Biochimica et Biophysica Acta 1787 (2009) 1112–1121

Contents lists available at ScienceDirect

Biochimica et Biophysica Acta

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

Photo-catalytic oxidation of a di-nuclear manganese centre in an engineeredbacterioferritin ‘reaction centre’

Brendon Conlan a,⁎, Nicholas Cox a, Ji-Hu Su b, Warwick Hillier a, Johannes Messinger b,1, Wolfgang Lubitz b,P. Leslie Dutton c, Tom Wydrzynski a,⁎a Australian National University, Canberra, Australiab Max Planck Institute for Bioinorganic Chemistry, Mülheim an der Ruhr, Germanyc University of Pennsylvania, Philadelphia, USA

Abbreviations: BFR, wild type bacterioferritin;mutant H46R, H112R without cofactors; BFR1–M, BFRbound; BFR1–Z, BFR1 with ZnCe6 bound; BFR1–ZM, BFRof ZnCe6 and manganese bound; EPR, electron pisothermal titration calorimetry; PSII, Photosystem II;⁎ Corresponding authors.

E-mail addresses: Brendon.Conlan@anu.edu.au (B. CTom.Wydrzynski@anu.edu.au (T. Wydrzynski).

1 Current address: Department of Chemistry, Umeå U

0005-2728/$ – see front matter © 2009 Elsevier B.V. Adoi:10.1016/j.bbabio.2009.04.011

a b s t r a c t

a r t i c l e i n f o

Article history:Received 13 February 2009Received in revised form 16 April 2009Accepted 21 April 2009Available online 3 May 2009

Keywords:Artificial photosynthesisEPRManganeseElectron transferProtein engineeringBacterioferritinZinc chlorin e6

Photosynthesis involves the conversion of light into chemical energy through a series of electron transferreactions within membrane-bound pigment/protein complexes. The Photosystem II (PSII) complex in plants,algae and cyanobacteria catalyse the oxidation of water to molecular O2. The complexity of PSII has thus farlimited attempts to chemically replicate its function. Here we introduce a reverse engineering approach tobuild a simple, light-driven photo-catalyst based on the organization and function of the donor side of thePSII reaction centre. We have used bacterioferritin (BFR) (cytochrome b1) from Escherichia coli as the proteinscaffold since it has several, inherently useful design features for engineering light-driven electron transport.Among these are: (i.) a di-iron binding site; (ii.) a potentially redox-active tyrosine residue; and (iii.) theability to dimerise and form an inter-protein heme binding pocket within electron tunnelling distance of thedi-iron binding site. Upon replacing the heme with the photoactive zinc–chlorin e6 (ZnCe6) molecule and thedi-iron binding site with two manganese ions, we show that the two Mn ions bind as a weakly coupled di-nuclear Mn2

II,II centre, and that ZnCe6 binds in stoichiometric amounts of 1:2 with respect to the dimeric formof BFR. Upon illumination the bound ZnCe6 initiates electron transfer, followed by oxidation of the di-nuclearMn centre possibly via one of the inherent tyrosine residues in the vicinity of the Mn cluster. The lightdependent loss of the MnII EPR signals and the formation of low field parallel mode Mn EPR signals areattributed to the formation of MnIII species. The formation of the MnIII is concomitant with consumption ofoxygen. Our model is the first artificial reaction centre developed for the photo-catalytic oxidation of a di-metal site within a protein matrix which potentially mimics water oxidation centre (WOC) photo-assembly.

© 2009 Elsevier B.V. All rights reserved.

1. Introduction

Oxygenic photosynthesis is the process whereby plants utilizesunlight to catalyse the reduction of plastoquinone and the oxidationof water into protons andmolecular oxygen. This process is initiated inthe pigment/protein complex called Photosystem II (PSII). Theessential elements of PSII that catalyse the oxidation of water include:(i.) the strongly oxidizing multi-chlorophyll complex termed P680;(ii.) a redox-active tyrosine termed YZ; and (iii.) the oxygen evolvingcentre (OEC) consisting of four μ-oxo bridgedmanganese ions and one

BFR1, bacterioferritin double1 with two manganese ions

1 with stoichiometric amountsaramagnetic resonance; ITC,ZnCe6, zincII chlorin e6

onlan),

niversity, Umeå, Sweden.

ll rights reserved.

calcium ion (Mn4OxCa) bound within the protein complex [1]. Uponlight activation of PSII, P680 forms the P680•+ radical cation(E0N1.2 V) by transferring an electron to a neighbouring pheophytinmolecule followed by rapid transfer of the electron to a boundquinone molecule (QA). The charge separated state (P+/QA

−) under-goes hole transfer from P680•+ to YZ, and then in turn to the Mn4OxCacluster, which couples the one electron photochemistry of P680 to thefour electron chemistry of water oxidation via the sequential advanceof the S-state intermediates (S0–S4) [1].

In the design of artificial photoactive proteins, there are twoapproaches that can be considered [2]. The first is the de novo designfollowed by organic synthesis. A growing number of de novo proteinsare being synthesized with increasing functionality [3–6]; however,their synthesis is currently limited to a polypeptide length of about100 amino acid residues. The second approach, and the one adoptedfor this work, involves the modification of natural protein scaffolds.Target scaffolds are selected based on pre-existing structural featuresthat may be useful in re-engineering the protein. In addition, thedesign of a light-activated redox catalyst has to take into consideration

1113B. Conlan et al. / Biochimica et Biophysica Acta 1787 (2009) 1112–1121

the physiological limitations of intra-protein electron transfer.Electron tunnelling between redox centres is reasonably accuratelydescribed by a simple exponential decay with distance [7,8]. Packingdensity and the secondary structure of the protein matrix play smallerroles in mediating electron tunnelling rates [9–11]. As such naturalproteins with productive electron transfer reactions have been foundto have cofactors separated by not more than ∼14 Å [8,12,13].

The complexity of the PSII enzyme has thus far discouragedattempts to reverse engineer a complete synthetic water oxidase [14].As a result, previous engineering efforts have focussed mostly onconstructing biological motifs that mimic one particular aspect of thewater oxidase; for example, chlorin excitation/oxidation [15] andmetal binding/complex formation [16]. Hay et al. (2004) showedlight-activated electron transfer between a chlorin (ZnCe6) and aquinone engineered into cytochrome b562. The quinonewas covalentlyattached approximately ∼10 Å away from the bound chlorinproducing surprisingly high efficiencies of electron transfer (∼20%).Metal binding/complex formationwas shown by Thielges et al. wherethe purple non-sulphur bacterial reaction centre was modified to binda single MnII ion. Upon illumination, the boundMn could reduce P870+

suggesting that the metal centre was oxidized [16]. Gray et al. haveshown electron transfer through ligation of a ruthenium complex tothe azurin protein. By differing the distance from a single copper ion,light-induced oxidation of the copper ion and reduction of theruthenium cation could be achieved [17].

In order to mimic the reactions of the PSII donor side, we haveengineered a di-nuclear manganese centre within close proximity of alight-active chlorin (ZnCe6). We have used a modified form ofbacterioferritin (BFR) from Escherichia coli as the protein scaffold tobind these cofactors. BFR is a soluble ∼18.5 kDa oligomeric proteinwhich contains both iron and heme b binding sites [18–20]. The

Fig. 1. (a) A ribbon diagram of the E. coli bacterioferritin homodimer with two identical subinterface of the subunits. Twomanganesemolecules bind per protein subunit (PDB 1BFR). (b)twelve homodimers. (c) Schematic showing the estimated distance between the cofactorsfound in (d) which shows the distance between the cofactors in PSII.

protein forms into a homodimer, binding a single heme groupsymmetrically at the interface between the two protein monomericsubunits via bis-methionine ligation, and each protein monomercontains a binuclear metal binding site [21,22]. The crystal structurefor BFR was initially produced with two MnII ions bound in the metalbinding site [21,23] (Fig. 1a.). Either Mn or Fe metal ions canselectively bind at the active enzymatic site. The BFR protein self-assembles to produce a large spherical-shaped shell, made up of 12homodimeric units, that enclose a hollow cavity with a diameter of∼8 nm [21,24,25]. The natural protein shell thus binds 12 heme groupsand 48 Fe ions [21] (Fig. 1b.). BFR presents an appealing starting pointfor engineering a multi-step, light-activated protein as: (i.) it is ahighly stable protein (ii.) the heme can be extracted and replaced witha photoactive chlorin [15,26]; (iii.) the binuclear metal site has ligandssimilar to manganese catalase where the di-manganese site can reachan oxidation state of Mn2

III,IV [27–30]; and (iv.) the in silico estimatedcofactor separationwithin the BFR protein is similar in distance to thatfound in PSII (Fig. 1c and d) [31].

Here we report the construction of a light-activated metallo-catalyst, which can oxidize MnII to MnIII upon illumination with atyrosine radical formed during the process. The light-inducedoxidation state changes were characterized by low temperatureperpendicular and parallel mode X-band EPR spectroscopy.

2. Materials and methods

2.1. Cloning and mutagenesis

The bfr gene encoding bacterioferritin was amplified from E. coliBL21 cells, using PCR primers 5′-GGTATTGAGGGTCGCATGAAAGG-TGATACTAAAGTTATAA (Forward) and 5′-AGAGGAGAGTTAGAGCCTC-

units each hosting di-nuclear metal centres and forming a heme binding pocket at theA diagram of the sphere structure formed by bacterioferritin. The sphere is composed ofin the engineered bacterioferritin BFR1 complex. The distances are very similar to that

Fig. 2. Circular dichroism spectra of the apo forms of BFR and the double mutant BFR1

(H46R, H112R). The BFR and BFR1 had 59% and 60% α-helix respectively, with 7% β-sheet and 33% random coil for both proteins.

2 ITC allows the accurate determination of binding constants (K), reactionstoichiometries (n), enthalpies (ΔH°), and entropies (ΔS°), thus providing a completethermodynamic profile of the molecular interaction.

1114 B. Conlan et al. / Biochimica et Biophysica Acta 1787 (2009) 1112–1121

ATCAACCTTCTTCGCGGAT (Reverse). Ligation Independent Cloning(LIC) regions in each primer (underlined) facilitated cloning into thepET30Xa/LIC vector (Novagen), for expression in E. coli BL21 cells. Twoexternal histidines (H46, H112) were changed to arginines using theStratageneQickchange site-directedmutagenesis kit and the followingPCR primers. H46R 5′-AATGATGTGGAGTATCGCGAATCCATTGATGAH112R 5′-GCCGATAGCGTTCGTGATTACGTCAGCC. This double mutant(BFR1) was utilized for all measurements carried out in this study.

2.2. Expression, purification

BFR1 was expressed in E. coli cultures with induction using 1 mMisopropyl-beta-D-thiogalactopyranoside (Sigma). Harvested cellswere pink in colour due to the high expression level of the hemecontaining protein. Cells were lysed and the proteinwas purified fromthe supernatant utilizing the His-tag and a nickel sepharose column(GE Healthcare). The His-tag was cleaved using Factor Xa, andremoved again using a nickel sepharose column.

The purified proteinwas dark blood red in colour due to the boundheme and was of single band purity on an overloaded SDS-PAGE gel.Non-heme iron was removed using the modified method ofBauminger et al. with exhaustive dialysis in the presence ofdithiothreitol and EDTA [32]. The heme was removed from theheme binding pocket of the dimeric holoprotein using the method ofTeale et al. [26]. Protein concentrations for the apo-protein weredetermined using ɛ280nm of 20340 M−1 cm−1. A 1 mM ZnCe6 stockwas prepared by the addition of zinc acetate to free base chlorin e6dissolved in methanol, added in a 1:1 molar ratio with UV–Visconfirmation of binding of the zinc to the chlorin [33]. UV–Visspectroscopy was carried out on a Cary 300 Spectrometer (Varian,USA) and CD spectroscopy was carried out on a Jobin Yvon type III+spectrometer (Horiba, Japan).

2.3. O2 measurements

All oxygen measurements were carried out at 25 °C on anoxytherm Clarke type oxygen electrode (Hansatech, UK) modified tocontain a side illumination port. Oxygen measurements were carriedout on samples containing 30 μM BFR, 15 μM ZnCe6, 60 μM or 300 μMMnCl2 or 300 µM FeSO4 in 50 mM MES pH 6.5. Samples wereilluminated for long periods of time until oxygen uptake plateaued.

2.4. Isothermal Titration Calorimetry (ITC)

ITC measurements were carried out on a Microcal calorimeter at25 °C and the data was fit to a least squares model with the MicrocalOrigin fitting software. The heat of dilution was subtracted from theraw data prior to analysis. Manganese binding to BFR1 was measuredby injecting 11 µl aliquots of 900 μM MnCl2 into a 30 μM solution ofBFR1. BFR1 binding to ZnCe6 was measured by injecting 11 μl aliquotsof 300 μM BFR1 into a 15 μM solution of ZnCe6. All samples wereprepared in 25 mM tricine, 100 mM KCl pH 7.7.

2.5. EPR measurements

Low temperature (5 K) X-band EPR spectra were acquired on aBruker ESP 300E spectrometer using either a TM011 cavity or anER4116 dual mode cavity, perpendicular mode microwave frequency9.6 GHz, parallel mode microwave frequency 9.3 GHz, modulationfrequency 100 kHz. The EPR spectrum of each sample was collected inboth parallel and perpendicular modes. To examine light-inducedchanges in oxidation state of the metal centre in BFR1 complexes,samples were prepared in the dark at room temperature then eitherleft for 1 h in the dark or illuminated for 1 h. The long illuminationtime used coincided with the time taken for the oxygen uptakereactions to go to completion (see previous section). Due to the

concentrated nature of the samples they were stirred under pureoxygen during illumination to prevent the samples becominganaerobic. Samples were then degassed and frozen to liquid nitrogenand stored until use. All samples were measured within 24–48 h.Samples were prepared in 50 mM MES pH 6.5 or in 25 mM tricine,100 mM KCl pH 7.7 with 400 mM sucrose added as cryoprotectant.

2.6. EPR simulations/molecular modelling

Spectral simulations were solved numerically from a Hamiltonian(36×36 matrix) using Scilab-4.4.1, an open source vector-based linearalgebra package (www.scilab.org). A least squares minimizationroutine was employed to find the optimal solutions for theparameters. A complete description of the EPR simulations can befound in the supporting information.

Molecular modelling of mutations was carried out with the aid ofHyPERCHEM (Hypercube, USA) using energy minimization in aperiodic box with the AMBER94 force field. Raytraced images wereprepared using PyMOL v0.99 (Delano Scientific, USA).

3. Results

3.1. Cofactor assembly in bacterioferritin

3.1.1. ZnCe6 binding to BFR1

Previous work has shown that ZnCe6 will bind at the heme site ofcytochrome b562 via axial ligation to histidine [15]. BFR has twosurface exposed histidines at positions 46 and 112. To prevent non-specific binding these two residues were replaced with arginines. Thesecondary structure and binding of ZnCe6 to the doublemutant (BFR1)was identical to the wild type as probed by UV–Vis and CircularDichroism (CD) spectroscopy (Fig. 2). The wild type BFR and the BFR1

double mutant contain 59% and 60% α-helix respectively, with 7% β-sheet and 33% random coil for both proteins. BFR1 was also found tostill form the spherical ∼444 kDa dodecamer of dimers as measuredby native gel electrophoresis. Thus for all the remaining experimentsBFR1 was used. The nomenclature employed throughout this manu-script is as follows: BFR1 (BFR1 devoid of all cofactors), BFR1–M (BFR1

with two Mn ions bound per protein unit, 4 per dimer), BFR1–Z (BFR1

with one ZnCe6 bound per dimer) or BFR1–ZM (BFR1 with fourMn andone ZnCe6 bound per protein dimer).

BFR1 cofactor binding was monitored using UV–Visible spectro-scopy (Figure S1) and isothermal titration calorimetry (ITC)2 (Fig. 3).Figure S1 shows the spectra of ZnCe6 titrated into a solution BFR1.

Fig. 3. Calorimetric titrations of cofactors with BFR1. The top of the figures is the raw ITCoutput for the titration and the lower half of the figure is derived from the integratedraw data. (a) Calorimetric titration of BFR1 binding to ZnCe6. The line represents thebest least squares fit to a model of a single binding site. (b) Calorimetric titration of MnII

binding to BFR1. The line represents the best least squares fit to a model of twoindependent binding sites.

Table 1Parameters of best fit for ITC measurements of BFR1 cofactor binding.

Mn titrated into BFR1 (two independent binding site model)

n/BFR1 Ka (M−1) ΔHbind

(kcal mol−1)ΔSbind(kcal mol−1)

Mn1 0.88±0.01 1.54 (±0.39) × 109 −7380±116 17.3Mn2 1.64±0.11 4.31 (±1.09) × 106 860±87 33.2

BFR1 titrated into ZnCe6 (single binding site model)

n/ZnCe6 Ka (M−1) ΔHbind

(kcal mol−1)ΔSbind(kcal mol−1)

1115B. Conlan et al. / Biochimica et Biophysica Acta 1787 (2009) 1112–1121

Binding ZnCe6 to BFR1 produces a shift in the absorbance maxima ofthe QY band of the ZnCe6 from632 to 639 nm. The association constantof the ZnCe6 was measured by monitoring this QY shift upon itstitration into a solution of BFR1 (Ka∼1.2×106). As a control a BFRmutant in which the axial heme ligating M52 was converted to H52was also measured. This mutant correctly assembled and was capableof oxidizing iron. As expected, no absorbance shift of the chlorine QY

band was observed, suggesting that the ZnCe6 no-longer binds to theBFR. This observation confirms the lack of heme binding activity ofBFR M52H previously reported by Andrews et al. [18].

Fig. 3a shows the injection heat for BFR1 titrated into ZnCe6 indilute solution (15 μM). The integrated heat (μCal) for each injection isplotted over the molar ratio of protein to ZnCe6 in the bottom of Fig.3a3. For this the heat of dilution of the proteinwas subtracted from theraw data. It was found that 0.82 ZnCe6 molecules bind per BFR1 dimer

3 Negative peaks correspond to an exothermic reaction and positive peakscorrespond to an endothermic reaction.

with a binding constant of 2.54 (±0.45)×106. All parameters arereported in Table 1.

3.1.2. Mn binding to BFR1

MnII binding to BFR1 was measured using ITC. Fig. 3b shows theinjection heat for MnII titrated into a solution of BFR1. Analysis of theintegrated heat using the best fit of a least squares model allowing fortwo independent binding sites revealed that BFR1 has both a highaffinity Mn site (Ka11.54 (±0.39)×109 M) and a second lower affinitysite (Ka2 4.31 (±1.09)×106 M) (Table 1). Titrations of Mn into BFR1

showed that binding of the first metal ion to site 1 is an exothermicprocess, while binding to the lower affinity site 2 is an endothermicprocess (Fig. 3b). According to the ITC fitting an average of 2.5(±0.12) mol of Mn ions bind per mole of BFR1.

EPR measurements were undertaken to determine the oxidationstate of manganese ions bound to the bacterioferritin protein and theircoupling environment. We have used conventional perpendicularmode EPR for the observation of half-integer spin systems (e.g. MnII,and organic radicals) [34,35], and parallel polarization EPR forobserving EPR signals issued from integer spin systems, in particularS=2 (e.g. MnIII, Mn2

III,III) [36–40].EPR spectra of BFR1 titrated with two Mn ions per protein unit

revealed a broad signal (∼400 mT wide), dominated by a largestructured g=2 derivative (Fig. 4a). The non-Curie temperaturedependence of this signal (Fig. 5a) suggested that it arises from aneven spin antiferromagnetically coupled system, most likely a Mn2

II,II

dimer [41–44]. As the line shape of the Mn2II,II dimer remained

constant for temperatures up to 50 K, the exchange coupling betweenthe two metal centres was expected to be small.

EPR simulations using the Spin Hamiltonian formalism of the lineshape and temperature dependence of the Mn2

II,II signal were under-taken to estimate the exchange/dipole couplings and zero-fieldparameters for this system. These calculations are described in detailin the supporting information. The exchange coupling was estimatedto be 1.3 K (0.9 cm−1), theMn–Mn distance∼4 Å, commensuratewithcrystallographic evidence [21]. Both D values for the Mn ions are small(Db0.2 K), typical for a MnII high spin (S=5/2) centre [44–46].

The addition of the ZnCe6 pigment to the BFR1–M complex (i.e.BFR1–ZM) led to a significant change in the dimer spectrum (Fig. 4b).The main intensity of the spectrum shifted to g∼2, and appeared lessstructured. As before this new signal demonstrated non-Curietemperature dependence (Fig. 5b) and its line shape was invariantat all temperatures up to 50 K. It was noted that while the spectrumabout g∼2 is altered, the wings (g∼2.5 etc., now smaller byapproximately a factor of 2) matched the structure seen in theoriginal MnII dimer spectrum, i.e., without ZnCe6. This suggested thatthe spectrum was heterogeneous, made up of both the original MnII

dimer spectrum and a new signal. The estimated contribution of theoriginal dimer spectrum to the total spectrum observed after theaddition of ZnCe6 was 59%. This suggested that the Zn–chlorin had aneffect on 41% of Mn clusters (i.e. 82% of reaction centres). This value is

BFR1 1.64±0.02 2.54 (±0.45) × 106 −4956±89 12.7

Stoichiometry of the interaction (n), association constant (Ka), enthalpy of binding(ΔHbind), and entropy of binding (ΔSbind).

Fig. 4. CW x-band EPR spectra and simulated spectra for a.) BFR1–M (BFR1 with twoMnions bound per subunit) b.) BFR1–ZM (BFR1 complex with twomanganese ions and oneZnCe6 bound per subunit). Protein concentration 400 μM, Mn 800 μM, ZnCe6 200 μM.Microwave frequency 9.44 GHz, microwave power 2 mW, modulation amplitude 2 mT,temperature 5 K.

Fig. 5. The temperature dependence of the signals. (a) (red ◊) and b) (blue □) and thesubsequent linear temperature dependence of (a) minus (b) producing (c) (green Δ).

1116 B. Conlan et al. / Biochimica et Biophysica Acta 1787 (2009) 1112–1121

approximately the same as the estimate from the number of boundZn–chlorins per Mn cluster as estimated by ITC. The spectrumgenerated by the subtraction of the original dimer spectrum appearedas a ‘simple’ derivative about g∼2. The spectra showed near lineartemperature dependence (Fig. 5c). Subsequent modelling of thissignal using the same formalism as described above suggested that itcould be explained by a substantial drop in the coupling between thetwometal centres, with the exchange component dropping to ∼0.01 Kand the Mn–Mn distance lengthening to ∼5 Å.

The binding of ZnCe6 also resulted in the formation of EPR signalscontaining hyperfine structure centred at g=4.5 and g=9.8 (Fig. 4a).These signals span from 50–185 mT, and were assigned to monomericMnII bound within a highly structured, weak ligand field. Thehyperfine splitting of this signal is approximately 9.5 mT, which isconsistent with MnII coordinated octahedrally with oxygen andnitrogen ligands, as commonly seen in a number of manganeseenzymes [47–49]. Interestingly, BFR1–ZM samples prepared in thedark exhibited an unusual multiline EPR signal (Figure S2) This signalwas super-imposed on the broad g=2 homovalent Mn2

II,II spectralcomponents observed in the dark. This multiline has 5 mT splittingand stretches from 290–410 mT, with up to 23 lines present. Thismultiline signal is commonly attributed to Mn2

II,II dimer systems [50].

3.2. Light-induced oxidation state changes

3.2.1. O2 consumption of the di-metal site of BFR1

Bacterioferritins are members of a class of spherical shell-like ironstorage proteins that contain a ferroxidase site within the proteinwhich catalyses the oxidation and hydrolysis of iron. The oxidationreaction involves the uptake of oxygen from solution and as aconsequence the reaction can be monitored using a Clarke typeoxygen electrode. Wild type BFR and BFR1 both oxidized FeII to FeIII

with the concomitant uptake of 0.25 O2 molecules per Fe atomwhen afive fold excess of iron was present (Table 2). Both forms oxidize ironwith ratios of oxygen to Fe identical to the literature [51] but the rateof iron oxidation for the mutant form is slightly higher than for thewild type. The addition of ZnCe6 to the BFR1 protein did not change theamount of oxygen taken up nor significantly affect the rate of ironoxidation in the dark. Illumination of the BFR1–Z protein with ironbound after iron oxidation had gone to completion produced furtheroxygen uptake at a slower rate. Additional light-induced oxygenuptake totalled approximately the same as dark iron oxygen uptake(see Fig. 6a, and Table 2).

The BFR1–ZM complex, when measured under the same condi-tions, did not auto-oxidize Mn in the dark. Interestingly though,subsequent illumination of this complex did lead to consumption ofoxygen (Fig. 6b–d). Control measurements demonstrated that only asmall amount of oxygen was consumed by the Mn free (BFR1–Z), andfor a protein free solution of ZnCe6 (Fig. 6e and f). Similarly, whenheme was bound to the protein in place of ZnCe6 and manganesebound at the di-metal site, no oxygen uptakewas observed in the darkor upon illumination (Table 2). Buffered pH 6.5 Mn2+ solutionsshowed very little oxygen uptake when illuminated (Fig. 6b). Theoxygen uptake rate of the BFR1–ZM complex was increased by theaddition of superoxide dismutase (SOD) and catalase but the totalamount of oxygen consumed was almost the same (Table 2). Thissuggests that SOD and catalase may have a protective function as theiraddition should slow ZnCe6 breakdown from reactionwith superoxideor peroxide. No oxygen was released upon dark adaptation. Moreoxygen was consumed when a five fold excess of Mn was addedsuggesting that more Mn was oxidized. The same amount of oxygenuptake was found for BFR1–Z with excess iron. Both consumed a totalof ∼0.5 mol of oxygen per metal ion.

3.2.2. Light-induced chlorin oxidation/tyrosine oxidation as measuredby EPR

Light-induced oxidation of the bound ZnCe6 was readily observedby EPR (Fig. 7a). Upon illumination of the BFR1–Z complex (i.e. no Mnpresent) with short actinic light flashes at room temperature anarrow 0.75 mT wide radical centred at g=2.0022 was generated.Under the same conditions the BFR1–ZM complex generated abroader signal (2.5 mT radical) centred at g=2.0058. This broaderradical resolved a 4–6 peak hyperfine structure with p–p spacing∼1.5 mT characteristic of an oxidized tyrosine (Fig. 7b) [52–54]. Acorrected spin count for each species revealed that the tyrosine signalarea was approximately 6 times more intense than that of the ZnCe6radical signal.

Table 2Oxygen uptake data.

Sample Total oxygenuptake (μM)

O2 uptake/metal ion

kmax

(μMol/s)

Controls Dark Fe 300 μM (no protein) 29 0.10 0.3Light BFR1–Z 39 – 0.3

ZnCe6 Mn 60 μM (no protein) 40 – 0.1BFR1–Mn Dark BFR1–ZM Mn 60 μM 9 0.14 0.1

Light BFR1 heme Mn 60 μM 7 0.12 0.1BFR1–ZM Mn 60 μM 127 2.12 0.6BFR1–ZM Mn 60 μMCatalase & SOD 133 2.22 1.3BFR1–ZM Mn 300 μM 147 0.49 0.7

BFR1–Fe Dark BFR1–Z Fe 300 μM 81 0.27 1.4wt-BFR Fe 300 μM 81 0.27 0.9BFR1 Fe 300 μM 71 0.24 1.3

Light BFR1–Z Fe 300 μM 156a (75)b 0.52a (0.25)b –a (0.2)b

Note. BFR 30 μM, ZnCe6 15 μM, samples with metal contained either stoichiometric amounts (60 μM) or a 5 fold excess (300 μM). aTotal O2 consumption light and dark cumulative.bAdditional O2 consumption in light (as compared to dark). Rates are an average over 10 s.

1117B. Conlan et al. / Biochimica et Biophysica Acta 1787 (2009) 1112–1121

3.2.3. Light-induced Mn2+ oxidation as measured by EPRBFR1–ZM complexes were prepared containing either stoichio-

metric (2Mn2+ per BFR unit) or 5-fold excessMn2+ (10Mn2+ per BFRunit) in the dark. Excess Mn2+ (unbound) appeared as a narrow

Fig. 6. Oxygen uptake measurements carried out on a Clarke type oxygen electrode. Thelight was turned on after 60 s for the measurement of all samples except (a) which wasmeasured in complete darkness and (f) where the light was turned on after oxygenuptake in the dark had ceased. (a) BZMMn 60 μMDark. (b) ZnCe6 Mn 60 μM. (c) BZ. (d)BZM Mn 60 μM. (e) BZM Mn 300 μM. (f) BZ Fe 300 μM. Protein concentration was30 μM, ZnCe6 15 μM in 50 mM MES pH 6.5.

Fig. 7. EPR spectra for the BFR1–Z complex after three flashes of light, withoutmanganese (a) andwithmanganese bound (b). Trace (a) reveals a narrow ZnCe6 radicalcation with a peak to peak width of 0.75 mT, centred at g=2.0022. Trace (b) displays abroad signal with a width of 2.5 mT, centred at g=2.0058. Figure is arbitrarily scaled.Protein concentration 150 μM, ZnCe6 75 μM, Mn 240 μM. Microwave frequency9.44 GHz, microwave power 1 μW, modulation amplitude 0.3 mT, temperature 5 K.

100 mT wide 6 line EPR signal centred at g∼2 characteristic ofMn2+(H2O)6 [55]. This signal overlaid the structured MnII

2 dimersignal discussed above (see Mn Binding to BFR1). The Mn2

II,II dimersignal was quantitatively the same for both samples.

Upon illumination of the sample containing stoichiometric Mn2+

the structured Mn2II,II dimer signal decreased by ∼1/3. No new Mn

signal was observed in perpendicular or parallel mode EPR. Smallspectral features observed in parallel mode EPR were attributed tobackground contaminant O2.

As with stoichiometric samples, illumination of samples thatcontained excess Mn2+, led to a reduction of the structured Mn2

II,II

dimer by ∼1/3. In these samples the additional Mn2+ six-line signalpresent was also partially reduced by∼25% (see Table 3). Concomitantwith the loss of unboundMn2+ (six-line) was the appearance of broadparallel polarized signal most likely arising from an even spin system(Fig. 8). This new signal resolves a hyperfine structure with p–pspacing of ∼15.6 mT. This spacing is larger than previously reportedfor monomeric MnIII [37–39].

4. Discussion

We have taken a bioengineering approach to study the light-activated electron transport in the water oxidation reactions duringoxygenic photosynthesis. Bacterioferritin from E. coli was geneticallyand biochemically modified to produce a minimalist model of the PSIIdonor side in which analogues of PSII cofactors were bound. Theseincluded a photo-oxidizable chlorin, a redox-active tyrosine, and a di-nuclear manganese centre (Fig. 9). This model system demonstratesboth porphyrin excitation and metal complex oxidation.

4.1. Cofactor-protein assembly

Upon titration of the BFR1 into a solution of ZnCe6, ITC analysisreveals that only 82% of the BFR1 binds ZnCe6 (Table 1). The smallfraction which does not bind may be due to partial aggregation of thechlorin in solution, even at the low concentrations used, or it mayreflect heterogeneity in the sample; i.e. some of the binding pockets

Table 3Relative metal ion EPR signal intensities.

Sample Dark Light

Structured g=2 Six-line Structured g=2 Six-line

BFR1ZM Mn 60 μM 100% 0 66% 0%BFR1ZM Mn 300 μM 100% 400% 65% 300%

Note. BFR 30 μM, ZnCe6 15 μM, metal concentration either stoichiometric amounts(60 μM) or a 5 fold excess (300 μM) concentration per BFR1 monomer.

Fig. 9. Diagram of the amino acid ligands to the di-manganese centre with ZnCe6 boundat the interface of the BFR dimeric complex. Residues from the A subunit are in blue andthe B subunit are in salmon. The three tyrosines have been labelled on either side of theZnCe6 group to show the orientation of the dimer subunits relative to each other. Thealpha helices are pictured as grey rods. The picture is modified from the PDB–1BFR.

Fig. 8. Parallel mode EPR spectra of the light minus dark signals of BFR1–ZM with a fivefold excess of Mn. The inset is a higher resolution scan of the multiline signal.Microwave frequency 9.34 GHz, microwave power 10mW, modulation amplitude 1 mT,temperature 5 K. Protein concentration 250 μM, 125 μM ZnCe6, 2500 μM MnCl2 in50 mM MES pH 6.5.

1118 B. Conlan et al. / Biochimica et Biophysica Acta 1787 (2009) 1112–1121

are perturbed. The control mutant (M52H) did not bind anyporphyrins (chlorin or heme) possibly due to steric hindrance throughthe introduction of two large histidine residues in the heme bindingpocket, preventing dimer formation. ITC analysis also revealed thatbinding of ZnCe6 to the hydrophobic heme pocket has a large enthalpyterm (Table 1). This large and negative ΔHb value reflects strong non-covalent interactions (van der Waals and hydrogen bonds) betweenZnCe6 and BFR1 relative to their interactionwith the solvent. The maindriving force for binding is most likely the hydrophobic interactionsbetween the porphyrin ring structure and the hydrophobic amino acidresidues lining the binding pocket.

As we showed earlier manganese ligates to the di-nuclear metalbinding site of bacterioferritin [2] via four glutamate residues (E18,E51, E94, E127) one from each helix and two histidine residues (H54,H130) [21]. E51 and E127 form di-μ-1,3-caboxylato bridges, whereeach oxygen atom is coordinated to a different metal ion, while E18and E94 form monodentate carboxylate ligands. H54 and H130coordinate to the metal ions separately through the Nδ nitrogen (Fig.9). ITC analysis of Mn binding to BFR1 revealed two distinct bindingsites in the darkwith very different association constants. Site 1 boundMn very tightly whilst site 2 binding was weaker by almost threeorders of magnitude. Binding of the first Mn ion has a large andnegative ΔHb1 value, which may reflect structural changes in theprotein upon metal ligation [56]. The second Mn binding has a smalland positive ΔHb but a larger ΔSb than site 1 indicating that binding

Mn to site 2 is entropically driven. Themost likely contributions to thispositive entropy are changes in the hydration of the protein and of themetal ion upon binding to the protein [57]. These observationscoincide with previous crystal structure findings for BFR withmanganese bound, where site 2 was found to have a lower incidenceof occupation than site 1 [21].

In comparison, during the photo-assembly of the tetra-nuclear Mncluster in PSII, the first MnII ion binds tightly at a high affinity site in theapo-complex [IM0 intermediate]. Light is then needed to start theassembly process in which the bound MnII is oxidized to MnIII and aproton is released [IM1]. The bound MnIII is unstable until a conforma-tional change occurs in the dark and allows the formation of the morestable MnIII–O–CaII [IM1⁎]. At this point the second MnII binds formingthe stable, spin-coupled cluster MnII–MnIII–O–CaII [IM2]. The formationof the μ-oxo bridge thus helps stabilize the intermediates until the fullassembly of the tetra-nuclear MnCa cluster [58].

The ITC titrations also showed that an average of 2.51 mol of Mnions bound per mole of BFR1 monomer. This is higher than expectedbut can be partly explained by the findings from the original crystalstructure where an extraneous Mn ion was shown to be bound at thefour-fold axis of the dodecamer of dimers by four Gln151 residuesinternal to the structure [23]. This would account for a further 0.25binding sites per mole of protein but still leaving 0.25 sitesunaccounted for. Nonspecific, low affinity binding sites have alsobeen detected in previous studies of the BFR protein. It is proposedthat these sites involve acidic residues located on the inner surface ofthe protein shell [59]. Under the conditions of these experiments theindividual thermodynamic parameters of the weak binding sitescannot be determined. This is due to the limiting form of the equationfor two classes of binding sites [60].

We characterized the oxidation states of the Mn centre by EPRmeasurements. Simulation of dark Mn EPR signals strongly suggeststhat manganese bound to the BFR1 protein remained in the 2+oxidation state and was present as a weakly spin-coupled Mn2

II,II

centre. The estimated Mn–Mn distance from EPR simulations was∼4 Å, commensurate with the crystal structure [21]. The addition ofZnCe6 appeared to alter approximately 50% of the Mn2

II,II complexes.This behaviourmay be rationalized as follows. Initially the homodimerhas two identical Mn clusters. Upon addition of ZnCe6 this symmetryis lost. The BFR proteinwas engineered to bind ZnCe6 between the twomonomeric BFR1 subunits via a methionine residue. Only onemonomer of the dimer subunits provides this ligand (the 5th axialcoordinate of the ZnCe6). The methionine residue is one amino acidfrom the crucial glutamic acid residue (E51) which ligates to both Mnions in the metal centre. It is possible that the binding of the ZnCe6disrupts the coupling through movement of the methionine and thusthe glutamic acid residue producing a carboxylate shift (presumablymoving the two metal centres further apart).

4.2. O2 consumption

BFR functions in biology to bind iron and oxidize Fe2II to Fe2III. Thisprocess involves the formation of a stabilizing μ-oxo bond betweenthe two metal ions. This reaction consumes dissolved O2 [61]. Thethree phases to this process involve (i.) metal binding and deprotona-tion of amino acid ligands followed by (ii.) oxidation of bound metalions with the formation of a single μ-oxo bond between the metal ionsand uptake of one oxygen atom per centre. Once all ferroxidase ironbinding sites are filled, and all the 48 iron atoms bound per dodecamerof dimers are oxidized, the complex then begins to oxidize excess FeII

within the ball and store it as ferric-oxy-hydroxide (iii.) [22,51,62].

i.) 2FeII+BFR→Fe2II,II–BFR +4H+

ii.) Fe2II,II–BFR +½O2→FeIII−O−FeIII–BFRiii.) 4FeII+O2+6H20→4FeO(OH)(core)+8H+

1119B. Conlan et al. / Biochimica et Biophysica Acta 1787 (2009) 1112–1121

The oxygen assay of the BFR1 protein (Fig. 6) with iron presentrevealed that 0.25 O2 molecules are taken up per iron oxidized. Thisagrees with the literature values for the wild type and the ZnCe6bound form, supporting the assumption that binding of the ZnCe6does not change the basic function of the protein. Minimal oxygenconsumption is observed for BFR1–ZM complex suggesting it cannotauto-oxidize MnII to MnIII in the dark.

Upon illumination the BFR1–ZM sample consumes ∼2 O2 per Mn.This value is 4 times larger than for the BFR with Fe bound and maysuggest that the di-Mn site constructs multiple μ-oxo bridges ascompared to when Fe is bound in the wild type enzyme and only asingle μ-oxo bridge is formed. Control samples – samples that lack Mni.e. BFR1–Z – did not consume a significant amount of oxygenconfirming that O2 is reacting with the Mn. In samples with a fivefold excess of manganese, a significant increase in O2 consumptionwas observed (∼20 μM). This likely indicates that ‘free’ Mn2+ isoxidized within the ball structure forming MnIII-oxy-hydroxide. ForMnIII oxy-hydroxide formation to occur multiple oxidation events arelikely occurring at the manganese binding site. The addition ofcatalase and SOD was found to increase the rate of oxygenconsumption but did not have any significant effect on the totalamount consumed4. We suggest that it acts to protect the chlorin,removing super-oxides and other reactive oxygen species.

4.3. Tyrosine oxidation

In the absence of manganese, illumination of the BFR1–Z producesa large ZnCe6•+ radical cation signal at g=2.0022. In contrast, theaddition of Mn results in the loss of the ZnCe6•+ signal and theappearance of a ‘tyrosine like’ signal. This new radical signal is unlikelyto be the ZnCe6•+ radical broadened due to through bond/throughspace interactions with the Mn site as the distance between the Mncentre and the ZnCe6 is large (∼14 Å). The intensity of the tyrosine likesignal is significantly in excess of the ZnCe6 (6 fold). This isunsurprising as tyrosine radicals are generally much longer livedthan chlorin radicals [63,64]. Here it is suspected that in the timebetween room temperature illumination and freezing the sampledown to 77 K (in the dark), a significant fraction of the chlorin radicaldecays away. The most likely mechanism for tyrosine oxidation iselectron hole migration from the oxidized ZnCe6•+ radical to a nearbytyrosine residue. ZnCe6 has a high estimated oxidation potential of∼+1.1 V [15] which is sufficient to oxidize a tyrosine. It is unclear whytyrosine oxidation does not occur when nomanganese ions are bound.When Mn (or any metal) binds to the BFR metal site deprotonation ofthe metal ligands occurs along with structural changes to the ligandsthemselves. These changes may alter the hydrogen bonding networkaround the tyrosine allowing proton coupled electron transfer.

There are three tyrosine residues in BFR (Y25, Y58, and Y45) thatare close enough to the ZnCe6 (10.6 Å, 11.0 Å, and 4.2 Å respectively)that they could potentially act as ‘efficient’ electron donors (Fig. 9). Ofthese only two (Y25 and Y58) are found at the metal binding site andboth are involved in the hydrogen bonding network of the site.Tyrosine 58 is not highly conserved and is commonly replaced by aleucine in other bacterioferritins. Tyrosine 25 is hydrogen bonded toglutamate 94 which is one of the metal ligands with the oxygen atomof this tyrosine placed 4.3 Å from the nearest Mn ion. Ferritins, whichhave very similar structure and function to bacterioferritin but do notbind heme, have a redox-active tyrosine within the hydrogen bondingnetwork of the metal binding site. This residue is highly conserved inferritins and bacterioferritins [65]. As a consequence, we suggest thattyrosine (Y25) is likely to be responsible for the tyrosine like radical inour system. This is currently being tested in our lab.

4 Even though this is the case it is possible that oxygen acts as the electron acceptorproducing hydrogen peroxide.

4.4. Mn2+ oxidation

The loss of MnII signals upon illumination of BFR1–ZM stronglysuggests thatMn oxidation occurs in this system. The broad structuredMn2

II,II signal centred at g=2 was found to decrease in BFR1–ZMsamples upon illumination. This structured Mn2

II,II signal decreasedon average by 34%.

No newMn EPR signals were observed upon loss of the MnII signalin BFR1–ZM.Wewould expect the one electron oxidation of an Mn2

II,II

to yield a mixed valence Mn2II,III dimer. Strongly coupled mixed

valance dimers (J≈10 cm−1) display characteristic ‘multiline’ signalsseen in perpendicular mode EPR centred at g∼2 with hyperfinesplittings of b90 G [30]. In contrast, theweakly coupledmixed valancedimer should still have broader perpendicular signals and couldpossibly resolve parallel polarization signals [30,66]. As neither ofthese is seen, it is suggested the Mn cluster undergoes two electronoxidation to an even spin Mn2

III,III system. It is not clear as to whetherboth proposed MnII oxidation events need to be photo-driven. As nointermediate state (Mn2

II,III) was observed, the second oxidation statemay not require light excitation of the chlorin. Alternatively the mixedvalance cluster may be more strongly oxidizing (electron donating) tothe oxidized tyrosine/chlorin than the Mn2

II,II precursor state. Oxygenbridged Mn2

III,III dimers with analogous ligand fields have beenobserved at approximately zero-field (g=20) [38]. We cannotdiscount the presence of these signals in our illuminated BFR complex.

The BFR1–ZM sample with a five fold excess of Mn revealed thesame decrease in amplitude of the structured g=2 Mn2

II,II signal uponillumination as seen in samples with stoichiometric Mn. The six-linesignal also showed a significant decrease upon illumination. Thissignal decreased by 25%, equivalent to 4 Mn oxidations per BFR dimer(48 Mn ions per dodecamer of dimers). Parallel polarization EPR onilluminated samples with excess Mn suggested the formation of amanganese species with even spin character, upon illumination of theBFR1–ZM complex; a broad signal centred at geff∼4–5 with clearhyperfine structure (Fig. 8). As this signal is not observed inilluminated samples with stoichiometric Mn, it is suggested that thissignal comes from a new Mn species unlikely to be bound to BFR.Comparing this to the native function of ferritins, it is tempting toassign the species to MnIII-oxy-hydroxide (MnIIIOOH) which isaccumulated in the core of the sphere structure [67].

MnIII is a 3d4 (S=2) transition ion [68]. There are potentially threegeometries that MnIII complexes in this protein ligand field can take: 6coordinate octahedral, 5 coordinate square-pyramidal or trigonal–bipyramidal [39]. The ground electronic state of the MnIII manifolddepending on the geometry of the ligand field can have two different3d4 configurations corresponding to the electronic hole position. Thehole either occupies the dx2–y2 orbital (ground state 5B1) or the dz2orbital (ground state 5A1). The hyperfine splitting for each of thesestates is different due to the effect of spin–orbit coupling [39,69–71]. Aground state of 5B1 yields a hyperfine splitting along AZ of ∼50 Gwhereas a ground state of 5A1 yields a hyperfine splitting of ∼100 G[39,70,72]. In this instance the sign of the Spin Hamiltonian parameterD indicates the ground state of the manifold. A positive D value yieldsan 5A1 ground state (as observed for the MnIII centre of the SODprotein) whereas a negative D values yields a 5B1 ground state (asobserved for the MnIII photo-assembly intermediate of PS II)[37,39,71,73].

Recently parallel polarization EPR signals have been reported forMnIII peroxo complexes (i.e. MnIIIOOH) [66]. These signals weregenerated by the oxidation of (L)MnII (where L=N-methyl-tris(2-pyridylmethyl)ethan-1,2-diamine) using a 1000 fold excess of H2O2.Two parallel polarization signals were observed: a structured signal atg∼8 with hyperfine splitting ∼50 G (analogous to MnIII signalsdescribed above [39,40,72]) and a less structured signal at g∼5 withhyperfine splitting ∼95 G. It was noted that the addition of base(triethyl-amine) led to the loss of the high field signal (g∼5).

1120 B. Conlan et al. / Biochimica et Biophysica Acta 1787 (2009) 1112–1121

Conversely the addition of acid (HClO4) leads to the loss of the lowfield signal (g∼8). As a consequence the low field signal (g∼8) wasassigned to the monocationic species [LMnIII(OO)]+. Both UV–Vis andESI-MS supported this assignment. XSophie simulations estimated thezero-field and hyperfine parameters of this complex to be: D=−2.9 cm−1, E/D=0.075, AZ=6.5×10−3 cm−1 [66]. Curiously, thesecond signal (g∼5) was not assigned to the corresponding protona-ted species [LMnIIIOOH]2+. It was noted that though the signalcould be modelled as arising from a MnIII signal (D=+2.9 cm−1,E/D=0.111), the resolved hyperfine structure (splittings of ∼95 G)was inconsistent with assignment to a MnIII [66]. From the abovediscussion this argument does not seem valid; MnIII canpotentially resolve hyperfine splitting of the order of ∼100 G.We note that the D value changes sign for these two complexes,which should reflect a re-ordering of electronic state manifold(ground state interchanges from 5B1 to 5A1) as seen between MnIII

photo-assembly intermediate of PS II and Mn-SOD protein, leadingto an apparent increase in hyperfine splitting observed [37,39].

Our newMn EPR signal observed in BFR bears a resemblance to theg∼5 signal seen in the above study [66] appearing at approximatelythe same g position (g∼4–5). It too has a large hyperfine spacing asseen for the potential [LMnIIIOOH]2+ species. We do acknowledgethough that the hyperfine spacing seen (∼150 G) in the ‘new Mn EPRsignal’ is exceedingly large for a monomeric MnIII. The unusually largehyperfine splitting may reflect the unique environment Mn centresoccupy in the core of the sphere protein structure (Mn–Mninteractions etc.). Alternatively, the new EPR signal may represent adi-Mn species, possibly a weakly coupled mixed valance dimer (i.e.MnIIMnIII).

5. Conclusion

In this study we have shown that the modified BFR1–ZM complexis capable of converting light energy into the stored higher oxidationstates of Mn. To our knowledge this work is the first example reportedin the literature of a bioengineered protein capable of photo-catalytic,multi-electron oxidation of a di-manganese centre. This representsone of the first steps in developing synthetic photo-catalytic, ‘greenenzymes’ that utilize light energy to catalyse oxygen evolution andhydrogen production.

Acknowledgements

BC was supported by an ANU Postgraduate Scholarship. Fundingwas supplied by ARC Grant DP0450421, by the Deutsche Forschungs-gemeinschaft (Me1629/2-4), Solar-H2, BioH2, and the Max-Planck-Gesellschaft which is gratefully acknowledged. PLD acknowledgessupport from the US Department of Energy grant DE-FG02-05ER46223. The authors would like to thank G.C. Dismukes for hishelpful discussions.

Appendix A. Supplementary data

Supplementary data associated with this article can be found, inthe online version, at doi:10.1016/j.bbabio.2009.04.011.

References

[1] T. Wydrzynski, K. Satoh, Photosystem II: The Light-Driven Water:PlastoquinoneOxidoreductase, Springer, Dordrecht, 2005.

[2] B. Conlan, Designing Photosystem II: molecular engineering of photo-catalyticproteins, Photosyn. Res. 98 (2008) 687–700.

[3] B.M. Discher, D. Noy, K.S. Reddy, S.X. Ye, J.D. Lear, C.C. Moser, P.L. Dutton, Self-assembly of membrane peptides and redox cofactors, Biophys. J. 84 (2003) 177A.

[4] B.R. Gibney, P.L. Dutton, Histidine placement in de novo-designed heme proteins,Protein Sci. 8 (1999) 1888–1898.

[5] D.A. Moffet, M.A. Case, J.C. House, K. Vogel, R.D. Williams, T.G. Spiro, G.L.McLendon, M.H. Hecht, Carbon monoxide binding by de novo heme proteins

derived from designed combinatorial libraries, J. Am. Chem. Soc. 123 (2001)2109–2115.

[6] W.F. DeGrado, Molecular design as an approach to understanding membraneprotein structure and function, FASEB J. 13 (1999) A1431.

[7] C. Moser, J.M. Keske, K. Warncke, R.S. Farid, P.L. Dutton, Nature of biologicalelectron-transfer, Nature 355 (1992) 796–802.

[8] C.C. Page, C.C. Moser, X.X. Chen, P.L. Dutton, Natural engineering principles ofelectron tunnelling in biological oxidation–reduction, Nature 402 (1999) 47–52.

[9] H.B. Gray, J.R. Winkler, Electron transfer in proteins, Annu. Rev. Biochem. 65(1996) 537–561.

[10] J.R. Winkler, A.J. Di Bilio, N.A. Farrow, J.H. Richards, H.B. Gray, Electron tunneling inbiological molecules, Pure. Appl. Chem. 71 (1999) 1753–1764.

[11] J.R. Winkler, H.B. Gray, Electron tunneling in proteins: role of the interveningmedium, J. Biol. Inorg. Chem. 2 (1997) 399–404.

[12] D. Noy, C.C. Moser, P.L. Dutton, Design and engineering of photosynthetic light-harvesting and electron transfer using length, time, and energy scales, Biochim.Biophys. Acta 1757 (2006) 90–105.

[13] C.C. Page, C.C. Moser, P.L. Dutton, Mechanism for electron transfer within andbetween proteins, Curr. Opin. Chem. Biol. 7 (2003) 551–556.

[14] T. Wydrzynski, W. Hillier, B. Conlan, Engineering model proteins for PhotosystemII function, Photosyn. Res. 94 (2007) 225–233.

[15] S. Hay, B.B. Wallace, T.A. Smith, K.P. Ghiggino, T. Wydrzynski, Protein engineeringof cytochrome b562 for quinone binding and light-induced electrons transfer, Proc.Natl. Acad. Sci. U. S. A. 101 (2004) 17675–17680.

[16] M. Thielges, G. Uyeda, A. Camara-Artigas, L. Kalman, J.C. Williams, J.P. Allen,Design of a redox-linked active metal site: manganese bound to bacterialreaction centers at a site resembling that of Photosystem II, Biochemistry 44(2005) 7389–7394.

[17] H.B. Gray, J.R. Winkler, Electron tunneling through proteins, Q. Rev. Biophys. 36(2003) 341–372.

[18] S.C. Andrews, N.E. Lebrun, V. Barynin, A.J. Thomson, G.R. Moore, J.R. Guest, P.M.Harrison, Site-directed replacement of the coaxial heme ligands of bacterioferritingenerates heme-free variants, J. Biol. Chem. 270 (1995) 23268–23274.

[19] S.C. Andrews, J.M.A. Smith, C. Hawkins, J.M. Williams, P.M. Harrison, J.R. Guest,Overproduction, purification and characterization of the bacterioferritin ofEscherichia coli and a C-terminally extended variant, Eur. J. Biochem. 213 (1993)329–338.

[20] G.D. Davis, C. Elisee, D.M. Newham, R.G. Harrison, New fusion protein systemsdesigned to give soluble expression in Escherichia coli, Biotechnol. Bioeng. 65(1999) 382–388.

[21] F. Frolow, A.J. Kalb, J. Yariv, Structure of a unique twofold symmetrical heme-binding site, Nat. Struct. Biol. 1 (1994) 453–460.

[22] N.E. LeBrun, S.C. Andrews, J.R. Guest, P.M. Harrison, G.R. Moore, A.J. Thomson,Identification of the ferroxidase center of Escherichia coli bacterioferritin,Biochem. J. 312 (1995) 385–392.

[23] A. Dautant, J.B. Meyer, J. Yariv, G. Precigoux, R.M. Sweet, A.J. Kalb, F. Frolow,Structure of a monoclinic crystal form of cytochrome b1 (bacterioferritin) from E.coli, Acta Crystallogr. D. 54 (1998) 16–24.

[24] P.M. Harrison, P. Arosio, Ferritins: molecular properties, iron storage function andcellular regulation, Biochim. Biophys. Acta 1275 (1996) 161–203.

[25] G.R. Moore, Bacterial 4-alpha-helical bundle cytochromes, Biochim. Biophys. Acta1058 (1991) 38–41.

[26] F.W.J. Teale, Cleavage of the haem-protein link by acid methylethylketone,Biochim. Biophys. Acta 35 (1959) 543.

[27] V.V. Barynin, M.M. Whittaker, S.V. Antonyuk, V.S. Lamzin, P.M. Harrison, P.J.Artymiuk, J.W. Whittaker, Crystal structure of manganese catalase from Lactoba-cillus plantarum, Structure 9 (2001) 725–738.

[28] D.W. Yoder, J. Hwang, J.E. Penner-Hahn, Chapter 16: Manganese Catalases, in: H.S.Sigel, A. (Eds.), Manganese and Its Role in Biological Processes, vol. 37, MarcellDekker, Inc., New York, 2000, pp. 527–557.

[29] K.O. Schafer, R. Bittl, F. Lendzian, V. Barynin, T. Weyhermuller, K. Wieghardt, W.Lubitz, MultifrequencyEPR , investigation of dimanganese catalase and related Mn(III)Mn(IV) complexes, J. Phys. Chem., B. 107 (2003) 1242–1250.

[30] C. Teutloff, K.O. Schafer, S. Sinnecker, V. Barynin, R. Bittl, K. Wieghardt, F. Lendzian,W. Lubitz, High-field EPR investigations of Mn(III)Mn(IV) and Mn(II)Mn(III)states of dimanganese catalase and related model systems, Magn. Reson. Chem.43. Spec. no. (2005) S51–64.

[31] B. Loll, J. Kern, W. Saenger, A. Zouni, J. Biesiadka, Towards complete cofactorarrangement in the 3.0 angstrom resolution structure of Photosystem II, Nature438 (2005) 1040–1044.

[32] E.R. Bauminger, P.M. Harrison, D. Hechel, I. Nowik, A. Treffry, Mossbauerspectroscopic investigation of structure–function relations in ferritins, Biochim.Biophys. Acta 1118 (1991) 48–58.

[33] A.R. Razeghifard, T. Wydrzynski, Binding of Zn–chlorin to a synthetic four-helixbundle peptide through histidine ligation, Biochemistry 42 (2003) 1024–1030.

[34] M.P. Hendrich, P.G. Debrunner, Electron-paramagnetic-res spectra of quintetferrous myoglobin and a model heme compound, J. Magn. Reson. 78 (1988)133–141.

[35] M.P. Hendrich, P.G. Debrunner, Integer-spin electron-paramagnetic resonance ofiron proteins, Biophys. J. 56 (1989) 489–506.

[36] S.L. Dexheimer, J.W. Gohdes, M.K. Chan, K.S. Hagen, W.H. Armstrong, M.P. Klein,Detection of EPR-spectra in S=2 states of trivalent manganese complexes, J. Am.Chem. Soc. 111 (1989) 8923–8925.

[37] K.A. Campbell, E. Yikilmaz, C.V. Grant, W. Gregor, A.F. Miller, R.D. Britt, ParallelpolarizationEPR , characterization of the Mn(III) center of oxidized manganesesuperoxide dismutase, J. Am. Chem. Soc. 121 (1999) 4714–4715.

1121B. Conlan et al. / Biochimica et Biophysica Acta 1787 (2009) 1112–1121

[38] M.M. Whittaker, V.V. Barynin, T. Igarashi, J.W. Whittaker, Outer sphere mutagen-esis of Lactobacillus plantarum manganese catalase disrupts the cluster core —mechanistic implications, Eur. J. Biochem. 270 (2003) 1102–1116.

[39] K.A. Campbell, D.A. Force, P.J. Nixon, F. Dole, B.A. Diner, R.D. Britt, Dual-mode EPRdetects the initial intermediate in photoassembly of the Photosystem II Mncluster: the influence of amino acid residue 170 of the D1 polypeptide on Mncoordination, J. Am. Chem. Soc. 122 (2000) 3754–3761.

[40] K.A. Campbell, M.R. Lashley, J.K. Wyatt, M.H. Nantz, R.D. Britt, Dual-modeEPR ,Study of Mn(III) salen and the Mn(III) salen-catalyzed epoxidation of cis-beta-methylstyrene, J. Am. Chem. Soc. 123 (2001) 5710–5719.

[41] T. Howard, J. Telser, V.J. DeRose, An electron paramagnetic resonance study of Mn2(H2O)(OAc)4(tmeda)2 (tmeda=N,N,N′,N′-Tetramethylethylenediamine): amodel for dinuclear manganese enzyme active sites, Inorg. Chem. 39 (2000)3379–3385.

[42] S.V. Khangulov, V.V. Barynin, N.V. Voevodskaya, A.I. Grebenko, ESR spectroscopyof the binuclear cluster of manganese ions in the active center ofMn-catalase fromThermus thermophilus, Biochim. Biophys. Acta 1020 (1990) 305–310.

[43] P.J. Pessiki, S.V. Khangulov, D.M. Ho, G.C. Dismukes, Structural and functionalmodels of the dimanganese catalase enzymes. 2. Structure, electrochemical,redox, and EPR properties, J. Am. Chem. Soc. 116 (1994) 891–897.

[44] B. Epel, K.O. Schafer, A. Quentmeier, C. Friedrich, W. Lubitz, Multifrequency EPRanalysis of the dimanganese cluster of the putative sulfate thiohydrolase SoxB ofParacoccus pantotrophus, J. Biol. Inorg. Chem. 10 (2005) 636–642.

[45] G.C. Dismukes, Manganese enzymes with binuclear active sites, Chem. Rev. 96(1996) 2909–2926.

[46] A.P. Golombek, M.P. Hendrich, Quantitative analysis of dinuclear manganese(II)EPR spectra, J. Magn. Reson. 165 (2003) 33–48.

[47] W.C. McGregor, S.I. Swierczek, B. Bennett, R.C. Holz, Characterization of thecatalytically active Mn(II)-loaded argE-encoded N-acetyl-L-ornithine deacetylasefrom Escherichia coli, J. Biol. Inorg. Chem. 12 (2007) 603–613.

[48] M.R. Schaab, B.M. Barney, W.A. Francisco, Kinetic and spectroscopic studies on thequercetin 2,3-dioxygenase from Bacillus subtilis, Biochemistry 45 (2006)1009–1016.

[49] A.K. Whiting, Y.R. Boldt, M.P. Hendrich, L.P. Wackett, L. Que, Manganese(II)-dependent extradiol-cleaving catechol dioxygenase from Archrobacter globiformisCM-2, Biochemistry 35 (1996) 160–170.

[50] A.J. Wu, J.E. Penner-Hahn, V.L. Pecoraro, Structural, spectroscopic, and reactivitymodels for the manganese catalases, Chem. Rev. 104 (2004) 903–938.

[51] X. Yang, N.E. Le Brun, A.J. Thomson, G.R. Moore, N.D. Chasteen, The iron oxidationand hydrolysis chemistry of Escherichia coli bacterioferritin, Biochemistry 39(2000) 4915–4923.

[52] G. Bleifuss, M. Kolberg, S. Potsch, W. Hofbauer, R. Bittl, W. Lubitz, A. Graslund, G.Lassmann, F. Lendzian, Tryptophan and tyrosine radicals in ribonucleotidereductase: a comparative high-field EPR study at 94 GHz, Biochemistry 40(2001) 15362–15368.

[53] W. Hofbauer, A. Zouni, R. Bittl, J. Kern, P. Orth, F. Lendzian, P. Fromme, H.T. Witt, W.Lubitz, Photosystem II single crystals studied by EPR spectroscopy at 94 GHz: thetyrosine radical Y–D(center dot), Proc. Natl. Acad. Sci. U. S. A. 98 (2001)6623–6628.

[54] S. Un, A. Boussac, M. Sugiura, Characterization of the tyrosine-Z radical and itsenvironment in the spin-coupled S(2)Tyr(Z)(center dot) state of Photosystem IIfrom Thermosynechococcus elongatus, Biochemistry 46 (2007) 3138–3150.

[55] C.W. Hoganson, D.F. Ghanotakis, G.T. Babcock, C.F. Yocum, Mn-2+ reduces YZ+ inmanganese-depleted Photosystem-II preparations, Photosyn. Res. 22 (1989) 285–293.

[56] N.E. LeBrun, A.M. Keech, M.R. Mauk, A.G. Mauk, S.C. Andrews, A.J. Thomson, G.R.Moore, Charge compensated binding of divalent metals to bacterioferritin: H+release associated with cobalt(II) and zinc(II) binding at dinuclear metal sites,FEBS Lett. 397 (1996) 159–163.

[57] F. Bou-Abdallah, P. Arosio, S. Levi, C. Janus-Chandler, N.D. Chasteen, Defining metalion inhibitor interactions with recombinant human H- and L-chain ferritins andsite-directed variants: an isothermal titration calorimetry study, J. Biol. Inorg.Chem. 8 (2003) 489–497.

[58] J. Dasgupta, G.M. Ananyev, G.C. Dismukes, Photoassembly of the water-oxidizingcomplex in Photosystem II, Coord. Chem. Rev. 252 (2008) 347–360.

[59] S. Baaghil, A.J. Thomson, G.R. Moore, N.E. Le Brun, Studies of copper(II)-binding tobacterioferritin and its effect on iron(II) oxidation, J. Chem. Soc. Dalton (2002)811–818.

[60] E. Freire, O.L. Mayorga, M. Straume, Isothermal titration calorimetry, Anal. Chem.62 (1990) A950–A959.

[61] H.L. Liu, H.N. Zhou, W.M. Xing, H.F. Zhao, S.X. Li, J.F. Huang, R.C. Bi, 2.6 Angstromresolution crystal structure of the bacterioferritin from Azotobacter vinelandii,FEBS Lett. 573 (2004) 93–98.

[62] S. Baaghil, A. Lewin, G.R. Moore, N.E. Le Brun, Core formation in Escherichia colibacterioferritin requires a functional, Biochemistry 42 (2003) 14047–14056.

[63] X.S. Tang, M. Zheng, D.A. Chisholm, G.C. Dismukes, B.A. Diner, Investigation ofthe differences in the local protein environments surrounding tyrosine radicalsY–Z(center dot) and Y–D(center dot) in Photosystem II using wild-type and theD2-Tyr160Phe mutant of Synechocystis 6803, Biochemistry 35 (1996)1475–1484.

[64] Y. Miloslavina, M. Szczepaniak, M.G. Muller, J. Sander, M. Nowaczyk, M. Rogner,A.R. Holzwarth, Charge separation kinetics in intact Photosystem II coreparticles is trap-limited. A picosecond fluorescence study, Biochemistry 45(2006) 2436–2442.

[65] M.A. Carrondo, Ferritins, iron uptake and storage from the bacterioferritinviewpoint, EMBO J. 22 (2003) 1959–1968.

[66] S. Groni, P. Dorlet, G. Blain, S. Bourcier, R. Guillot, E. Anxolabehere-Mallart,Reactivity of an aminopyridine [LMnII](2+) complex with H2O2. Detection ofintermediates at low temperature, Inorg. Chem. 47 (2008) 3166–3172.

[67] B. Zhang, J.N. Harb, R.C. Davis, J.W. Kim, S.H. Chu, S. Choi, T.Miller, G.D.Watt, Kineticand thermodynamic characterization of the cobalt and manganese oxyhydroxidecores formed in horse spleen ferritin, Inorg. Chem. 44 (2005) 3738–3745.

[68] A. Abragam, B.I. Bleaney, Electron Paramagnetic Resonance of Transition Ions,Oxford University Press, Oxford, 1970.

[69] S.A. Al'tshuler, B.M. Kozyrev, Electron Paramagnetic Resonance in Compounds ofTransition Elements, 2nd ed. Halsted-Wiley, New York, NY, USA, 1974.

[70] H.J. Gerritsen, E.S. Sabisky, Paramagnetic resonance of trivalent manganese inrutile (TiO2), Phys. Rev. 132 (1963) 1507–1512.

[71] J.S. Griffith, Theory of Transition: Metal Ions, Cambridge University Press, 1971.[72] A.M. Tyryshkin, R.K. Watt, S.V. Baranov, J. Dasgupta, M.P. Hendrich, G.C. Dismukes,

Spectroscopic evidence for Ca2+ involvement in the assembly of the Mn4Cacluster in the photosynthetic water-oxidizing complex, Biochemistry 45 (2006)12876–12889.

[73] J.W. Whittaker, M.M. Whittaker, Active-site spectral studies on manganesesuperoxide-dismutase, J. Am. Chem. Soc. 113 (1991) 5528–5540.

SUPPORTING INFORMATION

The general Hamiltonian used for the MnII dimer takes the form:

IASSDSSgHSDSSSJ iiiiZFiii2121H Eq.1

Where: J exchange tensor (assumed to be isotropic)

D the dipolar tensor (fixed by the geometry of the system i.e. the Mn-Mn

inter-distance)

DZFi zero-field tensors

Ai Hyperfine tensors

The orientation of the two zero-field tenors, dipolar tensor, etc., were assumed to be co-

linear.

The hyperfine component was approximated by an isotropic line-broadening. The resultant

Hamiltonian (36x36 matrix) was solved numerically using Scilab-4.4.1, an open source

vector-based linear algebra package available from www.scilab.org. This package can be

implemented on most computer platforms. Typical powder pattern averages used 10-100

angles (defined below). A least squares minimization routine was employed to find the

optimal solutions for the above parameters.

Powder Pattern Averaging

The paramagnetic centers of a frozen ESR sample take all possible orientations (relative to

the field axis), often referred to as a powder pattern. The simulation code allows for this

effect by using combination of Euler rotations ( and ). This system only requires two

Euler rotations to define all unique orientations as rotation about the field axis does not alter

the eigenvalues of the system.

A generalized powder pattern code was adopted (shown below). The typical number of

angles used (n) ranged from 20-100. Dummy variables x and y were used to uniformly

sample the hemisphere. For a given x () angle 1 to p(x+1) y () angles were sampled (with

the entries of p(x) as defined below). Note: for x =0 (, was also set to zero.

For x = 0 to n-1

For y = 1 to p(x+1)

x)21n(

90sin

)21n(90

360)1x(p

)1x(p

3602

1y

21n

90x

and angles define the rotation matrix as below:

)cos()sin()sin()cos(

0)cos()sin(

)cos()cos()sin()cos()cos(

R

The unitary transformation 'RTRT* was applied to the tensors:

'RJRJ

'RgRg

'RDRD

*

Fe*

Fe

*

Transition Intensities

The intensity of transition between the ith and jth state is given the equation below:

2Q1QFeFe1 jSHgSgHiI

Where the oscillating H1 field lies in the x-y plane (perpendicular to H). All of its possible

orientations must be included to correctly estimate transition intensities (requires all three

Euler angles). This is achieved by averaging over the third Euler angle (). H1 can be

expressed as:

zyzyyyxyyyx

zxzxyxxxyxx

222

1

Sg2SiggSigg

Sg2SiggSigg

where

jsincos2

HiI

hence

0

sin

cos

H

Integrating with respect to yields:

)(

)2cos(2

))2sin(21(

2))2sin(2

1(2

d)2sin())2cos(1(2

))2cos(1(2

d))sin()cos((

22

2

0

22

20

22

220

The transition intensity can therefore be expressed as:

22222j

2

HiI

Simulation parameters for the MnII dimers

Mn2II,II dimer

(structured no ZnCe6)

Mn2II,II dimer

(unstructured with ZnCe6)

J 1.3 K ~0.001

r 3.95 Å 5.30 Å

D1 0.15 K 0.15 K1

E1/D1 0.33 0.33

D2 0.02 K 0.02 K4

E2/D2 0.12 0.14

width 280 G 280 G

1 D and E values are assumed to be the same for the simulation of the unstructured Mn2

II,II dimer as for the structured Mn2

II,II dimer signal. 2 Can take any value 0-0.33 (i.e. all allowed values for E/D). The parameter is non-unique.

Figure S1: UV-Visible spectrum of ZnCe6 free in solution (red) and bound to BFR1 (black).

All samples were in 25mM tricine, 100mM KCl pH7.7.

Figure S2: a.) Perpendicular X-band EPR spectra of the BFR1-ZM complex with an

underlying multiline signal apparent. b.) Cubic spline subtraction of the baseline to more

clearly visualize the multiline signal (magnified 2 times). Protein concentration 400μM,

ZnCe6 200μM, Mn 800μM. Microwave frequency 9.44 GHz, microwave power 2 mW,

modulation amplitude 2 mT, temperature 5 K.