Biochimica et Biophysica Acta 1817 (2012) 2118–2127

Contents lists available at SciVerse ScienceDirect

Biochimica et Biophysica Acta

j ourna l homepage: www.e lsev ie r .com/ locate /bbabio

Understanding the FMN cofactor chemistry within the AnabaenaFlavodoxin environment

Isaias Lans, Susana Frago, Milagros Medina ⁎Departamento de Bioquímica y Biología Molecular y Celular, Facultad de Ciencias, Universidad de Zaragoza, 50009, Zaragoza, SpainInstitute for Biocomputation and Physics of Complex Systems (BIFI)‐Joint Unit BIFI‐IQFR (CSIC), Universidad de Zaragoza, 50018, Zaragoza, Spain

Abbreviations: ET, electron transfer; ox, oxidised; sq, sEox/sq, midpoint reduction potential for the ox/sq couple; Esqfor the sq/hq couple; Em,midpoint reduction potential for thFaraday constant; AnFld, Flavodoxin from Anabaena; Fld, Fand in its oxidised, semiquinone and hydroquinone statesFMNsq, FMNhq, FMN in its oxidised, neutral semiquinone aLM, LMox, LMsq, LMhq, lumiflavin and in its oxidised, neutralquinone states; FNR, FNRox, FNRsq, FNRhq, Ferredoxin-NADPsemiquinone and hydroquinone states; PSI, Photosystem I;parent observed rate constant; S, softness;η, hardness;ω, eleical potential; f-, f+ and f0 electrophilic, nucleophilic and rquantum mechanics/molecular mechanics; MD, molecularCG, Conjugate Gradient; 8-Cl-FMN, 8-nor-Cl-FMN; 7,8-7-Cl,8-H-FMN, 7-nor-7-Cl,8-nor-FMN; 7-H,8-Cl-FMN, 7-nor⁎ Corresponding author. Tel.: +34 976762476; fax: +

E-mail address: mmedina@unizar.es (M. Medina).

0005-2728/$ – see front matter © 2012 Elsevier B.V. Alhttp://dx.doi.org/10.1016/j.bbabio.2012.08.008

a b s t r a c t

a r t i c l e i n f o

Article history:Received 18 May 2012Received in revised form 26 August 2012Accepted 31 August 2012Available online 7 September 2012

Keywords:FlavodoxinIsoalloxazineFMN analogueFMN affinityQM/MM calculationChemical reactivity

The chemical versatility of flavin cofactors within the flavoprotein environment allows them to play main roles inthe bioenergetics of all type of organisms, particularly in energy transformation processes such as photosynthesisor oxidative phosphorylation. Despite the large diversity of properties shown by flavoproteins and of the biologicalprocesses in which they are involved, only two flavin cofactors, FMN and FAD (both derived from the 7,8-dimethyl-10-(1′-D-ribityl)-isoalloxazine), are usually found in these proteins. Using theoretical and experimentalapproaches we have carried out an evaluation of the effects introduced upon substituting the 7- and/or 8-methyls of the isoalloxazine ring in the chemical and oxido-reduction properties of the different atoms of the ringon free flavins and on the photosynthetic Anabaena Flavodoxin (a flavoprotein that replaces Ferredoxin as electroncarrier from Photosystem I to Ferredoxin-NADP+ reductase). In Anabaena Flavodoxin both the protein environmentand the redox state contribute tomodulate the chemical reactivity of the isoalloxazine ring.Anabaena apoflavodoxinis shown to be designed to stabilise/destabilise each one of the FMN redox states (but not of the analogues producedupon substitution of the 7- and/or 8-methyls groups) in the adequate proportions to provide Flavodoxin with theparticular properties required for the functions in which it is involved in vivo. The 7- and/or 8-methyl groups ofthe ixoalloxazine can be discarded as the gate for electrons exchange in Anabaena Fld, but a key role in this processis envisaged for the C6 atom of the flavin and the backbone atoms of Asn58.

© 2012 Elsevier B.V. All rights reserved.

1. Introduction

Flavodoxins (Flds) are flavin-dependent proteins participating in alarge number of electron transfer (ET) reactions in the main energytransformation processes in prokaryotic organisms and certain algae[1,2], while in eukaryotic proteins containing Fld-like modules areimplicated in key metabolic functions [3,4]. The FMN cofactor confersthese electron transferases the ability to act as low-potential electroncarriers in one-ET processes. This is a consequence of its isoalloxazine

emiquinone; hq, hydroquinone;/hq, midpoint reduction potentiale ox/hq couple;WT,wild-type; F,ldox, Fldsq, Fldhq, holoflavodoxin; ApoFld, apoflavodoxin; FMNox,nd anionic hydroquinone states;semiquinone and anionic hydro-+ reductase and in its oxidised,Kd, dissociation constant; kap, ap-ctrophilicity; μ, electronic chem-adical Fukui functions; QM/MM,dynamics; SD, Steepest Descent;diCl-FMN, 7,8-nor-7,8-Cl-FMN;-8-nor-8-Cl-FMN34 976762123.

l rights reserved.

ring being able to exist in three different redox states within the proteinenvironment; oxidised (ox), neutral semiquinone (sq) and anionichydroquinone (hq), characterised by a different number of electronsand protons. The isoalloxazine establishes several non-covalent interac-tions with the apoflavodoxin (ApoFld) moiety, and upon interaction, theFMNmidpoint reduction potentials for the two semi-reactions, Eox/sq andEsq/hq, result inverted andwell-separated, allowing for the stabilisation ofthe protonated one-electron reduced state, the otherwise thermody-namically unstable neutral semiquinone [5,6]. The stabilisation of thisstate allows flavoproteins in general, and Fld in particular, to mediateobligatory processes of a single electron with those of two-electrons,being therefore key intermediaries in the photosynthetic chain of severalalgae and cyanobacteria, as well as in many other metabolic processesinvolving energy transformation in all type of organisms. Stretches ofaminoacids 56–59 and 93–95 of ApoFld (Anabaena Fld (AnFld) number-ing used throughout the text) contribute to the interaction with the iso-alloxazine ring and to tune the midpoint reduction potentials [7–16]. InAnFld the Asn58-Ile59 peptide bond is hypothesised to flip between an“O-Down” conformation, in the oxidised and the hydroquinone states,and an “O-Up” one in the semiquinone, by changing its H-bond networkwith the flavin N5 (as shown in the highly homologous Fld fromAnacystis nidulans (AyFld)) (Fig. 1) [17]. The fact that the semiquinonestate is less stable in AnFld and AyFld than in other species correlateswith: i) the presence in AnFldox and AyFldox of a H-bond between the

2119I. Lans et al. / Biochimica et Biophysica Acta 1817 (2012) 2118–2127

un-protonated N5 of the flavin and the amine group of residue 59 that isnot observed in other Fldox, and ii) the H-bond established between theN5H and the carbonyl of Asn58 beingweaker in AnFldsq and AyFldsq thanthose formed in Flds with smaller side-chains at this position [6,15,17].However, despite the “O-Down” and “O-Up” flip provides with a versa-tile device tomodulatemidpoint reduction potentials and chemical reac-tivity within the flavin ring, in AnFld this structural rearrangement stillneeds to be structurally proved [14–16].

The 7- and 8-methyls of FMN are the only portions of the isoallox-azine ring accessible to the solvent in Fld. This confers particularmechanistic interest on these positions because of the role theymight play in the interaction with protein partners, and, particularly,as possible gates for electrons entering and leaving the flavin ring[18]. Taking advantage of the reversibility of the FMN:ApoFld com-plex [15,19], we experimentally produced AnFld variants where the7- and/or 8-methyls of FMNwere replaced by chlorine and/or hydro-gen. These variants are able to interact, and even to exchange elec-trons, with the AnFld physiological partners, Photosystem I (PSI)and Ferredoxin-NADP+ reductase (FNR), but their midpoint reduc-tion potentials are less negative than expected (Fig. SP1) [20]. The latterobservation suggests the chemical reactivity of the flavin ring mightalso contribute to modulate the effects induced in the isoalloxazineproperties by the protein environment. Here, we analyse the chemicaland structural changes produced by the substitutions in the differentoxido-reduction states of the isoalloxazine, either free or bound toAnApoFld, by combining experimental and theoretical approaches.These results contribute to better understand the flavin reactivity with-in the protein environment.

AV59/I59

Y94

N58

W57

T56

N

N

N-

NH

O

O

C

CHO

C

C

CH2

HO

HO

H

H

H

O

P

O-

O-O

1 2

34a4

10a

5

10

5a

9a

6

98

7

1' HH

H

2'

D

Fig. 1. Details of the isoalloxazine binding site in Flds. Conformational changes associated wishows: (A) “O-Down” conformation in AyFldox (pale green, PDB 1CZU). (B) “O-Up” conform1D04). For the purpose of comparison (A) includes superimposed AnFldox (pink, PDB 1FLV),(D) Definition of the QM region (atoms in orange) used for the QM/MM calculations. (E) De

2. Materials and methods

2.1. Biological material and FMN analogues

Wild-type (WT) AnFld was over-expressed in E. coli and purified asdescribed [21]. Its ApoFld was prepared by treatmentwith 3% trichloro-acetic acid at 4 °C in the presence of dithiothreitol. The precipitatedapoprotein was separated from FMN by centrifugation and dissolvedin 500 mMMOPS pH 7.0 before dialysis against 50 mMMOPS pH 7.0.8-nor-Cl-FMN, 7,8-nor-7,8-Cl-FMN, and 7-nor-7-Cl,8-nor-FMN and7-nor-8-nor-8-Cl-FMN (herein 8-Cl-FMN, 7,8-diCl-FMN, 7-Cl,8-H-FMN, 7-H,8-Cl-FMN) were produced from the corresponding ribofla-vin analogues as previously described [20].

2.2. Determination of the free binding energy for the flavin:ApoFldcomplexes of the FMN analogues in the different redox states

Dissociation constants (Kd) for the FMNox:ApoFld complexes weredetermined fluorometrically by following the quenching of the flavinfluorescence upon titration with ApoFld in 50 mM MOPS pH 7.0, at25 °C (Fig. SP2). In a typical experiment, 1 ml of ~200 nM of theFMN analogue was titrated with aliquots of 10–20 μM ApoFld. Aftereach addition, the system was allowed to equilibrate for 2 min. Exci-tation was at 445 nm and emission was monitored at 525 nm. Fittingof the experimental data to the theoretical equation for a 1:1 com-plex, as previously described, allowed the calculation of Kd for theFMNox:ApoFld complex and the starting concentration of flavin (CF)[15,22]. Standard deviation between replicates during determination

B C

N58

E

th the reduction of AyFld. Detail of the FMN (coloured in orange CPK) environment thatation in AyFldsq (blue, PDB 1CZL). (C) “O-Down” conformation in AyFldhq (green, PDBwhile (B) and (C) include AyFldox. Arrows indicate positions of observed displacements.tail of the main interactions stabilising the isoalloxazine ring in the AnFld environment.

2120 I. Lans et al. / Biochimica et Biophysica Acta 1817 (2012) 2118–2127

of Kd, including numerical error after fitting analysis, were within 10%of the media value (Table 1). The free energy for the formation of theFMNox:ApoFld complex (ΔGox) was obtained directly from the exper-imental Kd, with an error below±0.2 kcal/mol (in general larger thanerror propagation for each individual variant). Free energy values forFMNsq:ApoFld (ΔGsq) and FMNhq:ApoFld (ΔGhq) were calculatedfrom the equations derived from a thermodynamic cycle (Fig. SP3A)[15,22,23]:

ΔGsq ¼ ΔGox−F Eox=sq–Eox=sqfree

� �ð1Þ

ΔGhq ¼ ΔGox−F Eox==sq þ Esq=hq–2Emfree

� �ð2Þ

where F is the Faraday's constant, Eox/sq and Esq/hq are the previouslyreported semi-reduction midpoint potentials for the different recon-stituted Fld variants (data from Table 1 in [20] with an error within ±5 mV of the given value), and Eox/sq

free and Esq/hqfree are the semi-

reductionmidpoint potentials for the free FMN format pH7.0. For thedif-ferent FMN analogues only midpoint reduction potentials for the ox/hqcouple, Emfree, were known [24,25], being their Eox/sq

free and Esq/hqfree

values calculated using the equations

Eox=sq þ Esq=hq� �

=2 ¼ Em ð3Þ

Eox=sq−Esq=hq ¼ 2:303RT=F logK ð4Þ

In all cases the semiquinone formation constant (K) has been as-sumed the same as that determined for FMN using pulse radiolysis(0.00022) [26]. Determination of this value is not trivial and evenfor free FMN it has only been measured with low accuracy [27].Therefore, errors in the determined Eox/sq

free and ΔGsq, and, particu-larly, in Esq/hq

free, are difficult to quantify. ΔGhq is not affected by thevalue of K and was calculated below ±0.3 kcal/mol for all thevariants.

2.3. Production of structural models for AnFlds reconstituted with FMNanalogues in the oxidised, semiquinone and hydroquinone states

Starting in silico AnFldox and AnFldhq models were generated byreplacing the FMNox cofactor in the three-dimensional structure ofAnFldox (PDB code: 1FLV) with each of the corresponding FMNox orFMNhq analogues, since the “O-Down” conformation is expected forboth states. The starting model for AnFldsq was produced in the“O-Up” conformation by using the AnFld sequence, the GENO-3D

Table 1Dissociation constants for the FMNox:ApoFld complexes and free energies for the for-mation of the corresponding oxidised, semiquinone and hydroquinone complexes.Data obtained in 50 mM MOPS pH 7.0 at 25 °C.

Flavinform

Kd

(nM)aΔGox

(kcal/mol)bΔGsq

(kcal/mol)cΔGhq

(kcal/mol)d

FMN 0.7±0.06 −12.48 −15.23 −7.628-Cl-FMN 0.69±0.06 −12.49 −15.35 −8.717,8-diCl-FMN 0.76±0.02 −12.43 −15.71 −11.057-H,8-Cl-FMN 0.34±0.08 −12.91 −15.72 −8.257-Cl,8-H-FMN 0.85±0.07 −12.37 −15.43 −9.07

a Determined from fluorometric titrations of the different FMNox analogues withApoFld. Standard deviation in the experimental determination of Kd is indicated.

b Calculated from data ina and ΔGox=−RT ln 1/Kd. Errors below ±0.2 kcal/mol, avalue in general larger than error propagation for each individual variant.

c Calculated as described in Eq. (1) [22], with Eox/sqfree and Esq/hq

free values from [20].Errors difficult to quantify due to assumptions in the K constant.

d Calculated as described in Eq. (2) [22], with errors below ±0.3 kcal/mol.

platform [28] and the crystal structure of AyFldsq (PDB code: 1CZL,“O-Up” conformation) as template. Then, each of the FMNsq ana-logues was situated in the model generated for AnApoFldsq at theposition of FMNsq in AyFldsq.

Quantum mechanics/molecular mechanics (QM/MM) moleculardynamics (MD) simulations were performed with AMBER 9.0 [29],using the amber94 force field and the semiempirical method AustinModel 1 (AM1) [30]. The QM subsystem included the lumiflavin(LM) ring of each Fld. The QM/MM interface was handled by includ-ing a link atom between the C1’ and C2’ atoms of the FMN ribitylchain (Fig. 1D) [31]. Each molecular system was neutralised by theaddition of sodium ions, and solvated with the TIP3P water modelin a cubic box. The cut-off distance for the non-bonded interactionswas 10 Å. Solvent molecules and counter ions were relaxed andallowed to redistribute around the restrained protein molecule byminimisation with 1000 steps of Steepest Descent (SD) and 2000steps of Conjugate Gradient (CG), while the protein atoms wereconstrained with a harmonic force constant of 500 kcal/mol Å2. Asecond energy minimisation was carried out with 2000 steps of SDand 3000 steps of CG also allowing protein relaxation. The resultingsystemwasheated from0 K to300 Kat constant volumewith theproteinatoms constrained with a harmonic force constant of 10 kcal/mol Å2.The system was then equilibrated during 100 ps at 300 K by using aLangevine temperature equilibration algorithm at 1 atm with periodicboundary conditions, and the Particle Mesh Ewald method to treatlong-range electrostatic interactions. Production runs consisted of3–4 ns of MD with 2 fs steps. Temperature was kept at 300 K using aBerendsen constant temperature algorithm. The pressure was kept at1 atm using a weak-coupling pressure algorithm. System coordinateswere collected every 2 ps. During equilibration and simulations theleapfrog Verlet integration scheme and the SHAKE algorithm wereused [32]. Three-dimensional structures and trajectories were visuallyinspected using PyMOL and VMD [33,34]. Interatomic distances and an-gles, as well as the root mean square deviation from a given structure,were monitored using PTRAJ [29].

2.4. Gas-phase calculation of the structures for the different freeanalogues

The LM form was chosen as a model in the investigation of theelectronic properties of the free flavin ring in the different redoxstates. Calculations were carried out with Gaussian 03 [35]. The struc-tures of all the LM analogues were optimised using the hybrid B3LYPfunctional and 6–31G* basis set [36,37].

2.5. Determination of reactivity indices

Chemical reactivity indices arising from density functional theorywere calculated for the different redox states of the flavin variants[38]. These indices include: the electronic chemical potential (μ), amea-sure of the directions of charge transfers during a chemical reaction; thehardness (η), ameasure of the system resistance to exchange electroniccharge; the softness (S), a measure of the system polarisability; andthe electrophilicity (ω), an indicator of the stabilisation energy of thesystem when obtaining electronic charge from the environment. Anapproximation using the method of finite differences allowed calculat-ing their global values in the flavin ring in terms of the ionisation poten-tial, I, and of the electron affinity, A, [38–40],

−μ≈ Iþ Að Þ=2; ð3Þ

η≈ I–Að Þ=2; ð4Þ

S≈1=2η; ð5Þ

ω≈μ2=2η; ð6Þ

2121I. Lans et al. / Biochimica et Biophysica Acta 1817 (2012) 2118–2127

where I and A are obtained from the electronic energy calculations onthe N−1, N and N+1 electron systems.

I ¼ EN−1−EN; A ¼ EN−ENþ1 ð7Þ

Local reactivity of the flavin atoms was predicted using the Fukuifunction f(r). Their electrophilic (f -), nucleophilic (f +) and radical (f 0)components provide an indication of the preferred sites for an attackby an electrophilic, a nucleophilic, and a radical agent, respectively. Ahighly different value between the electrophilic and nucleophilic localindices is associated with a high reactivity at that point in the molecularregions. Fukui functions of free LM analogues were obtained using analgorithm described by Contreras et al. [41]. For Flds, an approximationwas used and Fukui functions for their FMNmolecules were determinedin terms of atomic charges according to [42]:

fþk ≈qk Nð Þ−qk N þ 1ð Þ ð8Þ

f−k ≈qk N−1ð Þ−qk Nð Þ ð9Þ

f 0k≈12

fþk þ f−kh i

ð10Þ

where fk is the Fukui function at atom k, and qk is the charge over atom kfor the systems with N+1, N and N−1 electrons. Local softness andelectrophilicity were calculated according to the expressions [38,42]:

S�k ¼ f�k S ð11Þ

ωk ¼ fþk ω ð12Þ

3. Results

3.1. Strength of the FMN:ApoFld complexes with FMN analogues

The experimentally determined affinities of AnApoFld for all theFMNox analogues were found to be similar to that for FMN (Table 1).

BA

ED

ox

FMN:Fld

8-Cl-FMN:Fld

Fig. 2. Isoalloxazine environment showing selected residues at the equilibrium position(D) An8-Cl-FMN:Fldox, (E) An8-Cl-FMN:Fldsq and (F) An8-Cl-FMN:Fldhq.

The FMN:ApoFld interaction energy profiles as a function of the flavinredox state had the same V-shape for the different analogues as thatfor WT AnFld (Fig. SP3B) [14,15,22]. Uncertainties in binding energiesfor the semiquinone state when using the analogues make difficult topredict changes regarding Fldsq. However, an increase in the relativestabilisation of the hydroquinone complexes is detected for the Fldsreconstituted with the analogues (Fig. SP3B, Table 1), particularly for7,8-diCl-Fldhq (by 3.4 kcal/mol compared to Fldhq).

3.2. Dynamics of the FMN:ApoFld complexes

QM/MM MD simulations for all the Fldox and Fldhq variants main-tain the typical “O-Down” conformation of the AnFldox crystal struc-ture (Figs. 1 and 2, Table 2). Simulations of WT Fldsq and 8-Cl-Fldsqvariants keep the starting “O-Up” conformation for this redox state.However, analysis of the evolution of the position of the nitrogenand oxygen backbone atoms of Asn58 regarding the C6 atom of thesemiquinone flavin ring showed an intermediate situation betweenthe “O-Down” and “O-Up” conformations for the rest of the variants(Fig. 3), with the Asn58-Ile59 peptide bond in a perpendicular arrange-ment with regard to the flavin ring. The main interactions between theoxidised isoalloxazine ring and the protein environment are conservedalong the simulations in all the variants, including: i) stabilisation of thepyrimidine ring by polar contactswith backbone atomsof Gly60, Asp80,Asn97 and Gln99, ii) the weak H-bond between the N5 position of theflavin ring and the nitrogen atom of Ile59, iii) a π–π stacking interactionof Tyr94 at the si-face of the pyrazine moiety of the flavin ring, iv) themain carbonyl of Thr56 both stacking at the re-face of the pyrimidineand pyrazine rings and H-bonding the 2′OH of the ribityl, and v) thestacking of the aromatic side-chain of Trp57 at the re-face of the ben-zene portion (Figs. 1E and 2, Table 2). Most of these interactions are sim-ilarlymaintained in the semiquinone and hydroquinone states. However,in these sates the side-chain of Trp57 slightly changes its relative positionregarding both the isoalloxazine ring and its position in Fldox, increasingthe distance between the mass centres of the isoalloxazine and theTrp57 side-chain. This occurs in all the Fld variants (particularly in WTFldsq), with the only exception of 8-Cl-Fldsq where, contrarily, it is closer

C

F

sq hq

of the QM/MM MD simulation of (A) AnFldox, (B) AnFldsq and (C) AnFldhq and of

Table2

Calculated

averag

edistan

ces(Å

)be

twee

nselected

atom

sof

theflav

inring

andseve

ralr

esidue

sof

theproteinen

vironm

entalon

gQM/M

MMD

simulations

ofthedifferen

treco

nstitutedFlds

.

Fldform

O1-Na

(G60

)O2-N

a

(Q99

)N3-Oa

(N97

)N5-Na

(I59

)N5-Oa

(N58

)N5-Na

(N58

)Y9

4-FM

Nb

N5-

OH(Y

94)

OH(Y

94)-

Oa (N58

)W

57-

FMNa

C7-

NE(

W57

)Oa (T5

6)-

NE(

W57

)N1-

Oa (T5

6)N3-

Oa (T5

6)N5-

Oa (T5

6)N10

-Oa (T5

6)C4

a-O(T56

)

PDB-1F

LVc

2.80

2.90

2.90

3.71

5.71

3.59

3.44

4.70

8.30

5.38

6.70

4.00

3.80

4.30

3.60

3.40

3.31

WTFld o

x3.36

3.62

3.08

3.72

5.78

3.80

3.77

4.87

8.16

5.83

6.71

4.17

3.80

4.37

3.75

3.46

3.38

8-Cl-Fld

ox

3.53

3.14

2.98

3.91

6.03

4.54

3.61

4.87

8.90

5.73

6.75

4.14

4.06

4.69

3.53

3.43

3.41

7,8-diCl-Fld

ox

3.34

3.93

3.25

3.72

5.79

3.85

3.48

4.75

8.55

5.10

6.70

4.06

3.67

4.08

3.81

3.58

3.32

7-H,8-Cl-Fld o

x3.21

3.47

3.04

3.86

5.89

4.15

3.79

4.62

7.92

5.82

7.19

4.37

3.91

4.47

3.53

3.47

3.38

7-Cl,8-H

-Fld

ox

3.49

3.22

2.98

3.99

6.08

4.53

3.70

4.87

8.57

5.62

6.38

4.08

3.95

4.51

3.47

3.44

3.31

WTFld s

q3.11

3.44

3.17

4.58

3.67

3.90

3.99

4.04

4.09

7.40

4.20

5.89

4.10

4.89

3.88

3.35

3.66

8-Cl-Fld

sq3.01

2.97

3.10

4.50

3.56

3.79

3.73

4.36

4.35

5.37

6.75

4.07

3.67

4.42

3.95

3.48

3.52

7,8-diCl-Fld

sq2.92

3.49

3.15

4.53

3.33

3.73

3.92

4.17

4.00

6.33

3.87

5.61

3.90

4.75

4.04

3.38

3.69

7-H,8-Cl-Fld s

q2.97

3.37

3.08

4.56

3.45

3.73

3.89

4.33

4.24

6.81

3.85

5.67

3.83

4.60

3.99

3.41

3.61

7-Cl,8-H

-Fld

sq3.48

2.94

3.07

4.85

4.26

4.22

3.61

4.61

5.31

5.50

6.42

4.04

3.71

4.43

3.61

3.30

3.33

WTFld h

q3.11

3.13

3.20

3.36

5.49

3.73

3.85

5.00

8.04

6.77

3.97

5.87

3.95

4.73

4.47

3.83

3.98

8-Cl-Fld

hq

3.19

3.88

3.46

3.50

5.59

3.72

4.00

4.67

8.03

6.95

4.18

6.00

3.99

4.80

4.36

3.70

3.91

7,8-diCl-Fld

hq

2.95

3.97

3.37

3.23

5.32

3.49

3.82

4.99

7.84

5.82

4.15

5.84

3.56

4.29

4.68

3.76

3.91

7-H,8-Cl-Fld h

q3.07

3.21

3.26

3.48

5.57

3.65

4.03

4.96

7.84

6.25

4.06

5.45

3.71

4.39

4.20

3.61

3.66

7-Cl,8-H

-Fld

hq

3.11

3.69

3.48

3.50

5.55

3.55

3.76

4.78

8.14

6.49

3.72

5.86

3.88

4.52

4.48

3.90

3.91

aTh

eseN

andO

atom

sbe

long

totheba

ckbo

neof

theco

rrespo

ndingresidu

e.b

Distanc

esbe

twee

nthemasscentre

oftheco

rrespo

ndingside

-cha

inarom

atic

ring

san

dthat

oftheisoa

lloxa

zine

.cValue

sob

tained

from

theX‐ray

crystals

truc

ture

ofAnF

ldoxareinclud

edforco

mpa

rison.

2122 I. Lans et al. / Biochimica et Biophysica Acta 1817 (2012) 2118–2127

to the isoalloxazine ring (Table 2, Fig. 2E). Additionally, a slight displace-ment is also observed in the position of Tyr94with regard to the pyrazinering, getting the aromatic side-chain particularly centred on the N5Hposition in Fldsq.

3.3. Effects of substitutions in the chemical reactivity of free LManalogues

The molecular electrostatic potentials (MEP) calculated for thefree LM analogues indicate that, independently of the redox state,substitutions at the 7- or 8-methyls produce an electron withdrawaleffect on the pyrazine and pyrimidine rings (Fig. 4). Such effects arelarger when the 7- and 8-methyls are simultaneously replaced, eitherone with a chlorine and the other one with a proton, but particularly,if both of them are simultaneously replaced with chlorine. The neutralLMsq state shows a zone of low charge density at the protonated N5H,not observed in the oxidised or hydroquinone states, which is highlysensitive to substitutions at positions 7- and 8- of the ring (Fig. 4). Theentrance of the second electron to generate the anionic LMhq pro-duces an increase of the negative charge density throughout the iso-alloxazine ring, but particularly in the pyrazine and pyrimidinerings. Global reactivity indices indicate that electrophilicity increasesin LMsq upon replacement of the methyl groups, while softness ishardly influenced (Table 3). This latter parameter also indicates theneutral semiquinone state as the one suffering the larger polarisationby the protein environment, being the hydroquinone state the lessaffected.

Fig. 5 indicates the flavin atoms with the highest Fukui functionfor the different free analogues (numerical values in Table SP1). AllLMox analogues show considerably higher values for f + than f− atC4a and, particularly, N5. This indicates an electrophilic character atthese positions, predicting preferential entrance of electrons overthe N5-C4a region. The highest Fukui function in the anionic LMhq

analogues are similarly ascribed to C4a and N5, being in this case f−

considerably larger than f+. Therefore, in the anionic LMhq variantsthe C4a-N5 region is the preferential site for electrons leaving thering. The LMsq state shows some notable differences: i) N3 and C8are predicted as the most reactive positions in LMsq for both nucleo-philic or radical attacks, suggesting they are the preferred positionsfor the second electron entrance, and ii) substitutions at the 7- and8-methyls modulate reactivity of the different positions turning N10into the most reactive (with the exception of 8-Cl-LMsq, CM7). There-fore, the electron withdrawing effect introduced by the substituentsin C7 and C8 will, in general, displace the LMsq isoalloxazine electro-philic character towards the N10 position of the flavin ring.

3.4. Protein influence of the electronic distribution of the FMN analogues

Fukui functions were also determined for the Fld variants by usingthe Mulliken charges obtained from QM/MM minimisations (Fig. 5and Table SP1). Despite all the Fldox variants maintain the N5 positionas the preferred site for entrance of electrons, the second possiblereactive site follows closely showing small differences in the magni-tude of the Fukui function. In WT Fldox and 8-Cl-FMN:Fldox the C8 po-sition appears as highly probable for electron entrance, while the C6atom would be preferred for the rest of the Fld variants.

The protein environment also has a particular impact on the reac-tivity in the semiquinone state. The first interesting observationregarding f+, f− and f0 is that they show quite similar values amongthem. This might indicate that these atoms could be involved in eitheraccepting electrons, donating electrons or reacting with radical spe-cies, a fact probably related with the high percentage of maximumsemiquinone stabilised in the apoprotein environment. Additionally,the entrance/exit of the electron is predicted to occur over differentatoms of the flavin ring in the Fldsq variants compared to the corre-sponding free LMsq analogues, and also to depend on the flavin

2

3

4

5

6

2

3

4

5

6

0 500 1000 1500 20002

3

4

5

6

2

3

4

5

6

2

3

4

5

6

7,8-diCl-Fldsq

7-H-8-Cl-Fldsq

7-Cl-8-H-Fldsq

Fldsq

8-Cl-Fldsq

Time (ps)

Dis

tan

ce (

Å)

7-Cl-8-H-Fldsq

Fld sq

B

AC

Fig. 3. Relative disposition of the C6 atom of the flavin ring regarding the nitrogen (red line) and oxygen (black line) backbone atoms of Asn58 in the semiquinone state for the differentreconstituted variants. Snapshots at the QM/MMMD equilibrium for (A) Fldsq and (B) 7-Cl-8-H-Fldsq. (C) Evolution of the C6-N-Asn58 and C6-O-Asn58 distances along the QM/MMMDfor the different reconstituted variants.

- +

LMhq 8-Cl-LMhq

7,8-diCl-LMhq7-Cl-8-H-LMhq 7-H-8-Cl-LMhq

7,8-diCl-LM ox7-H-8-Cl-LMox7-Cl-8-H-LMox

LMox 8-Cl-LMox8-Cl-LM sq

7,8-diCl-LMsq7-Cl-8-H-LMsq 7-H-8-Cl-LMsq

LMsq

Fig. 4. MEP, calculated using a Hartree–Fock method and 6–31G* basis set, of the different LM analogues in the oxidised, neutral semiquinone, and anionic hydroquinone states.

2123I. Lans et al. / Biochimica et Biophysica Acta 1817 (2012) 2118–2127

Table 3Global softness (S) and electrophilicity (ω) of the free LM analogues.

Flavin form S (au) ω (au)

LMox 4.18 0.138-Cl-LMox 4.20 0.147,8-diCl-LMox 4.22 0.167-H,8-Cl-LMox 4.11 0.157-Cl,8-H-LMox 4.17 0.15LMsq 5.38 0.138-Cl-LMsq 5.39 0.157,8-diCl-LMsq 5.41 0.167-H,8-Cl-LMsq 5.32 0.157-Cl,8-H-LMsq 5.30 0.15LMhq 4.14 0.018-Cl-LMhq 4.16 0.017,8-diCl-LMhq 4.31 0.007-H,8-Cl-LMhq 4.12 0.017-Cl,8-H-LMhq 4.09 0.01

2124 I. Lans et al. / Biochimica et Biophysica Acta 1817 (2012) 2118–2127

substituents (Fig. 5). Thus, the C2 and, particularly, C6 atoms becomethe most reactive in all reconstituted Fldsq variants. In Fldhq, C4a ap-pears the predicted region for electron exit for all the variants, butthe parameters also indicate that the protein environment consider-ably displaces the nucleophilic character exhibited by position N5 inthe free LMhq analogues towards the pyrimidine ring (N1 atom) forFldhq and towards N10 for 7-H,8-Cl-FMN-Fldhq and 7-Cl,8-H-FMN-Fldhq.

4. Discussion

4.1. Redox state and protein environment modulate the isoalloxazinechemical reactivity

Our calculations for free LM identify the N5-C4a region as the pre-ferred for electron exchange in LMox and LMhq, while the C8 and N3are predicted for the exchange in LMsq (Fig. 5). These data also indi-cate the neutral LMsq as the state with the most homogeneous chargedistribution, while LMox and anionic LMhq show, respectively, lowerand higher charge density in the pyrimidine and pyrazine ringswith respect to the benzene one (Figs. 4 and 5). Reactivity indicesfor the FMN atoms in Fldox, Fldsq and Fldhq indicate that the non-covalent interactions between the isoalloxazine and ApoFld displacethe most reactive positions and are modulated by the differentialcharge distribution among redox states (Fig. 5). These effects are ob-served particularly in Fldox and Fldhq, where the benzene–pyrazine

A

Oxidized

Free LM ApoFld bound

Semiquin

Free LM

Fig. 5. Atoms with the highest Fukui functions (marked with red spheres) within the isoa(columns 3 and 4) and hydroquinone (columns 5 and 6) states for the different LM analogspond, from top to bottom, to the isoalloxazine rings of LM, 8-Cl-LM, 7,8-diCl-LM, 7-H,8-C7-H,8-Cl-AnFld and 7-Cl,8-H-AnFld in columns 2, 4 and 6.

moiety is predicted as preferred for electron entrance and the pyr-azine–pyrimidine for its exit, and are in agreement with experimentalevidences suggesting that the methyl groups on the benzene ring arenot the main points for electrons exchange in Fld [20]. AnFld crystalstructures have only been reported in the oxidised state, which adoptan “O-Down” conformation. Our QM/MM simulations support the hy-pothesis of the “O-Down” conformation being preferred also for AnFld-hq, while AnFldsq would adopt the “O-Up” conformation. Thesesimulations also predict a slight displacement of the Tyr94 side-chaintowards the N5 position (also observed in crystal structures for AyFldsqand AyFldhq [17]) and, particularly, of the position of Trp57 in AnFldsqand AnFldhq that modulates the solvent accessibility of the pyrazine–pyrimidine region (Fig. 2, Table 2). X-ray structures for Fld and ApoFldfrom Anabaena and Helicobacter pylori [43,44] indicated that the Apoforms already present the overall protein fold. The most significant dif-ferences were confined to the 56–59 isoalloxazine binding loop. Al-though the stabilisation in the latter ApoFld of the empty FMNbinding site is not reached by an aromatic–aromatic interaction asin Anabaena, in both cases the mechanism similarly involves exten-sive rearrangements in dihedral angles in the 56–59 loop. NMR evi-dences additionally support that the FMN binding loops might bemore flexible than seen in the X-ray structures [45,46]. However,changes in these dihedral angles are not observed in our simulations inany of the redox states, in agreement with FMN binding inducing rigidityin the 56–59 binding loop. These data support the fact that the aromaticstacking of the Trp57 side-chain against the isoalloxazine is not criticalto modulate midpoint reduction potentials or FMN binding, while, as ex-perimentally suggested, these parameters are rather influenced by therelative strength of the H-bond between the N5 of the flavin and theAsn58-Ile59 residues [15,22,47]. These observations indicate that in Fldboth the protein environment and the redox state of the cofactor contrib-ute tomodulate the chemical reactivity of atomswithin the isoalloxazinering.

4.2. Chemical reactivity of the isoalloxazine upon substitution of the 7-and 8-methyls

The dimethylbenzene moiety of FMN interacts with hydrophobicApoFld areas, leaving the 7- and 8-methyls as the only portions acces-sible to the solvent [48,49], in agreement with modifications at thesepositions only producing minor effects in the FMNox:ApoFld affinityand in the network of non-covalent interactions maintaining thisassociation (Figs. SP2, SP3B, Tables 1 and 2). However, modificationsincrease the FMNhq:ApoFld affinity (Table 1). The chloride electron-

poFld bound ApoFld bound

one Hydroquinone

Free LM

lloxazine ring of the flavin analogues in the oxidized (columns 1 and 2), semiquinoneues and the corresponding reconstituted AnFlds. Structures shown in each line corre-l-LM and 7-Cl,8-H-LM in columns 1, 3 and 5 and of AnFld, 8-Cl-AnFld, 7,8-diCl-AnFld,

BA

0,12 0,14 0,16

-450

-360

-270

m=5360LM

7-H,8-Cl-LM

8-Cl-LM

7-Cl,8-H-LM

7,8-diCl-LM

0,12 0,14 0,16

-180

-120

-60

E F

ldo

x/sq

(mV

)

E F

ldsq

/hq

(mV

)

ωox(au) ωsq(au)

7-Cl,8-H-LM

7-H,8-Cl-LM

7,8-diCl-LM

8-Cl-LM

LM m=3203

Fig. 6. Midpoint reduction potentials and global electrophilicity correlations as a function of the Fld redox state. Correlation between (A) Eox/sqFld and global electrophilicity of LMox

analogues (R2 0.91) and (B) Esq/hqFld and global electrophilicity of LMsq analogues (R2 0.87).

2125I. Lans et al. / Biochimica et Biophysica Acta 1817 (2012) 2118–2127

withdrawing effect at position 7- and/or 8- causes removal of negativecharge from all atoms, but especially in the pyrazine and pyrimidinerings, building-up some negative character at the edge of the benzenering (Fig. 4). Since formation of the anionic FMNhq is energeticallyunfavourable in the ApoFld negative binding site, removal of negativecharge from the pyrimidine in the FMNhq analogues appears as one ofthe main reasons contributing to the strengthen of the FMNhq:ApoFldinteraction (Fig. SP3B, Table 1) [8,13,50–52].

Isosteric substitutions on the isoalloxazine ring modulate mid-point potentials of free flavins [53,54], but if substituents did not pro-duce significant effects on the flavin-apoprotein interaction shifts inpotentials upon protein binding should be predictable by a linearrelationship between Efree and EFld with a slope ~1.0 [55–57]. Evenconsidering the low accuracy in the estimation of Eox/sqfree and Esq/hq

free forthe analogues, midpoint potentials for the reconstituted Flds appearedless shifted to negative values than in WT Fld (Fig. SP1). Moreover, thecorrelation between the global isoalloxazine electrophilicity of freeanalogues and the midpoint potential values shows that the proteininfluence is roughly proportional to the isoalloxazine electrophilicity(Fig. 6, Fig. SP4), suggesting that differential electronic properties ofthe analogues contribute to the observed effects. Thus, the strongerthe substituents withdrawal effect is, the stronger the FMNhq:ApoFldinteraction and the smaller the ApoFld influence in setting midpointreduction potentials, particularly Esq/hq. Thus, the protein environmentenhances the substituents withdrawal effect, in agreement with the

1.00 1.02 1.04 1.06

0

90

180

Fldox

7H,8-Cl-Fldox

7-Cl,8-H-Fldox

8-Cl-Fld ox

7,8-diCl-Fldox

S +N5 (au)

k ap

1(s-1

)

Fig. 7. Correlation between the apparent rate constants, (A) kap1 and (B) kap2, for the reduccases the lines correspond to the linear fitting excluding 7-H,8-Cl-FMN:Fld.

polarity and electrostatic properties of the flavin environment beingtuned in each protein to modulate its midpoint potentials [13,50,58,59].These data indicate that AnApoFld is optimised to destabilise the interac-tion with FMNhq to much larger extent than with any other of the ana-logues here described.

4.3. Chemical reactivity of Fld as electron transferase in photosynthesis

Previous kinetics characterisations indicated that substitutions atthe 7- and 8-methyls allowed Fldox to interact with PSI and FNR,while interaction of 7-H,8-Cl-Fldsq, 7-Cl,8-H-Fldsq and 7,8-diCl-Fldsqwith PSI was poor. Moreover, despite the theoretical increase in thedriving force for Fld accepting electrons, drastic changes in reactivityupon reduction by PSI were observed (see Fig. 7 from [20]). Fukuifunction predicts that the most probable position for electron en-trance for all Fldsq variants is the C6 atom of the flavin ring (Fig. 5and Table SP1). Thus, a change in the environment of this positionmight alter the PSI-Fld competent complex and, therefore, influencethe ET mechanism. In fact WT Fldsq and 8-Cl-Fldsq are the only Fldsqvariants i) with a methyl substituent at C7, ii) showing an “O-Up”conformation for the Ans58-Ile59 backbone and iii) accepting elec-trons from PSI through the formation of a reorganisation complex[20]. Therefore, the methyl at position C7 and the Asn58-Ile59 back-bone conformation appear as a determinant for the formation of thecompetent complex for ET between PSIrd and Fldsq. These results

1.00 1.02 1.04 1.06

0

10

20

30

S +N5 (au)

k ap

2(s-1

)

7-Cl,8-H-Fldox

7,8-diCl-Fldox

7-H,8-Cl-Fldox

8-Cl-Fld ox

Fldox

tion of the Fldox variants by FNRhq and the local softness of the N5 flavin atom. In both

2126 I. Lans et al. / Biochimica et Biophysica Acta 1817 (2012) 2118–2127

additionally support an electron exchange pathway including the C6of the flavin ring and Asn58 backbone atoms in Fldsq.

A positive correlation was reported for the ET rates from FNRhq orFNRsq to the Fldox variants and their midpoint potentials (see Fig. 7from [20]). Some tendency for a reverse correlation might also beenenvisaged between the N5 local softness of the LMox variants andthe apparent ET rates reported for the reduction of the Fldox variants byFNRhq (kap1 and kap2, corresponding to the Fldox+FNRhq→Fldsq+FNRsqand Fldox+FNRsq→Fldsq+FNRox processes, respectively) (Fig. 7). Thissuggests contribution of the N5 atom of the flavin, as well as of the ther-modynamic driving force, in the overall ET. Since reactions involvingsoft-soft or hard-hard reactants will be preferred over those involvingsoft-hard ones [60], the LM global softness might contribute as an addi-tional factor to explain why, in general, kap1 is considerably faster thankap2: softness of LMox analogues is more similar to LMhq than that of theLMsq (Table 3). In the WT system the initial orientation driven by thealignment of the FNR and Fld dipole moments contributes to the forma-tion of a number of alternative binding modes competent for the ET,where the substituent at C7 has one of the largest propensities for beingat the interaction surface [61]. The large divergence of 7-H,8-Cl-FMN:Fldin Fig. 7 agrees with its reactivity with FNR being slower than expectedfrom the thermodynamics (see Fig. 7 from [20]). Changes in the directionof the molecular dipole moment and the increase in the accessible sur-face area of the flavin reactive region (caused by the smaller volume ofthe C7 substituent) might favour orientations between 7-H,8-Cl-FMN:Fld and FNR less-productive than those favoured in theWT system. Sim-ilar situations have already been reported upon mutation of some resi-dues in the isoalloxazine environment on the Fld surface [16,62],where modulation of the ET processes by different orientations and dis-tances between the redox centres explained different reactivities. Thus,several factors contribute to the efficiency of the reactions involvingFld, including dipole moment orientation, surface complementarity,thermodynamic driving forces or electronic factors. Among them, contri-bution of structural and electronic factors is key in the ET from PSI to Fld,while in general the ET between FNR and Fld has amajor dependence onthe thermodynamics.

Analysis along QM/MM MD simulations also indicate that inde-pendently of the redox state and of the Fld variant, the backbones ofThr56 (particularly its carbonyl group situated at the re-face of theisoalloxazine ring) and Asn58 are the closest atoms to the preferredsites for the exchange of electrons, the C6-N5-C4a region of the isoal-loxazine (Fig. 5 and Table 2). Therefore, the data here presented indicatethat the Thr56-Trp57-Asn58-Ile59 backbone has a key contribution in theET efficiency. This is in agreement with previous site-directedmutagene-sis studies indicating these residues are the key tomodulate the Fld abilityto bind and to exchange electronswith its physiological partners, FNRandPSI [63]. Additionally changes in the dihedral angles in this backboneregionmight also contribute to the slight displacement of Trp57, openingthe access towards the C6-N5-C4a region for electron exchange.

5. Conclusions

Altogether, the data here presented indicate that substitutions inthe isoalloxazine ring will modulate its chemical reactivity, as wellof that of the flavoproteins they are part from. Both the apoproteinenvironment and the redox state modulate the chemical reactivityof the isoalloxazine ring within flavoproteins. Moreover, the proteinportion of AnFld is particularly optimised to stabilise the charge dis-tribution of the neutral FMNsq and to destabilise the negative chargeof the pyrazine ring of FMNhq, critical aspects to modulate Eox/sq andEsq/hq within the values required for the physiological functions inwhich Fld is involved. Using Fld as model we have proven that theuse of different derivatives of the isoalloxazine ring in flavoproteinsmight potentially increase their versatility. However, this would pro-vide the cell with a number of isoforms of each particular flavoproteinexhibiting different redox properties. Additionally, the cell would

require to increase, or to adapt, its machinery to produce, or to obtainfrom the environment, the different flavin derivatives. Instead, Naturehas optimised its cellular resources by using a single organic molecule,riboflavin (7,8-dimethyl-10-(1′-D-ribityl)-isoalloxazine or Vitamin B2),as precursor of all the flavin cofactors used by flavoproteins. The single7,8-dimethyl-isoalloxazine ring has been selected as the reactive portionin all flavoproteins, being the protein environment in each flavoprotein(as here proven for Fld) carefully tuned to provide to the flavin ringwith the chemical reactivity, midpoint reduction potentials and capaci-ties required for the different reactions in which it is involved withineach particular protein environment.

Acknowledgements

This work has been supported by the Spanish Ministry of Science andInnovation (Grant BIO2010-1493 to M.M). Riboflavin analogues were agenerous gift from Dr. D. Edmondson. We thank Dr. R. Contreras for pro-viding with the algorithm for the determination of Fukui functions. I.L. isthe recipient of a JAE-CSIC fellowship associated to the Instituto Química-Física Rocasolano (CSIC).

Appendix A. Supplementary data

Supplementary data to this article can be found online at http://dx.doi.org/10.1016/j.bbabio.2012.08.008.

References

[1] S.G. Mayhew, M.L. Ludwig, Flavodoxins and electron-transferring flavoproteins,Enzymes 12 (1975) 57.

[2] S.G.Mayhew, G. Tollin, General properties of flavodoxins, in: F.Müller (Ed.), Chemistryand Biochemistry of Flavoenzymes, vol. 3, CRC Press, Boca Raton, Florida, 1992,pp. 389–426.

[3] T. Iyanagi, Structure and function of NADPH-cytochrome P450 reductase andnitric oxide synthase reductase domain, Biochem. Biophys. Res. Commun. 338(2005) 520–528.

[4] D.C. Haines, I.F. Sevrioukova, J.A. Peterson, The FMN-binding domain of cytochromeP450BM-3: resolution, reconstitution, and flavin analogue substitution, Biochemistry39 (2000) 9419–9429.

[5] S.G. Mayhew, G.P. Foust, V. Massey, Oxidation–reduction properties of flavodoxinfrom Peptostreptococcus elsdenii, J. Biol. Chem. 244 (1969) 803–810.

[6] M.L. Ludwig, K.A. Pattridge, A.L. Metzger, M.M. Dixon,M. Eren, Y. Feng, R.P. Swenson,Control of oxidation–reduction potentials inflavodoxin from Clostridium beijerinckii:the role of conformation changes, Biochemistry 36 (1997) 1259–1280.

[7] A. Lostao, C. Gómez-Moreno, S.G. Mayhew, J. Sancho, Differential stabilization of thethree FMNredox forms by tyrosine 94 and tryptophan57 inflavodoxin fromAnabaenaand its influence on the redox potentials, Biochemistry 36 (1997) 14334–14344.

[8] L.H. Bradley, R.P. Swenson, Role of glutamate-59 hydrogen bonded to N(3)H ofthe flavin mononucleotide cofactor in the modulation of the redox potentials ofthe Clostridium beijerinckii flavodoxin. Glutamate-59 is not responsible for thepH dependency but contributes to the stabilization of the flavin semiquinone,Biochemistry 38 (1999) 12377–12386.

[9] L.H. Bradley, R.P. Swenson, Role of hydrogen bonding interactions to N(3)H of theflavin mononucleotide cofactor in the modulation of the redox potentials of theClostridium beijerinckii flavodoxin, Biochemistry 40 (2001) 8686–8695.

[10] F. Chang, L.H. Bradley, R.P. Swenson, Evaluation of the hydrogen bonding interac-tions and their effects on the oxidation–reduction potentials for the riboflavin com-plex of the Desulfovibrio vulgaris flavodoxin, Biochim. Biophys. Acta 1504 (2001)319–328.

[11] F.C. Chang, R.P. Swenson, Regulation of oxidation–reduction potentials throughredox-linked ionization in the Y98H mutant of the Desulfovibrio vulgaris[Hildenborough] flavodoxin: direct proton nuclear magnetic resonance spectro-scopic evidence for the redox-dependent shift in the pKa of Histidine-98,Biochemistry 36 (1997) 9013–9021.

[12] L.J. Druhan, R.P. Swenson, Role of methionine 56 in the control of theoxidation-reduction potentials of the Clostridium beijerinckii flavodoxin: effectsof substitutions by aliphatic amino acids and evidence for a role of sulfur–flavininteractions, Biochemistry 37 (1998) 9668–9678.

[13] A.A. McCarthy, M.A. Walsh, C.S. Verma, D.P. O'Connell, M. Reinhold, G.N.Yalloway, D. D'Arcy, T.M. Higgins, G. Voordouw, S.G. Mayhew, Crystallographicinvestigation of the role of aspartate 95 in the modulation of the redox potentialsof Desulfovibrio vulgaris flavodoxin, Biochemistry 41 (2002) 10950–10962.

[14] I. Nogues, L.A. Campos, J. Sancho, C. Gomez-Moreno, S.G. Mayhew, M. Medina,Role of neighboring FMN side chains in the modulation of flavin reduction poten-tials and in the energetics of the FMN:apoprotein interaction in Anabaenaflavodoxin, Biochemistry 43 (2004) 15111–15121.

[15] S. Frago, G. Goñi, B. Herguedas, J.R. Peregrina, A. Serrano, I. Perez-Dorado, R. Molina,C. Gomez-Moreno, J.A. Hermoso, M. Martinez-Julvez, S.G. Mayhew, M. Medina,

2127I. Lans et al. / Biochimica et Biophysica Acta 1817 (2012) 2118–2127

Tuning of the FMN binding and oxido-reduction properties by neighboring sidechains in Anabaena flavodoxin, Arch. Biochem. Biophys. 467 (2007) 206–217.

[16] G. Goñi, B. Herguedas, M. Hervas, J.R. Peregrina, M.A. De la Rosa, C. Gomez-Moreno,J.A. Navarro, J.A. Hermoso, M. Martinez-Julvez, M. Medina, Flavodoxin: a compro-mise between efficiency and versatility in the electron transfer from Photosystem Ito Ferredoxin-NADP+ reductase, Biochim. Biophys. Acta 1787 (2009) 144–154.

[17] D.M. Hoover, C.L. Drennan, A.L. Metzger, C. Osborne, C.H. Weber, K.A. Pattridge, M.L.Ludwig, Comparisons of wild-type and mutant flavodoxins from Anacystis nidulans.Structural determinants of the redox potentials, J. Mol. Biol. 294 (1999) 725–743.

[18] M. Medina, Structural and mechanistic aspects of flavoproteins: photosyntheticelectron transfer from photosystem I to NADP+, FEBS J. 276 (2009) 3942–3958.

[19] J.I. Martinez, P.J. Alonso, C. Gomez-Moreno, M. Medina, One- and two-dimensionalESEEM spectroscopy of flavoproteins, Biochemistry 36 (1997) 15526–15537.

[20] S. Frago, I. Lans, J.A. Navarro, M. Hervas, D.E. Edmondson, M.A. De la Rosa, C.Gomez-Moreno, S.G. Mayhew, M. Medina, Dual role of FMN in flavodoxinfunction: electron transfer cofactor and modulation of the protein-proteininteraction surface, Biochim. Biophys. Acta 1797 (2010) 262–271.

[21] J.L. Casaus, J.A. Navarro, M. Hervas, A. Lostao, M.A. De la Rosa, C. Gomez-Moreno, J.Sancho, M. Medina, Anabaena sp. PCC 7119 flavodoxin as electron carrier fromphotosystem I to ferredoxin-NADP+ reductase. Role of Trp(57) and Tyr(94),J. Biol. Chem. 277 (2002) 22338–22344.

[22] A. Lostao, C. Gomez-Moreno, S.G. Mayhew, J. Sancho, Differential stabilization of thethree FMNredox formsby tyrosine 94 and tryptophan57 inflavodoxin fromAnabaenaand its influence on the redox potentials, Biochemistry 36 (1997) 14334–14344.

[23] M. Dubourdieu, J. le Gall, V. Favaudon, Physicochemical properties of flavodoxinfrom Desulfovibrio vulgaris, Biochim. Biophys. Acta 376 (1975) 519–532.

[24] M.T. Stankovich, Redox properties of flavins and flavoproteins, in: F. Müller (Ed.),Chemistry and Biochemistry of Flavoenzymes, vol. I, CRC Press, Inc., Boca Raton,Fl., 1991, pp. 401–425.

[25] F. Muller, V. Massey, Flavin–sulfite complexes and their structures, J. Biol. Chem.244 (1969) 4007–4016.

[26] R.F. Anderson, Energetics of the one-electron reduction steps of riboflavin, FMN andFAD to their fully reduced forms, Biochim. Biophys. Acta 722 (1983) 158–162.

[27] S.G. Mayhew, The effects of pH and semiquinone formation on the oxidation–reduction potentials of flavin mononucleotide. A reappraisal, Eur. J. Biochem. 265(1999) 698–702.

[28] C. Combet, M. Jambon, G. Deléage, C. Geourjon, Geno3D: automatic comparativemolecular modelling of protein, Bioinformatics 18 (2002) 213–214.

[29] D.A. Case, T.E. Cheatham III, T. Darden, H. Gohlke, R. Luo, K.M. Merz Jr., A.Onufriev, C. Simmerling, B. Wang, R.J. Woods, The AMBER biomolecular simula-tion programs, J. Comput. Chem. 26 (2005) 1668–1688.

[30] M.J.S. Dewar, E.G. Zoebisch, E.F. Healy, J.J.P. Stewart, Development and use ofquantum mechanical molecular models. 76. AM1: a new general purpose quan-tum mechanical molecular model, J. Am. Chem. Soc. 107 (1985) 3902–3909.

[31] U.C. Singh, P.A. Kollman, A combined ab initio quantum mechanical and molecu-lar mechanical method for carrying out simulations on complex molecularsystems: applications to the CH3Cl + Cl− exchange reaction and gas phase pro-tonation of polyethers, J. Comp. Chem. 7 (1986) 718–730.

[32] J.-P. Ryckaert, G. Ciccotti, H.J.C. Berendsen, Numerical integration of the cartesianequations of motion of a system with constraints: molecular dynamics ofn-alkanes, J. Comput. Phys. 23 (1977) 327–341.

[33] W.L. Delano, They PyMOL molecular graphics system, DeLano Scientific, SanCarlos, CA, USA, 2002. (http://www.pymol.org).

[34] W. Humphrey, A. Dalke, K. Schulten, VMD - Visual Molecular Dynamics, J. Mol.Graph. 14 (1996) 33–38.

[35] M.J. Frisch, G.W. Trucks, H.B. Schlegel, G.E. Scuseria, M.A. Robb, J.R. Cheeseman, V.G.Zakrzewski, J.A. Montgomery, R.E. Stratmann, J.C. Burant, Gaussian 03 (Revision B.04), Gaussian, Inc., Pittsburgh PA, 2003.

[36] A.D. Becke, Density-functional thermochemistry. III. The role of exact exchange,J. Chem. Phys. 98 (1993) 5648–5652.

[37] C. Lee, W. Yang, R.G. Parr, Development of the Colle–Salvetti correlation-energyformula into a functional of the electron density, Phys. Rev. B 37 (1988) 785–789.

[38] R.G. Parr, W. Yang, Density functional theory of atoms and molecules, OxfordUniversity Press, New York, 1989.

[39] J.L. Gazquez, F. Mendez, The hard and soft acids and bases principle: an atoms inmolecules viewpoint, J. Phys. Chem. 98 (1994) 4591–4593.

[40] R.G. Parr, L.V. Szentpály, S. Liu, Electrophilicity index, J. Am. Chem. Soc. 121(1999) 1922–1924.

[41] R.R. Contreras, P. Fuentealba, M. Galván, P. Pérez, A direct evaluation of regionalFukui functions in molecules, Chem. Phys. Lett. 304 (1999) 405–413.

[42] W. Yang, W.J. Mortier, The use of global and local molecular parameters for theanalysis of the gas-phase basicity of amines, J. Am. Chem. Soc. 108 (1986)5708–5711.

[43] M. Martínez-Júlvez, N. Cremades, M. Bueno, I. Pérez-Dorado, C. Maya, S.Cuesta-Lopez, D. Prada, F. Falo, J.A. Hermoso, J. Sancho, Common conformationalchanges in flavodoxins induced by FMN and anion binding: the structure ofHelicobacter pylori apoflavodoxin, Proteins 69 (2007) 581–594.

[44] C.G. Genzor, A. Perales-Alcon, J. Sancho, A. Romero, Closure of a tyrosine/tryptophanaromatic gate leads to a compact fold in apo flavodoxin, Nat. Struct. Biol. 3 (1996)329–332.

[45] E. Steensma, C.P. van Mierlo, Structural characterisation of apoflavodoxin showsthat the location of the stable nucleus differs among proteins with a flavodoxin-like topology, J. Mol. Biol. 282 (1998) 653–666.

[46] G.M. Langdon, M.A. Jimenez, C.G. Genzor, S. Maldonado, J. Sancho, M. Rico,Anabaena apoflavodoxin hydrogen exchange: on the stable exchange core ofthe alpha/beta (21345) flavodoxin-like family, Proteins 43 (2001) 476–488.

[47] A. Lostao, M. El Harrous, F. Daoudi, A. Romero, A. Parody-Morreale, J. Sancho,Dissecting the energetics of the apoflavodoxin-FMN complex, J. Biol. Chem. 275(2000) 9518–9526.

[48] L.A. Campos, J. Sancho, Native-specific stabilization of flavodoxin by the FMN cofac-tor: structural and thermodynamical explanation, Proteins 63 (2006) 581–594.

[49] S.T. Rao, F. Shaffie, C. Yu, K.A. Satyshur, B.J. Stockman, J.L. Markley, M. Sundarlingam,Structure of the oxidized long-chain flavodoxin from Anabaena 7120 at 2 A resolu-tion, Protein Sci. 1 (1992) 1413–1427.

[50] Z. Zhou, R.P. Swenson, The cumulative electrostatic effect of aromatic stackinginteractions and the negative electrostatic environment of the flavin mononucle-otide binding site is a major determinant of the reduction potential for theflavodoxin from Desulfovibrio vulgaris [Hildenborough], Biochemistry 35 (1996)15980–15988.

[51] Z. Zhou, R.P. Swenson, Electrostatic effects of surface acidic amino acid residueson the oxidation–reduction potentials of the flavodoxin from Desulfovibriovulgaris (Hildenborough), Biochemistry 34 (1995) 3183–3192.

[52] J.D. Walsh, A.-F. Miller, Flavin reduction potential tuning by substitution andbending, J. Mol. Struct. (Thoechem) 623 (2003) 185–195.

[53] S. Ghisla, V. Massey, New flavins for old: artificial flavins as active site probes offlavoproteins, Biochem. J. 239 (1986) 1–12.

[54] D.E. Edmondson, S. Ghisla, Electronic effects of 7 and8 ring substituents as predictors offlavin oxidation-reduction potentials, in: S. Ghisla, P. Kroneck, P. Macheroux, H. Sund(Eds.), Flavins and Flavoproteins, Rudolf Weber, Berlin, 1999, pp. 71–76.

[55] L.M. Schopfer, V. Massey, A. Claiborne, Active site probes of flavoproteins. Determi-nation of the solvent accessibility of the flavin position 8 for a series of flavoproteins,J. Biol. Chem. 256 (1981) 7329–7337.

[56] L.M. Schopfer, A. Wessiak, V. Massey, Interpretation of the spectra observed dur-ing oxidation of p-hydroxybenzoate hydroxylase reconstituted with modified fla-vins, J. Biol. Chem. 266 (1991) 13080–13085.

[57] K. Yorita, H. Misaki, B.A. Palfey, V. Massey, On the interpretation of quantitativestructure–function activity relationship data for lactate oxidase, Proc. Natl.Acad. Sci. U. S. A. 97 (2000) 2480–2485.

[58] R.P. Swenson, G.D. Krey, Site-directed mutagenesis of tyrosine-98 in the flavodoxinfrom Desulfovibrio vulgaris (Hildenborough): regulation of oxidation–reductionproperties of the bound FMN cofactor by aromatic, solvent, and electrostatic interac-tions, Biochemistry 33 (1994) 8505–8514.

[59] S.M. Geoghegan, S.G. Mayhew, G.N. Yalloway, G. Butler, Cloning, sequencing andexpression of the gene for flavodoxin fromMegasphaera elsdenii and the effects ofremoving the protein negative charge that is closest to N(1) of the bound FMN,Eur. J. Biochem. 267 (2000) 4434–4444.

[60] R.G. Pearson, Hard and soft acids and bases, HSAB, part 1: fundamental principles,J. Chem. Educ. 45 (1968) 581.

[61] M. Medina, R. Abagyan, C. Gomez-Moreno, J. Fernandez-Recio, Docking analysisof transient complexes: interaction of ferredoxin-NADP+ reductase with ferre-doxin and flavodoxin, Proteins 72 (2008) 848–862.

[62] G. Goñi, A. Serrano, S. Frago, M. Hervas, J.R. Peregrina, M.A. De la Rosa, C.Gomez-Moreno, J.A. Navarro, M. Medina, Flavodoxin-mediated electrontransfer from photosystem I to ferredoxin-NADP+ reductase in Anabaena:role of flavodoxin hydrophobic residues in protein–protein interactions, Bio-chemistry 47 (2008) 1207–1217.

[63] I. Nogues, M. Hervas, J.R. Peregrina, J.A. Navarro, M.A. de la Rosa, C. Gomez-Moreno,M. Medina, Anabaena flavodoxin as an electron carrier from photosystem I toferredoxin-NADP+ reductase. Role of flavodoxin residues in protein–protein inter-action and electron transfer, Biochemistry 44 (2005) 97–104.