; NDO-O,
terminal oxygenase component of NDO; NMR, nuclear magnetic resonance; PCB, polychlorinated biphenyl; RO, Rieske non-heme iron oxygenase;
SDS–PAGE, sodium dodecyl sulfate-polyacrylamide gel electrophoresis; TLC, thin-layer chromatography; WT, wild type

3238
H. UCHIMURA et al.
4
1
A
3
2
H
OH
Anthracene
1
H
OH
cis-1,2-Dihydroxy-1,2-dihydroanthracene
H
OH
2
3
OH
H
4
6
5
Fluoranthene
cis-2,3-Dihydroxy-2,3-dihydrofluorenthene
4
3
B
2
9a
1
4
H
N
N
H
H
OH
OH
NH2 OH
OH
OH
Carbazole
2'-Aminobiphenyl-2,3-diol
O
O
O
3
2
10a
O
O
1
OH
OH
H
Dibenzo-p-dioxin
OH
2,2',3-Trihydroxydiphenyl ether
OH
4
3
C
2
9
1
Fluorene
1
H
OH
9-Hydroxyfluorene
2
3
5
S
4
S
Dibenzothiophene
O
Dibenzothiophene-5-oxide
Fig. 1. Diverse Types of Oxygenation Catalyzed by CARDO.
CARDO catalyzes cis-dihydroxylation at lateral positions (lateral dioxygenation, LD) for anthracene and fluoranthene (A), cis-dihydroxylation at
angular positions (angular dioxygenation, AD) for carbazole and dibenzo-p-dioxin (B), and monooxygenation for methylene carbon of fluorene and
sulfide sulfur of dibenzothiophene (C). Only the main products are shown for the respective substrates. Absolute stereochemistry is not intended, with
the exception of the AD product of carbazole (1,9a-dihydroxy-1-hydrocarbazole), for which the stereochemistry is proposed based on the carbazole-
bound structure at the active site of CARDO-O.11) The oxygenation products shown in brackets are unstable and were not detected directly.
agrochemicals, medicines, and industrial raw materials,
is a key phenomenon in removing previously released
contaminants from subsurface areas such as surface
soils, river water, and seashores. Compared with
anaerobic bacterial degradation, aerobic bacterial deg-
radation generally progresses more rapidly because of
the enhanced metabolic status, and has been studied for
various xenobiotic compounds to provide basic infor-
mation on its potential use in the removal of aromatic
contaminants from the environment.
Bacterial metabolism of aromatic compounds is
usually initiated by dioxygenation catalyzed by Rieske
non-heme iron oxygenase (RO). RO members are
multicomponent-type enzymes that consist of terminal
oxygenase and electron transfer components. RO mem-
bers constitute an important class of enzymes, not only
as a tool for the bioremediation of aromatic pollutants,
but also as regio- and stereoselective aromatic-modify-
ing enzymes. With very few exceptions, RO members
catalyze the incorporation of both oxygen atoms of
molecular dioxygen to tandemly linked carbon atoms in
an aromatic ring as two hydroxyl groups in the cis-
configuration.1,2) This type of dioxygenation is called
lateral dioxygenation (LD). For example, LD for
anthracene and fluoranthene is shown in Fig. 1A.
Carbazole is an N-heterocyclic aromatic compound
found in coal tar creosote that is highly recalcitrant,
mutagenic, and toxic.3) We have isolated various
carbazole degraders, such as Pseudomonas resinovorans
CA10, Janthinobacterium sp. J3, Novosphingobium sp.
KA1 (formerly Sphingomonas sp. KA1), and Nocar-
dioides aromaticivorans IC177, and their carbazole
degradation pathway and enzymes have been elucidated
extensively.4) One of the RO members, carbazole 1,9a-

Structure-Based Mutagenesis of Angular Dioxygenase
3239
dioxygenase (CARDO), is an initial enzyme in the
degradation pathway of carbazole. CARDO attacks at the
angular (C-9a) and adjacent (C-1) carbons of carbazole,
and the resulting unstable hemiaminal intermediate,
1,9a-dihydroxy-1-hydrocarbazole, is cleaved spontane-
ously to 20-aminobiphenyl-2,3-diol (Fig. 1B).5,6) This
type of reaction is termed angular dioxygenation
(AD).4,7) CARDO consists of the terminal oxygenase
component (CARDO-O, a homotrimer of CarAa pro-
teins) and electron transfer components.5,8) In the
CARDO systems of P. resinovorans CA10 and Janthi-
nobacterium sp. J3, electron transfer from NAD(P)H
to terminal oxygenase is mediated by ferredoxin
(CARDO-F, a monomer of CarAc protein) and ferre-
doxin reductase (CARDO-R, a monomer of CarAd
protein) components.8) AD is also involved in the
degradation pathways of carbazole-related compounds
such as dibenzo-p-dioxin, dibenzofuran, and 9-fluore-
none,4,7) and CARDO can catalyze AD for some of these
compounds (Fig. 1B).5,6,9) CARDO also catalyzes LD
for several polycyclic aromatic hydrocarbons, and
monooxygenation for methylene carbon of fluorene
and sulfide sulfur of dibenzothiophene (Fig. 1A and
C).6,9) Although numerous RO members have been
isolated and characterized to date, only a limited number
of these catalyze AD.
We have determined the crystal structure of CARDO-
O from J3 and have performed docking simulation of
carbazole to the substrate-binding pocket.10) In addition,
we have succeeded in determining the CARDO-O:
CARDO-F:carbazole ternary complex structure, and the
docking position of carbazole in the crystal structure.11)
The docking structure suggested that the binding of
carbazole to CARDO-O was stabilized by hydrogen
bonding between the imino nitrogen of carbazole and
the carbonyl oxygen of Gly178, and, in this particular
case, the C-9a and C-1 carbon atoms of carbazole were
situated more closely to the ferrous iron active site. This
manner of binding is consistent with the fact that
CARDO-O catalyzes AD for carbazole.
were grown in lysogeny (L)-medium (1% tryptone
peptone, 0.5% yeast extract, and 1% NaCl)12) at 37 ꢁC
with reciprocal shaking (300 strokes/min). Ampicillin
(Ap; 50 mg/ml) was added to the selective medium.
For plate cultures, the medium was solidified with 1.6%
(w/v) agar.
Standard DNA techniques. Restriction endonucleases
(Takara Bio, Shiga) and Ligation High (Toyobo) were
used according to the manufacturers’ instructions. DNA
fragments were extracted from agarose gels using an
E.Z.N.A. gel extraction kit (Omega Bio-Tek, Doraville,
GA). Determination of nucleotide sequences was per-
formed as described previously.13) Other DNA manipu-
lations were performed according to standard methods.12)
Substrate docking simulation to the substrate-binding
pocket of CARDO-O. The complex structure of dibenzo-
p-dioxin and wild-type (WT) CARDO-O of J3 (PDB
code, 1WW9) was simulated as described previously.10)
Figures were constructed using PyMOL.14)
Site-directed mutagenesis in the carAa gene. Plasmid
pUCARAJ3, which contains the carAa, carAc, and
carAd genes from Janthinobacterium sp. J3, was
constructed as the reference plasmid used for expression
of WT CARDO-O. In brief, pBJ300115) was digested
with EcoRI, and the resulting 5.6-kb fragment was
inserted into the corresponding restriction sites of
modified pUC119, from which the PstI site had
previously been removed by T4 DNA polymerase
treatment followed by self-ligation. The plasmid was
then digested with PstI and self-ligated to form plasmid
pUCARAJ3, which contains the carAaAcAd genes and
lacks the carBaBbC genes.
To replace the target codon of F275, Q282, or F329 in
the carAa gene with the desired amino acid residues,
pTemplateJ, which contains the partial carAa gene
fragment, was constructed. Initially, a 1.9-kb BamHI-
PstI fragment of pUCARAJ3 containing the entire
carAa gene and the 50 end of the carBa gene was
ligated with the corresponding restriction sites of
pBluescript II SK(ꢀ) to give pJ31BP. The 628-bp
EcoRV fragment from pJ31BP was inserted into the
same site of pBluescript II SK(ꢀ) to create plasmid
pTemplateJ. For replacement of the I262 codon, pTem-
plateJ-II was constructed as follows: pJ31BPC was
formed by ligation of the 1.2-kb ClaI fragment of
pJ31BP with ClaI-treated pBluescript II SK(ꢀ). Plasmid
pJ31BPC was then digested with ApaI and self-ligated,
and the resulting plasmid was designated pTemplateJ-II.
Site-directed mutagenesis was performed using
the QuickChange XL Site-Directed Mutagenesis Kit
(Stratagene) according to the manufacturer’s instruc-
tions. A template plasmid, viz., pTemplateJ or pTem-
plateJ-II, and two complementary oligonucleotide pri-
mers containing the desired mutation were used.
Detailed information on the primers used is available
Amino acid residues, which constitute the wall of the
substrate-binding pocket, are obviously a determinant of
carbazole orientation suitable for AD catalysis. We
generated 17 CARDO-O mutants that had single amino
acid substitutions at the substrate-binding pocket. Based
on the oxygenation activities for several substrates, we
discuss below the amino acid residues involved in
determining the substrate specificity of CARDO-O.
Materials and Methods
Bacterial strains, media, and culture conditions.
Escherichia coli DH5ꢀ (Toyobo, Tokyo) and JM10912)
were used as host strains for plasmid preparations of
pUC119,12) pT7Blue T-vector (Merck Japan, Tokyo),
pBluescript II SK(ꢀ) (Stratagene, La Jolla, CA), and
their derivatives. E. coli JM109 was also used as a host
strain in biotransformation experiments. E. coli strains

3240
H. UCHIMURA et al.
upon request. Following mutagenesis, preferred nucleo-
tide replacements were confirmed by sequencing, and
the resulting plasmids were labeled as shown in
Table 1 (pTemplateJ-F275A or pTemplateJ-II-I262A).
According to the reverse scheme from pUCARAJ3 to
pTemplateJ/J-II, we constructed pUCARAJ3 deriva-
tives, which had the desired replacements in the carAa
gene, to express the site-directed mutations of CARDO-
O (Table 1).
under a nitrogen stream and redissolved into 100 ml of
methanol solution (40% methanol by volume of water).
The resulting sample was injected into a column
equilibrated with methanol/water (4:6 v/v) and eluted
with an increasing gradient of methanol/water. The
methanol in the mobile phase was increased to 60% for
25 min, followed by an increase to 100% for 20 min.
Identification of products formed by CARDO-O
mutants. In carbazole biotransformation experiments
using several CARDO-O mutants, we detected mono-
hydroxycarbazole having an Rf of 0.61 by analytical
thin-layer chromatography (TLC) with a solvent system
of hexane-ethyl acetate-acetic acid (6:4:0.5 by volume).
To purify this compound, preparative TLC was per-
formed as follows: TLC was developed on a precoated
silica gel plate (1 mm in thickness, 20 ꢂ 20 cm, Silica
gel 60 F254, Merck Japan) with the same solvent system
as analytical TLC. The desired band was scraped off and
eluted with H2O-saturated ethyl acetate. The resulting
elute was dried with Na2SO4 before evaporation and
used in nuclear magnetic resonance (NMR) spectrom-
Biotransformations. E. coli JM109 harboring
pUCARAJ3 or its derivatives was precultured in 5 ml
of L-medium supplemented with Ap at 37 ꢁC for 12 h,
followed by 1-ml aliquot transfer of the resulting culture
to 250 ml of L-medium. After incubation at 25 ꢁC for
5 h, isopropyl-ꢁ-D-thiogalactopyranoside was added to a
final concentration of 0.1 mM. After an additional 15 h of
incubation, the cells were harvested by centrifugation,
washed twice with MM3 (2.2 g/l Na2HPO4, 0.8 g/l
.
.
KH2PO4, 500 mg/l MgSO4 7H2O, 10 mg/l FeSO4
.
7H2O, 10 mg/l CaCl2 2H2O, 50 mg/l yeast extract), and
resuspended in MM3 adjusted to an OD600 of 20. As
substrates, we used carbazole, dibenzo-p-dioxin, anthra-
cene, fluorene, fluoranthene, and dibenzothiophene.
Each substrate, except for anthracene, was dissolved
in dimethylsulfoxide (10 mg/ml); 50 ml of the result-
ing solution was added to 5 ml of cell suspension.
Anthracene was dissolved in N,N-dimethylformamide
(10 mg/ml) and similarly added to the cell suspension.
The reaction mixtures were incubated on a reciprocal
shaker (300 strokes/min) at 30 ꢁC for 18 h.
1
etry. The H and 13C NMR spectra of the compound
were recorded using an ECA-600 spectrometer (JEOL,
Tokyo) operated at 600 and 150 MHz respectively, with
TMS as an internal standard. Two-dimensional NMR
experiments for the determination of direct 1H–13C
connectivity (HSQC), long-range (two- and three-bond)
1H–13C connectivity (HMBC), measurements of nuclear
Overhauser enhancement spectroscopy (NOESY), and
1H–1H shift-correlated spectroscopy (1H–1H DQF-
COSY) were done under the above conditions.
The reaction mixtures were extracted with ethyl
acetate, and the ethyl acetate layer was dried over
anhydrous Na2SO4. A portion of each extract was
directly analyzed by gas chromatography-mass spec-
trometry (GC-MS) after trimethylsilylation with N-
methyl-N-trimethylsilyltrifluoroacetamide (MSTFA) at
70 ꢁC for 20 min, except for the quantification of
dibenzothiophene and its oxygenation products.
GC-MS was performed as described previously.6) The
products were identified by comparison of their mass
spectra with data from previous reports.6,9,16,17) To
quantify the products formed and the remaining sub-
strate, we compared the peak areas for the total ion
current of the respective compounds extracted from the
reaction mixtures containing E. coli cells that harbored
pUCARAJ3 derivatives. The product formation rate was
calculated from the peak area of the product divided by
the sum of the peak areas of the product formed and the
remaining substrate.
Quantification of the remaining dibenzothiophene and
its oxygenation products was done by monitoring their
respective A220s using high-performance liquid chroma-
tography (HPLC). HPLC was done using a Senshu-Pak
ODS 1251-D column (4.6 by 250 mm; Senshu Scientific,
Tokyo). The solvent flow rate was 0.8 ml/min, and the
column was maintained at 30 ꢁC. A portion of the ethyl
acetate extract from the reaction mixture was dried
We observed a marked increase in the putative
monohydroxyl derivative of fluorene on GC-MS analy-
sis of the biotransformation product by the F275W
mutant enzyme. A preparative TLC with a solvent
system of hexane-ethyl acetate (2:1, by volume) gave
Rf ¼ 0:6 for this product. The desired product was
scraped off, purified, and identified by NMR, as
described above.
Western blots. Crude cell extract of E. coli in which
the expression of CARDO-O mutant was induced as
described above was prepared by sonication and
subsequent centrifugation. Protein concentrations were
estimated using a protein assay kit (Bio-Rad Laborato-
ries, Tokyo) with BSA as standard by the method
described by Bradford.18) A 100 to 1,000-mg aliquot of
denatured crude cell extract was subjected to SDS-12%
polyacrylamide gel electrophoresis (SDS–PAGE) and
electrophoretically transferred onto PVDF membranes
using the iBlot Gel Transfer System (Invitrogen,
Tokyo). The CarAa protein (CARDO-O subunit) was
detected immunologically using a rabbit polyclonal
antibody as the primary antibody and horseradish
peroxidase-linked anti-rabbit Ig (GE Healthcare Bio-
Sciences, Tokyo) as the secondary antibody. A CarAa-
specific polyclonal antibody was prepared by Sigma

Structure-Based Mutagenesis of Angular Dioxygenase
3241
Table 1. Plasmids Used in This Study
Plasmid
pBJ3001
Relevant characteristics
Source or reference
Apr, pBluescript II KS(ꢀ) with 5.6-kb EcoRI fragment from J3, carAaBaBbCAcAd
15
pBluescript II SK(ꢀ) Apr, lacZ, pMB9 replicon
Stratagene
This study
This study
This study
This study
This study
This study
This study
This study
This study
This study
This study
This study
This study
This study
This study
This study
This study
This study
This study
This study
This study
This study
This study
Merck Japan
This study
This study
This study
This study
This study
This study
This study
This study
This study
This study
This study
This study
This study
This study
This study
This study
This study
This study
This study
12
pJ31BP
pJ31BP-F275A
pJ31BP-F275I
pJ31BP-F275L
pJ31BP-F275V
pJ31BP-F275W
pJ31BP-F329A
pJ31BP-F329I
pJ31BP-F329L
pJ31BP-F329V
pJ31BP-F329W
pJ31BP-I262A
pJ31BP-I262L
pJ31BP-I262V
pJ31BP-I262W
pJ31BP-Q282N
pJ31BP-Q282S
pJ31BP-Q282Y
pJ31BPC
pJ31BPC-I262A
pJ31BPC-I262L
pJ31BPC-I262V
pJ31BPC-I262W
pT7Blue T-vector
pTemplateJ
pTemplateJ-F275A
pTemplateJ-F275I
pTemplateJ-F275L
pTemplateJ-F275V
pTemplateJ-F275W
pTemplateJ-F329A
pTemplateJ-F329I
pTemplateJ-F329L
pTemplateJ-F329V
pTemplateJ-F329W
pTemplateJ-Q282N
pTemplateJ-Q282S
pTemplateJ-Q282Y
pTemplateJ-II
Apr, pBluescript II SK(ꢀ) with 1.9-kb BamHI-PstI fragment of J3, carAa
pJ31BP derivative, Triplet codon for F275 is replaced with GCG
pJ31BP derivative, Triplet codon for F275 is replaced with ATT
pJ31BP derivative, Triplet codon for F275 is replaced with CTG
pJ31BP derivative, Triplet codon for F275 is replaced with GTG
pJ31BP derivative, Triplet codon for F275 is replaced with TGG
pJ31BP derivative, Triplet codon for F329 is replaced with GCG
pJ31BP derivative, Triplet codon for F329 is replaced with ATT
pJ31BP derivative, Triplet codon for F329 is replaced with CTG
pJ31BP derivative, Triplet codon for F329 is replaced with GTG
pJ31BP derivative, Triplet codon for F329 is replaced with TGG
pJ31BP derivative, Triplet codon for I262 is replaced with GCG
pJ31BP derivative, Triplet codon for I262 is replaced with CTG
pJ31BP derivative, Triplet codon for I262 is replaced with GTG
pJ31BP derivative, Triplet codon for I262 is replaced with TGG
pJ31BP derivative, Triplet codon for Q282 is replaced with AAT
pJ31BP derivative, Triplet codon for Q282 is replaced with TCG
pJ31BP derivative, Triplet codon for Q282 is replaced with TAC
Apr, pBluescript II SK(ꢀ) containing 1.2-kb ClaI insert of pJ31BP
pJ31BPC derivative, Triplet codon for I262 is replaced with GCG
pJ31BPC derivative, Triplet codon for I262 is replaced with CTG
pJ31BPC derivative, Triplet codon for I262 is replaced with GTG
pJ31BPC derivative, Triplet codon for I262 is replaced with TGG
Apr, lacZ, f1 ori, T7 promoter, pUC/M13 priming sites
Apr, pBluescript II SK(ꢀ) with 628-bp EcoRV fragment from pJ31BP
pTemplateJ derivative, Triplet codon for F275 is replaced with GCG
pTemplateJ derivative, Triplet codon for F275 is replaced with ATT
pTemplateJ derivative, Triplet codon for F275 is replaced with CTG
pTemplateJ derivative, Triplet codon for F275 is replaced with GTG
pTemplateJ derivative, Triplet codon for F275 is replaced with TGG
pTemplateJ derivative, Triplet codon for F329 is replaced with GCG
pTemplateJ derivative, Triplet codon for F329 is replaced with ATT
pTemplateJ derivative, Triplet codon for F329 is replaced with CTG
pTemplateJ derivative, Triplet codon for F329 is replaced with GTG
pTemplateJ derivative, Triplet codon for F329 is replaced with TGG
pTemplateJ derivative, Triplet codon for Q282 is replaced with AAT
pTemplateJ derivative, Triplet codon for Q282 is replaced with TCG
pTemplateJ derivative, Triplet codon for Q282 is replaced with TAC
Apr, pBluescript II SK(ꢀ) containing 833-bp ClaI-ApaI insert of pJ31BPC
pTemplateJ-II-I262A pTemplateJ-II derivative, Triplet codon for I262 is replaced with GCG
pTemplateJ-II-I262L pTemplateJ-II derivative, Triplet codon for I262 is replaced with CTG
pTemplateJ-II-I262V pTemplateJ-II derivative, Triplet codon for I262 is replaced with GTG
pTemplateJ-II-I262W pTemplateJ-II derivative, Triplet codon for I262 is replaced with TGG
pUC119
pUCARA
Apr, lacZ, f1 ori, M13 priming sites
Apr, modified pUC119 (lacking PstI site in multi-cloning site) with 5.6-kb EcoRI insert of pUCA1
with the deletion of 1.3-kb PstI-BglI fragment, carAaAaAcAd
Apr, modified pUC119 (lacking PstI site in multi-cloning site) with 3.8-kb EcoRI fragment from J3
in which 1.7-kb PstI fragment containing carBaBbC region was deleted, carAaAcAd
pUCARAJ3 derivative, expressing F275A CARDO-O mutant
pUCARAJ3 derivative, expressing F275I CARDO-O mutant
pUCARAJ3 derivative, expressing F275L CARDO-O mutant
pUCARAJ3 derivative, expressing F275V CARDO-O mutant
pUCARAJ3 derivative, expressing F275W CARDO-O mutant
pUCARAJ3 derivative, expressing F329A CARDO-O mutant
pUCARAJ3 derivative, expressing F329I CARDO-O mutant
pUCARAJ3 derivative, expressing F329L CARDO-O mutant
pUCARAJ3 derivative, expressing F329V CARDO-O mutant
pUCARAJ3 derivative, expressing F329W CARDO-O mutant
pUCARAJ3 derivative, expressing I262A CARDO-O mutant
pUCARAJ3 derivative, expressing I262L CARDO-O mutant
pUCARAJ3 derivative, expressing I262V CARDO-O mutant
pUCARAJ3 derivative, expressing I262W CARDO-O mutant
pUCARAJ3 derivative, expressing Q282N CARDO-O mutant
pUCARAJ3 derivative, expressing Q282S CARDO-O mutant
pUCARAJ3 derivative, expressing Q282Y CARDO-O mutant
5
This study
pUCARAJ3
pUCARAJ3-F275A
pUCARAJ3-F275I
pUCARAJ3-F275L
pUCARAJ3-F275V
pUCARAJ3-F275W
pUCARAJ3-F329A
pUCARAJ3-F329I
pUCARAJ3-F329L
pUCARAJ3-F329V
pUCARAJ3-F329W
pUCARAJ3-I262A
pUCARAJ3-I262L
pUCARAJ3-I262V
pUCARAJ3-I262W
pUCARAJ3-Q282N
pUCARAJ3-Q282S
pUCARAJ3-Q282Y
This study
This study
This study
This study
This study
This study
This study
This study
This study
This study
This study
This study
This study
This study
This study
This study
This study

3242
H. UCHIMURA et al.
Genosys (Ishikari, Japan) using the native (non-His-
tagged) form of the CarAa protein from CA10 that was
overexpressed by E. coli cells and purified as described
previously.8) Western blots were performed using
Immobilon Western (Nihon Millipore, Tokyo) according
to the manufacturer’s instructions. CarAa protein signals
were visualized using Luminescent Image Analyzer
LAS-1000 plus (Fuji Photo Film, Tokyo).
ranthene, and dibenzothiophene. The quantification of
the products formed by all CARDO-O mutants is
summarized in Tables 2–4. Compared to our previous
results using the WT CARDO system from P. resin-
ovorans CA10,6,9) we did not detect any significant
difference in the substrate preference or product for-
mation capability of the WT CARDO system from J3.
Dibenzo-p-dioxin and anthracene
Results
In biotransformation experiments using several mu-
tant and WT enzymes, we detected production of 2,20,3-
trihydroxydiphenyl ether formed by AD with dibenzo-p-
dioxin, followed by spontaneous central ring-cleavage
reactions (Table 2). No other products, such as cis-
dihydrodiols of dibenzo-p-dioxin or their putative dehy-
dration product and monohydroxydibenzo-p-dioxin,
were detected.
In comparison with previous data from GC-MS,
anthracene was converted to anthracene-1,2-dihydro-
diol, a major product of CARDO from CA10,6) by WT
CARDO from J3 and several of its mutants. In addition,
another anthracene cis-dihydrodiol-like compound was
detected by GC-MS [356 (Mþ, 43), 341 (25), 266 (40),
178 (25), 73 (100)] and was assumed to be 2,3-
dihydrodiol, because anthracene has two tandemly
linked vicinal carbon atom pairs for which LD is
possibly catalyzed (Table 2).
The oxygenation capabilities of the CARDO-O
mutants observed for dibenzo-p-dioxin had a tendency
similar to those for anthracene, although these two
substrates were catalyzed primarily by AD and LD
respectively. Generally, mutants having the replacement
at F275 or F329 had no or dramatically reduced
oxygenation activity for these substrates, except for the
F275W mutant. In contrast, the I262L, I262V, F275W,
and Q282Y mutants showed relatively high oxygenation
activity for these substrates among the I262- and Q282-
substituents, although the oxygenation activity of I262A,
I262W, and Q282N decreased markedly. It is note-
worthy that production of the putative 2,3-dihydrodiol of
anthracene, a minor product of the WT enzyme, was
highly increased in F275W as compared to the produc-
tion of anthracene-1,2-dihydrodiol.
Preparation of CARDO-O mutants by site-directed
mutagenesis
To identify the possible amino acid residues affecting
the binding of substrate to CARDO-O, we first per-
formed a docking simulation of dibenzo-p-dioxin, for
which CARDO catalyzed AD. The computer-predicted
docking positions of dibenzo-p-dioxin rarely differed
(Fig. 2A), and the root-mean-square difference of non-
hydrogen atoms of the 10 members from its cluster
˚
center was less than 1.08 A. This manner of binding for
dibenzo-p-dioxin is in accord with the catalysis of AD
(cis-1,10a-dihydroxylation) by CARDO-O, and was well
superposed on the positions of the carbazole binding in
the CARDO-O:CARDO-F:carbazole ternary complex
structure11) (Fig. 2B).
Except for the ligands for ferrous iron active site,
˚
there are six amino acids within 3.6 A distance from any
atoms of carbazole molecule. Among them, I184 is the
amino acid neighboring to the active site ligand, H183,
and Nꢂ2 of N330 forms hydrogen bond with Oꢂ2 of
D333, the ligand of the ferrous ion active site. In this
situation, their replacements were expected to affect the
active site geometry and to affect the CARDO-O
catalytic activity itself. Based on this consideration, we
selected amino acid residues I262, F275, Q282, and
F329, which can affect the catalysis of AD by CARDO-
O. Accordingly, we generated site-directed mutants by
replacing a single amino acid residue with similar amino
acid residues of different size in the side chain in each
one of the above four residues (viz., for Ile: Ala, Val,
Leu, and Trp; for Phe: Ala, Val, Leu, Ile, and Trp; and
for Gln: Ser, Tyr, and Asn).
Production of CARDO-O mutants in E. coli cells
Using the E. coli JM109 expressing the CARDO-O
mutant in conjunction with CARDO-F and CARDO-R
from Janthinobacterium sp. J3, crude cell extracts were
prepared and subjected to Western blotting using CarAa-
specific antibody. The WT and all of the CARDO-O
mutants produced a full-length soluble protein with a
small variation in the amount of each mutant protein
obtained (Fig. 3).
Carbazole
Although F329-replaced mutants showed only negli-
gible oxygenase activity, most mutant enzymes pro-
duced 20-aminobiphenyl-2,3-diol, which is generated by
AD followed by spontaneous rearomatization and
hetero-ring cleavage, as the main product from carba-
zole (Table 2). In addition to the AD product, I262A,
I262L, I262V, F275W, and Q282-replaced mutants
yielded another product from carbazole. In the case of
the I262V and Q282Y mutants, the amount of new
product detected equal to or was of higher than that of
20-aminobiphenyl-2,3-diol. GC-MS showed that the
TMS derivative of the new product had the presumed
molecular ion peak at m=z 255, implying that this
Oxygenating activity of the site-directed mutants of
CARDO-O
We carried out biotransformation experiments with
carbazole, dibenzo-p-dioxin, anthracene, fluorene, fluo-

Structure-Based Mutagenesis of Angular Dioxygenase
3243
A
A
B
I262
I262
A259
A259
I231
I231
Q282
F275
Q282
I262
F275
F275
C10a
2+
2+
Fe
Fe
C1
F204
F204
D333
D333
H187
F329
F329
C9a
2+
O1
O1
Q282
H183
H183
C1
O2
2+
Fe
O2
L200
L200
H187
Fe
F329
V272
V272
B
C
L307
L307
Fig. 2. Computer-Predicted Docking Structures of Dibenzo-p-dioxin
in the CARDO-O Active Site (A) and a Comparison with the
Carbazole-Binding Position in the CARDO-O:CARDO-F:Carbazole
Ternary Complex Structure (B).11)
Q282
F352
Q282
F275
V260
F275
V260
Ten representative binding conformations of dibenzo-p-dioxin to
dioxygen-bound CARDO-O are shown in panel A. In panel B, the
cluster center of the 10 dibenzo-p-dioxin-docking structures and
carbazole-docking crystal structure are colored magenta and green
respectively. In both panels, the catalytic mononuclear irons, the
modeled dioxygen bound to the active site, and the ligand water in
the crystal structure are shown in the CPK model, and are colored
brown, red, and marine blue respectively. Ile262, Phe275, Gln282,
and Phe329 residues, which were replaced with similar amino acid
residues in this study, are shown in the wire-frame model, and are
colored gold.
F352
F329
F329
W358
W358
V272
V272
I326
I326
Q282
F368
Q282
F368
F275
F275
F329
F329
T367
T367
50 kDa
37 kDa
Fig. 4. Stereoview of the Amino Acid Residues Possibly Affecting
the Substrate Specificity of CARDO from Janthinobacterium sp. J3
(A), and a Comparison with Those in NDO-O from Pseudomonas sp.
NCIB9816-4 (B) and BDO-O from Rhodococcus jostii RHA1 (C).
In panel A, intended amino acid residues around the ferrous iron
active sites, including ligands (gray), target amino acids for site-
directed mutagenesis (gold), and other possible determinants of
substrate specificity (wheat) are shown in the wire-frame model.
Catalytic mononuclear irons are shown as spheres. Carbazole is also
shown in the wire-frame model, colored gray. In panels B and C,
carbazole, ferrous iron, and CARDO-O amino acid residues are
shown as in panel A. The ligands of the ferrous iron active site,
substrate, and amino acid residues affecting substrate specificities of
NDO-O (in panel B) and BDO-O (in panel C) are colored blue and
green respectively.
50 kDa
37 kDa
50 kDa
37 kDa
50 kDa
37 kDa
Fig. 3. Western Blots of E. coli Cell Lysates Expressing CARDO-O
Mutants Using the CARDO-O-Specific Antibody.
A rabbit polyclonal antibody specific for the CarAa protein, which
is a subunit of CARDO-O, was used, as described in ‘‘Materials and
Methods.’’ A 0.15-mg aliquot of soluble proteins isolated from
E. coli cells was loaded for SDS–PAGE. ‘‘P’’ and ‘‘N’’ mean
positive control (soluble protein of E. coli JM109(pUCARAJ3)) and
negative control (soluble protein of E. coli JM109(pUC119))
respectively.
7.45 (d, J ¼ 8:24 Hz, H-8), 7.65 (d, J ¼ 7:93 Hz, H-4),
8.04 (d, J ¼ 7:78 Hz, H-5). In addition, the 13C NMR for
this compound showed signals at ꢂ 110.70 (C-2), 111.00
(C-8), 112.95 (C-4), 119.42 (C-6), 119.80 (C-3), 120.64
(C-5), 123.73 (C), 125.32 (C), 125.91 (CH), 129.18 (C),
139.51 (CH), and 141.51 (C-1). Based on the results of
the 1H–1H COSY spectra (data not shown), we
identified the compound as 1-hydroxycarbazole. There-
fore, the CARDO-O mutants mentioned above acquired
the ability to convert carbazole to 1-hydroxycarbazole,
which was assumed to be created by the spontaneous
dehydration of primary cis-dihydroxylation products.
A comparison of oxygenation activity for dibenzo-p-
dioxin and carbazole showed that the F275V, Q282N, and
Q282S mutants retained relatively high AD activity for
carbazole, but showed only low or negligible AD activity
for dibenzo-p-dioxin, although the WT enzyme cata-
lyzes AD for these two substrates as a primary reaction.
product was monohydroxycarbazole [255 (Mþ, 100),
239 (52), 224 (46), 209 (20), 73 (33)]. Its fragmentation
pattern was similar to those of the TMS derivatives of
the authentic 2-hydroxy- and 4-hydroxycarbazole, but
its GC retention time was different from those of these
TMS derivatives (data not shown). The 1H NMR spectra
of the product in CDCl3 showed chemical shifts (ꢂ ppm)
with chemical shift multiplicities (d, doublet; t, triplet,
coupling constants, J in Hz, in parentheses), as follows:
6.83 (d, J ¼ 7:78 Hz, H-2), 7.06 (t, J ¼ 7:78 Hz, H-3),
7.21 (t, J ¼ 7:01 Hz, H-6), 7.40 (t, J ¼ 7:02 Hz, H-7),

3244
H. UCHIMURA et al.
Table 2. Products Formed from Carbazole, Dibenzo-p-dioxin, and Anthracene by CARDO-O Mutantsa
From carbazole
From dibenzo-p-dioxin
From anthracene
20-Aminobiphenyl-
1-Hydroxy-
carbazoleb
2,20,3-Trihydroxy-
diphenyl etherb
Anthracene-
1,2-dihydrodiolb
Putative anthracene
2,3-dihydrodiolb
2,3-diolb
Wild-type
99.1 (0.07)
Trace
88.7 (1.56)
20.8 (1.82)
6.17 (0.40)
I262A
I262L
I262V
I262W
32.0 (1.26)
95.3 (0.24)
26.3 (9.27)
31.5 (2.18)
1.67 (1.74)
4.71 (0.24)
16.2 (1.26)
N.D.
20.0 (7.15)
90.5 (1.32)
37.5 (1.80)
18.8 (5.60)
1.06 (0.06)
83.7 (5.22)
8.25 (0.46)
1.69 (1.49)
Trace
1.84 (0.24)
Trace
N.D.
F275A
F275I
F275L
F275V
F275W
24.4 (0.78)
45.3 (2.62)
3.14 (0.95)
18.1 (7.22)
91.8 (1.62)
N.D.
N.D.
N.D.
Trace
8.15 (1.63)
9.63 (5.60)
N.D
Trace
1.04 (0.31)
97.8 (0.27)
3.24 (1.81)
N.D.
N.D.
N.D.
18.1 (0.10)
N.D.
N.D.
N.D.
N.D.
22.7 (1.65)
Q282N
Q282S
Q282Y
63.9 (1.93)
33.8 (5.00)
39.3 (1.91)
3.90 (0.95)
2.63 (2.90)
46.8 (5.97)
2.38 (0.68)
12.9 (9.60)
43.8 (2.77)
N.D.
2.34 (2.21)
11.1 (9.73)
N.D.
2.21 (0.10)
1.11 (0.98)
F329A
F329I
F329L
F329V
F329W
Trace
Trace
Trace
Trace
Trace
Trace
Trace
Trace
Trace
N.D.
N.D.
Trace
Trace
N.D.
N.D.
N.D.
N.D.
N.D.
N.D.
N.D.
N.D.
N.D.
N.D.
N.D.
N.D.
a‘‘Trace’’ means that the product formation estimated was less than 1%. ‘‘N.D.’’ means that product formation was not detected. Values in parentheses represent
standard deviations calculated from at least three assays.
bTo quantify the products formed in GC-MS analyses, we compared the peak areas for the total ion current of the respective compounds extracted from the reaction
mixtures containing E. coli cells that harbored pUCARAJ3 derivatives. The product formation rate was calculated from the peak area of the product divided by the
sum of the peak areas of the product formed and the remaining substrate.
Table 3. Products Formed from Fluorene and Fluoranthene by CARDO-O Mutantsa
From fluorene
From fluoranthene
7-Hydroxy-
fluorantheneb
cis-2,3-Dihydroxy-2,3-
dihydrofluorantheneb
Putative cis-dihydrodiol
of fluorantheneb
9-Hydroxyfluoreneb
4-Hydroxyfluoreneb
3.03 (0.70)
Wild-type
5.86 (0.18)
Trace
Trace
N.D.
I262A
I262L
I262V
I262W
Trace
6.80 (0.16)
Trace
N.D.
Trace
Trace
N.D.
N.D.
Trace
N.D.
N.D.
N.D.
Trace
Trace
N.D.
N.D.
N.D.
N.D.
N.D.
1.01 (0.02)
F275A
F275I
F275L
F275V
F275W
4.12 (0.98)
1.24 (0.06)
Trace
1.07 (0.30)
3.90 (1.81)
N.D.
N.D.
N.D.
N.D.
Trace
N.D.
N.D.
N.D.
Trace
Trace
N.D.
N.D.
N.D.
Trace
9.72 (1.60)
N.D.
N.D.
N.D.
N.D.
22.6 (1.47)
Q282N
Q282S
Q282Y
Trace
Trace
1.19 (0.11)
Trace
N.D.
Trace
N.D.
N.D.
N.D.
Trace
Trace
Trace
N.D.
N.D.
N.D.
F329A
F329I
F329L
F329V
F329W
2.67 (0.02)
2.23 (1.20)
2.47 (1.26)
2.95 (0.41)
2.91 (1.46)
N.D.
Trace
N.D.
N.D.
N.D.
N.D.
N.D.
N.D.
N.D.
N.D.
N.D.
N.D.
N.D.
N.D.
N.D.
N,D.
N.D.
N.D.
N.D.
N.D.
a‘‘Trace’’ means that the product formation estimated was less than 1%. ‘‘N.D.’’ means that product formation was not detected. Values in parentheses represent
standard deviations calculated from at least three assays.
bTo quantify the products formed in GC-MS analyses, we compared the peak areas for the total ion current of the respective compounds extracted from the reaction
mixtures containing E. coli cells that harbored pUCARAJ3 derivatives. The product formation rate was calculated from the peak area of the product divided by the
sum of the peak areas of the product formed and the remaining substrate.
Fluorene
showed relatively high oxygenating activity for fluorene.
Although F275W catalyzed 9-hydroxylation of fluorene
at an efficiency as low as that of the WT, the yield of
another product was increased to approximately 22.6%.
The oxygenation activity of the WT and mutant
CARDO-O enzymes for fluorene were noticeably lower
than those for carbazole (Table 3), but only F275W

Structure-Based Mutagenesis of Angular Dioxygenase
3245
Table 4. Products Formed from Dibenzothiophene by Wild-Type
and CARDO Mutantsa
CARDO-O enzymes, most mutant enzymes lost all
fluoranthene-oxygenating activity (Table 3), but the
F275A mutant yielded a new and noteworthy product.
GC-MS of the TMS derivative of this new product
provided a fragmentation pattern 380 (Mþ, 78), 259
(10), 202 (28), 147 (26), 73 (100) and the putative
molecular ion peak, suggesting that the new product was
cis-dihydrodiol of fluoranthene. However, definite iden-
tification of this product by NMR could not be made due
to a low conversion ratio.
From dibenzothiophene
Putative
Dibenzothiophene
monohydroxy-
dibenzothiophenesb,c
5-oxideb
Wild-type
70.8 (1.58)
10.7 (2.02)
I262A
I262L
I262V
I262W
47.2 (3.77)
89.2 (1.20)
84.6 (2.08)
21.3 (3.52)
45.3 (8.16)
7.70 (1.23)
5.33 (2.08)
41.9 (6.54)
Dibenzothiophene
F275A
F275I
F275L
F275V
F275W
30.6 (8.15)
N.D.
N.D.
N.D.
90.2 (2.15)
21.6 (7.64)
N.D.
N.D.
N.D.
7.38 (1.61)
In a biotransformation experiment using the WT
enzyme, we detected dibenzothiophene-5-oxide as the
major product from dibenzothiophene, similarly to
previous results (Table 4).6,9) The overall tendency in
oxygenation capability for dibenzothiophene observed in
mutants was roughly similar to those for other substrates.
A marked decrease or complete loss of oxygenating
activity was detected in some of the F275-replaced
mutants and in all of the F329-replaced mutants. Active
mutants could be divided into two groups: mutants
exhibiting activity similar to WT activity (I262L, I262V,
F275W, Q282S, and Q282Y) and mutants producing
rather large amounts of other products from dibenzo-
thiophene (I262A, I262W, and F275A). The new
products displayed UV spectrum maxima at 256 and
285 nm. GC-MS of the HPLC-fractionated new products
(HPLC retention time, 40.5 min) suggested that this
fraction contained three products that had similar MS
fragmentation patterns, as follows: 272 (Mþ, 73), 257
(100), 197 (12), 139 (16), 73 (21); 272 (Mþ, 100), 257
(89), 184 (17), 139 (18), 73 (17); 272 (Mþ, 63), 257 (65),
117 (56), 71 (68), 57 (100). In comparison with the
previous results using WT CARDO-O from CA10,6,9)
these products are proposed to be monohydroxy deriv-
atives of dibenzothiophene.
Q282N
Q282S
Q282Y
12.2 (3.23)
65.2 (6.84)
53.0 (2.33)
7.12 (2.96)
8.33 (1.53)
6.34 (2.72)
F329A
F329I
F329L
F329V
F329W
N.D.
N.D.
N.D.
N.D.
N.D.
N.D.
N.D.
N.D.
N.D.
N.D.
a‘‘N.D.’’ means that product formation was not detected.
bTo quantify the products formed, we compared the peak areas in HPLC
analyses for the respective A220s of the compounds extracted from reaction
mixtures containing E. coli cells that harbored pUCARAJ3 derivatives. The
product formation rate was calculated from the peak area of the product
divided by the sum of the peak areas of the product formed and the
remaining substrate.
cAt least three monohydroxyl derivatives were included in the products.
This second product was also produced by the WT, but
the production yield was markedly low (approximately
3%). GC-MS showed that the TMS derivative of the
unidentified product had the assumed molecular ion peak
at m=z 254, implying that this product was monohydro-
xyfluorene [254 (Mþ, 90), 239 (41), 223 (100), 165 (12),
152 (10)]. Although its fragmentation pattern was similar
to that of the TMS derivatives of authentic 2-hydro-
Discussion
1
xyfluorene, its GC retention time was different. The H
NMR spectra of the product in CDCl3 showed chemical
shifts (ꢂ ppm, chemical shifts multiplicities and coupling
constants in parenthese) as follows: 6.73 (d, J ¼ 6:0 Hz,
H-3), 7.14 (d, J ¼ 7:0 Hz, H-1), 7.15 (t, J ¼ 6:8 Hz,
H-2), 7.27 (t, J ¼ 7:3 Hz, H-7), 7.37 (t, J ¼ 7:5 Hz, H-6),
7.51 (d, J ¼ 7:5 Hz, H-8), 8.10 (d, J ¼ 7:5 Hz, H-5). In
addition, the 13C NMR for this compound showed signals
at ꢂ 113.67 (C-3), 117.61 (C-1), 123.45 (C-5), 124.44 (C-
8), 125.92 (C-7), 126.77 (C-6), 127.49 (C-2), 128.60
(CH), 140.62 (C), 142.49 (C), 145.86 (CH), and 151.51
(C-4). Based on the results of the COSY spectra, the
NOESY spectra, and HMBC (data not shown), we
identified the compound as 4-hydroxyfluorene.
To understand the molecular mechanism governing
the substrate specificity of CARDO, we generated
CARDO-O mutants that had different substrate specif-
icities from WT CARDO-O. Detailed studies of the
substrate specificity of CARDO were performed using
CARDO from P. resinovorans CA10.6,9) Because the
first trial in determining the crystal structure of CARDO-
O from CA10 failed, we used CARDO-O from
Janthinobacterium sp. J3 in X-ray crystallography, and
succeeded in obtaining the three-dimensional structures
of CARDO-O and its ternary complex form with
CARDO-F and carbazole.10,11) The structure of the
substrate-binding pocket and the manner of carbazole
binding around the ferrous iron active site provided us
with information on the amino acid residues constituting
the substrate-interacting surface around the ferrous iron
active site and an important cue to the targets for amino
acid substitution to improve or modify the activity of
Fluoranthene
Although conversion of fluoranthene to 7-hydroxy-
fluoranthene and/or cis-2,3-dihydroxy-2,3-dihydrofluor-
anthene was detected in the WT and some mutant

3246
H. UCHIMURA et al.
this novel enzyme (Fig. 2). Based on this information,
we created site-directed mutants at the F275, Q282, and
F329 residues and the I262 residue of CARDO-O from
J3 and determined their oxygenation activities for
several substrates. The former three amino acid residues
are located near the bottom of the substrate-binding
pocket and are close to the aromatic ring to be oxidized.
The latter amino acid residue is located near the entrance
of the substrate-binding pocket and is close to another
aromatic ring (Figs. 2 and 4). All mutants expressed
CARDO-O at similar efficiency as the WT CARDO-O,
implying that most substitutions did not affect the
folding or the stability of the CARDO-O protein.
Accordingly, the product-forming activities detected
in the biotransformation experiments indicated the
substrate specificities of the mutant enzymes.
Based on our results using the WT enzyme, we did
not detect significant differences between the substrate
specificities of CARDO-O from CA106,9) and that from
J3. This result may be because the two CARDO-O
molecules of different origin have only three amino acid
substitutions in the regions farthest from the substrate-
binding pocket and ferredoxin-binding surface (data not
shown).
The crystal structure of the carbazole-bound form of
CARDO-O clearly suggests that two benzene moieties
of carbazole interact with the hydrophobic wall in the
substrate-binding pocket. In addition, the binding of
carbazole is stabilized in a single orientation by hydro-
gen-bonding between imino nitrogen and carbonyl
oxygen of G178.11) Unlike carbazole, dibenzo-p-dioxin
and anthracene have no functional group affecting their
orientation in the substrate-binding pocket, and the
binding of these substrates appeared to be stabilized
only by hydrophobic interactions between the substrate
and the wall of the substrate-binding pocket. Consider-
ing that these two substrates have similar three-ring
fused planar structures, it is reasonable to suggest that
they bind in a similar manner around the ferrous iron
active site of the WT CARDO-O and most CARDO-O
mutants. This coincides with the fact that the oxygen-
ation capability of the mutant CARDO-Os for these two
substrates was essentially unaffected.
Among the amino acid residues in which replacement
was done, the phenyl moiety of F275 appears to be
directly involved in the stacking interaction with the
benzene ring of carbazole, which is hydroxylated. In fact,
the F275 replacement with amino acid residues that have
aliphatic side chains dramatically reduced the oxygen-
ation activity for the substrates tested (Tables 2–4). The
above-mentioned role of F275 in substrate binding
appears to coincide with our results for the fluorene
biotransformation experiments. When fluorene was
added as a substrate for WT CARDO-O and its mutants,
we detected only weak oxygenation activity (Table 3),
but the F275W mutant showed elevated oxygenation
activity, and 4-hydroxyfluorene was detected as a major
product. These results imply that fluorene does not bind
tightly around the ferrous iron active site in the WT and
in most of the mutant CARDO-Os. The indole moiety of
W275 in the F275W mutant formed a strengthened
stacking interaction with the benzene moiety of the
substrate. The bulky indole moiety might change the
orientation of the fluorene bound around the ferrous iron
active site, resulting in the formation of 4-hydroxyfluor-
ene as a major product. Another interesting CARDO-O
mutant is F275A, which showed unique oxygenation
activity for fluoranthene (Table 3). Although the oxy-
genation activity of the WT and mutants of CARDO-O
for this bulky substrate was extremely low, the F275A
mutant clearly showed elevated oxygenation activity
producing putative cis-dihydrodiol from fluoranthene,
although the oxygenating site has not been identified.
Alanine substitution for the F275 might enlarge the space
of the substrate-binding pocket, and this structural
change might be beneficial for binding of the bulky
aromatic substrate. Similar enlargement of the substrate-
binding pocket has been reported in other RO oxygenase
components to accommodate larger substrates, such as
benzo[a]pyrene, benzo[a]anthracene, and chrysene.19,20)
His183, His187, and Asp333 coordinate the ferrous
iron active site,10) forming a 2-His-1-carboxylate facial
triad (Fig. 4A).21) The main chain carbonyl of F329
formed hydrogen bonds to imidazole Nꢂ1 of H187.10,11)
The F329V mutation can lead to a disorder of the
coordination geometry (viz., a distorted octahedral
bipyramid) in the CARDO-O active site. This phenom-
enon can result in the loss of the ferrous iron active site,
resulting in negligible or lower activities of F329
mutants. The F329 residue corresponds to the Trp358
residue of the large (ꢀ) subunit (the NahAc protein)
of the terminal oxygenase component of naphthalene
1,2-dioxygenase (NDO-O) from Pseudomonas sp.
NCIB9816-4 (Fig. 4A and B). A similar dramatic
decrease in oxygenation activity has also been reported
for the Ala substitution for W358,22) although no
hydrogen bonding occurred between W358 and the
ligand active sites.23)
To the best of our knowledge, the amino acid residues
located near the entrance of the active site pocket, such
as I262 of CARDO-O, have not been targets of regional
design for the alteration of substrate specificities of ROs.
This amino acid is located in close proximity to F204
and I231 (Fig. 4A). These amino acid residues are
located on the loops, which move to the entrance of the
substrate-binding pocket upon carbazole binding.11)
These conformational changes presumably result in
trapping carbazole at the substrate-binding site and
excluding water from the active site during turnover, and
prevent the leakage of partially processed substrates. In
addition, the sec-butyl side chain of I262 forms hydro-
phobic interactions with the portion of the aromatic ring
of the carbazole skeleton in which oxygenation is not
occurring (Figs. 2 and 4A). This suggests the possibility
that this residue affects not only the orientation of the
substrate in the substrate-binding pocket, but also the

Structure-Based Mutagenesis of Angular Dioxygenase
3247
introduction of the substrate into the active site cavity or
the leakage of substrate/intermediate from the cavity. In
fact, the noteworthy changes in substrate specificity due
to I262 replacement are improved selectivity and oxy-
genation activity for anthracene by I262L mutant
(Table 2) and increased aromatic-ring hydroxylating
activity for dibenzothiophene by I262A and I262W
mutants (Table 4).
change in the regio- and stereoselectivity of product
formation. A comparison of the substrate-binding
structures of NDO-O with the terminal oxygenase of
nitrobenzene 1,2-dioxygenase from Comamonas sp.
JS765 also indicates the role of amino acid residues,
affecting regioselectivity in nitroarene substrate oxy-
genation and stereoselectivity in naphthalene oxygen-
ation.26,27) Structural information from CARDO-O mu-
tants should help in understanding the role of replaced
amino acids in the regulation of substrate orientation.
Such detailed studies of the relationship between
structure and substrate specificity should provide critical
cues to the modification and modulation of the substrate
pocket structure to obtain an enhanced catalyst for two
distinct areas of scientific endeavor: bioremediation
(degradation) of toxic contaminants and enzymatic
synthesis of chiral precursors in the production of
specialty compounds. In addition, other amino acid
residues that possibly interact directly with the substrate,
such as I184, L200, A259, V272, and E284 (Fig. 4), are
future targets of site-directed mutation to generate
CARDO-O mutants that have noteworthy oxygenating
activity for various substrates, thus helping to obtain
further information on the mechanisms determining the
substrate specificity of CARDO.
We have found that the shape of the substrate-binding
pocket of CARDO-O is quite different from those of the
terminal oxygenase component of other ROs.10)
However, the carbazole-interacting wall consisting of
F275, Q282, and F329 in the CARDO-O active site
virtually corresponds to the naphthalene-interacting
wall, consisting of V260, L307, F352, and W358, in
the active site of NDO-O from NCIB9816-4 (all amino
acid residues are in the ꢀ subunit of the NDO-O
molecule; Fig. 4B). Of these residues, F352 is critical in
determining both the regio- and enantioselectivity of the
enzyme with a variety of substrates.22,24) In nitrobenzene
1,2-dioxygenase from Comamonas sp. JS765 and 2-
nitrotoluene 2,3-dioxygenase from Acidovorax sp. JS42,
the amino acid residues corresponding to V260 and
F352 in the NDO-O active site are important in
determining regiospecificity with nitroarene substrates
and enantiospecificity with naphthalene.25–27) Suenaga
and coworkers28–30) performed mutagenesis of the
terminal oxygenase component of biphenyl 2,3-dioxy-
genase (BDO-O) from Pseudomonas pseudoalcaligenes
KF707 and generated many BDO-O mutants that had
different oxygenation activity for polychlorinated bi-
phenyl (PCB) congeners and other aromatic compounds.
They found that I335, T376, and F377 in the active site
of BDO-O from KF707 are important for regioselectiv-
ity with various PCB congeners. These amino acid
residues correspond to I326, T367, and F368 respec-
tively in BDO from Rhodococcus jostii RHA1, whose
crystal structure has been reported.31) Superposition of
the carbazole-bound structures of CARDO-O11) and
biphenyl-bound BDO-O31) clearly shows that such
important amino acid residues are located in the wall
corresponding to F275, Q282, and F329 in the CARDO-
O structure (Fig. 4C). Overall, amino acid residues that
alter substrate specificity in CARDO-O and those that
affect the regio- and stereoselectivity of product for-
mation in the above-mentioned terminal oxygenases lay
in similar regions. The observation of many crystal
structures of the terminal oxygenase components of ROs
indicates that the orientation of substrate binding at the
active site is the primary determinant of product regio-
and stereoselectivity, as suggested by Ferraro et al.32)
These regions tend to form the topology of the active site
that influences substrate orientation.
Acknowledgments
This study was supported by a Grant-in-Aid for
Scientific Research (17380052, to HN) from the Ministry
of Education, Culture, Sports, Science, and Technology
of Japan, and by the Program for Promotion of Basic
Research Activities for Biosciences (PROBRAIN).
References
1) Boyd, D. R., and Sheldrake, G. N., The dioxygenase-
catalyzed formation of vicinal cis-diols. Nat. Prod. Rep.,
15, 309–325 (1998).
2) Parales, R. E., and Resnick, S. M., Aromatic ring
hydroxylating dioxygenases. In ‘‘Pseudomonas Vol-
ume 4,’’ eds. Ramos, J.-L., and Levesque, R. C.,
Springer, New York, pp. 287–340 (2006).
3) Arcos, J. C., and Argus, M. F., Molecular geometry and
carcinogenic activity of aromatic compounds. New
perspectives. Adv. Cancer Res., 11, 305–471 (1968).
4) Nojiri, H., and Omori, T., Carbazole metabolism by
pseudomonads. In ‘‘Pseudomonas Volume 5,’’ eds.
Ramos, J.-L., and Filloux, A., Springer, New York,
pp. 107–145 (2007).
5) Sato, S., Nam, J.-W., Kasuga, K., Nojiri, H., Yamane,
H., and Omori, T., Identification and characterization of
genes encoding carbazole 1,9a-dioxygenase in Pseudo-
monas sp. strain CA10. J. Bacteriol., 179, 4850–4858
(1997).
Recently, the structure of the F352V NDO-O mutant
was reported as native, phenanthrene-bound, and
anthracene-bound forms.33) This achievement clearly
indicated how the Val substitution for F352 affects the
substrate orientation in the active site and results in a
6) Nojiri, H., Nam, J.-W., Kosaka, M., Morii, K.,
Takemura, T., Furihata, K., Yamane, H., and Omori,
T., Diverse oxygenation catalyzed by carbazole 1,9a-
dioxygenase from Pseudomonas sp. strain CA10.
J. Bacteriol., 181, 3105–3113 (1999).

3248
H. UCHIMURA et al.
7) Nojiri, H., and Omori, T., Molecular bases of aerobic
20) Jakoncic, J., Jouanneau, Y., Meyer, C., and Stojanoff, V.,
The catalytic pocket of the ring-hydroxylating dioxyge-
nase from Sphingomonas CHY-1. Biochem. Biophys.
Res. Commun., 352, 861–866 (2007).
bacterial degradation of dioxins: involvement of angular
dioxygenation. Biosci. Biotechnol. Biochem., 66, 2001–
2016 (2002).
8) Nam, J.-W., Nojiri, H., Noguchi, H., Uchimura, H.,
Yoshida, T., Habe, H., Yamane, H., and Omori, T.,
Purification and characterization of carbazole 1,9a-
dioxygenase, a three-component dioxygenase system of
Pseudomonas resinovorans strain CA10. Appl. Environ.
Microbiol., 68, 5882–5890 (2002).
9) Takagi, T., Nojiri, H., Yoshida, T., Habe, H., and Omori,
T., Detailed comparison between the substrate spe-
cificities of two angular dioxygenases, dibenzofuran
4,4a-dioxygenase from Terrabacter sp. and carbazole
1,9a-dioxygenase from Pseudomonas resinovorans.
Biotechnol. Lett., 24, 2099–2106 (2002).
10) Nojiri, H., Ashikawa, Y., Noguchi, H., Nam, J.-W.,
Urata, M., Fujimoto, Z., Uchimura, H., Terada, T.,
Nakamura, S., Shimizu, K., Yoshida, T., Habe, H., and
Omori, T., Structure of the terminal oxygenase compo-
nent of angular dioxygenase, carbazole 1,9a-dioxyge-
nase. J. Mol. Biol., 351, 355–370 (2005).
11) Ashikawa, Y., Fujimoto, Z., Noguchi, H., Habe, H.,
Omori, T., Yamane, H., and Nojiri, H., Electron transfer
complex formation between oxygenase and ferredoxin
components in Rieske nonheme iron oxygenase system.
Structure, 14, 1779–1789 (2006).
12) Sambrook, J., and Russell, D. W., ‘‘Molecular Cloning,
a Laboratory Manual’’ 3rd ed., Cold Spring Harbor
Laboratory Press, Cold Spring Harbor (2001).
13) Nojiri, H., Sekiguchi, H., Maeda, K., Urata, M., Nakai,
S., Yoshida, T., Habe, H., and Omori, T., Genetic
characterization and evolutionary implications of car
gene cluster in the carbazole degrader Pseudomonas sp.
strain CA10. J. Bacteriol., 183, 3663–3679 (2001).
14) DeLano, W. L., ‘‘The PyMOL User’s Manual,’’ DeLano
Scientific, San Carlos (2001).
15) Inoue, K., Widada, J., Nakai, S., Endoh, T., Urata, M.,
Ashikawa, Y., Shintani, M., Saiki, Y., Yoshida, T.,
Habe, H., Omori, T., and Nojiri, H., Divergent structures
of carbazole degradative car operon isolated from Gram-
negative bacteria. Biosci. Biotechnol. Biochem., 68,
1467–1480 (2004).
16) Grifoll, M., Selifonov, S. A., Gatlin, C. V., and
Chapman, P. J., Actions of a versatile fluorene-degrading
bacterial isolate on polycyclic aromatic compounds.
Appl. Environ. Microbiol., 61, 3711–3723 (1995).
17) Kasuga, K., Habe, H., Chung, J.-S., Yoshida, T., Nojiri,
H., Yamane, H., and Omori, T., Isolation and character-
ization of the genes encoding a novel oxygenase
component of angular dioxygenase from Gram-positive
dibenzofuran-degrader Terrabacter sp. strain DBF63.
Biochem. Biophys. Res. Commun., 283, 195–204 (2001).
18) Bradford, M. M., A rapid and sensitive method for the
quantitation of microgram quantities of protein utilizing
the principle of protein dye binding. Anal. Biochem., 72,
248–254 (1976).
21) Hegg, E. L., and Que, L., Jr., The 2-His-1-carboxylate
facial triad: an emerging structural motif in mononuclear
non-heme iron(II) enzymes. Eur. J. Biochem., 250, 625–
629 (1997).
22) Parales, R. E., Lee, K., Resnick, S. M., Jiang, H.,
Lessner, D. J., and Gibson, D. T., Substrate specificity of
naphthalene dioxygenase: effect of specific amino acids
at the active site of the enzyme. J. Bacteriol., 182, 1641–
1649 (2000).
23) Kauppi, B., Lee, K., Carredano, E., Parales, R. E.,
Gibson, D. T., Eklund, H., and Ramaswamy, S.,
Structure of an aromatic-ring-hydroxylating dioxyge-
nase-naphthalene 1,2-dioxygenase. Structure, 6, 571–
586 (1998).
24) Parales, R. E., Resnick, S. M., Yu, C.-L., Boyd, D. R.,
Sharma, N. D., and Gibson, D. T., Regioselectivity and
enantioselectivity of naphthalene dioxygenase during
arene cis-dihydroxylation: control by phenylalanine 352
in the alpha subunit. J. Bacteriol., 182, 5495–5504
(2000).
25) Lee, K.-S., Parales, J. V., Friemann, R., and Parales,
R. E., Active site residues controlling substrate specificity
in 2-nitrotoluene dioxygenase from Acidovorax sp. strain
JS42. J. Ind. Microbiol. Biotechnol., 32, 465–473 (2005).
26) Friemann, R., Ivkovic-Jensen, M. M., Lessner, D. J., Yu,
C.-L., Gibson, D. T., Parales, R. E., Eklund, H., and
Ramaswamy, S., Structural insight into the dioxygena-
tion of nitroarene compounds: the crystal structure of
nitrobenzene dioxygenase. J. Mol. Biol., 348, 1139–1151
(2005).
27) Ju, K.-S., and Parales, R. E., Control of substrate
specificity by active-site residues in nitrobenzene dioxy-
genase. Appl. Environ. Microbiol., 72, 1817–1824
(2006).
28) Suenaga, H., Goto, M., and Furukawa, K., Emergence of
multifunctional oxygenase activities by random priming
recombination. J. Biol. Chem., 276, 22500–22506 (2001).
29) Suenaga, H., Mitsuoka, M., Ura, Y., Watanabe, T., and
Furukawa, K., Directed evolution of biphenyl dioxyge-
nase: emergence of enhanced degradation capacity for
benzene, toluene, and alkylbenzenes. J. Bacteriol., 183,
5441–5444 (2001).
30) Suenaga, H., Watanabe, T., Sato, M., Ngadiman, and
Furukawa, K., Alteration of regiospecificity in biphenyl
dioxygenase by active-site engineering. J. Bacteriol.,
184, 3682–3688 (2002).
31) Furusawa, Y., Nagarajan, V., Tanokura, M., Masai, E.,
Fukuda, M., and Senda, T., Crystal structure of the
terminal oxygenase component of biphenyl dioxygenase
derived from Rhodococcus sp. strain RHA1. J. Mol.
Biol., 342, 1041–1052 (2004).
32) Ferraro, D. J., Gakhar, L., and Ramaswamy, S., Rieske
business: structure-function of Rieske non-heme oxy-
genases. Biochem. Biophys. Res. Commun., 338, 175–
190 (2005).
33) Ferraro, D. J., Okerlund, A. L., Mowers, J. C., and
Ramaswamy, S., Structural basis for regioselectivity and
stereoselectivity of product formation by naphthalene
1,2-dioxygenase. J. Bacteriol., 188, 6986–6994 (2006).
19) Ferraro, D. J., Brown, E. N., Yu, C.-L., Parales, R. E.,
Gibson, D. T., and Ramaswamy, S., Structural inves-
tigations of the ferredoxin and terminal oxygenase
components of the biphenyl 2,3-dioxygenase from
Sphingomonas yanoikuyae B1. BMC Struct. Biol., 7,
10 (2007).