In most cyanobacteria, and some algae under low iron conditions, flavo-doxin Fld an FMN flavoprotein, in particular Keywords electron transfer; ferredoxin; ferredoxin– NADP+reductase; flav
Trang 1Structural and mechanistic aspects of flavoproteins:
photosynthetic electron transfer from photosystem I
Milagros Medina
Departamento de Bioquı´mica y Biologı´a Molecular y Celular and BFIF, Universidad de Zaragoza, Spain
Introduction
Many electron-transfer reactions in biological systems
depend on redox chains that involve flavoproteins [1]
In these chains, questions remain regarding not only
the mechanisms of electron transfer and hydride
transfer, but also the role that flavins might play in
these events The primary function of photosystem I
(PSI) is to reduce NADP+to NADPH, which is then
used in the assimilation of CO2 [2,3] In plants, this occurs via reduction of the soluble [2Fe–2S] ferre-doxin (Fd) by PSI Subsequent reduction of NADP+
by Fdrd is catalysed by FAD-containing ferredoxin– NADP+ reductase (FNR) [4] In most cyanobacteria, and some algae under low iron conditions, flavo-doxin (Fld) (an FMN flavoprotein), in particular
Keywords
electron transfer; ferredoxin; ferredoxin–
NADP+reductase; flavodoxin; hydride
transfer; NAD(P) + ⁄ H; photosystem I;
protein–flavin complexes; protein–protein
and protein–ligand interaction; redox
potential regulation
Correspondence
M Medina, Departamento de Bioquı´mica y
Biologı´a Molecular y Celular, Facultad de
Ciencias, Pedro Cerbuna 12, Universidad de
Zaragoza, 50009-Zaragoza, Spain
Fax: +34 976 762123
Tel: +34 976 762476
E-mail: mmedina@unizar.es
(Received 28 January 2009, revised 22
April 2009, accepted 4 May 2009)
doi:10.1111/j.1742-4658.2009.07122.x
This minireview covers the research carried out in recent years into differ-ent aspects of the function of the flavoproteins involved in cyanobacterial photosynthetic electron transfer from photosystem I to NADP+,
flavodox-in and ferredoxflavodox-in–NADP+ reductase Interactions that stabilize protein– flavin complexes and tailor the midpoint potentials in these proteins, as well as many details of the binding and electron transfer to protein and ligand partners, have been revealed In addition to their role in photosyn-thesis, flavodoxin and ferredoxin–NADP+ reductase are ubiquitous fla-voenzymes that deliver NAD(P)H or low midpoint potential one-electron donors to redox-based metabolisms in plastids, mitochondria and bacteria They are also the basic prototypes for a large family of diflavin electron transferases with common functional and structural properties Under-standing their mechanisms should enable greater comprehension of the many physiological roles played by flavodoxin and ferredoxin–NADP+ reductase, either free or as modules in multidomain proteins Many aspects
of their biochemistry have been extensively characterized using a combina-tion of site-directed mutagenesis, steady-state and transient kinetics, spec-troscopy and X-ray crystallography Despite these considerable advances, various key features of the structural–function relationship are yet to be explained in molecular terms Better knowledge of these systems and their particular properties may allow us to envisage several interesting applica-tions of these proteins beyond their physiological funcapplica-tions
Abbreviations
2¢P-AMP, 2¢-phospho-AMP portion of NADP + ⁄ H; CTC, charge-transfer complex; Fd, ferredoxin; Fdox, oxidized ferredoxin; Fdrd,reduced ferredoxin; Fld, flavodoxin; Fldhq,hydroquinone flavodoxin; Fldox,oxidized flavodoxin; Fldsq,semiquinone flavodoxin; FNR, ferredoxin–NADP + reductase; FNRox, oxidized ferredoxin–NADP+reductase; FNRsq, semiquinone ferredoxin–NADP+reductase; NMN, nicotinamide
mononucleotide portion of NAD(P) + ⁄ H; PSI, photosystem I.
Trang 2Fldsq⁄ Fldhq, substitutes for the Fdox⁄ Fdrd pair in this
reaction [5,6]
PSIrdþ Fldsq! PSI + Fldhq
NADPþþ 2Fldhq¡
FNRNADPH + 2Fldsq Two Fldsq molecules transfer two electrons from
two PSI molecules to one FNR FNR becomes fully
reduced through formation of the intermediate,
FNRsq, and later transfers both electrons
simulta-neously to NADP+[7,8] During this process, the Fld
molecule must move between its docking site in PSI
and the docking site in FNR, and the formation of
short-life transient complexes is required Mutational
and structural studies that characterize such
interac-tions are the subject of this minireview Studies on
proteins from the cyanobacterium Anabaena are
pref-erentially considered because it is the most thoroughly
investigated system containing both Fld and FNR
(hereafter, AnFld and AnFNR) [5,8]
PSI architecture
Cyanobacterial PSI exists as either a monomer or tri-mer embedded in the thylacoid membrane It contains
12 subunits, 96 chlorophylls, 22 carotenoids, 2 phyllo-quinones, 4 lipids and 3 [4Fe–4S] clusters per mono-mer [9,10] Protein subunits with more relevant functions are conserved between plant and cyanobacte-ria [11] When light strikes one of the antenna chloro-phylls and the exciton is transferred to the pair of chlorophylls in the PSI reaction centre, charge separa-tion occurs Low-potential electrons are transferred across the membrane by a chain that ends in three [4Fe–4S] clusters, FX, FA and FB FX is coordinated
by cysteines located in both of the large PSI subunits, PsaA and PsaB, via a loop that also plays a role in the attachment of PsaC [12] PsaC, PsaD and PsaE are located at the cytosolic site (Fig 1A) [2,7,13–16] PsaC carries the terminal FA and FB clusters After binding
of the protein carrier to this PSI site, the electron is
A
F B
F A
R39
(PsaE)
K106 (PsaD) K34
(PsaC)
I59
W57 N58 Y94
D146
D90
C
K2, K3 T56, W57, N58
Y94
E61, D65 E67 E16
E20 T12
D150 D144, E145
I59
I92 D96, N97
B
R264 K75
L76 L78
E301
K72 R16
Y303
Fig 1 (A) Molecular surface with the electrostatic potential of the putative Fd ⁄ Fld-binding site of Synechococcus elongatus PSI (PDB code 1jb0) [10] The surface is transparent to show the internal position of the FA and FB centres (represented as spheres) in the PsaC subunit of PSI (S elongatus numbering is used) (B) Molecular surface with the electrostatic potential of Anabaena FNR at the Fld-docking site (PDB code 1que) [62] The FAD group is drawn in CPK with carbons shown in orange PSI and FNR positions for the interaction with the protein carrier are indicated (C) Molecular surface with electrostatic potential of Anabaena Fld (PDB code 1flv) [28] Detail of residues in the close FMN environment in the oxidized Fld O-down conformation is shown on the right FMN is drawn in CPK (balls or sticks), with carbons shown in orange Figure 1(A,B) is reproduced from the supplementary material in Gon˜i et al [48]
Trang 3transferred from FB to Fld (or Fd), which
subse-quently leaves the PSI site bringing the electron to
FNR
Flavins: key cofactors in protein
electron transfer reactions
Free flavins stabilize very little of their one-electron
reduced form, the semiquinone, because the midpoint
potential for reduction of the oxidized state to the
semiquinone, Eox⁄ sq, is more negative than that for
reduction of the semiquinone to the hydroquinone,
Esq⁄ hq [17] Binding of FAD or FMN to the
apopro-tein usually displaces Eox⁄ sq to a less negative value,
whereas Esq⁄ hqshifts to a more negative value,
stabiliz-ing the semiquinone [1,18,19] This allows flavoproteins
to function as key intermediates at the interface
between one- and two-electron transfers [20,21] and
allows flavoproteins to participate in many biological
processes [19,21,22]
Flavodoxin
The redox activity of Fld derives from its FMN
cofac-tor The Fld semiquinone is exceptionally stable and its
midpoint potentials are quite negative This is a direct
consequence of the differing stability of the oxidized,
semiquinone and reduced ApoFld:FMN complexes
[23–25] In AnFld, the values are Eox⁄ sq=)266 mV
and Esq⁄ hq=)439 mV at pH 8.0 and 25 C, with a
maximum stabilization of the semiquinone of 96%
[25,26] Therefore, Fld is proposed to replace Fd
(Eox ⁄ rd=)384 mV) by exchanging electrons between
its semiquinone and hydroquinone states [5,8]
Oxidized Flds fold into a five-stranded parallel
b sheet sandwiched between five a helices [27–31], with
the FMN group located at the edge of the globular
protein and its two isoalloxazine methyls solvent
acces-sible (Fig 1C) The polypeptide chain in the loops
sur-rounding amino acid residue 50 and amino acid
residue 90 (50’s and 90’s loops) make close contact
with the isoalloxazine and modulate its reduction
properties [24–26,32–37] The H-bond network
observed in the AnFld FMN environment is conserved
in Flds across species [38], but specific interactions
around the flavin vary [39–41] The 90’s loop usually
provides a Tyr stacked against the FMN si-face (Y94
in AnFld) which makes a large contribution to the
midpoint potential [23–25] The residue from the 50’s
loop that stacks at the re-inner face is commonly a
Trp (W57 in AnFld) [20,28,29,39,42,43], but
nonaro-matic residues (L, H, M or A) have also been found
[29,40,44,45] Both residues ensure that the flavin is in
an electronegative environment which allows tight FMNhq binding while making formation of its anion thermodynamically unfavourable [24,25]
In several Flds, rearrangement of the peptide bond equivalent to 58–59 in AnFld allows a main chain car-bonyl to flip from an ‘O-down’ conformation to an
‘O-up’ conformation In the ‘O-up’ conformation, an H-bond occurs between this carbonyl and N(5)H from the neutral semiquinone [46,47] In Anacystis nidulans Fld, the flip involves breaking a weak H-bond present
in the oxidized state between the FMN N(5) and the
NH of V59, in favour of a stronger H-bond between the carbonyl of N58 (‘O-up’ conformation) and FMN N(5)H [41] The semiquinone states of A nidulans and Anabaena Flds are less stable than those from other species because the semiquinone H-bond with the CO
of Asn is weaker than the bond formed with the sma-ller Gly, and because of the presence in the oxidized state of a N(5)–HN59 H-bond that is absent in other Flds Replacements at T56, W57, N58, I59 and E61
in AnFld regulate the ability of the N58–I59 peptide
to H-bond with the N(5) or N(5)H, and modulate the energy of its conformational change [25,26,48] There-fore, the backbone rearrangements of N58–I59 provide a versatile device for modulating the strength
of FMN binding and Eox⁄ sq and Esq⁄ hq in AnFld [25,26,48]
Fld has a large excess of acidic residues which pro-duce a strong dipole that orients its negative end towards the FMN isoalloxazine The importance of electrostatic repulsion in the control of Eox⁄ sq, and particularly of Esq⁄ hq, has also been demonstrated with several Flds [25,37,41,48–53] Electron nuclear double resonance and 1D and 2D electron spin echo envelope modulation spectroscopies applied to AnFldsq also led
to assignment of the interaction parameters of N(1), N(3), H(5), H(6), CH3(8) and N(10) with the electron spin [54,55] Analysis of mutants indicated that the stacking of a bulky residue at the re-face of the flavin decreases the electron-spin density in the benzene ring, whereas an aromatic residue at the si-face increases the spin density at N(5) and C(6) [56]
The first structure obtained for a photosynthetic FNR was from spinach (spFNR) spFNR folds in two domains, one of which presents a noncovalently bound FAD molecule and the other binds NADP+ [57,58] Structures from other species have also been reported [59–62] The FAD-binding domain in AnFNR includes residues 1–138 and is made up of six antiparallel b strands arranged in two
Trang 4perpendicu-lar b sheets, with a short a helix at the bottom and
another a helix and a long loop that is maintained by
a small two-stranded antiparallel b sheet at the top
(Fig 1B) The NADP+-binding domain includes
resi-dues 139–303 and consists of a core of five parallel
b strands surrounded by seven a helices [62] FAD is
bound outside the antiparallel b barrel, and its
isoal-loxazine lies between two tyrosines, Y79 and the
C-terminal Y303 in AnFNR
The two one-electron midpoint potentials of the
fla-vin in FNRs are close to each other, therefore these
proteins stabilize only 10–20% of the maximal amount
of semiquinone [63,64] An Eox⁄ hq value of )325 mV
has been reported for recombinant AnFNR at pH 8.0
and 10C [63,65] Despite differences in buffer,
tem-perature and pH, the reported values for AnFNR are
in good agreement, however, slightly more negative
midpoint potentials are reported for other species
[64,66,67] Replacements for Y79 have been reported
for the pea and spinach enzymes, Y89 and Y95,
respectively, suggesting that the aromaticity of this
res-idue is essential for FAD binding and that H-bonding
through the Tyr-OH is involved in the correct
posi-tioning of the NADP+substrate for efficient catalysis
[61,68] Replacement of Y303 with Ser and of E301
(situated at the active site) with Ala shift the flavin
midpoint potential to considerably less negative values,
although semiquinone stabilization is severely
ham-pered, introducing constraints into one-electron
trans-fer processes [26,63] In addition, L76, L78 and
particularly K75 at the FAD-domain modulate Eox⁄ hq
[63,65] FNR:Fd complexation correlates with the
AnFd midpoint potential becoming 15 mV more
nega-tive and that of AnFNR becoming 27–40 mV less
neg-ative [69] As seen with the spinach proteins,
complexation makes electron transfer
thermodynami-cally more favourable [70] Similarly, the midpoint
potential for reduction of NADP+ in complex with
FNR is 40 mV less negative than that of the free
NADP+⁄ NADPH pair [66,71] Assignment of
hyper-fine couplings to nuclei of the isoalloxazine
semi-quinone have also been reported for AnFNRsq and
pFNRsq[72] These studies indicated that the net effect
of the C-terminal Tyr is withdrawal of electron density
from the benzene ring towards the pyrazine ring,
plac-ing the accepted electron nearer to a site where it can
best be neutralized by protonation – the N5 position
Electron transfer from PSI to
flavodoxin
Fd and Fld differ in size and in the chemical nature of
their redox cofactors There is no sequence homology
between them, but structural alignment based on their surface electrostatic potentials shows cofactor superpo-sition in the region where both proteins accumulate the negative end of their molecular dipole moments [73] Their biding site on PSI was analysed by studying the kinetic behaviour of site-directed mutants and by electron microscopy on cross-linked complexes [12,15,16,74,75] The cytosolic subunits of Synechococ-cus elongatus (Sy) PSI, PsaC, PsaD and PsaE, and the extrinsic loop of PsaA, present a positively charged surface potential (Fig 1A) and are proposed to partici-pate in electrostatic docking of the negatively charged
Fd or Fld (Fig 1C) [13,16] The PsaC subunit cannot
be deleted without loss of PSI activity because it car-ries the FBdonor [14] PsaD contributes to the electro-static steering of Fd toward its binding site [76,77], whereas several roles are proposed for PsaE [78,79] K35 from the PsaC subunit of Chlamydomonas rein-hardtiiis critical for the interaction and, therefore, effi-cient electron transfer [15,80] The residues of PSI and
Fd facing each other have not yet been identified, with the exception of K106 in SyPsaD, which interacts with E93 in SyFd (E95 in AnFd) [77,81] The scarce data about the interaction of Fld and PSI suggest similar functions for PsaC, PsaD and PsaE, but there are insufficient data to propose a Fld docking site [13,16,82,83] Analysis of different AnFd and SyFd mutants revealed that E31, R42, T48, D67, D68, D69, E94, and particularly D59, D62 and E95 (AnFd num-bering) influence electron transfer and are involved in either the binding process or electron transfer itself [84–86]
Although the Fldsq⁄ Fldhq pair is involved in shut-tling electrons between PSI and FNR, a physiological role for the Fldox⁄ Fldsq pair cannot be precluded [7,14] Reduction of AnFldox to the semiquinone state
by PSI has been a useful model with which to ana-lyse the interaction forces and electron transfer parameters involved in the physiological reaction [48] Wild-type AnFld forms a transient PSI:Fldox complex prior to electron transfer [87] Site-directed mutagene-sis has been used to find the role of specific AnFld side chains in the interaction and electron transfer with PSI [53,84,88–90] Many of these Fld mutants (T12V, E16Q, T56G, W57 replaced by K, R, F, L, A and Y, I59 replaced by A and K, Y94 replaced by A and F, N97K, I59A⁄ I92A and I59E ⁄ I92E) accept electrons from PSI following transient complex for-mation [53,87,90,91] For some (T12V, W57Y and Y94F), ket was lowered considerably, suggesting that the complex is not optimal for electron transfer Changes in the midpoint potential are proposed to be responsible for the W57Y Fld behaviour [90], but
Trang 5changes in the electrostatics within the FMN
environ-ment which favour a complex less efficient for
elec-tron transfer explain the results for the other mutants
[88] By contrast, ket was enhanced considerably for
most of the remaining mutants More or less negative
values of Eox⁄ sq for these Flds compared with
wild-type Fld might influence this kinetic behaviour by
decreasing or increasing, respectively, the driving
force of the reaction In addition, analysis of multiple
charge reversal Fld mutants has recently indicated
that changes in the orientation and magnitude of the
molecular dipole moment have a critical effect on
electron transfer [48]
However, for some Fld mutants (E20K, T56S,
W57E, N58C, N58K, I59E, E61A, E61K, I92 replaced
by A, E and K and D96N and also some multiple
charge reversal mutants), electron transfer from PSI to
Fldox takes place via a collisional-type mechanism
[48,53,91] Noticeably, some mutants show rates higher
than those obtained for the wild-type in all the
PSI⁄ Fld ratios assayed [88] These effects are not
related to a change in Fld midpoint potentials They
might be interpreted as being caused by a modification
in the accessibility of the flavin, but more generally
can be explained by a conformational change in the
orientation of the interacting PSI and Fld surfaces,
leading to a smaller edge-to-edge distance between FB
and the flavin ring [48] However, for other mutations,
rates at a given concentration are lower than the
corre-sponding rate for wild-type Fld Because, in general,
the introduced mutations are not in the direct
isoall-oxazine coordination, this is unlikely to be because of
differences in the structural FMN environment, but
rather because the orientation between the protein
dipoles is not optimal for electron transfer or because
of a change in the electrostatic potential of the protein
[48]
In conclusion, subtle changes in the isoalloxazine
environment influence Fld binding ability and
modu-late the electron-exchange process by producing
differ-ent oridiffer-entations and distances between redox cdiffer-entres
Observations indicate that these side chains contribute
to the orientation of AnFld on the PSI, producing a
wild-type complex that is not the most optimal for
electron transfer Mutation of these residues changes
Fld surface topology, and the module and orientation
of the molecular dipole, contributing to their altered
behaviour Mutational studies on I59 and I92 AnFld
indicate that their hydrophobicity is far from critical,
suggesting that either hydrophobic interactions do not
play a crucial role or that the hydrophobic surface of
Fld must be provided by the solvent-exposed portion
of FMN
Electron flow from flavodoxin to
Interaction and electron transfer between Fld and FNR
Crystal structures of Fd:FNR complexes have been reported for Anabaena [92] and maize leaf [93] Despite the two structures exhibiting a different orientation for
Fd [94], the [2Fe–2S] cluster lies close to the FAD of FNR in both Mutants of the two partners have also contributed to the identification of residues essential for complex formation and electron transfer [5,65,95– 102] Both electrostatic and hydrophobic interactions play an important role in the association and dissocia-tion processes in these complexes [5,8,92] K72 (K88 in spFNR), K75, L76, L78 and V136 in AnFNR are key for the interaction with Fd [5,97,100,103] Similarly, residues on the AnFd surface have a moderate effect
on complex stability and electron transfer with the reductase, with E94, F65 and S47 being crucial [69,104–106] Despite proposals that FNR interacts with Fld and Fd using the same region [100,101], in general, replacing some FNR residues had more dras-tic effects in processes involving Fld, suggesting that the individual residues do not contribute equally to complex formation with both partners This was the case for R16, K72, and particularly K75 [65,89,100,107] In addition, K138 and R264 in the NADP+-binding domain of AnFNR are more impor-tant in establishing interactions with AnFld than with AnFd [100,108] Moreover, although removal of the E139 AnFNR negative charge has a deleterious effect
on electron transfer reactions with AnFd, it appears to enhance electron transfer with AnFld [109] Electron transfer with Fld is severely diminished upon the intro-duction of negatively charged side chains at L76, L78 and V136 in AnFNR [89] Therefore, these nonpolar residues participate in the establishment of interactions with both AnFld and AnFd With this in mind, it was expected that one or more negatively charged or hydrophobic residues on the Fld surface would interact with some of the above specified residues on FNR
A number of AnFld variants containing replace-ments, either at the putative interaction surface with FNR or in the FMN environment, have been analy-sed None of the E16, E20, T56, I59, E61, D65, I92, Y94, D96 and N97 positions is key, but they do con-tribute cooperatively to the orientation and strengthen-ing of the FNR:Fld complexes [53,88] Simultaneous replacement of I59 and I92 indicated that they are not involved in crucial specific interactions [53,89] T12, W57 and N58 seem to be more important in the
Trang 6inter-action [53,88–90] and, in addition, FMN might be
crit-ical [53] It is somehow unclear whether the residues
stacking the FMN ring, W57 and Y94, truly affect
protein binding, or if the altered electron transfer
properties in mutations be explained in terms of
altered flavin accessibility and⁄ or thermodynamic
parameters [24,25,90] Therefore, electron transfer
pro-cesses between FNR and Fld resulted in the
modula-tion (either negatively or positively) of some mutants,
but in no case was electron transfer prevented
Consid-ering also that the crystallographic structure of the
AnFNR:AnFld interaction appears highly elusive, these
observations suggest that the interaction of Fld with
FNR is less specific than that of Fd
Theoretical models of this interaction, one based on
the rat NADPH–cytochrome P450 reductase (CPR)
structure [95] and the other obtained by docking [110],
confirm that a mutual orientation between FNR and
Fld, similar to the corresponding binding domains in
CPR, is highly probable and places the redox centres
closer than observed in the FNR:Fd structures
[92,93,95] The docking model fits well with the
experi-mental data, showing that all Fld residues important
for the interaction with FNR are in contact with the
FAD cofactor Different interface propensities for the
same FNR residues, with either Fd or Fld, are
consis-tent with experimental observations indicating that,
although FNR uses the same site for interaction with
Fd and Fld, each individual residue does not partici-pate to the same extent in interactions with each of the partners [100] This is in agreement with the fact that, although multiple chemical modifications produced Flds less suitable for electron transfer [111], site-direc-ted mutagenesis has not revealed any residues critical for the interaction with FNR [53,88–90] Replacement
of the few Fld positions, T12, W57, N58 and Y94, with a high interface propensity produced opposing effects: some Fld:FNR complexes can be either weaker
or stronger and less optimal for electron transfer than those with wild-type Fld, but others can appear more optimal for a particular electron transfer process Docking suggests that wild-type Fld could adopt different orientations on the FNR surface without sig-nificantly altering the distance between the methyl groups of FAD and FMN (Fig 2A) This might explain why subtle changes in the Fld still produce functional complexes Moreover, the enhanced or hindered reactivity can also be explained if there is a single orientation of Fld in the complex that is retained and changes either the overall interaction or the electron transfer parameters Recent analysis of multiple charge-reversal mutations on the Fld surface concluded that interactions do not rely on a precise complementary surface in the reacting molecules In
E301
L263 R264
Y79 S80
B A
Fig 2 Proposed interactions in the Anabaena FNR active site leading to electron transfer ⁄ hydride transfer (A) Model of a ternary Fld:FNR:NADP + complex FNR is shown as a grey surface with the atoms of Y303 CPK coloured, with C shown in violet The position of NADP + on the FNR surface corresponds to the X-ray FNR:NADP + complex (PDB code 1gjr), in which Y303 prevents stacking of the flavin and nicotinamide rings The figure also shows several positions determined by docking of Fld onto FNR (Fld in green, light orange and pink correspond to docking solutions ranked 1, 3 and 5 respectively) (B) Detail of the proposed FNR active site centre in a model of ternary com-plexes competent for hydride transfer FNR active site residues are given as sticks and CPK coloured, with C in grey The nicotinamide por-tion of the coenzyme presents the posipor-tion derived from the structure in complex with Y303S FNR (PDB code 2bsa) [117] The dotted surface around the nicotinamide indicates the position of Y303 in wild-type FNR For both structures, FAD in FNR, FMN in Fld and NADP+ are show as sticks and are CPK coloured, with carbons in yellow, orange and pink, respectively.
Trang 7fact, analysis indicates that the initial orientation,
dri-ven by alignment of the Fld molecule dipole moment
with that of FNR, contributes to the formation of
sev-eral alternative binding modes competent for the
effi-cient electron transfer reaction [48] Similar behaviours
have been reported in other electron transfer systems,
where dynamic ensembles, as opposed to single
confor-mations, contribute to the electron transfer process
[112,113]
Electrostatic nonspecific interactions as major
determinants of the efficient interaction between
Fld and its counterparts
Some mutations appear to favour single orientations
which improve the association and electron transfer
with a particular partner and native Fld complexes are
not the most optimal for electron transfer Such
obser-vation, in agreement with docking analysis [110],
sug-gests that the flavin atoms might be mainly involved in
the interaction and be solely responsible for electron
transfer Therefore, subtle changes in the isoalloxazine
environment not only influence Fld-binding abilities,
but also modulate the electron transfer process by
pro-ducing different orientations and distances between the
redox centres This further confirms that Fld interacts
with different structural partners through nonspecific
interactions, which in turn decrease the potential
effi-ciency that could be achieved if unique and more
favourable orientations were produced with a reduced
number of partners During Fld-dependent
photosyn-thetic electron transfer, the Fld molecule must move
from its docking site in PSI to that in FNR In vivo,
the formation of transient complexes of Fld with PSI
and FNR is useful, but not critical, during this process
to promote electron transfer and avoid the reduction
of oxygen by the donor centres [7,8,14,48,53] Thus,
electrostatic alignment appears to be one of the major
determinants of the orientation of Fld on the partner
surface The fact that simultaneous replacement on the
Fld surface did not hinder or enhance processes with
PSI and FNR also suggests a different interaction
mode with each partner [48]
Interaction of FNR with the NADP+coenzyme
and the hydride transfer event
Once the FAD cofactor of FNR has accepted two
electrons, they have to be transferred to NADP+ The
FNR protein portion has a dual role in this process
by: (a) modulating the FAD midpoint potential to a
value that makes the hydride transfer reversible, and
(b) providing the environment for an efficient
encoun-ter between the N5 of the flavin and the C4 of the nic-otinamide FNR is highly specific for NADP+⁄ H versus NAD+⁄ H and different studies have established
a role for several FNR residues in determining coen-zyme binding, specificity and enzymatic efficiency [114–120] Three FNR regions appear to be mainly responsible for the interaction: 2¢-phospho-AMP (2¢P-AMP) and pyrophosphate of the NADP+⁄ H binding sites, and the position occupied by the C-terminal resi-due where the nicotinamide portion of NADP+ (NMN) is proposed to bind for hydride transfer [114– 117,119] S223 and Y235 at the AnFNR 2¢P-AMP site are critical in determining the specificity and efficient coenzyme orientation [99,115,121] The 155–160 and 261–268 loops, which accommodate the coenzyme pyrophosphate portion, also confer specificity and the volume of residues in the latter loop fine-tunes FNR catalytic efficiency [114,116,119,120] R100 (K166 in spFNR), situated at the FAD-binding domain, allows its guanidinium group to H-bond to the NADP+ pyrophosphate, providing the necessary flexibility to address the NMN moiety of NADP+ towards the active site [99,108,114] Finally, the Tyr at the si-face contributes to the correct positioning of the substrate NADP+ [68], whereas the C-terminal Tyr at the re-face is surely critical for modulating NADP+⁄ H biding affinity and selectivity [117,118,122–126] Structural studies have allowed us to postulate a stepwise mechanism in which the nucleotide must bind
to a bipartite site [59,62,114,117,127] The first stage is recognition of the 2¢P-AMP moiety [62] The interme-diate state represents a narrowing of the cavity to match the adenine and the pyrophosphate, whereas the nicotinamide is placed in a pocket near the FAD [114] However, in this arrangement, the C-terminal Tyr pre-vents interaction of nicotinamide with the isoalloxazine and its energetically unfavourable displacement is then expected if the hydride transfer optimal interaction is
to be achieved (Fig 2A) [59,114,118,127] Only FNR variants in which the C-terminal Tyr has been replaced produced structures with a rearrangement between fla-vin and nicotinamide that was compatible with hydride transfer, for example Y303S in AnFNR, and Y308W
or Y308S in pFNR (Fig 2B) [59,117] These FNRs improved the affinity for NADP+ and produced a close interaction between flavin and nicotinamide [117,118] However, because of this strong binding, they show low catalytic efficiency [59,117,118] All the data indicate a fine-tuning of the FNR efficiency pro-duced by minor structural changes in the regions involved in coenzyme binding [117,119]
Reduction of NADP+ by FNRhq occurs by a formal hydride transfer from the flavin anionic
Trang 8hydro-quinone to the nicotinamide In vitro, the reaction
takes place via a two-step mechanism, in which the
first observed process is related to formation of the
FNRhq–NADP+charge-transfer complex (CTC-2) via
an intermediate Michaelis–Menten complex (MC-2),
followed by hydride transfer to produce an equilibrium
mixture of the CTC-2 and FNRox–NADPH (CTC-1)
CTCs Both CTCs are also detected for the reverse
reaction, although the mechanism has some
differ-ences Spectroscopic properties for these CTCs and
their hydride transfer rates for interconversion have
been estimated [66,121] In AnFNR, CTC-2
accumu-lates during the reaction and at equilibrium, whereas
CTC-1 evolves rapidly into other FNR states
Forma-tion of these CTCs appears necessary for efficient
hydride transfer and the relative conformation and
ori-entation of FNR and NADP+⁄ H during the
interac-tion are critical [121] Hydride transfer in systems
involving flavins and pyridine nucleotides is highly
dependent on the approach and colinear orientation
of the N5 of the flavin, the hydride to be transferred,
and C4 of the nicotinamide In FNR, displacement
of the C-terminal Tyr appears to be required for the
interaction to occur [117,118] The data reported to
date suggest good agreement between CTC formation
and hydride transfer rates [96,121] However, this
hypothesis is based on a limited data set and
preli-minary characterization of the process for Y303S
AnFNR suggests that, at least for this mutant, there
might not be a direct correlation [128] Therefore, it
may be that for wild-type FNR the most favourable
orientation between the nicotinamide and the flavin
might present considerably less overlap of the rings
than in the mutant structure (Fig 2B), and other
relative orientations that maintain C4, H and N5
colinearity might account for the efficient hydride
transfer Further work is needed to clarify these
points
Most data indicate a similar interaction in higher
plant and cyanobacterial FNRs, but NADP+
disor-ders their spectra differently [71] The different
spec-tra of higher plant FNRs show a peak at 510 nm
indicative of a stacking interaction between the
nico-tinamide and the isoalloxazine, which is not seen in
cyanobacterial FNRs [71] Spectra obtained upon
addition of NADP+ to C-terminal Tyr mutants
pro-duce prominent peaks in AnFNR and pFNR,
sug-gesting greater nicotinamide occupancy of the active
site [117] Thus, although the AnFNR UV spectra
and electron spin density distribution of AnFNRsq
are perturbed by NADP+ [55,72], lower nicotinamide
occupancy of the active site is expected in AnFNR
relative to higher plant FNRs Therefore, differences
in nicotinamide binding to the active sites cannot be discounted
FNR catalytic site The structure of the catalytically competent FNR:NADP+conformation indicates that in AnFNR, S80, C261, E301 and Y303 constitute the FNR cata-lytic site [59,117] Y303 plays distinct and complemen-tary roles during the catalytic cycle by lowering the affinity for NADP+⁄ H to levels compatible with turn-over, by stabilizing the flavin semiquinone required for electron splitting and by modulating the electron trans-fer rates [59,117,118] Moreover, a role in providing adequate orientation between the reacting rings might
be envisaged [128] S80 and C261 contribute to the efficient flavin:nicotinamide interaction through the production of CTCs during hydride transfer (in spFNR S96 and C272) [96,129] This Ser also contrib-utes to semiquinone stabilization, and the volume of the Cys residue modulates the enzyme catalytic effi-ciency [119] E301 has been studied in AnFNR and spFNR (E312) [98,130] Structural and functional dif-ferences were found when the same mutants were pro-duced in both species, again suggesting differences in their mechanisms [131] E301 was more critical for sta-bilization of the semiquinone and midpoint potential
in AnFNR [63,98,130] Studies in spFNR concluded that E301 does not act as a proton donor [98], but whether it transfers protons in AnFNR could not be determined [130] In fact, in the AnFd:AnFNR and AnFld:AnFNR dockings, the carboxylic group of E301
is no longer exposed to solvent and it is one of the res-idues with highest propensity for being at the interface [110] Similar observations have been extracted from the Fd:FNR crystal structure [92] This suggests a pos-sible pathway for proton transfer between the external medium and the AnFNR isoalloxazine N5 via S80 [58,62] In addition, in both enzymes, this residue is critical for proper binding of the nicotinamide to the active centre, CTC stabilization and efficient flavin reduction by NADPH [98,130]
FNR catalytic cycle: the ternary complex The ability of FNR, Fd and NADP+ to form a ter-nary complex is fully accepted, indicating that NADP+is able to occupy a site on FNR without dis-placing Fd [70,71,114,127], and similar behaviour also applies for Fld [132] During catalysis, the order in which substrates are added is not important, although
Fd and Fld lower the affinity for NADP+and occupa-tion of the NADP+-binding site weakens the Fd:FNR
Trang 9and Fld:FNR complexes [70,132] The two binding
sites are not completely independent, and the overall
reaction is proposed to work in an ordered
two-sub-strate process, with the pyridine nucleotide binding
first [70,133] Complex formation between Fdrd and
FNR:NADP+ was found to increase the rate of
elec-tron transfer by facilitating the rate-limiting step of the
process – dissociation of the product (Fdox) [127]
Thus, in the system involving Fd, negative
cooperativi-ty in the ternary interaction is translated into positive
cooperativity at the kinetic level However, some key
features of the process remain to be explained in
molecular terms because the expected molecular
move-ments are not apparent and, although reversibile,
dif-ferent mechanisms seem to apply in each direction
[127] Binding equilibrium and steady-state studies in
wild-type and mutant proteins envisage similar
mecha-nisms for Fld [5,90,130,132] Fld, reduced to Fldhq by
PSI, will cycle between Fldhqand Fldsqupon passing a
single electron to FNR [5] Fast kinetic methods have
been used to analyse binding and electron transfer
between FNR and Fld, but to date the reactions have
been followed at single wavelengths and have not
involved NADP+⁄ H [90] Therefore, it remains to the
future to better evaluate the intermediate and final
spe-cies of the equilibrium mixture in the electron transfer
process in the binary Fld:FNR and ternary
Fld:FNR:NADP+electron transfer systems
Flds and FNRs in nonphotosynthetic
organisms and as building blocks for
more complex proteins
Flds are electron transfer proteins involved in a variety
of photosynthetic and nonphotosynthetic reactions in
bacteria [134] In eukaryotes, only a few Flds have
been reported [135–137], but a descendant of the Fld
gene helps to build multidomain proteins [134,138] A
photosynthetic function was first proposed for FNR,
but flavoproteins with FNR activity have been
described in chloroplasts, phototropic and
heterotro-phic bacteria, apicoplasts, and animal and yeast
mito-chondria [139] Two unrelated families of proteins can
be found in these enzymes: the plant type and the
glu-tathione reductase type [126] Based on their structural
and functional properties, plant-type FNRs are
classi-fied as plastidic type and bacterial type Plastidic
FNRs efficiently catalyse electron transfer and hydride
transfer between low-potential one-electron carriers
and NADP+⁄ H, usually participating in the
produc-tion of NADPH Bacterial FNRs generally exhibit
considerably slower turnover, provide the cell with
reduced electron carriers and are examples of novel
methods of FAD and NADP+⁄ H binding However, their structures, the particular residues involved in FAD binding and the residues at the catalytic centre are well conserved [91] In addition, all plant-type FNRs may share a similar catalytic mechanism [140] The general fold found in FNR is also present in other enzymes Many of these enzymes are multidomain proteins that, in addition to the FNR-like domain, also contain Fd- or Fld-like domains These proteins contain FAD (or FMN) and a FMN or Fe–S protein, and shut-tle electrons from NAD(P)H to the metal centres via their FNR and Fd⁄ Fld domains [138,141–144] The Fld and FNR domains in diflavin reductases appear to have evolved independently [141,142,144] Despite most charged residues in Anabaena proteins being conserved
in these domains, the dipole moment orientations between the FNR and Fld domains are far from colin-ear [48] Long-range electrostatic forces to attract their interaction surfaces have been decreased However, resi-dues on the Fld- and FNR-domain interaction surfaces may have been conserved to orientate the Fld domain when pivoting between the FNR domain and the electron acceptor [144]
The FNR family also contains NAD+⁄ H- and NADP+⁄ H-dependent members Some NAD+⁄ H-dependent members do not present the C-terminal aromatic residue stacking the flavin and a cavity appears open at its re-face [138,145,146], but the NMN moiety is usually not observed in the structures
of their complexes and, when observed an interaction between the flavin and the nicotinamide compatible with hydride transfer is not present [114,138,145,147] Catalytic differences between NADP+-dependent members are related to the different energies required
to produce stacking of the nicotinamide at the re-face
to FAD [119,138] These observations are compatible with a mechanism in which the initial interactions between the enzyme and 2¢P-AMP must evolve towards the production of alternative structures for each protein The fine-tuning of the enzyme catalytic efficiency is governed by the distance between and mutual orientation of the N5 of FAD and the nicotin-amide C4 Therefore, it is reasonable to suppose that ancestral FNR adapted its NAD(P)+⁄ H-binding site, modulating unique orientations to adapt its efficiency
to the coenzyme oxidation or reduction rates required
in each particular electron transfer chain
Applications of current knowledge about Flds and FNRs
The NADPH-producing electron transfer chain has been used to explore the possibility of redesigning
Trang 10existing electron transfer systems so that they can
per-form functions other than that for which they were
synthesized [148,149] Strategies to engineer stress
tol-erance in plants based on the typical stress response of
photosynthetic micro-organisms are underway [150–
152] The change in the enzyme specificity with respect
to its coenzyme is another example of redesign FNR
regions involved in coenzyme binding have been
mod-elled to mimic the site in NAD+⁄ H-dependent
enzymes [115–118,120], but further work is needed to
improve their catalytic efficiency Finally, Flds and
FNRs are essential for the survival of some human
pathogens and so may be important in the field of
drug design [126,153–155]
Acknowledgements
This work has been supported by Ministerio de
Educa-cio´n y Ciencia, Spain (Grant BIO2007-65890-C02-01)
I would like to thank Drs J Sancho and C
Go´mez-Moreno who initially introduced me in the
Flavopro-teins and Photosynthesis In particular, I appreciate all
the support and collaboration received from Dr
Go´mez-Moreno over more than 20 years I also thank
to Drs M L Peleato, M F Fillat and T Bes, as well
as all those collaborators that over the years have
helped us to obtain X-ray structures and kinetic data:
Drs J A Hermoso, G Tollin, J K Hurley, M A
Rosa, M Herva´s and J A Navarro Finally, I must
thank to my current and former PhD students, Dr M
Martı´nez-Ju´lvez, Dr J Tejero, Dr I Nogue´s, Dr S
Frago, J R Peregrina, G Gon˜i, A Serrano, B
Her-guedas and I Lans for their collaboration and interest
to better understand the FNR system and for all they
teach me everyday
References
1 Mu¨ller F (1990) Chemistry and Biochemistry of
Flavo-enzymes CRC Press, Boca Raton, FL
2 Golbeck JH (2006) Photosystem I The Light-driven
Platocyanin:Ferredoxin Oxidoreductase Springer,
Dordrecht
3 Vishniac W & Ochoa S (1952) Fixation of carbon
dioxide coupled to photochemical reduction of pyridine
nucleotides by chloroplast preparations J Biol Chem
195, 75–93
4 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
5 Medina M & Go´mez-Moreno C (2004) Interaction of
ferredoxin–NADP+reductase with its substrates:
opti-mal interaction for efficient electron transfer
Photosyn-th Res 79, 113–131
6 Bottin H & Lagoutte B (1992) Ferredoxin and flavo-doxin from the cyanobacterium Synechocystis sp PCC 6803 Biochim Biophys Acta 1101, 48–56
7 Se`tif P (2006) Electron transfer from the bound iron– sulfur clusters to ferredoxin⁄ flavodoxin: kinetic and structural properties of ferredoxin⁄ flavodoxin reduc-tion by photosystem I In Photosystem I The Light-dri-ven Platocyanin:Ferredoxin Oxidoreductase (Golbeck
JH, ed.), pp 439–454 Springer, Dordrecht
8 Hurley JK, Tollin G, Medina M & Go´mez-Moreno C (2006) Electron transfer from ferredoxin and
flavodox-in to ferredoxflavodox-in-NADP+reductase In Photosystem I The Light-driven Platocyanin:Ferredoxin Oxido-reductase (Golbeck JH, ed.), pp 455–476 Springer, Dordrecht
9 Fromme P, Jordan P & Kranb N (2001) Structure of photosystem I Biochim Biophys Acta 1507, 5–31
10 Jordan P, Fromme P, Witt hydride transfer, Klukas O, Saenger W & Kranb N (2001) Three-dimensional structure of cyanobacterial photosystem I at 2.5 A˚ resolution Nature 411, 909–917
11 Ben-Shem A, Frolow F & Nelson N (2003) Crystal structure of plant photosystem I Nature 426, 630–635
12 Antonkine ML, Jordan P, Fromme P, Kranb N, Golbeck JH & Stehlik D (2003) Assembly of protein subunits within the stromal ridge of photosystem I Structural changes between unbound and sequentially
PS I-bound polypeptides and correlated changes of the magnetic properties of the terminal iron sulfur clusters
J Mol Biol 327, 671–697
13 Mu¨hlenhoff U, Kruip J, Bryant DA, Rogner M, Se`tif
P & Boekema E (1996) Characterization of a redox-active cross-linked complex between cyanobacterial photosystem I and its physiological acceptor
flavodox-in EMBO J 15, 488–497
14 Se`tif P (2001) Ferredoxin and flavodoxin reduction by photosystem I Biochim Biophys Acta 1507, 161–179
15 Se`tif P, Fischer N, Lagoutte B, Bottin H & Rochaix
JD (2002) The ferredoxin docking site of photosys-tem I Biochim Biophys Acta 1555, 204–209
16 Lelong C, Boekema EJ, Kruip J, Bottin H, Rogner
M & Se`tif P (1996) Characterization of a redox active cross-linked complex between cyanobacterial photosystem I and soluble ferredoxin EMBO J 15, 2160–2168
17 Draper RD & Ingraham LL (1968) A potentiometric study of the flavin semiquinone equilibrium Arch Bio-chem Biophys 125, 802–808
18 Mayhew SG (1999) The effects of pH and semiquinone formation on the oxidation–reduction potentials of fla-vin mononucleotide A reappraisal Eur J Biochem 265, 698–702