Modulation of the enzymatic efficiency of
ferredoxin-NADP(H) reductase by the amino acid
volume around the catalytic site
Matı
´
as A. Musumeci, Adria
´
n K. Arakaki, Daniela V. Rial, Daniela L. Catalano-Dupuy and
Eduardo A. Ceccarelli
Molecular Biology Division, Instituto de Biologı
´
a Molecular y Celular de Rosario (IBR), Facultad de Ciencias Bioquı
´
micas y Farmace
´
uticas,
Universidad Nacional de Rosario, Argentina
Ferredoxin (flavodoxin)-NADP(H) reductases (FNRs,
EC 1.18.1.2) are a widely distributed class of flavoen-
zymes that have non-covalently bound FAD cofactor
as a redox center. FNRs participate in a wide variety
of redox-based metabolic reactions, transferring elec-
trons between obligatory one- and two-electron carri-
ers and therefore functioning as a general electron
splitter. In non-phototrophic bacteria and eukaryotes,
the reaction is driven towards ferredoxin (Fd) reduc-
tion, providing reducing power for multiple metabolic
pathways, including steroid hydroxylation in mamma-
lian mitochondria, nitrite reduction and glutamate
synthesis in heterotrophic tissues of vascular plants,
radical propagation and scavenging in prokaryotes,

uticas,
Universidad Nacional de Rosario, Suipacha
531, S2002LRK Rosario, Argentina
Fax: +54 341 4390465
Tel: +54 341 4351235
E-mail:
(Received 1 November 2007, revised 8
January 2008, accepted 16 January 2008)
doi:10.1111/j.1742-4658.2008.06298.x
Ferredoxin (flavodoxin)-NADP(H) reductases (FNRs) are ubiquitous
flavoenzymes that deliver NADPH or low-potential one-electron donors
(ferredoxin, flavodoxin, adrenodoxin) to redox-based metabolic reactions in
plastids, mitochondria and bacteria. Plastidic FNRs are quite efficient
reductases. In contrast, FNRs from organisms possessing a heterotrophic
metabolism or anoxygenic photosynthesis display turnover numbers 20- to
100-fold lower than those of their plastidic and cyanobacterial counterparts.
Several structural features of these enzymes have yet to be explained. The
residue Y308 in pea FNR is stacked nearly parallel to the re -face of the fla-
vin and is highly conserved amongst members of the family. By computing
the relative free energy for the lumiflavin–phenol pair at different angles
with the relative position found for Y308 in pea FNR, it can be concluded
that this amino acid is constrained against the isoalloxazine. This effect is
probably caused by amino acids C266 and L268, which face the other side
of this tyrosine. Simple and double FNR mutants of these amino acids were
obtained and characterized. It was observed that a decrease or increase in
the amino acid volume resulted in a decrease in the catalytic efficiency of
the enzyme without altering the protein structure. Our results provide exper-
imental evidence that the volume of these amino acids participates in the
fine-tuning of the catalytic efficiency of the enzyme.
Abbreviations

the isoalloxazine moiety and interacts extensively with
it (Fig. 1A) ([1,2,13] and references therein). This tyro-
sine has been implicated in catalysis, modulation of
the FAD reduction potential, inter- and intra-protein
electron transfer processes [14–18] and determination
of the specificity and high catalytic efficiency [15,17–
19]. Using NMR techniques, it has been shown that
the maize FNR homolog Y314 is perturbed on
NADP
+
binding, as is the carboxyl terminal region of
the protein [20]. Recently, experimental evidence for
the mobility of the carboxyl terminal backbone region
of FNR and, mainly, Y308 has been provided [19],
indicating that this movement is essential for obtaining
an FNR enzyme with high catalytic efficiency.
During catalysis, the nicotinamide ring must move
to the re-face of the isoalloxazine moiety for electron
transfer to occur. Thus, Bruns and Karplus [3] have
proposed that the aromatic side-chain of the carboxyl
terminal tyrosine should be displaced to allow the sub-
strate to move into the correct position (named ‘in’
conformation). The interaction of the phenol ring of
Y308 with the isoalloxazine should be precisely
adjusted to facilitate the ‘in’ and ‘out’ conformations
of the NADP(H) nicotinamide. A strong interaction of
Y308 with the flavin would impede the ability of nico-
tinamide to go into the site; meanwhile, a slight inter-
action would favor the stacking of the nicotinamide
onto the isoalloxazine, thus decreasing the turnover

decreases the catalytic efficiency, suggesting that these
steric considerations may be a requirement for high
catalytic efficiency. The mutations did not produce a
significant perturbation of the overall protein structure
and did not affect the oxidase activity of the flavo-
enzyme. Our results suggest that these amino acids
participate in the fine-tuning of enzyme efficiency,
modulating the interaction of Y308 and ⁄ or the nicotin-
amide with the isoalloxazine. This type of modulation
of aromatic residue interactions could be a general
strategy occurring in enzyme structures.
Results
Ab initio molecular orbital calculations
The geometries of aromatic amino acids facing the re-
face of the flavin were determined using high-resolu-
tion plant-type FNR crystal structures. It was observed
that these tyrosines always interact in face-to-face posi-
tions (Fig. 1A; Table 1). The B ring of the flavin is
always involved in this interaction in a nearly parallel
position in which the angle formed with the tyrosine
phenol and isoalloxazine varies from 0° to 6° in all
high-efficiency plastidic FNRs and 15° for the ferre-
doxin-NADP(H) reductase from E. coli. To gain a
better understanding of this interaction, the geometric
preferences of the above-mentioned interaction were
analyzed using model molecules and ab initio mole-
cular orbital calculations with the restricted Hartree
Fock theory level and a 6-311 + G(d,p) basis set. A
simplified system was constructed containing lumiflavin
(7,8,10-trimethylisoalloxazine), which is an accepted

[24].
A global energy minimum was theoretically detected
between 11° and 22° at a distance between centroids of
3.6 A
˚
. The angle found in E. coli FNR was the closest
to the minimum of the plot. In all plastidic FNRs, the
position of the tyrosine was near the minimum (repre-
sented in Fig. 2B with open circles and a number indi-
cating the enzyme). Any position that does not fall
within )10° to 37° notoriously decreases the stability of
the pair, increasing repulsion, probably as result of steric
constraints between the two aromatic rings. When the
total energy of the system was analyzed at a centroid–
centroid distance of 4.6 A
˚
, a minimum was observed at
40° and a shallow low-energy region was detected from
20° to 55°. Moreover, all differences in potential energy
values obtained for arrangements at 4.6 A
˚
between
angles from )20° to 85° were equal or lower than the
energy calculated for the observed arrangements found
in plastidic FNR enzymes in nature (Fig. 2B). All FNRs
Table 1. Angles and distances between the tyrosine interacting
with the re-face of the flavin and the isoalloxazine B ring obtained
from FNR crystal structures.
FNR source Type
Maximal

flavin falling near or into the minimum energy valley
with a centroid separation of about 3.6 A
˚
. However, if
the tyrosine were able to move away from the flavin, a
more stable arrangement was possible between both
aromatic rings, allowing them to gain up to approxi-
mately 5.8 kcalÆmol
)1
of stabilization energy, as calcu-
lated from Fig. 2B. Thus, it may be inferred that the
position of the re-face tyrosine in FNRs is not governed
by the energetic minimum of the pairwise flavin–phenol
interaction. By analyzing the crystal structure of pea
FNR, it was deduced that Y308 is constrained against
the isoalloxazine in an unfavorable conformational
arrangement by the influence of amino acids C266 and
L268. These residues face the other side of this tyrosine
and are members of a conserved loop (266CGLKG270)
that shapes part of the FNR catalytic site (see
Fig. 1A,B). They may force Y308 to adopt a more pla-
nar orientation with respect to the flavin. The overall
result is a less stable conformational arrangement.
Design and construction of C266, G267 and L268
single and double FNR mutants
Five single mutants of C266, G267 and L268 and a
double mutant of C266 and L268 were successfully
constructed and confirmed by DNA sequencing. The
design of the mutants was intended to preserve the
amino acid character and to modify only the relative

)1
Æ(lmolÆFNR)
)1
], as determined by measuring the
increase in FAD fluorescence [22] after incubating the
enzymes for 5 h at 25 °C. These observations suggest a
weaker FAD interaction with the apoprotein, and may
explain the difficulties in obtaining these enzymes in
soluble form during protein expression in E. coli.
Attempts to purify mutant enzyme G267V were unsuc-
cessful and no further analysis was possible.
All reductase variants were excised from the carrier
protein using thrombin protease and, after chromato-
graphy on nickel-nitrilotriacetic acid agarose, were
obtained in homogeneous form as judged by SDS-
PAGE (not shown).
FAD content and spectral properties
Analysis of the UV–visible absorption properties of
the different FNR mutants showed small changes,
AB
Fig. 2. Computed relative free energy calculations for the lumiflavin–phenol interaction. (A) Scheme of the coordinate system used to define
the relative positions of phenol and lumiflavin, as found for Y308 and flavin in pea FNR. a, Dihedral interplanar angle between rings (for clar-
ity, only three positions are shown); d, distance between ring centroids. (B) Relative free energy of the arrangement shown in (A) as a func-
tion of the stated a angles at fixed distances of 3.6 A
˚
(
, full line) and 4.6 A
˚
( , broken line). Open symbols indicate the observed values for
the different plastidic FNRs as follows: 1, paprika; 2, spinach; 3, Anabaena ; 4, pea; 5, maize; 6, Synechococcus sp.; 7, E. coli. Ab initio

and visible CD spectra (Fig. 3C) of the proteins were
also very similar, showing the typical spectrum for
FNR [26], with positive ellipticity in the region of the
first flavin visible absorption band, and with a peak at
approximately 380 nm for the wild-type enzyme and
370 nm for the mutant proteins. This is consistent with
the alteration observed in the absorbance spectra of
the mutants. A less intense band of negative ellipticity
was observed in the region of the second flavin visible
band at 470 nm for the wild-type enzyme and mutant
proteins (Fig. 3C). In the near-UV region, all FNRs
exhibited very strong, sharp positive and negative sig-
nals at 271 and 286 nm, respectively. A similar strong
signal at 272 nm, observed in the CD spectrum of
E. coli Fld oxidoreductase, has been attributed to the
stacked interactions between FAD and one or more
aromatic residues [27]. The introduced mutations in
FNR did not alter the position of this near-UV band.
Some changes in intensity were observed in the FNR
mutants, indicating some perturbation of the symmetry
relationships between the isoalloxazine chromophore
Fig. 3. Absorbance and CD spectra of wild-type and mutant FNRs.
Absorbance (A) and CD (B, C) spectra of wild-type FNR (thick line),
L268V (thin line), C266AL268A (thick dotted line), C266A (thin dot-
ted line), C266L (thick broken line) and C266M (thin broken line).
For spectra at 200–250 nm (B), the optical path length was 0.2 cm
and the protein concentration of FNRs was 0.5 l
M. For spectra at
250–600 nm (C), the optical path length was 1 cm and the protein
concentration of FNRs was 5 l

constants for NADP
+
were not significantly affected
in any of the mutants for either NADP
+
or Fd
(Tables 2 and 3, respectively). The only exception was
the double mutant, which showed an increase in the
K
d
value for the enzyme–NADP
+
complex. It has
been documented that the intensity of the FNR–
NADP
+
differential spectrum peak at about 510 nm
correlates with the nicotinamide interaction on the
re-face of the isoalloxazine [17,25,28,29]. It may be
inferred from the spectral data presented that the inter-
action of the NADP
+
nicotinamide with the flavin is
considerably disturbed, probably as a result of changes
introduced by the mutations in the environment of the
prosthetic group. As a result of the important changes
observed for each mutant in the differential spectra
elicited by NADP
+
, it was decided to use an alterna-

reactions. The observed values for k
cat
and K
m
and
the calculated k
cat
⁄ K
m
value for NADPH and Fd are
summarized in Tables 2 and 3.
L268V FNR displayed k
cat
and K
m
values for the
diaphorase reaction in the region of 0.8 and 2.0 times
those observed for the wild-type enzyme. Similarly, the
decrease in Fd was about 0.7 times that observed with
the wild-type enzyme.
By contrast, mutations in C266 produced a more
dramatic effect on the catalytic properties of FNR.
Replacement of C266 with a methionine, which implies
a volume increase of 55.5 A
˚
3
, decreased the k
cat
value
by more than 99.8% and increased the K

M NADP
+
.
M. A. Musumeci et al. Enzyme efficiency modulated by amino acid volume
FEBS Journal 275 (2008) 1350–1366 ª 2008 The Authors Journal compilation ª 2008 FEBS 1355
neutral amino acid by non-polar residues without a
change in the net charge. The catalytic efficiencies of
the different enzymes were plotted as a function of the
absolute change of hydropathy according to Kyte and
Doolittle [30] (Fig. 5A), the octanol–water partition
coefficient (log P) [31] and volume [32]. No correla-
tions were found with changes in hydropathy (Fig. 5A)
and log P (Fig. 5B). In contrast, the absolute change
in volume correlated with the decrease in catalytic effi-
ciency (Fig. 5C). An alteration (increase or decrease)
in volume of the amino acid at position 266 induced a
decrease in catalytic efficiency of the enzyme. Mutants
with higher volume changes in this residue were more
affected. The value for the reduction in catalytic effi-
ciency previously obtained by replacement of spinach
FNR C272 (homolog to pea FNR C266) with a serine
[28] was included (open symbols in Fig. 5A–C and
white bar in Fig. 5D).
Mutations in C266 that decreased the amino acid
volume resulted in a moderate increase in catalytic effi-
ciency for the activity of cytochrome c reductase.
These results originate from the higher relative
decrease in K
m
than k

)
k
cat
⁄ K
m
(lM
)1
Æs
)1
)
DDG
MUT ⁄ WT
b
(kcalÆmol
)1
)
K
d
(NADP
+
)
c
(lM)
K
d
(NADP
+
)
d
(lM)

in 50 m
M Tris ⁄ HCl (pH 8.0) at 25 °C. Absorbance differences (DA at 510 nm for the wild-type and L268V mutant FNRs, and at 390 nm for
the C266A, C266AL268A, C266L and C266M mutant FNRs) were measured and plotted against increasing NADP
+
concentration. The data
were fitted to a theoretical equation for a 1 : 1 complex.
d
Determined by fluorescence spectroscopy using oxidized flavoproteins at 8.5 lM
in 50 mM Tris ⁄ HCl (pH 8.0) at 25 °C, as described in Materials and methods.
Table 3. Kinetic parameters for cytochrome c reductase of the wild-type (WT) and mutant FNRs, and dissociation constants for the com-
plexes of the different FNR forms with Fd. Cytochrome c reduction was followed at 550 nm (e
550
=19mM
)1
Æcm
)1
) as described in Materials
and methods. ND, not determined.
FNR form
DV
a
(A
˚
3
)
K
m
(Fd)
(l
M)

b
K
dP
⁄ K
dA
WT 0 2.2 ± 0.4 1.62 ± 0.14 0.73 ± 0.19 5.16 ± 0.25 0.68 ± 0.02 7.6
C266A –21.5 0.20 ± 0.03 0.74 ± 0.03 3.70 ± 0.70 2.78 ± 0.20 1.02 ± 0.13 2.8
L268V –25.0 4.6 ± 0.8 1.08 ± 0.08 0.23 ± 0.05 2.80 ± 0.30 2.69 ± 0.10 1.0
C266AL268A –96.2 0.004 ± 0.001 0.016 ± 0.0009 4.00 ± 1.22 2.74 ± 0.24 2.20 ± 0.26 1.3
C266L 53.2 ND ND ND 2.86 ± 0.22 2.88 ± 0.19 1
C266M 55.5 ND ND ND 3.53 ± 0.39 2.77 ± 0.15 1.3
a
Volume change of the R amino acid groups introduced by mutations was determined following the standard radii and volumes calculated
by Tsai et al. [32].
b
Determined by fluorescence spectroscopy using oxidized flavoproteins at 3 lM in 50 mM Tris ⁄ HCl (pH 8.0) at 25 °Cin
the absence or presence of 0.3 m
M NADP
+
, as described in Materials and methods.
Enzyme efficiency modulated by amino acid volume M. A. Musumeci et al.
1356 FEBS Journal 275 (2008) 1350–1366 ª 2008 The Authors Journal compilation ª 2008 FEBS
saturating NADPH concentration. As shown in
Table 4, wild-type and mutant enzymes displayed simi-
lar oxidase activities, indicating that no changes are
evident in this process on mutation of the FNR resi-
dues under study.
Thermal analysis of protein unfolding for
wild-type and mutant FNRs
Thermal denaturation determined by CD was used to

55.5
–96.2
ΔV (Å
3
)
FNRs
ΔV (Å
3
) (absolute value)
1
ΔHydropathy (absolute value)
ΔlogP (absolute value)
0 20 40 60 80 100
0.01
0.1
10
100
0234
0.01
0.1
1
10
100
0.0 0.1 0.2 0.3 0.4 0.5
k
cat
/K
m

(%)

FNR form DV (A
˚
3
) Oxidase activity (s
)1
)
WT 0 0.10 ± 0.01
C266A )21.5 0.09 ± 0.01
L268V )25 0.09 ± 0.01
C266AL268A )96.2 0.08 ± 0.009
C266L 53.2 0.11 ± 0.01
C266M 55.5 0.14 ± 0.01
M. A. Musumeci et al. Enzyme efficiency modulated by amino acid volume
FEBS Journal 275 (2008) 1350–1366 ª 2008 The Authors Journal compilation ª 2008 FEBS 1357
direct effect of mutations that replace native amino
acids with alanine on the overall stability of the pro-
tein was evaluated. A theoretical DDG value of
)1.02 kcalÆmol
)1
was obtained for the C266A mutant,
in complete agreement with our experimental results.
When the amino acid mutation induced a volume
increase, important destabilizations were experimen-
tally observed. C266L and C266M exhibited lower
DDG values: )8.50 and )6.80 kcalÆmol
)1
, respectively.
These outcomes indicate that although little influence
is exerted by residue substitutions on the destabiliza-
tion of the secondary and tertiary structure (see Fig. 3)

and L268. The data support the hypothesis that the
aromatic interaction between the flavin, Y308 and the
nicotinamide of NADP
+
is precisely tuned by selecting
amino acids that face the other side of the tyrosine
phenol ring. The specific volumes of the above-men-
tioned residues condition the arrangement of Y308
and the nicotinamide of NADP
+
in the catalytic site.
Non-covalent aromatic interactions are essential to
protein–ligand recognition [39]. Furthermore, they are
widespread in biomolecules, clusters, organic ⁄ biomo-
lecular crystals and, more recently, in the building of
nanomaterials [40]. In proteins, the rings of trypto-
phan, tyrosine, phenylalanine and histidine participate
either in the interaction with hydrogen donors (p–H
interaction) or binding with other aromatic rings (p–p
interactions) [41]. The latter interactions are observed
in a great variety of geometries. The edge–face geome-
try is commonly found between aromatic residues in
proteins. Other two-stacked orientations are also estab-
lished, including one in which the interacting rings are
offset and stacked near-planar, and arrangements of
face-to-face stacked aromatic rings [42].
By analyzing the crystal structure of FNRs, it was
found that the inter-ring orientational angles between
the re-face aromatic ring and flavins were quite con-
stant and always positioned at a limiting distance of

C266A )21.5 63.2 ± 0.3 0.71 ± 0.03 )0.94
L268V )25.0 63.5 ± 0.1 )0.40 ± 0.01 )0.76
C266AL268A )96.2 63.5 ± 0.1 0.69 ± 0.01 )0.77
C266L 53.2 50.8 ± 0.1 0.38 ± 0.01 )8.50
C266M 55.5 53.6 ± 0.2 0.43 ± 0.02 )6.80
a
Volume change of the R amino acid groups introduced by the
mutations was determined following the standard radii and vol-
umes calculated by Tsai et al. [32].
Fig. 6. Thermal unfolding of wild-type and mutant FNRs monitored
by CD. CD melting curves were recorded at 280 nm, using a pro-
tein concentration of 3 l
M in 50 mM potassium phosphate (pH 8.0),
whilst the temperature of the sample was increased at a uniform
rate of 1 °CÆmin
)1
(from 25 to 80 °C). Wild-type FNR (thick line),
L268V (thin line), C266AL268A (thick dotted line), C266A (thin dot-
ted line), C266L (thick broken line) and C266M (thin broken line) are
shown.
Enzyme efficiency modulated by amino acid volume M. A. Musumeci et al.
1358 FEBS Journal 275 (2008) 1350–1366 ª 2008 The Authors Journal compilation ª 2008 FEBS
energy for a distance of 3.6 A
˚
. However, when calcula-
tions were performed with aromatic rings stacked at
4.6 A
˚
, a lower energy minimum was obtained. These
results suggest that, if more freedom were available for

sine–flavin bacterial arrangement in E. coli FNR is
1.24 kcalÆmol
)1
more stable than that observed for the
same pair in plastidic pea FNR (open circle numbered
4 in Fig. 2). Thus, tyrosine displacement for nicotin-
amide binding should be easier in pea FNR than in
the bacterial enzyme. As this movement was postulated
to be the rate-limiting step for catalysis [18], the differ-
ences in stability may account for the distinct turnover
numbers that are 20- to 100-fold lower for bacterial
enzymes than their plastidic and cyanobacterial coun-
terparts.
Our mutants enabled the observed results to be
interpreted in terms of protein structure, thermody-
namics and function. The C266 mutants are of particu-
lar interest because this residue has functional
homologs in all FNR-like structures. Moreover, the
cysteine and glycine at this position are part of one of
the consensus sequences that define the structural fam-
ily [1,11]. As anticipated, the final tertiary structure of
the mutants, with the exception of G267V, was rela-
tively unchanged, as shown by the fact that mutations
in FNR did not alter the near-UV band of the
CD spectra. A small perturbation of isoalloxazine was
detected by CD and UV–visible spectrophotometry.
Flavin electronic transitions in the 300–600 nm region
originate from p–p transitions [26]. Thus, changes in
the CD spectra are expected to occur on modification
of the interaction of Y308 with the flavin. Our mutants

+
binding to L268V are coincident with the dif-
ferential spectra previously obtained for the Anabaena
variabilis FNR mutant L263A [47]. The K
d
values
obtained for NADP
+
binding to the mutants were
only slightly modified, with the exception of the double
mutant. It can be concluded that the interactions with
the adenine and phosphate regions of NADP
+
are
conserved, and that the observed alteration is probably
the result of a change in the position or extent of inter-
action between the flavin and the nicotinamide.
Kinetic analysis of the mutants indicates that the
cysteine sulfhydryl group is by no means essential for
catalysis, as documented previously [28]. Replacement
of C266 by any aliphatic residue produced enzymes
that, even when notoriously affected in catalysis, were
still active. When the cysteine was substituted with a
methionine, providing a sulfur atom in a nearby posi-
tion, a functional enzyme was also obtained. Sulfur–
flavin interactions have been proposed and analyzed
by computational studies and experimental means
[48,49]. These studies have indicated the existence of
M. A. Musumeci et al. Enzyme efficiency modulated by amino acid volume
FEBS Journal 275 (2008) 1350–1366 ª 2008 The Authors Journal compilation ª 2008 FEBS 1359

volume decrease. Interestingly, the double mutant
C266AL268A, in which both amino acid replacements
reduced the amino acid volume, displayed a greater
change in K
m
. A correlation was observed between the
volume change introduced by the mutation in position
266 and the decrease in enzyme catalytic efficiency.
Remarkably, the value for the catalytic efficiency pre-
viously obtained by replacement of the homologous
residue in spinach FNR (C272) with a serine [28] fits
perfectly on our graph, and follows the trend of the
other mutants. Our theoretical calculations indicate
that a more stable arrangement of the flavin and tyro-
sine would provide up to approximately 5.8 kcalÆmol
)1
of stabilization energy. This value is in good agreement
with the energy barrier introduced by the mutations to
the catalytic efficiency of FNR, as shown in Table 2.
Data from the L268V mutant also support our
hypothesis, although the change in k
cat
was smaller
than that observed for enzymes mutated at position
266. L268V FNR shows a catalytic efficiency of about
40% with respect to that of the wild-type enzyme. Sim-
ilar observations have been made previously by mutat-
ing the equivalent residue L263 in Anabaena FNR. In
the latter case, L263A (change in volume, )74.7 A
˚

observed only when C266 was mutated with an amino
acid that induced an important volume increase, but
not when substituted with smaller amino acids. L268V
is only 0.76 kcalÆmol
)1
less stable than wild-type FNR.
Thus, C266 has not been evolutionarily selected to
merely stabilize the protein structure. The importance
of amino acid volume in relation to non-synonymous
substitutions in proteins was envisaged several years
ago [51]. When globin sequences were analyzed, it was
observed that the total sequence volume in conserved
proteins was quite constant, with variations of 2–3%
[52]. The variation in amino acid volume per internal
position is in the region of 13% and up to 21% in sur-
face residues [52]. Recently, it has been observed that
the probabilities of compensatory mutations that
involve small changes in amino acid volumes are
higher [53]. These observations may be taken to sup-
port the intuitive idea that small changes may produce
a lesser effect on protein structure, consistent with that
observed in the protein stability of our mutants. There-
fore, these conclusions strengthen our hypothesis that
the catalytic efficiencies of the recombinant enzymes
used in our study were related to the volume of the
mutated amino acids. It has been proposed that, in
Anabaena FNR, the 261–265 loop (which is equivalent
to the 266–270 region in pea FNR, see Fig. 1) is
involved in determining coenzyme specificity [47]. A
triple mutant of amino acids T155G ⁄ A160T ⁄ L263P

the final observed result. At present, there are no
experimental data to clarify this issue.
Another scenario should also be considered. C266
participates in a hydrogen bond net that includes the
essential amino acids S90 and E306 and the B side of
the nicotinamide C4 atom, when NADP
+
is bound in
a productive position [21]. These residues are primarily
involved in nicotinamide binding rather than being
directly involved in the hydride transfer reaction [21].
Our results, together with the amino acid arrangement
observed in the catalytic site, indicate that C266 may
constrain the nicotinamide and or the terminal tyrosine
against the flavin. It has been described that the bind-
ing of NADH to lactate dehydrogenase conditions the
nicotinamide glycosidic bond torsion angle, altering the
distribution of conformations, and thus promoting
the catalytic reaction [54]. Residues C266 and L268
may influence the conformational freedom of the sub-
strate, favoring a reactive conformation. In other
enzymes, the role of pressing the nicotinamide against
the flavin may be carried out by other residues. For
example, in bacterial reductases from Azotobacter vine-
landii [9] and Rhodobacter capsulatus [8], which belong
to subclass I [1], the interacting amino acid facing the
re-side of the flavin is a conserved alanine. Similarly, in
human glutathione reductase [55,56] and thioredoxin
reductase [45] a tyrosine residue, which changes from a
T-shaped aromatic interaction to a planar position,

+
nor the residues C266, G267 and L268 are
adjacent to the Fd binding site [2], it may be concluded
that a conformational change involving distant regions
of the protein may occur on NADP
+
binding. As
mentioned previously, Hermoso et al. [5] proposed that
the loop including C266, G267 and L268 suffers a
structural rearrangement on NADP
+
binding. Taken
together, it is suggested that this conformational
change in protein structure may allow enzyme negative
co-operativity to occur between substrates, with C266
being the key residue for initiating this process.
Materials and methods
Ab initio molecular orbital force field calculations
and amino acid physicochemical properties
The original FAD and tyrosine arrangements were based
on X-ray diffraction data for the pea enzyme [21]. Angles
and distances were calculated using hyperchem version
6.01 (HyperCube Inc., Gainesville, FL, USA) and gopen-
mol version 3.00, written by Leif Laaksonen and available
at .fi/gopenmol/. The effect of alanine substi-
tutions was estimated using foldx software [36]. Figures
were built using pymol, available at rce-
forge.net/. Ab initio molecular theory calculations were car-
ried out at the Restricted Hartree Fock theory level with pc
gamess V7.0 accessible at />gamess/index.html using a 6-311 + G(d,p) basis set. The

digested with NheI and EcoRI enzymes and the resulting
fragment was ligated into pET205 vector digested with the
same enzymes. FNR mutants C266L and C266M were
obtained using the overlap extension PCR method [62].
In this case, the coding sequence for mature FNR was
amplified using oligonucleotides FNRC266L (5¢-GACAA
CACTTTTGTCTACATGTTGGGACTGAAAGG-3¢)or
FNRC266M (5¢-GACAACACTTTTGTCTACATGATGG
GACTGAAAGG-3¢) and FNRlw1 (5¢-GTAATCTATCTA
CAGAATACAGGAGGGTGATA-3¢) as primers and plas-
mid pCV105 [63] as template. In addition, oligonucleotides
FNRup1 (5¢-AACAAGTTCAAACCTAAGGAACCATA
CG-3¢) and FNRC266Llw (5¢-CCTTTCAGTCCCAACA
TGTAGACAAAAGTGTTGTC-3¢) or FNRC266Mlw
(5¢ -C CTTTCAGTCC
CATCATGTAGACAAAAGTGTTG
TC-3¢) were used as primers in a second PCR. After ampli-
fication, the products were used as template in a third PCR
with oligonucleotides FNRup1 and FNRlw1 as primers.
The amplified product was digested with ClaI and NheI
enzymes and the resulting fragment was ligated into
pCV105 vector [63] digested with the same enzymes.
Finally, this plasmid was digested with NheI and EcoRI
enzymes and the fragment of 880 bp obtained was ligated
into pET205 vector digested with the same enzymes. In the
primer sequences, the bold letters indicate a silent mutation
that generates an AccI recognition site which was used for
mutant screening. The italic letters indicate mutations that
produce the amino acid variants. Finally, all constructions
were verified by DNA sequencing.

Spectral analyses
Absorption spectra were recorded on a Shimadzu (Kyoto,
Japan) UV-2450 spectrophotometer. CD spectra were
obtained using a JASCO (Tokyo, Japan) J-810 spectropola-
rimeter at 25 °C. The spectra were recorded on solutions
having protein concentrations of 5.0 lm for the near-UV
and visible regions (250–600 nm) and 0.5 l m for the far-
UV region (200–250 nm). Samples were filtered through a
G25 Sephadex spin column equilibrated with 50 mm potas-
sium phosphate (pH 8.0) before measurements. Extinction
coefficients of the FNR forms were determined by releasing
FAD from the protein by treatment with 0.2% (w ⁄ v) SDS
and quantifying the flavin spectrophotometrically [25].
Determination of dissociation constants of the
FNR–NADP
+
complex
The K
d
values of the complexes between different FNR
variants and NADP
+
were determined either by difference
absorption spectroscopy, essentially as described previously
[19], or by fluorescence spectroscopy monitoring FAD fluo-
rescence [22,66]. Fluorescence spectra were monitored using
a Varian (Palo Alto, CA, USA) Cary Eclipse fluorescence
spectrophotometer interfaced with a personal computer.
For difference absorption spectroscopy,  15 lm flavopro-
tein in 50 mm Tris ⁄ HCl (pH 8.0) was titrated at 25 °C with

Thermal unfolding transitions
Protein stock solutions were diluted to a final concentration
of 3 lm in 50 mm potassium phosphate (pH 8.0). The CD
signal was registered by excitation at 280 nm whilst the
temperature of the sample was increased at a uniform rate
of 1 °CÆmin
)1
(from 25 to 80 °C). Thermal unfolding transi-
tions were analyzed assuming a two-state approximation in
which only the native and unfolded states are significantly
populated. T
m
, DS
m
and DDG values were determined as
described previously [67].
Enzymatic assays
FNR-dependent diaphorase and cytochrome c reductase
activities were determined using published methods [68].
The cytochrome c reductase activity of FNR was assayed
in a reaction medium (1 mL) containing 50 mm Tris ⁄ HCl
(pH 8.0), 0.3 mm NADP
+
,3mm glucose-6-phosphate, 1 U
of glucose-6-phosphate dehydrogenase, 50 lm cytochrome c
and 5 lm Fd. After the addition of  15–100 nm FNR, the
reaction was monitored spectrophotometrically by follow-
ing cytochrome c reduction at 550 nm (e
550
=19mm

´
fica y Tecnolo
´
gica (ANPCyT, Argentina). EAC
is a staff member of CONICET. MAM and DLCD
are fellows of the same institution.
References
1 Ceccarelli EA, Arakaki AK, Cortez N & Carrillo N
(2004) Functional plasticity and catalytic efficiency in
plant and bacterial ferredoxin-NADP(H) reductases.
Biochim Biophys Acta 1698, 155–165.
2 Carrillo N & Ceccarelli EA (2003) Open questions in
ferredoxin-NADP
+
reductase catalytic mechanism. Eur
J Biochem 270, 1900–1915.
3 Bruns CM & Karplus PA (1995) Refined crystal struc-
ture of spinach ferredoxin reductase at 1.7 A
˚
resolution:
oxidized, reduced and 2¢-phospho-5¢ -AMP bound states.
J Mol Biol 247, 125–145.
4 Serre L, Vellieux FM, Medina M, Gomez-Moreno C,
Fontecilla-Camps JC & Frey M (1996) X-ray structure
of the ferredoxin:NADP
+
reductase from the cyanobac-
terium Anabaena PCC 7119 at 1.8 A
˚
resolution, and

11740.
9 Sridhar PG, Kresge N, Muhlberg AB, Shaw A, Jung
YS, Burgess BK & Stout CD (1998) The crystal struc-
ture of NADPH:ferredoxin reductase from Azotobacter
vinelandii. Protein Sci 7, 2541–2549.
10 Ingelman M, Bianchi V & Eklund H (1997) The three-
dimensional structure of flavodoxin reductase from Esc-
herichia coli at 1.7 A
˚
resolution. J Mol Biol 268, 147–157.
11 Arakaki AK, Ceccarelli EA & Carrillo N (1997) Plant-
type ferredoxin-NADP
+
reductases: a basal structural
framework and a multiplicity of functions. FASEB J
11, 133–140.
M. A. Musumeci et al. Enzyme efficiency modulated by amino acid volume
FEBS Journal 275 (2008) 1350–1366 ª 2008 The Authors Journal compilation ª 2008 FEBS 1363
12 Arakaki AK, Orellano EG, Calcaterra NB, Ottado J &
Ceccarelli EA (2001) Involvement of the flavin
si-face tyrosine on the structure and function of
ferredoxin-NADP
+
reductases. J Biol Chem 276,
44419–44426.
13 Karplus PA & Bruns CM (1994) Structure–function
relations for ferredoxin reductase. J Bioenerg Biomembr
26, 89–99.
14 Calcaterra NB, Pico GA, Orellano EG, Ottado J, Car-
rillo N & Ceccarelli EA (1995) Contribution of the

NADP(+) reductase in hydride transfer processes with
NAD(P)
+
⁄ H. Biochemistry 44, 13477–13490.
19 Catalano-Dupuy DL, Orecchia M, Rial DV & Ceccarel-
li EA (2006) Reduction of the pea ferredoxin-NADP(H)
reductase catalytic efficiency by the structuring of a car-
boxyl-terminal artificial metal binding site. Biochemistry
45, 13899–13909.
20 Maeda M, Lee YH, Ikegami T, Tamura K, Hoshino
M, Yamazaki T, Nakayama M, Hase T & Goto Y
(2005) Identification of the N- and C-terminal substrate
binding segments of ferredoxin-NADP
+
reductase by
NMR. Biochemistry 44, 10644–10653.
21 Deng Z, Aliverti A, Zanetti G, Arakaki AK, Ottado J,
Orellano EG, Calcaterra NB, Ceccarelli EA, Carrillo N
& Karplus PA (1999) A productive NADP
+
binding
mode of ferredoxin-NADP
+
reductase revealed by pro-
tein engineering and crystallographic studies. Nat Struct
Biol 6, 847–853.
22 Catalano Dupuy DL, Rial DV & Ceccarelli EA (2004)
Inhibition of pea ferredoxin-NADP(H) reductase by
Zn-ferrocyanide. Eur J Biochem 271, 4582–4593.
23 Zheng YJ & Ornstein RL (1996) A theoretical study of

29 Aliverti A, Deng Z, Ravasi D, Piubelli L, Karplus PA
& Zanetti G (1998) Probing the function of the invari-
ant glutamyl residue 312 in spinach ferredoxin-NADP
+
reductase. J Biol Chem 273, 34008–34015.
30 Kyte J & Doolittle RF (1982) A simple method for dis-
playing the hydropathic character of a protein. J Mol
Biol 157, 105–132.
31 Black SD & Mould DR (1991) Development of hydro-
phobicity parameters to analyze proteins which bear
post- or cotranslational modifications. Anal Biochem
193, 72–82.
32 Tsai J, Taylor R, Chothia C & Gerstein M (1999) The
packing density in proteins: standard radii and volumes.
J Mol Biol 290, 253–266.
33 Eisenthal R, Danson MJ & Hough DW (2007) Cata-
lytic efficiency and k
cat
⁄ K
M
: a useful comparator?
Trends Biotechnol. 25, 247–249.
34 Li M, Binda C, Mattevi A & Edmondson DE (2006)
Functional role of the ‘‘aromatic cage’’ in human
monoamine oxidase B: structures and catalytic pro-
perties of Tyr435 mutant proteins. Biochemistry 45,
4775–4784.
35 Pace CN, Shirley A & Thomson JA (1989) Protein
Structure. A Practical Approach. IRL Press, Oxford,
New York and Tokyo.

Sci USA 94, 8411–8416.
44 Gruez A, Pignol D, Zeghouf M, Coves J, Fontecave M,
Ferrer JL & Fontecilla-Camps JC (2000) Four crystal
structures of the 60 kDa flavoprotein monomer of the
sulfite reductase indicate a disordered flavodoxin-like
module. J Mol Biol 299, 199–212.
45 Biterova EI, Turanov AA, Gladyshev VN & Barycki JJ
(2005) Crystal structures of oxidized and reduced mito-
chondrial thioredoxin reductase provide molecular
details of the reaction mechanism. Proc Natl Acad Sci
USA 102, 15018–15023.
46 Marohnic CC, Crowley LJ, Davis CA, Smith ET &
Barber MJ (2005) Cytochrome b5 reductase: role of the
si-face residues, proline 92 and tyrosine 93, in structure
and catalysis. Biochemistry 44, 2449–2461.
47 Tejero J, Martinez-Julvez M, Mayoral T, Luquita A,
Sanz-Aparicio J, Hermoso JA, Hurley JK, Tollin G,
Gomez-Moreno C & Medina M (2003) Involvement of
the pyrophosphate and the 2¢-phosphate binding regions
of ferredoxin-NADP
+
reductase in coenzyme specific-
ity. J Biol Chem 278, 49203–49214.
48 Breinlinger EC, Keenan CJ & Rotello VM (1998) Mod-
ulation of flavin recognition and redox properties
through donor atom–pi interactions. J Am Chem Soc
120, 8606–8609.
49 Rotello VM (1998) The donor atom–pi interaction of
sulfur with flavin. A density functional investigation.
Heteroatom Chem 9, 605–606.

57 Lu G, Campbell WH, Schneider G & Lindqvist Y
(1994) Crystal structure of the FAD-containing frag-
ment of corn nitrate reductase at 2.5 A
˚
resolution: rela-
tionship to other flavoprotein reductases. Structure 2,
809–821.
58 Nishida H, Inaka K, Yamanaka M, Kaida S, Kobay-
ashi K & Miki K (1995) Crystal structure of NADH-
cytochrome b5 reductase from pig liver at 2.4 A
˚
resolu-
tion. Biochemistry 34, 2763–2767.
59 Shirabe K, Yubisui T, Nishino T & Takeshita M (1991)
Role of cysteine residues in human NADH-cytochrome
b5 reductase studied by site-directed mutagenesis. Cys-
273 and Cys-283 are located close to the NADH-bind-
ing site but are not catalytically essential. J Biol Chem
266, 7531–7536.
60 Batie CJ & Kamin H (1984) Ferredoxin:NADP
+
oxido-
reductase. Equilibria in binary and ternary complexes
with NADP+ and ferredoxin. J Biol Chem 259, 8832–
8839.
61 Barik S (1996) Site-directed mutagenesis in vitro by meg-
aprimer PCR. In In Vitro Mutagenesis Protocols (Trower
MK, ed.), pp. 203–215. Humana Press, Totowa, NJ.
62 Aiyar A & Xiang YLJ (1996) Site-directed mutagenesis
using overlap extension PCR. In In Vitro Mutagenesis

site of ferredoxin-NADP reductase. Biochim Biophys
Acta 445, 14–24.
Enzyme efficiency modulated by amino acid volume M. A. Musumeci et al.
1366 FEBS Journal 275 (2008) 1350–1366 ª 2008 The Authors Journal compilation ª 2008 FEBS