Báo cáo khoa học: Thermodynamic and kinetic analysis of the isolated FAD domain of rat neuronal nitric oxide synthase altered in the region of the FAD shielding residue Phe1395 docx

Thermodynamic and kinetic analysis of the isolated FAD domain
of rat neuronal nitric oxide synthase altered in the region
of the FAD shielding residue Phe1395
Adrian J. Dunford, Ker R. Marshall, Andrew W. Munro and Nigel S. Scrutton
Department of Biochemistry, University of Leicester, UK
In rat neuronal nitric oxide synthase, Phe1395 is positioned
over the FAD isoalloxazine ring. This is replaced by Trp676
in human cytochrome P450 reductase, a tryptophan in
related diflavin reductases (e.g. methionine synthase reduc-
tase and novel reductase 1), and tyrosine in plant ferredoxin-
NADP
+
reductase. Trp676 in human cytochrome P450
reductase is conformationally mobile, and plays a key role in
enzyme reduction. Mutagenesis of Trp676 to alanine results
in a functional NADH-dependent reductase. Herein, we
describe studies of rat neuronal nitric oxide synthase FAD
domains, in which the aromatic shielding residue Phe1395 is
replaced by tryptophan, alanine and serine. In steady-state
assays the F1395A and F1395S domains have a greater
preference for NADH compared with F1395W and wild-
type. Stopped-flow studies indicate flavin reduction by
NADH is significantly faster with F1395S and F1395A
domains, suggesting that this contributes to altered prefer-
ence in coenzyme specificity. Unlike cytochrome P450
reductase, the switch in coenzyme specificity is not attributed
to differential binding of NADPH and NADH, but prob-
ably results from improved geometry for hydride transfer in
the F1395S– and F1395A–NADH complexes. Potentio-
metry indicates that the substitutions do not significantly
perturb thermodynamic properties of the FAD, although

considerable changes in electronic absorption properties are
observed in oxidized F1395A and F1395S, consistent with
changes in hydrophobicity of the flavin environment. In
wild-type and F1395W FAD domains, prolonged incuba-
tion with NADPH results in development of the neutral blue
semiquinone FAD species. This reaction is suppressed in
the mutant FAD domains lacking the shielding aromatic
residue.
Keywords: coenzyme specificity; cytochrome P450 reduc-
tase; electron transfer; nitric oxide synthase; redox
potential.
The nitric oxide synthases (NOS) catalyse the NADPH- and
oxygen-dependent conversion of
L
-arginine to
L
-citrulline
and nitric oxide (NO) [1–3]. They are dimeric flavohaem
enzymes and each monomer comprises a C-terminal diflavin
reductase domain and an N-terminal oxygenase domain
[4–7]. The reductase domain is related structurally and
functionally to cytochrome P450 reductase (CPR) [8,9],
methionine synthase reductase (MSR [10]); and the cancer-
associated protein NR1 [11]. The N-terminal oxygenase
domain of NOS contains one mole equivalent of haem and
possesses binding sites for
L
-arginine and (6R)-5,6,7,8-
tetrahydrobiopterin [12]. The reductase and oxygenase
domains are linked by a calmodulin (CaM) binding

sequence [8,13–15], and CaM acts by releasing an
NADPH-dependent conformational lock [16]. CaM bind-
ing has been proposed to enhance the rate of interflavin
electron transfer [17–19], although this remains a contro-
versial aspect of CaM regulation of electron transfer [20].
Enhanced steady-state rates of cytochrome c reduction by
NOS reductase by CaM binding are, in the main, attributed
to faster FMN to cytochrome c electron transfer rates in the
presence of CaM, through release of the NADPH-depend-
ent conformational lock [16]. Of the three NOS isoforms,
the inducible NOS isoform is expressed with CaM tightly
bound [14] and regulation of activity is primarily through
transcriptional processes; the activities of endothelial NOS
and neuronal NOS (nNOS) are regulated by CaM binding,
which in turn is controlled by intracellular calcium levels
[4,5,7] and is mediated by an autoinhibitory sequence in the
FMN domain [21].
NADPH is the preferred reducing coenzyme for nNOS
and the other NOS isoforms and this property is shared by
other members of the diflavin reductase family of enzymes,
including P450 BM3 [22], CPR [23], MSR [24] and NR1
[11]. The structure of the NOS FAD domain indicates that
Phe1395 stacks over the FAD isoalloxazine ring [25]. This
residue is equivalent to Trp677 in rat CPR which likewise
stacks over the FAD isoalloxazine ring [9]. On binding
NADPH, this residue must move to allow hydride transfer
from NADPH to FAD. That this residue is mobile has been
confirmed by stopped-flow kinetic analysis of FAD reduc-
tion with wild-type human CPR and the W676H mutant
(equivalent to W677 in rat CPR) [26,27]. W676 facilitates

Correspondence to A. W. Munro and N. S. Scrutton, Department of
Biochemistry, University of Leicester, University Road, Leicester,
LE1 7HR, UK. Fax: + 44 116252 3369, Tel.: + 44 116223 1337 or
+ 44 116252 3464, E-mail: or
Abbreviations: NOS, nitric oxide synthase; CPR, cytochrome P450
reductase; MSR, methionine synthase reductase; CaM, calmodulin;
BM3, Bacillus megaterium flavocytochrome P450.
(Received 25 February 2004, revised 22 April 2004,
accepted 26 April 2004)
Eur. J. Biochem. 271, 2548–2560 (2004) Ó FEBS 2004 doi:10.1111/j.1432-1033.2004.04185.x
release of NADP
+
following hydride transfer, a process
that is impaired in the W676H mutant enzyme [27].
Moreover, the coenzyme specificity of human CPR is
switched in favour of NADH in the W676A enzyme,
indicating a role for W676 in coenzyme discrimination [28].
A similar switch in coenzyme specificity has been reported
for the F1395S mutant of nNOS [29] suggesting that
Phe1395 in nNOS plays a role similar to that of W676 in
human CPR as regards coenzyme binding. Similar studies
on Y308 mutants of the pea choloroplast ferredoxin
reductase enzyme indicated large changes in pyridine
nucleotide selectivity towards NADH in the Y308G and
Y308S mutants. The absence of the tyrosine was conjec-
tured to stabilize interaction with the nicotinamide group
common to both NADH and NADPH [30].
In rat nNOS, residue Phe1395 is located close to the
re face of the FAD isoalloxazine ring [25], and is structurally
equivalent to Trp677 in rat CPR (Fig. 1A) [9,47]. An

aromatic residue is found in all sequence-related mamma-
lian diflavin reductases sequenced to date, and also in plant
ferredoxin reductases that are structurally similar to the
FAD domains of NOS and CPR (Fig. 1B). In an attempt to
understand in detail the role of Phe1395 in the catalytic
mechanism of nNOS, we have isolated wild-type and
mutant forms of the nNOS FAD domain in which Phe1395
was exchanged for a serine (F1395S), alanine (F1395A) and
tryptophan (F1395W) residue. The value of studying
the thermodynamic and kinetic properties of individual
domains of complex multidomain enzymes has been
demonstrated in previous work with P450 BM3 [31], rat
nNOS [32], MSR [33,34], human NR1 [35] and human CPR
[26,36]. Clear evidence for the multidomain nature of these
enzymes has been obtained through the stable expression of
domains that bind their cognate cofactors and retain
catalytic properties typical of the parental diflavin reduc-
tases and flavocytochromes (e.g. [35,37]). In particular,
domain dissection has facilitated precise determination of
the midpoint reduction potentials of flavin cofactors, and
also of the haems in the case of NOS and P450 BM3. In
spectroelectrochemical titrations of the isolated domains,
the lack of overlapping spectral contributions from other
cofactors present in the full-length enzymes has enabled:
(a) deconvolution of the contributions of individual flavin
cofactors to the overall absorption changes observed in the
intact enzymes; (b) determination of the relative tendencies
of individual flavin cofactors to stabilize semiquinone
intermediates; and (c) precise determination of the reduction
potentials of the one- and two-electron redox couples

associated with each flavin (e.g. [32,33,36,38]). Studies of
individual domains have subsequently assisted in assign-
ment and determination of mid-point reduction potentials
for each redox couple in full-length enzymes.
Stuehr and coworkers have reported that exchange of
Phe1395 for serine in full-length rat NOS improves activity
with NADH. They propose that Phe1395 forms part of
a conformational ÔtriggerÕ mechanism that positively or
Fig. 1. The coenzyme-binding site in nNOS
FAD domain and sequence alignments around
the conserved aromatic residue in this domain.
(A) The nicotinamide-binding site of nNOS
FAD domain showing the position of
Phe1395 in relation to the structurally equiv-
alentW677inratCPR.TheFADofCPR
(PDB code 1AMO) and rat NOS (PDB code
1F2O) are shown in yellow. NADP
+
in the
ÔoffÕ conformation (PDB code 1JA1) is shown
in blue, and in the ÔonÕ conformation (PDB
code 1J9Z) in green (see [48] for details).
Phe1395 (rat nNOS) is shown in purple;
Trp677 (rat CPR) is shown in pink. (B)
Alignment of sequences for mammalian di-
flavin reductases and plant ferredoxin reduc-
tase in the region of the conserved aromatic
residues that shield the FAD isoalloxazine
ring. nNOS, rat neuronal nitric oxide syn-
thase; eNOS, rat endothelial nitric oxide syn-

thase; iNOS, rat inducible nitric oxide
synthase; CPR, human cytochrome P450 re-
ductase; NR1, human novel oxidoreductase 1;
MSR, human methionine synthase reductase;
BM3, Bacillus megaterium flavocytochrome
P450 BM3; FNR, spinach ferredoxin
NADP
+
reductase. The relevant flavin
shielding aromatic residue is underlined in
bold text in all cases.
Ó FEBS 2004 Kinetic and thermodynamic properties of nNOS FAD domains (Eur. J. Biochem. 271) 2549
negatively regulates NO synthesis depending on whether
CaM is bound [29]. In this paper, we have extended the
studies of Stuehr and coworkers by investigating the effects
of exchanging Phe1395 with serine, alanine and tryptophan
in the isolated FAD domain of rat nNOS (F1395W, F1395S
and F1395A) and by probing the thermodynamic and
kinetic consequences of these mutations. We have studied
the thermodynamic and kinetic properties of the isolated
FAD domain of rat nNOS and three mutant forms. The
isolated FAD domains have enabled: (a) potentiometric
analysis of the wild-type and mutant proteins to probe any
thermodynamic consequences of mutation; and (b) studies
of the kinetics of flavin reduction by reducing coenzymes in
the absence of spectral change arising from electron transfer
to the FMN domain. Clear differences in steady-state
kinetic properties are observed in the mutants, along with a
considerable shift in pyridine nucleotide specificity towards
NADH for the F1395S and F1395A proteins. However,

steady-state kinetic differences are not attributed to gross
changes in the flavin redox potentials, although effects on
the kinetics of FAD reduction are observed. Our data are
discussed in the light of results from mutagenic studies of
related enzymes (particularly CPR), and indicate that there
aresubtledifferencesintherolesofthestackingaromatic
residues in the different diflavin reductase enzymes and in
how they regulate pyridine nucleotide coenzyme specificity
and enzymatic properties.
Experimental procedures
Cloning of rat nNOS FAD/NADPH domain
The rat nNOS FAD/NADPH domain, amino acid residues
987–1463, was amplified from plasmid pCRNNR [39]
comprising a pKK223-3 clone of the rat nNOS reductase
domain. PCR amplification was performed using Pfu Turbo
DNA polymerase (Stratagene) and the forward primer
5¢-GCAATCATATGAGCTGGAAGAGGAACAAGTT
CCG-3¢ and the reverse primer 5¢-GGATCCTTAGGA
GCTGAAAACCTCATCTGCG-3¢, containing NdeIand
BamHI restriction sites, respectively. The resultant fragment
was gel-purified (QIAquick gel extraction kit, Qiagen) and
then A-tailed using Taq DNA polymerase prior to being
cloned into pGEM-T Easy (Promega). Clones were verified
by automated DNA sequencing prior to being subcloned
into NdeI- and BamHI-cut expression vector, pET11a.
Site-directed mutagenesis of rat nNOS FAD/NADPH
domain
Residue F1395 of the rat nNOS FAD/NADPH domain
was mutated to either A1395, S1395 or W1395 using the
nonstrand-displacing DNA polymerase Pfu Turbo and the

following mutagenic primer combinations: F1395A, for-
ward primer 5¢-CACGAG
GATATCGCTGGAGTCAC
CCTC-3¢ and the reverse complement thereof; F1395S,
forward primer 5¢-CACGAG
GATATCTCTGGAGTCA
CCCTCAG-3¢ and its reverse complement; F1395W,
5¢-CCGGTACCACGAG
GATATCTGGGGAG-3¢ toge-
ther with the reverse complementary primer. All primers
incorporated silent mutations to introduce an EcoRV
restriction site (underlined) to assist in mutant screening.
Mutated bases are given in bold type. Cycling parameters
for mutagenesis reactions were 95 °Cfor30sfollowedby
16 cycles of 95 °C for 30 s, 55 °Cfor1minand68°Cfor
9 min. Nonmutated template DNA was then removed by
DpnI digestion and mutant DNA transformed into Escheri-
chia coli JM109. Selected clones were first assessed by
EcoRV digestion and then verified by automated DNA
sequencing.
Purification of the isolated FAD domains
Transformed cells were grown in Terrific Broth [40].
Expression of the isolated FAD-domains was induced by
addition of isopropyl thio-b-
D
-galactoside (1 m
M
)ata
culture optical density of 0.8 at 600 nm; cells were grown for
afurther24hat30°C. Harvested cells were resuspended in

lysisbuffer[50mL;50m
M
Tris/HCl pH 7.4 containing
10% (v/v) glycerol, 1 m
M
CaCl
2
and a Complete
TM
EDTA-
free protease inhibitor tablet (Roche)]. Cells were disrupted
by sonication, the cell extract clarified by centrifugation
(15 000 g, 50 min) and fractionated with ammonium sulfate
(FAD domain was recovered in the 30–50% saturation
fraction). Enzyme was dialysed exhaustively against lysis
buffer, and applied to an anion exchange resin (DE-52)
previously equilibrated with lysis buffer. The column was
washed with lysis buffer (500 mL) and FAD domain
was recovered by developing the column with a gradient
(0–0.5
M
) of KCl. Fractions containing FAD domain were
pooled, and applied to an affinity resin (2¢5¢-ADP Seph-
arose) equilibrated with lysis buffer containing 100 m
M
NaCl. After washing ( 250 mL lysis buffer 100 m
M
NaCl
and then  250 mL lysis buffer, 250 m
M

NaCl), FAD
domain was recovered by the application of lysis buffer
containing 500 m
M
NaCl. Enzyme was dialysed exhaust-
ively against lysis buffer and stored at )20 °Cinthe
presence of 20% (v/v) glycerol.
Potentiometry
Redox titrations for the nNOS FAD domains (wild-type,
F1395A, F1395S and F1395W) were performed in a Belle
Technology glove box under a nitrogen atmosphere,
essentially as described previously [36]. All solutions were
degassed under vacuum with argon. Oxygen levels were
maintained at < 2 p.p.m. The protein solution [ 50 l
M
in
5mL100m
M
potassium phosphate pH 7.0 in the presence
and absence of 10% (v/v) glycerol] was titrated electro-
chemically according to the method of Dutton [41] using
sodium dithionite as reductant and potassium ferricyanide
as oxidant. Mediators (2 l
M
phenazine methosulfate, 5 l
M
2-hydroxy-1, 4-naphthoquinone, 0.5 l
M
methyl viologen,
and 1 l

M
benzyl viologen) were included to mediate in the
range between +100 and )480 mV as described previously
[35,41]. At least 15 min was allowed to elapse between each
addition of reductant/oxidant to allow stabilization of the
electrode. Spectra (300–800 nm) were recorded using a Cary
UV-50 Bio UV-Visible scanning spectrophotometer, using a
fibre optic probe immersed in the protein solutions and
connected externally to the spectrophotometer. The electro-
chemical potential of the solution was measured using a
Hanna pH 211 meter coupled to a Pt/Calomel electrode
(Thermo Russell Ltd) at 25 °C. The electrode was calibrated
2550 A. J. Dunford et al. (Eur. J. Biochem. 271) Ó FEBS 2004
using the Fe
3+
/Fe
2+
EDTA couple as a standard
(+108 mV). A factor of +244 mV was used to correct
relative to the standard hydrogen electrode. Spectral data
were imported into
ORIGIN
(Microcal) and spectral subtrac-
tions performed to correct for baseline drift during the
titrations (bringing absorption back to zero at 800 nm,
where there is no significant absorption from the cofactor in
oxidized or reduced states). Spectral data were fitted to
appropriate Nernst functions in
ORIGIN
to derive the

relevant midpoint reduction potentials of the flavins, as
described previously [36].
Stopped-flow kinetic measurements
Stopped-flow studies were performed using an Applied
Photophysics SX.18 MX stopped-flow spectrophotometer
contained within an anaerobic glove box (Belle Technol-
ogy). Measurements were carried out at 25 °Cin50m
M
Tris/HCl pH 7.4 containing 10% (v/v) glycerol. Protein
concentration was 5 l
M
(reaction cell concentration) for
single wavelength work and 10 l
M
for photodiode array
studies. All buffers were made oxygen-free by evacuation
and extensive bubbling with argon before use. Buffers were
then placed in the glove box overnight before use. Prior to
stopped-flow studies, protein samples were treated with
potassium hexacyanoferrate, and excess cyanoferrate was
removed by rapid gel filtration [Sephadex G25 equilibrated
in 50 m
M
Tris/HCl pH 7.4, 10% (v/v) glycerol].
Stopped-flow, multiple-wavelength absorption studies
were carried out using a photodiode array detector and
X
-
SCAN
software (Applied Photophysics Ltd). Spectral

deconvolution was performed by global analysis and
numerical integration methods using
PROKIN
software
(Applied Photophysics Ltd). In single wavelength studies,
flavin reduction by NADPH was observed at 458 nm.
Stopped-flow fluorescence experiments used excitation
wavelengths of 340 nm (NADPH) and 295 nm (trypto-
phan). Emission bands were selected using the appropriate
band pass filter.
Steady-state enzyme assays
The steady-state activities of wild-type and mutant enzymes
using ferricyanide as electron acceptor were determined
using a Jasco V-550 UV/visible double-beam spectro-
photometer. Assays were performed in the double-beam
spectrophotometer to take account of any nonenzyme-
mediated reduction of electron acceptor. The reference
cuvette contained the same mix as the sample cuvette, but
thesamevolumeofbufferreplacedtheenzyme.The
reaction was initiated by the simultaneous addition of
NAD(P)H to both cuvettes.
Potassium ferricyanide reduction was monitored at
420 nm (De
420nm (red-ox)
¼ 1020
M
)1
Æcm
)1
). Reactions were

performed in 50 m
M
Tris/HCl pH 7.4, 10% glycerol at
25 °C. With ferricyanide as exogenous electron acceptor,
saturating concentrations of NADPH (500 l
M
)wereused
to determine the K
m
for the substrate. The K
m
for
NAD(P)H was determined at a fixed, saturating concen-
tration of ferricyanide (2 m
M
).
ORIGIN
software (Microcal)
wasusedindatafittingtoderiveK
m
and k
cat
values from
steady-state assays.
Results
Domain isolation and mutagenesis of Phe1395
Each of the F1395S/A/W FAD domains were purified in
high yield (typically 15 mg from 1 L recombinant cells) and
in pure form as judged by SDS/PAGE. The spectral
properties of each are shown in Fig. 2. Major alterations in

electronic absorption spectra are apparent for the F1395A
and F1395S FAD domains. In these two mutants, the
shorter wavelength absorption band of the FAD is inten-
sified and blue-shifted with respect to the wild-type. The
F1395W spectrum is virtually indistinguishable from that
for wild-type, suggesting a conservative effect of the
aromatic replacement on the flavin electronic properties.
Flavin spectral maxima are located at 457 nm and 398 nm
for wild-type and F1395W mutants, with a pronounced
shoulder on the longer wavelength band at  480 nm.
Similarly, the electronic absorption spectra of F1395A and
F1395S FAD domains are strongly similar to one another,
but distinct from those of wild-type and F1395W proteins.
Absorption maxima are at 456.5 nm and 387.5 nm for
F1395A/S. There is a marked increase in the relative
intensity of the shorter wavelength band in the F1395A/S
mutants, and the shoulder on the longer wavelength band is
also much less pronounced than in wild-type and F1395W
(Fig. 2). The spectral perturbations observed as a conse-
quence of the nonaromatic amino acid substitutions in
F1395A/S are consistent with a less hydrophobic environ-
ment for the FAD isoalloxazine ring compared to wild-type
and F1395W.
Steady-state kinetic analysis of wild-type and mutant
FAD domains
Steady-state analysis of nNOS FAD domain-dependent
ferricyanide reduction indicates that there are only moderate
effects on K
m
for NADPH induced by the replacement of

Fig. 2. UV-Visible absorption spectra for wild-type and mutant forms of
nNOS FAD domain (all  7 l
M
). Protein samples were contained in
50 m
M
Tris/HCl buffer, 10% (v/v) glycerol, pH 7.4, at 25 °C. Spectra
are F1395S (dashed line), F1395A (dotted line), WT (solid, thick line)
and F1395W (solid, thin line).
Ó FEBS 2004 Kinetic and thermodynamic properties of nNOS FAD domains (Eur. J. Biochem. 271) 2551
the FAD-stacking phenylalanine with either aromatic
(tryptophan) or nonaromatic (alanine, serine) sidechains
(Table 1). There is an apparent small decrease in K
M
for
NADPH in the F1395W mutant (K
m
¼ 15.4 l
M
cf 28.2 l
M
for wild-type) and a larger increase for the F1395A mutant
(K
m
¼ 83.5 l
M
). The value for the F1395S mutant is within
error of that for the wild-type nNOS FAD domain
(Table 1). There are also effects on the k
cat

values for
ferricyanide reduction following mutation, with diminution
in k
cat
for the F1395W and F1395S mutants compared to
wild-type (70% and 31%, respectively, of wild-type k
cat
);
interestingly, the k
cat
for F1395A is increased by 30% over
wild-type. The net effects on the second order rate constant
(k
cat
/K
m
) describing the overall efficiency of NADPH-
dependent ferricyanide reduction is that the F1395W
mutant shows a modest improvement (30%) over wild-
type, while the nonaromatic substituted mutants are
decreased to 44% (F1395A) and 39% (F1395S) of the
wild-type value (Table 1).
Much more marked effects are seen in NADH-dependent
catalysis. The apparent K
m
values are lower in all mutants
than in wild-type, with F1395S showing the greatest
improvement (K
m
¼ 1830 l

M
compared with 5890 l
M
for
wild-type). While the F1395W mutant has a diminished k
cat
value (78%) compared to wild-type, both of the F1395A/S
mutants show considerable improvements in k
cat
values
(5.9-fold and 2.6-fold, respectively). This leads to even
greater improvements in the k
cat
/K
m
ratio for the F1395A/S
mutants over wild-type (10.7/8.4-fold), with a minor
improvement also observed for the F1395W mutant (1.3-
fold) (Table 1). The large increases in catalytic efficiency
of the nonaromatic substituted mutants mirror the results
observed in earlier studies with the human CPR [28]. The
increases in efficiency of ferricyanide reduction are also
consistent with the results of Stuehr and coworkers in their
analysis of the effects of the F1395S mutation in full-length
CaM-bound nNOS [29]. For full-length F1395S nNOS, a
 30-fold increase over wild-type was obtained for the the
apparent k
cat
for NADH-dependent ferricyanide reduction.
The rather smaller enhancement of k

cat
reported here for the
isolated FAD domain F1395S mutant may represent a
more accurate representation of the effect of the mutation
on electron transfer from NADH through to ferricyanide,
since effects of CaM (particularly in view of the interplay
between the F1395 sidechain and CaM indicated by the
studies of Adak et al.[29])andinterflavinelectrontransfer
on the steady-state kinetics can be ruled out. Steady-state
assays were repeated using potassium phosphate (50 m
M
,
pH 7.0) instead of Tris/HCl. Results were very similar to
those obtained in Tris/HCl for wild-type and mutant FAD
domains, indicating that the presence of phosphate ions
(that potentially could occupy the 2¢-phosphate binding site
for the coenzyme) does not affect catalysis in nNOS FAD
domain.
The relative catalytic efficiencies with NADPH and
NADH as electron donors (k
cat
/K
m[NADH]
/k
cat
/K
m[NADPH]
)
for the nNOS FAD domains are shown in the final column
of Table 1 (NADPH/NADH) and detail the extent of the

ÔswitchÕ in specificity for the two pyridine nucleotide
coenzymes. These data reveal that there is negligible
alteration in the relative selectivity between wild-type and
the F1395W mutant. These data give confidence that the
F1395W mutant can be used as a wild-type mimic in
mechanistic studies to observe changes in the environment
of the aromatic stacking residue close to the FAD [this
residue must move to facilitate docking of NAD(P)H in its
catalytically relevant conformation, and for hydride transfer
to occur]. A similar approach exploiting changes in
tryptophan fluorescence has been used to characterize
conformational events in human CPR [26,27]. Although
F1395W is virtually identical to wild-type as regards
comparative efficiencies with NADH/NADPH, the
F1395A/S mutants show large changes in favour of NADH
(Table 1). The relative efficiencies (k
cat
/K
m[NADH]
/k
cat
/
K
m[NADPH]
) are changed > 24-fold for the F1395A variant,
and > 21.5-fold for the F1395S mutant. Clearly, the
replacement of F1395 with nonaromatic residues has a
major effect on the ability of the nNOS FAD domain to
discriminate against NADH, and this leads to large
improvements in efficiency of F1395A/S mutants in cata-

lytic turnover with NADH. The steady-state kinetic data
indicate that small overall changes in the K
m
values for
NADH contribute partially to the specificity switch towards
NADH, but that the major effect is on the k
cat
parameter
(Table 1).
A further interesting observation from these kinetic
data is that the replacement of the aromatic stacking
residue does not lead to enhanced binding of NADPH,
contrary to observations made with the related pea
ferredoxin reductase and human CPR enzymes [28,30].
Clearly this reflects differences in structural features of the
NAD(P)H binding site in NOS. This is maybe not
surprising in view of the known differences by which
electron transfer is regulated in NOS isoforms compared
Table 1. Steady-state kinetic parameters for ferricyanide reduction by wild-type and mutant forms of the nNOS FAD domain. Kinetic parameters
were determined using both NADPH and NADH as electron donors. The K
m
and k
cat
values for ferricyanide (Fe[(CN)
6
]
3–
) were determined at a
fixed and saturating concentration of NADPH (500 l
M

). The K
m
and k
cat
values for NADPH and NADH were determined at a fixed and
saturating concentration of ferricyanide (2 m
M
). All reactions were performed in 50 m
M
Tris/HCl,10%glycerol,pH7.4,at25°C.
nNOS
NADPH NADH
K
m
(l
M
) k
cat
(s
)1
)
k
cat
/K
m
(l
M
)1
Æs
)1

) K
m
(l
M
) k
cat
(s
)1
)
k
cat
/K
m
(l
M
)1
Æs
)1
)
k
cat
/K
NADPH
m
:
k
cat
/K
NADH
m

WT 28.2 ± 4. 5 161.6 ± 7.2 5.7 5890 ± 270 44.1 ± 2.7 0.0075 760
F1395W 15.4 ± 4.5 113.5 ± 10.1 7.4 3550 ± 480 34.6 ± 1.0 0.0097 763
F1395A 83.5 ± 11.9 209.8 ± 6.6 2.5 3250 ± 830 259.8 ± 26.7 0.0799 31.3
F1395S 22.9 ± 4.2 50.8 ± 2.4 2.2 1830 ± 160 114.8 ± 2.6 0.0627 35.1
2552 A. J. Dunford et al. (Eur. J. Biochem. 271) Ó FEBS 2004
with other ferredoxin reductase and diflavin reductase
enzymes.
Thermodynamic basis for electron transfer
Anaerobic spectroelectrochemical methods were used to
determine the reduction potentials for the FAD flavin in
wild-type and all three F1395 mutants. In all cases, the
development of a spectral signal typical of a neutral, blue
semiquinone was observed during the course of the redox
titration, with absorption maximum close to 600 nm. As in
previous studies of the redox properties of human CPR,
isosbestic points were observed in the titrations, at
 500 nm (oxidized-to-semiquinone transition) and at
 411 nm (semiquinone-to-hydroquinone transition) [36].
Plots of the change in absorption vs. reduction potential
were made at both 475 nm and 600 nm in order to
determine the midpoint reduction potentials for the flavin
transitions. Data are presented in Table 2. The midpoint
potentials for the oxidized/semiquinone (E
1
) and semiqui-
none/hydroquinone (E
2
) couples in the mutants are not
grossly altered from wild-type values. No hysteresis was
observed in any of the titrations, and similar spectra were

obtained at the same potentials in both reductive and
oxidative directions. The E
1
values for the F1395S/A
mutants ()156 mV/)157 mV) are slightly more positive
than those of wild-type and F1395W enzymes ()177 mV
and )167 mV, respectively), but the overall changes in the
thermodynamic properties of the flavins are rather minimal,
as judged by the close proximity of the midpoint potentials
for the two-electron couples determined from the A
475
data
(Table 2). Replacement of an aromatic residue for the
aliphatic serine/alanine thus has relatively little effect on the
potentials for the FAD cofactor, and does not destabilize
the semiquinone to any significant degree. However, clear
differences in spectral properties are observed between the
aromatic (wild-type and F1395W) and nonaromatic
(F1395A/S) mutants. As discussed above, there are signi-
ficant differences in the spectra for the oxidized forms of the
enzymes (Fig. 2), probably resulting from differences in the
environment of the FAD cofactor in the different proteins.
These differences might also be manifest in the semiquinone
forms, as the relative intensities of the semiquinone signal in
F1395A/S diminished compared with those observed for
wild-type and F1395W (Fig. 3). This might reflect a change
in the absorption characteristics (i.e. extinction coefficient)
of the semiquinone species in the nonaromatic mutants.
However, small changes in the separations between the
E

1
and E
2
couples could also give rise to changes in the
amounts of semiquinone signal detected during redox
titration. This aspect is under further investigation. The
semiquinone formation constants (K values) were deter-
mined as described previously [42,43]. These yielded values
of 177.5 for wild-type and F1395W mutants, 170.8 for
F1395A, and 344.2 for F1395S. These values reinforce the
assertion that the neutral blue semiquinone in wild-type and
mutant enzymes is strongly stabilized. The proteins all
showed some tendency to aggregate at lower potentials, as
has been observed with the FAD domain of P450 BM3 [38],
and this caused some problems in obtaining good quality
data for the F1395A/S mutants at potentials < )350 mV,
due to the rather small changes in absorption in the
semiquinone absorption coupled to the development of
some turbidity in the solutions. Notwithstanding these
problems, near-identical values were obtained from dupli-
cated redox titrations in all cases, and values derived from
fitting at two different wavelengths produced consistent
results. Thus, despite qualitative differences in the absorp-
tion properties of the flavins in these mutants, there are
relatively small changes in the reduction potentials. Conse-
quently, alterations in the kinetic properties of the wild-type
and mutant FAD domains can not simply be explained in
terms of large-scale changes in the potentiometric properties
of their cofactors.
Stopped-flow kinetic analysis of electron transfer

Reduction of the wild-type and mutant FAD domains was
investigated by stopped-flow methods using both a photo-
diode array detector and single wavelength detection. The
spectral changes accompanying flavin reduction following
rapid mixing with NADPH are shown for the wild-type and
F1395S FAD domains in Fig. 4. The spectral changes for
the wild-type enzyme revealed a rapid bleaching of flavin
absorption in the early time domain (Fig. 4A) consistent
with flavin reduction. These spectral changes were followed
by the development of long wavelength signature over an
Table 2. Reduction potentials for the FAD cofactor in wild-type nNOS
FAD domain, the F1395A/S/W mutants and related diflavin reductase
FAD domains. Reduction potentials for the oxidized/semiquinone (E
1
),
semiquinone/hydroquinone (E
2
) and oxidized/hydroquinone couples
of the wild-type and mutant forms of nNOS reductase FAD domain
are shown. Experiments and data fitting were performed as described
in Experimental Procedures. E
1
and E
2
values were determined by
fitting A
600
(near semiquinone absorption maximum) data to a two-
electron Nernst function, as described [36,38]. The E
12

value was de-
rived by fitting the A
475
(near oxidized flavin absorption maximum)
data to the Nernst equation. SHE, Standard hydrogen electrode.
FAD domain
Reduction potential (vs. SHE)
A
600
A
475
E
1
E
2
E
12
nNOS wild-type )177 ± 5 )310 ± 8 )229 ± 5
nNOS F1395A )157 ± 4 )289 ± 6 )227 ± 4
nNOS F1395S )156 ± 4 )306 ± 9 )237 ± 5
nNOS F1395W )167 ± 4 )300 ± 6 )222 ± 3
Intact nNOS
a
)250 )260 –
Human CPR )286 ± 6
b
)371 ± 7
b
)329 ± 7
b

Human MSR )222 ± 4 )288 ± 7 )272 ± 8
c
Human NR1 )315 ± 5
d
)365 ± 5
d
)340 ± 5
d
P450 BM3 )264 ± 3 )375 ± 3 )320 ± 4
e
a
Values for the FAD in intact rat nNOS were determined by si-
mulations at various wavelengths. No statistical errors are reported
on these data [44];
b
E
1
,E
2
and E
12
for human CPR FAD domain
were determined similarly at 583 nm and 474 nm [36];
c
E
12
value
for human MSR FAD comes from A
450
analysis for the intact

reductase [33];
d
E
1
and E
2
data for human NR1 are from data
fitting at 585 nm, with the E
12
value being the midpoint of these E
1
and E
2
values [35]; and that for the FAD domain of P450 BM3;
e
E
12
value cited for the FAD domain of P450 BM3 is the midpoint
of the E
1
and E
2
values determined at 600 nm.
Ó FEBS 2004 Kinetic and thermodynamic properties of nNOS FAD domains (Eur. J. Biochem. 271) 2553
extended time domain, indicating the formation of a blue
neutral semiquinone species (Fig. 4B). Qualitatively similar
data were obtained for the F1395W mutant FAD domain.
By contrast, reduction of the F1395S FAD domain (and
also the F1395A FAD domain; data not shown) is also
relatively rapid (Fig. 4C), but over an extended time base

the semiquinone signature is not developed (Fig. 4D).
Given the inferior signal-to-noise ratio of the photodiode
array detector compared with data acquisition at single
wavelengths using a photomultiplier, we also performed
single wavelength studies at 600 nm for the wild-type and
F1395S FAD domains. This confirmed that relatively small
absorption changes that might indicate formation of a blue
neutral semiquinone are not observed at 600 nm for the
F1395S (Fig. 4C,D, insets) and F1395A (data not shown)
domains. Flavin reduction in the wild-type (Fig. 4A, inset)
and F1395W (data not shown) domains is accompanied by
a rapid bleaching of absorption at 600 nm, which we
attribute to the loss of charge-transfer character in an
E–NADPH complex. The fast formation and decay of a
charge-transfer species at 600 nm in wild-type and F1395W
FAD domains is consistent with our previous assignment of
rapid 600 nm absorption changes in the isolated reductase
domain of rat nNOS (which were likewise attributed to
formation and decay of a charge-transfer species [20]).
Reactions over an extended time base illustrate the forma-
tion of the blue semiquinone species (Fig. 4B, inset). The
lack of any spectral change in the early time domain for the
F1395S and F1395A domains suggests that a spectrally
distinct charge-transfer species is not formed (Fig. 4C,
inset).
Observed reaction rates for flavin reduction were calcu-
lated by analysis of single wavelength reaction transients
recorded at 454 nm (Fig. 5). Reaction transients were
biphasic, and data were analysed using a standard double-
exponential function; the amplitude of the first phase was

found to contribute  70% of the total absorption change.
This phase was shown to represent hydride transfer from
NADPH to FAD through stopped-flow fluorescence ana-
lysis, since the loss of NADPH fluorescence accompanying
oxidation of the coenzyme on mixing wild-type domain with
NADPH occurs with similar kinetics to the fast kinetic
Fig. 3. Potentiometric analysis of wild-type
and F1395 mutant nNOS reductase FAD do-
mains. (A) Spectral changes observed during
the redox titration of wild-type FAD domain
( 45 l
M
). The most intense spectrum is that
for the oxidized enzyme. Reduction is associ-
ated with bleaching of the absorption in the
region of the two major absorption bands of
the flavin, while there is development, and
then decay, of absorption at longer wave-
length due to the formation of the semiqui-
none species, followed by its reduction to
hydroquinone. Arrows indicate the direction
of absorption in these regions of the spectrum
during reductive titration. The inset shows a fit
of the A
600
(semiquinone) data to a two-elec-
tron Nernst function, as described previously
[35,37,40]. (B) A similar set of spectra and the
relevant A
600

vs. reduction potential data fit
from the titration of the F1395S mutant
( 65 l
M
).
2554 A. J. Dunford et al. (Eur. J. Biochem. 271) Ó FEBS 2004
phase observed in absorption measurements Fig. 6A. The
fluorescence transient also has a second phase with kinetics
comparable to the slow phase seen in absorption mode,
suggesting further oxidation of NADPH. The mechanistic
origin of the second phase is uncertain, but one possibility is
that the slow phase might represent further reduction of the
flavin following release of NADP
+
, as the equilibrium
distribution of enzyme species adjusts to favour further
reduction of the enzyme. Similar arguments have been
advanced in studies of the isolated FAD domain of human
CPR [26]. Alternatively, different conformational states of
the FAD domain might also account for the biphasic nature
of the reaction transients and this would be consistent with
the observed NADH concentration dependence for the fast
and slow phases in flavin reduction using NADH as
reducing coenzyme (see below). Although the mechanistic
origin of the two phases remains uncertain, it is clear that
both phases report on flavin reduction by reducing coen-
zyme. The observed rates of flavin reduction (fast phase) as
a function of NADPH concentration are plotted in Fig. 6B.
As with the isolated diflavin reductase of rat nNOS [20],
these rates are independent of NADPH concentration in the

pseudo first-order regime. Kinetic constants for reactions
with NADPH are collated in Table 3.
We also performed stopped-flow studies of flavin reduc-
tion in wild-type and mutant FAD domains using NADH
as the reducing coenzyme. Reaction transients were again
biphasic (fast phase  70% and slow phase  30% of the
total amplitude change). Unlike with NADPH, observed
rates for each phase displayed a hyperbolic dependence on
NADH concentration. Derived kinetic constants for each
phase of the reaction transient are collated in Table 3. The
most striking result from these series of stopped-flow studies
is that the fast phase for the F1395A and F1395S FAD
domains is faster than the wild-type and F1395W FAD
domains (Table 3). Likewise, the slow phase of the kinetic
transient is markedly more rapid in the F1395A and F1395S
domains compared with the wild-type and F1395W
Fig. 4. Spectral changes observed during the reduction of wild-type and F1395S FAD domain on mixing with NADPH. Photodiode array data
collected were obtained for the reaction of wild-type (10 l
M
) and F1395S (10 l
M
) FAD domain mixed with NADPH (100 l
M
). Reactions were
performed in 50 m
M
Tris/HCl pH 7.4, 10% glycerol at 25 °C.Upperpanels:dataforwild-typeFADdomainrecordedfrom1.28msto1s(A)and
from 1.28 ms to 200 s (B). Lower panels: data for F1395S FAD domain recorded from 1.28 ms to 1 s (C) and from 1.28 ms to 200 s (D). Insets
show typical stopped-flow transients monitored at 600 nm for reactions of wild-type (5 l
M

, A and B) and F1395S (5 l
M
,CandD)FADdomains
with NADPH (100 l
M
). Transients were recorded over the same time periods as the respective spectral changes shown in the main panels.
Ó FEBS 2004 Kinetic and thermodynamic properties of nNOS FAD domains (Eur. J. Biochem. 271) 2555
domains, suggesting that the NMN ring of NADH is in a
more favourable geometry for hydride transfer to the FAD
following removal of the aromatic shielding residue.
With human CPR we have demonstrated that the FAD
shielding residue (Trp676) is conformationally mobile using
fluorescence stopped-flow methods [26,27]. Fluorescence
stopped-flow studies with the wild-type nNOS FAD
domain indicated essentially no change in tryptophan
emission on mixing with NADPH, consistent with the lack
of a tryptophan residue in the NADPH-binding site. Large
changes in tryptophan fluorescence emission were observed,
however, on mixing the F1395W FAD domain with
NADPH. Two kinetic phases were observed: the first
(increase in fluorescence,  200 s
)1
) occurs on a timescale
consistent with the kinetics of flavin reduction; the second
phase (decrease in fluorescence emission, 0.04 s
)1
) accom-
panies development of the flavin semiquinone observed in
stopped-flow absorption measurements (see Fig. 4 insets
and Table 3). Clearly, the environment of Trp1395 in the

mutant FAD domain is perturbed on reduction of the fla-
vin, and also on subsequent disproportionation of the
domain to yield the blue neutral semiquinone form.
Discussion
This study on the kinetic and thermodynamic features of the
wild-type and F1395 mutants of nNOS reductase indicates
an important role for the aromatic residue that shields FAD
in rat nNOS. However, mutation of this residue in rat
nNOS and the equivalent residue in other members of the
diflavin reductase family has revealed different functional
characteristics. Previous studies on both human CPR and
NOS revealed that the replacement of this aromatic residue
with nonaromatic substitutes influenced the specificity for
the reducing pyridine nucleotide coenzyme in favour of
NADH [28,29]. An  1000-fold switch in coenzyme speci-
ficity was achieved for the W676A mutant of human CPR
[28]. Also, in recent studies, we have demonstrated that a
similar switch in specificity occurs in another member of
the diflavin reductase family, flavocytochrome P450 BM3
(R. Neeli, O. Roitel, N. S. Scrutton and A. W. Munro,
unpublished data). Herein, we show that the switch in
specificity towards NADH occurs in the nNOS FAD
domain, in which the aromatic shielding residue has been
mutated to an aliphatic side chain, although the extent of
Fig. 5. Stopped-flow transients of wild-type and mutant FAD domains monitored at 454 nm. Kinetic transients show the reaction of each FAD
domain (5 l
M
)withNADPH(100l
M
). Reactions were performed in 50 m

M
Tris/HCl pH 7.4, 10% glycerol at 25 °C. In all cases transients were
fitted to a standard biphasic expression; fits are shown in A–D. (A) Wild-type domain. (B) F1395W. (C) F1395A. (D) F1395S.
2556 A. J. Dunford et al. (Eur. J. Biochem. 271) Ó FEBS 2004
the conversion is less marked than that observed for the full-
length nNOS F1395A mutant in previous studies [29]. An
 25-fold switch in specificity towards NADH is achieved
for the F1395A FAD domain [by comparing relative k
cat
/
K
m
values for NAD(P)H-dependent ferricyanide reduction]
(Table 1). Our reductionist approach in studying the
isolated FAD domain of nNOS has enabled us to analyse
in detail the effects of these mutations in the absence of
other influences (most notably those induced by CaM
binding and by interflavin electron transfer). This has clear
advantages over studies with the full-length NOS protein,
which are compromised by multiple spectroscopic signals
from additional cofactors and conformational effects
induced by CaM binding. Thus, for the first time we are
able to report the effects of these mutations on the
thermodynamic properties of the FAD and also their
consequences on the kinetics of FAD reduction.
The thermodynamic properties of the wild-type and
mutant nNOS FAD domains are compared with the
potentials for the FAD domains of other members of the
diflavin reductase class of enzymes in Table 2. While there
are clearly distinct differences in the absorption properties of

the FAD on removal of the aromatic shielding residue
(Fig. 2), there are only relatively small differences in the
thermodynamic properties of the wild-type and mutant
FAD domains (a maximum of 21 mV between E
1
couples,
21 mV between E
2
couples, 15 mV between E
12
couples).
Potentiometric studies suggest that differences in spectral
properties are also a feature of the semiquinone forms of the
FAD, with the decreased intensity of the semiquinone
observed in redox titrations of the F1395A/S mutants
possibly being attributed to an approximately twofold
change in the extinction coefficient for this species in these
mutant domains. When compared with the other diflavin
reductases, there are very clear differences in the thermo-
dynamic properties of the nNOS FAD flavin. All of these
proteins have in common the ability to stabilize the FAD
blue semiquinone, but the wild-type and mutant nNOS
domains have a more positive potential for the oxidized/
semiquinone couple (E
1
) compared with other members of
the family. Specifically, for wild-type nNOS FAD domain,
the E
1
value is between 45 and 138 mV more positive than

E
1
for the other flavoproteins (Table 2). By contrast, the
potentials for the nNOS FAD domain semiquinone/
hydroquinone couples (E
2
) are very similar to those for
the other diflavin reductases. The net effect is that the
overall potential for the oxidized/hydroquinone couple (E
12
)
of wild-type nNOS FAD domain is 43–111 mV more
positive than any of the other diflavin reductase enzymes.
A further point to note from the potentiometric analysis
of the wild-type and mutant domains is that there is some
variance with the previously reported data for the potentials
of the wild-type nNOS FAD measured in the reductase
(diflavin) domain of the enzyme [32]. The data reported here
for the FAD domain are E
1
¼ )177±5mV; E
2
¼
)310 ± 8 mV; (E
12
¼ )229 ± 5 mV), whereas the values
reported from an absorption vs. potential fit to the
inherently more complex four-electron Nernst function
are E
1

¼ )232 ± 7 mV; E
2
¼ )280 ± 6 mV (E
12
¼
)256 ± 7 mV). In recent studies, Gao et al.haveused
even more complex simulations to derive estimates of the
flavin and haem reduction potentials in intact nNOS. While
Fig. 6. Reduction of wild-type FAD domain monitored by absorption
and fluorescence spectroscopy. The reduction of wild-type FAD
domain (5 l
M
) was monitored by absorption spectroscopy (454 nm)
and fluorescence emission spectroscopy (excitation 340 nm). Reactions
were performed in 50 m
M
Tris/HCl pH 7.4, 10% glycerol at 25 °C.
The reaction transient in absorption mode (A, trace a) is fitted to a
two-exponential equation [ 450 s
)1
( 70% of total amplitude) and
95 s
)1
( 30% amplitude), respectively]. Likewise, the reaction tran-
sient in fluorescence mode (A, trace b) is fitted to a two-exponential
equation [ 400 s
)1
( 60% of total amplitude) and 90 s
)1
( 40%

amplitude), respectively]. (B) Plot of the observed rate constant k
obs
for
the fast phase of flavin reduction as a function of NADPH concen-
tration for wild-type and mutant FAD domains. Conditions as for (A).
Theplotshowsthat,inallcases,therateisindependentofthe
concentration of NADPH used. Symbols: (j)WT;(d) F1429W;
(m) F1429A; (.) F1429S.
Ó FEBS 2004 Kinetic and thermodynamic properties of nNOS FAD domains (Eur. J. Biochem. 271) 2557
standard errors are not reported on these data, the results
are broadly similar to those reported previously for the
diflavin reductase form [32,44]. It is possible that the further
genetic dissection of the reductase to produce the isolated
FAD domain results in some perturbation of the redox
properties of the flavin. However, no such changes were
observed in similar domain dissection and potentiometric
studies of the P450 BM3 diflavin reductase [38], although a
small shift in potential was reported for the FAD domain in
studies of full-length MSR and its genetically excised FAD
domain [33]. That said, regardless of the origin of the
apparent small difference in the potentials of the FAD
cofactor in full-length nNOS reductase and the isolated
FAD domain, it is clear that the mutations do not perturb
the overall redox properties of the FAD (E
12
)toany
significant degree. The origin of the kinetic differences
between wild-type and mutant FAD domains is therefore
unlikely to be attributable to altered thermodynamic effects.
Exchange of the aromatic shielding residue (Trp676) in

human CPR for alanine substantially compromises ( 400-
fold) the rate of FAD reduction by NADPH [27]. This
contrasts markedly with the properties of the F1395A and
F1395S nNOS FAD domains, in which the rates of FAD
reduction are similar to wild-type and F1395W FAD
domain. In stopped-flow studies, reduction of the isolated
FAD of CPR is dominated by a slow kinetic phase
( 3.5 s
)1
), and it has been postulated that FAD reduction
in this phase is linked to the release of NADP
+
that
displaces the equilibrium distribution of enzyme species
towards further flavin reduction [26,45]. A smaller but more
rapid phase ( 200 s
)1
) that precedes the major flavin
reduction step involves the establishment of an equilibrium
distribution predominantly involving an enzyme–NADPH
charge-transfer species and a small amount of reduced FAD
domain bound to NADP
+
. In CPR, Trp676 is required to
accelerate NADP
+
release; in the W676H FAD domain of
CPR, a reduced enzyme-NADP
+
charge-transfer species is

stabilized compared with wild-type FAD domain, indica-
ting that Trp676 is actively involved in displacing NADP
+
from reduced FAD domain [27]. Mutagenesis of the
aromatic shielding tyrosine residue in pea ferredoxin-
NADP
+
reductase to serine stabilizes the E-NADP
+
complex and allows the nicotinamide ring of the coenzyme
to bind productively, close to the isoalloxazine ring [46].
This does not occur in wild-type enzyme, indicating that the
aromatic shielding residue competes with the nicotinamide
ring for the productive binding site. By analogy, one might
expect F1395 in nNOS reductase to compete with nicotin-
amide coenzyme for productive binding, and that produc-
tive binding might be favoured in the F1395S and F1395A
FAD domains. In full-length nNOS reductase, however, the
situation is more complicated owing to the effects of CaM
on conformational events in the nicotinamide-binding site.
The recent studies of Adak et al. indicate differential effects
of CaM binding on full-length wild-type NOS and full-
length F1395S NOS, suggesting interplay between CaM
binding, nicotinamide binding and F1395 in coenzyme
binding/FAD reduction [29].
The switch in coenzyme specificity towards NADH in
CPR is consistent with a bipartite mode of binding
nicotinamide coenzyme and weaker binding of NADH in
wild-type enzyme compared with the W676A enzyme,
owing to the flipping motion of W676 [27]. By analogy with

CPR, one might expect F1395 to flip on binding nicotin-
amide coenzyme. In the F1395S and F1395A domains this
model would suggest that the thermodynamically unfa-
vourable flipping motion would no longer be required, and
both binding sites in the bipartite recognition site for
nicotinamide coenzyme should be readily accessible in the
F1395S and F1395A domains. In the wild-type and
F1395W domains, interaction with the adenosine ribose
moiety is probably the first binding step for the coenzyme,
with subsequent occupation of the nicotinamide site occur-
ring only after flipping of the aromatic shielding residue.
However, although the removal of the aromatic residue in
the F1395S and F1395A FAD domains clearly switches the
coenzyme specificity towards NADH, the effect is predom-
inantly attributed to an improved turnover number with
NADH and not an improved Michaelis constant, unlike
with CPR [28]. This suggests that a more favourable binding
geometry for hydride transfer to FAD, and not an improved
affinity for NADH, is the key determinant in switching
specificity. This assertion is consistent with the improved
rate constants observed in stopped-flow studies for reduc-
tion of the F1395S and F1395A FAD domains by NADH
(Table 3). Also, the Michaelis constant for NADPH is not
Table 3. Kinetic constants and apparent enzyme–NADH dissociation constants obtained from analysis of stopped-flow kinetic data for wild-type and
mutant FAD domains following mixing with NADPH and NADH. For reactions with NADPH, k
obs1
and k
obs2
are independent of NADPH
concentration over the range used in stopped-flow studies and represent rate constants for the fast and slow phases, respectively, observed at

454 nm; k
obs3
is the observed rate constant for the formation of the neutral blue semiquinone. For reactions with NADH, data for the fast phase
and slow phase observed at 454 nm are shown. k
lim
is the limiting rate constant for flavin reduction and K
NADH
is the coenzyme concentration that
yields k
lim
/2. Values were derived from hyperbolic fits of rate data vs. NADH concentration for wild-type and mutant FAD domains. All reactions
wereperformedin50m
M
Tris/HCl, 10% glycerol, pH 7.4, at 25 °C for wild-type and mutant enzymes, respectively, as indicated in Experimental
procedures. N/A, not applicable due to lack of semiquinone formation.
NADH
NADPH Fast phase Slow phase
k
obs1
(s
)1
) k
obs2
(s
)1
) k
obs3
(s
)1
) k

lim1
(s
)1
) K
NADH
(m
M
) k
lim2
(s
)1
) K
NADH
(m
M
)
WT 496 ± 29 95 ± 7 0.031 ± 0.002 82.6 ± 4.3 6.2 ± 0.8 3.8 ± 0.2 1.40 ± 0.4
F1395W 500 ± 31 89 ± 6 0.032 ± 0.002 70.3 ± 3.2 25.7 ± 1.7 5.6 ± 0.5 2.9 ± 0.7
F1395S 125 ± 9 10 ± 1 N/A 114.0 ± 6.7 3.5 ± 0.6 67.5 ± 4.4 7.8 ± 1.2
F1395A 680 ± 43 55 ± 3 N/A 139.6 ± 12.3 9.3 ± 1.3 91.0 ± 9.5 29.9 ± 6.7
2558 A. J. Dunford et al. (Eur. J. Biochem. 271) Ó FEBS 2004
substantially affected on removal of the aromatic shielding
residue (Table 1), and this is in stark contrast with the
effects on K
m
on mutating Trp676 to alanine in human CPR
with both NADPH and NADH which resulted in  75-fold
and 150-fold decrease in K
m
, respectively. Again, this

emphasizes in nNOS FAD domain that the geometry for
hydride transfer from NADH to FAD is improved on
removing the aromatic shielding residue, and that differen-
tial binding affinities are not the major factor in coenzyme
discrimination.
In conclusion, using a domain dissection approach we
have been able to demonstrate the functional consequences
of removal of the conserved aromatic shielding residue in
nNOS FAD domain. Our approach has released us from
complications associated with CaM binding and internal
electron transfer to the FMN and haem domains. In
contrast with work on CPR, removal of the shielding
residue does not adversely affect the kinetics of flavin
reduction, although mutation results in a switch in coen-
zyme specificity as reported also for human CPR. The basis
of this switch appears to have its origins in an improved
geometry for NADH binding rather than on improved
affinity for the coenzyme (as seen with CPR). Mutagenesis
to a nonaromatic counterpart protects against dispro-
portionation of the oxidized and reduced FAD domains
to form the neutral blue semiquinone, which might in turn
be a consequence of tighter binding to the oxidized
nicotinamide coenzyme. Similar disproportionation be-
tween NADP(H)-bound quinone and hydroquinone forms
of eukaryotic and bacterial adrenodoxin reductase enzymes
produces large amounts of blue semiquinone species once
these systems reach equilibrium (e.g. [47]). Removal of the
aromatic side chain impacts on the electronic absorption
features of the FAD, but not on the thermodynamic
properties of the FAD. The work emphasizes the different

functional roles of the conserved aromatic residues across
the family of diflavin reductases.
Acknowledgements
The work was funded by the UK Biotechnology and Biological
Sciences Research Council., N.S.S. is a Lister Institute Research
Professor. A.W.M. is a Royal Society Leverhulme Trust Senior
Research Fellow.
References
1. Ignarro, L.J., Buga, G.M., Wood, K.S., Byrns, R.E. & Chaudhuri,
G. (1987) Endothelium-derived relaxing factor produced and
released from artery and vein is nitric oxide. Proc. Natl Acad. Sci.
USA 84, 9265–9269.
2. Palmer, R.M., Ferrige, A.G. & Moncada, S. (1987) Nitric oxide
release accounts for the biological activity of endothelium-derived
relaxing factor. Nature 327, 524–526.
3. Garthwaite, J., Charles, S.L. & Chess-Williams, R. (1988)
Endothelium-derived relaxing factor release on activation of
NMDA receptors suggests role as intercellular messenger in the
brain. Nature 336, 385–388.
4. Griffith, O.W. & Stuehr, D.J. (1995) Nitric Oxide Syn-
thases.Properties Catalytic Mechanism Annu. Rev. Physiol. 57,
707–736.
5. Stuehr, D.J. (1997) Structure-Function Aspects in the Nitric Oxide
Synthases Annu. Rev. Pharmacol. Toxicol. 37, 339–359.
6. Marletta, M.A. (1993) Nitric-oxide synthase structure and mech-
anism. J. Biol. Chem. 268, 12231–12234.
7. Bredt, D.S. & Snyder, S.H. (1994) Nitric oxide: a physiologic
messenger molecule. Annu. Rev. Biochem. 63, 175–195.
8.Bredt,D.S.,Hwang,P.M.,Glatt,C.E.,Lowenstein,C.,
Reed, R.R. & Snyder, S.H. (1991) Cloned and expressed nitric

oxide synthase structurally resembles cytochrome P-450 reductase.
Nature 351, 714–718.
9. Wang,M.,Roberts,D.L.,Paschke,R.,Shea,T.M.,Masters,B.S.
& Kim, J.J. (1997) Three-dimensional structure of NADPH-
cytochrome P450 reductase: prototype for FMN- and FAD-
containing enzymes. Proc. Natl Acad. Sci. USA 94, 8411–8416.
10.Leclerc,D.,Wilson,A.,Dumas,R.,Gafuik,C.,Song,D.,
Watkins, D., Heng, H.H., Rommens, J.M., Scherer, S.W.,
Rosenblatt, D.S. & Gravel, R.A. (1998) Cloning and mapping of
a cDNA for methionine synthase reductase, a flavoprotein
defective in patients with homocystinuria. Proc. Natl Acad. Sci.
USA 95, 3059–3064.
11. Paine,M.J.,Garner,A.P.,Powell,D.,Sibbald,J.,Sales,M.,Pratt,
N., Smith, T., Tew, D.G. & Wolf, C.R. (2000) Cloning and
characterization of a novel human dual flavin reductase. J. Biol.
Chem. 275, 1471–1478.
12. McMillan, K. & Masters, B.S.S. (1995) Prokaryotic expression
heme-binding flavin-binding domains rat neuronal nitric-oxide
synthase distinct polypeptides – identification heme-binding
proximal thiolate ligand cysteine-415. Biochemistry 34, 3686–3693.
13. Abu-Soud, H. & Stuehr, D. (1993) Nitric oxide synthases reveal a
role for calmodulin in controlling electron transfer Proc. Natl
Acad. Sci. USA 90, 10769–10772.
14. Cho, H.J., Xie, Q.W., Calaycay, J., Mumford, R.A., Swiderek,
K.M., Lee, T.D. & Nathan, C. (1992) Calmodulin is a subunit
of nitric oxide synthase from macrophages. J. Exp. Med. 176,
599–604.
15. Sheta, E.A., McMillan, K. & Masters, B.S.S. (1994) Evidence for a
bidomain structure of constitutive cerebellar nitric oxide synthase.
J. Biol. Chem. 269, 15147–15153.

16. Craig, D.H., Chapman, S.K. & Daff, S. (2002) Calmodulin acti-
vates electron transfer through neuronal nitric-oxide synthase
reductase domain by releasing an NADPH-dependent con-
formational lock. J. Biol. Chem. 277, 33987–33994.
17. Abu-Soud, H.M., Yoho, L.L. & Stuehr, D.J. (1994) Calmodulin
controls neuronal nitric-oxide synthase by a dual mechanism.
Activation of intra- and interdomain electron transfer. J. Biol.
Chem. 269, 32047–32050.
18. Matsuda, H. & Iyanagi, T. (1999) Calmodulin activates
intramolecular electron transfer between the two flavins of neu-
ronal nitric oxide synthase flavin domain Biochim. Biophys. Acta
1473, 345–355.
19. Matsuda, H., Kimura, S. & Iyanagi, T. (1999) In Flavins and
Flavoproteins (Ghisla, S., Kroneck, P., Macheroux, P. & Sund, H.,
eds), pp. 171–174. Weber, Berlin.
20. Knight, K. & Scrutton, N.S. (2002) Stopped-flow kinetic studies of
electron transfer in the reductase domain of neuronal nitric oxide
synthase: re-evaluation of the kinetic mechanism reveals new
enzyme intermediates and variation with cytochrome P450
reductase. Biochem. J. 367, 19–30.
21. Daff, S., Sagami, I. & Shimizu, T. (1999) The 42-amino acid insert
in the FMN domain of neuronal nitric-oxide synthase exerts
control over Ca2+/calmodulin-dependent electron transfer.
J. Biol. Chem. 274, 30589–30595.
22. Narhi, L.O., Wen, L.P. & Fulco, A.J. (1988) Characterization of
the protein expressed in Escherichia coli by a recombinant plasmid
containing the Bacillus megaterium cytochrome P-450BM-3 gene.
Mol. Cell. Biochem. 79, 63–71.
23. Sem, D.S. & Kasper, C.B. (1993) Interaction with arginine 597 of
NADPH-cytochrome P-450 oxidoreductase is a primary source of

Ó FEBS 2004 Kinetic and thermodynamic properties of nNOS FAD domains (Eur. J. Biochem. 271) 2559
the uniform binding energy used to discriminate between NADPH
and NADH. Biochemistry 32, 11548–11558.
24. Olteanu, H. & Banerjee, R. (2001) Human methionine synthase
reductase, a soluble P-450 reductase-like dual flavoprotein, is
sufficient for NADPH-dependent methionine synthase activation.
J. Biol. Chem. 276, 35558–35563.
25. Zhang, J., Martasek, P., Paschke, R., Shea, T., Siler Masters, B.S.
& Kim, J.J. (2001) Crystal structure of the FAD/NADPH-binding
domain of rat neuronal nitric-oxide synthase. Comparisons with
NADPH-cytochrome P450 oxidoreductase. J. Biol. Chem. 276,
37506–37513.
26. Gutierrez, A., Lian, L.Y., Wolf, C.R., Scrutton, N.S. & Roberts,
G.C. (2001) Stopped-flow kinetic studies of flavin reduction in
human cytochrome P450 reductase and its component domains.
Biochemistry 40, 1964–1975.
27. Gutierrez, A., Doehr, O., Paine, M., Wolf, C.R., Scrutton, N.S. &
Roberts, G.C. (2000) Trp-676 facilitates nicotinamide coenzyme
exchange in the reductive half-reaction of human cytochrome
P450 reductase: properties of the soluble W676H and W676A
mutant reductases. Biochemistry 39, 15990–15999.
28. Dohr,O.,Paine,M.J.,Friedberg,T.,Roberts,G.C.&Wolf,C.R.
(2001) Engineering of a functional human NADH-dependent
cytochrome P450 system. Proc. Natl Acad. Sci. USA 98, 81–86.
29. Adak, S., Sharma, M., Meade, A.L. & Stuehr, D.J. (2002) A
conserved flavin-shielding residue regulates NO synthase electron
transfer and nicotinamide coenzyme specificity Proc. Natl Acad.
Sci. USA 99, 13516–13521.
30. Piubelli, L., Aliverti, A., Arakaki, A., Carillo, N., Ceccarelli, E.A.,
Karplus, P.A. & Zanetti, G. (2000) Competition between

C-terminal tyrosine and nicotinamide modulates pyridine
nucleotide affinity and specificity in plant ferredoxin-NADP
+
reductase. J. Biol. Chem. 275, 10472–10476.
31. Roitel, O., Scrutton, N.S. & Munro, A.W. (2003) Electron transfer
in flavocytochrome P450 BM3: kinetics of flavin reduction and
oxidation, the role of cysteine 999, and relationships with mam-
malian cytochrome. P450 Reductase Biochem. 42, 10809–10821.
32. Noble, M.A., Munro, A.W., Rivers, S.L., Robledo, L., Daff, S.N.,
Yellowlees, L.J., Shimizu, T., Sagami, I., Guillemette, J.G. &
Chapman, S.K. (1999) Potentiometric analysis of the flavin
cofactors of neuronal nitric oxide synthase. Biochemistry 38,
16413–16418.
33. Wolthers, K.R., Basran, J., Munro, A.W. & Scrutton, N.S. (2003)
Molecular dissection of human methionine synthase reductase:
determination of the flavin redox potentials in full-length enzyme
and isolated flavin-binding domains. Biochemistry 42, 3911–3920.
34. Wolthers, K.R. & Scrutton, N.S. (2004) Electron transfer in
human methionine synthase reductase studied by stopped-flow
spectrophotometry. Biochemistry 43, 490–500.
35. Finn, R.D., Basran, J., Roitel, O., Wolf, C.R., Munro, A.W.,
Paine, M.J. & Scrutton, N.S. (2003) Determination of the redox
potentials and electron transfer properties of the FAD- and FMN-
binding domains of the human oxidoreductase NR1. Eur.
J. Biochem. 270, 1164–1175.
36. Munro, A.W., Noble, M.A., Robledo, L., Daff, S.N. &
Chapman, S.K. (2001) Determination of the redox properties of
human NADPH-cytochrome P450 reductase. Biochemistry 40,
1956–1963.
37. Smith, G.C., Tew, D.G. & Wolf, C.R. (1994) Dissection of

NADPH-cytochrome P450 oxidoreductase into distinct func-
tional domains. Proc. Natl Acad. Sci. USA 91, 8710–8714.
38. Daff, S.N., Chapman, S.K., Turner, K.L., Holt, R.A.,
Govindaraj, S., Poulos, T.L. & Munro, A.W. (1997) Redox
control of the catalytic cycle of flavocytochrome P-450 BM3.
Biochemistry 36, 13816–13823.
39. Newton, D.C., Montgomery, H.J. & Guillemette, J.G. (1998) The
reductase domain of the human inducible nitric oxide synthase is
fully active in the absence of bound calmodulin Arch. Biochem.
Biophys. 359, 249–257.
40. Tartof, K. & Hobbs, C. (1987) Improved media for growing
plasmid and cosmid clones. Bethesda Res. Lab. Focus 9,12.
41. Dutton, P.L. (1978) Redox potentiometry: determination of
midpoint potentials of oxidation-reduction components of biolo-
gical electron-transfer systems. Methods Enzymol. 54, 411–435.
42. Mayhew, S.G. (1999) The effects of pH and semiquinone forma-
tion on the oxidation-reduction potentials of flavin mononucleo-
tide. Eur. J. Biochem. 265, 698–702.
43. Pollegioni, L., Porrini, D., Molla, G. & Pilone, M.S. (2000) Redox
potentials and their pH dependence of
D
-amino acid oxidase of
Rhodotorula gracilis and Trigonopsis variabilis. Eur. J. Biochem.
267, 6624–6632.
44. Gao, Y.T., Smith, S.M.E., Bryce Weinberg, J., Montgomery, H.J.,
Newman, E., Guillemette, J.G., Ghosh, D.K., Roman, L.J.,
Martasek, P. & Salerno, J.C. (2004) Thermodynamics of oxida-
tion-reduction reactions in mammalian nitric oxide synthase iso-
forms. J. Biol. Chem. 279, 18759–18766.
45. Gutierrez, A., Grunau, A., Paine, M.J.I., Munro, A.W.,

Wolf, C.R., Roberts, G.C.K., & Scrutton, N.S. (2003) Electron
transfer in human cytochrome P450 reductase. Biochem. Soc.
Trans. 31, 497–501.
46. Deng, Z., Aliverti, A., Zanetti, G., Arakaki, A.K., Ottado, J.,
Orellano, E.G., Calcaterra, N.B., Ceccarelli, E.A., Carillo, N., &
Karplus, P.A. (1999) A productive NADP+ binding mode of
ferredoxin-NADP+ reductase revealed by protein engineering
and crystallographic studies. Nat. Struct. Biol. 6, 847–853.
47. McLean, K.J., Scrutton, N.S. & Munro, A.W. (2003) Kinetic,
spectroscopic and thermodynamic characterization of the Myco-
bacterium tuberculosis adrenodoxin reductase homologue FprA.
Biochem. J. 372, 317–327.
48. Hubbard,P.A.,Shen,A.L.,Paschke,R.,Kasper,C.B.&Kim,J.J.
(2001) NADPH-cytochrome P450 oxidoreductase. Structural
basis for hydride and electron transfer. J. Biol. Chem. 276,
29163–29170.
2560 A. J. Dunford et al. (Eur. J. Biochem. 271) Ó FEBS 2004