Rational design of a mononuclear metal site into the archaeal Rieske-type protein scaffold.

Proteins containing Rieske-type [2Fe-2S] clusters play essential functions in all three domains of life. We engineered the two histidine ligands to the Rieske-type [2Fe-2S] cluster in the hyperthermophilic archaeal Rieske-type ferredoxin from Sulfolobus solfataricus to modify types and spacing of ligands and successfully converted the metal and cluster type at the redox-active site with a minimal structural change to a native Rieske-type protein scaffold. Spectroscopic analyses unambiguously established a rubredoxin-type mononuclear Fe3+/2+ center at the engineered local metal-binding site (Zn2+ occupies the iron site depending on the expression conditions). These results show the importance of types and spacing of ligands in the in vivo cluster recognition/insertion/assembly in biological metallosulfur protein scaffolds. We suggest that early ligand substitution and displacement events at the local metal-binding site(s) might have primarily allowed the metal and cluster type conversion in ancestral redox protein modules, which greatly enhanced their capabilities of conducting a wide range of unique redox chemistry in biological electron transfer conduits, using a limited number of basic protein scaffolds.

Natural selection allows the utilization of a limited number of protein scaffolds to produce proteins with different types of active sites for various biological catalysis, molecular recognition, and metabolic needs. Metal ions add new functionality to proteins, facilitating some of the most difficult biological catalytic reactions (1)(2)(3). For this reason, design and engineering of a specific metalbinding site in natural and de novo protein scaffolds to promote new and specific functionalities are attractive targets in the field of basic and applied protein design research (4,5).
The metallosulfur redox sites, containing sulfurs, usually from cysteinyl side chains, are the most common in biological redox-active metalloproteins (1). In particular, the iron-sulfur (Fe-S) proteins are widely distributed over all living organisms and have been considered to be of early evolutionary origin (2,6,7). The mononuclear iron core is the simplest form of Fe-S redox sites, present in small modular proteins such as rubredoxin (Rd) 1 and desulforedoxin. These proteins have the iron atom coordinated in an approximately tetrahedral geometry to the sulfur atoms of four cysteinyl residues (1). Other major forms of protein-bound Fe-S redox sites are polynuclear clusters (such as [2Fe-2S], [3Fe-4S], and [4Fe-4S] clusters) for which some specific synthesis/assembly enzymes are required for in vivo cluster formation and maturation steps (7)(8)(9)(10). These sites are also found within complex metalloprotein molecules themselves where they form part of the internal electron transfer conduit to or from the catalytic site (e.g. in some membrane-bound respiratory complexes (11)(12)(13)) as a result of modular evolution.
Among the biological Fe-S clusters with at least one noncysteinyl ligand, "Rieske-type" [2Fe-2S] clusters are ubiquitous in a variety of organisms, playing crucial electron transfer functions in respiratory chains, photosynthetic chains, and multicomponent oxygenase systems for biodegradation of aromatic and alkene compounds (14 -16). In contrast to regular plant-and vertebrate-type ferredoxins having complete cysteinyl ligations, the Rieske-type cluster has an asymmetric [2Fe-2S] core with the S ␥ atom of each of the two cysteine residues coordinated to one iron site and the N ␦ atom of each of the two histidine residues coordinated to the other iron site. The crystal structure of a mitochondrial Rieske protein domain fragment suggests that its cluster-binding loops have similar geometry as those found in the Rd and zinc ribbon scaffolds (17). The presence of the two histidine ligands to the Rieske centers is considered to be essential in respiratory cytochrome bc complexes in which the protonation state of the histidine ligands plays a crucial role in the quinoloxidizing Q o -site turnover (see Ref. 18 for review).
We have recently addressed the influence of substitution of each of two histidine ligands (His-44 and His-64) by cysteine on the properties of a low potential Rieske-type cluster in archaeal Rieske-type ferredoxin (ARF) from the hyperthermophile Sulfolobus solfataricus strain P-1 (19 -22) as a new tractable model. The replacement of one of the histidine ligands, His-64, by cysteine allowed the assembly of a new low potential [2Fe-2S] cluster with one histidine plus three cysteine ligands in the archaeal Rieske-type protein scaffold, whereas replacement of the other ligand, His-44, by cysteine generated a protein that failed in cluster insertion and/or assembly (20) (see Figs. 1 and 2, a, b, and e). Here we demonstrate the successful rational design and characterization, for the first time, of an engineered mononuclear metal (iron/zinc) site in the Rieske-type protein scaffold. Together with the previously reported, unexpected reverse conversion of the iron site by introduction of an alanine residue into the Clostridium pasteurianum Rd sequence (i.e. the incorporation of an oxidized [2Fe-2S] cluster into a polypeptide chain that normally binds a mononuclear iron site) by Meyer et al. (23), we discuss possible prototypal evolutionary patterns of early redox protein modules in the biological electron transfer system.

EXPERIMENTAL PROCEDURES
Escherichia coli strains HB101 (TaKaRa) used for cloning were grown in LB or terrific broth medium, with 50 g/ml kanamycin, when required. The expression vector, pET28a, was purchased from Novagen. Water was purified by a Milli-Q purification system (Millipore). Other chemicals mentioned in this study were of analytical grade.
The nucleotide sequence determination was performed by the dideoxy chain termination method with an automatic DNA sequencer, ABI PRISM 310 Genetic Analyzer (PE Biosystems). The DNA sequence was processed with the DNASIS v3.6 software (Hitachi Software Engineering Co., Ltd.). The homology search against databases was performed with the BEAUTY and BLAST network service (24). The structural alignments were constructed using a Swiss-PdbViewer v3.7b2 (http://expasy.ch/spdbv/mainpage.htm).
The arf gene coding for the ARF (ORF c06009, DDBJ accession number, AB047031) of S. solfataricus strain P-1 (DSM 1616 T ) has been cloned and sequenced (20). Site-directed mutagenesis was performed by the PCR mutagenesis technique with a QuikChange site-directed mutagenesis kit (Stratagene), using a pET28aARF vector harboring the arf gene (20) as a long template. For the double mutant (replacements of His-44 by cysteine and His-64 by cysteine (H44C/H64C)), PCR mutagenesis was carried out in a stepwise manner, with new sets of the following PCR primers: 5Ј-GGT GCG ATT TAT GCG GAT ATG AAT ATA GTC TTG AAA ACG GTG-3Ј and 5Ј-CAC CGT TTT CAA GAC TAT ATT CAT ATC CGC ATA AAT CGC ACC-3Ј for H64C, and 5Ј-GCA TAT TGT CCT TGT AAG GGA AGG AAT CTG G-3Ј and 5Ј-CCA GAT TCC TTC CCT TAC AAG GAC AAT ATG C-3Ј for H44C. For the H44I/K45C mutant (replacements of His-44 by isoleucine, and Lys-45 by cysteine) and the triple mutant (replacements of His-44 by isoleucine, Lys-45 by cysteine, and His-64 by cysteine (H44I/K45C/H64C)), PCR mutagenesis was carried out using pET28aARF and pET28aARF-H64C vectors as a template, respectively, with a new set of the following PCR primers: 5Ј-GCA TAT TGT CCT ATT TGC GGA AGG AAT CTG GAA TAT GGA GAG G-3Ј and 5Ј-CCT CTC CAT ATT CCA GAT TCC TTC CGC AAA TAG GAC AAT ATG C-3Ј for H44I/K45C. Each amplified PCR product was individually treated with DpnI, and transformed into E. coli HB101 competent cells. In each case, the nucleotide sequences of the resultant vectors were confirmed for both strands before subjecting to the second set of PCR mutagenesis.
The pET28aARF vectors harboring the arf gene and its variants were transformed separately into the host strain, E. coli BL21-Codon-Plus(DE3)-RIL strain (Stratagene). The transformants were grown overnight at 25°C in LB medium containing 50 g/ml kanamycin and 100 -200 M FeCl 3 , and the recombinant holoprotein was overproduced with 1 mM isopropyl 1-thio-␤-D-galactopyranoside for 24 h at 25°C. The cells were pelleted by centrifugation and stored at Ϫ80°C until use.
Purification of each recombinant holoprotein having a hexahistidinetag at the N terminus was performed as described previously (20,21), except that the heat treatment step (at 65°C for 15-30 min) was omitted for the H44C, H44I/K45C, and double (H44C/H64C) mutants. The recombinant holoprotein was further purified by Sephadex G-75 gel filtration column chromatography (Amersham Biosciences). The purified recombinant holoprotein was stored at either 4 or Ϫ80°C until use.
Absorption spectra were recorded with a Hitachi U3210 spectrophotometer or a Beckman DU-7400 spectrophotometer equipped with a thermoelectric cell holder. EPR measurements were performed by using a JEX-RE3X spectrometer equipped with an ES-CT470 Heli-Tran cryostat system and a Scientific Instruments digital temperature indicator/controller Model 9650 (25). Spin concentrations of purified recombinant proteins were estimated by double integration, with Cu-EDTA (0.1 mM) and Sulfolobus tokodaii sulredoxin (0.05 mM) (28) as standards. Low temperature resonance Raman spectra were recorded at 77 K using a Spex 750M Raman spectrometer fitted with a Spectrum-One 2048 ϫ 512 CCD camera and a Spectra-Physics 2017 Ar ϩ laser (output, 500 milliwatt) by collecting 45°backscattering off the surface of a frozen sample in a quartz microcell (21). The slit width of the spectrometer is 80 m, and a multiscan signal-averaging technique was employed to improve the signal-to-noise ratio. The spectral data were processed using KaleidaGraph v3.51 (Synergy Software).
Purified recombinant proteins were concentrated with a Centriprep-10 apparatus (Amicon) to ϳ0.3-0.5 mM. Further concentration was achieved by placing the samples under a stream of dry argon gas. The resultant samples (ϳ1-2 mM), containing 30% (v/v) glycerol, were frozen in 24 ϫ 3 ϫ 2 mm polycarbonate cuvettes with a Mylar-tape front window for x-ray absorption spectra studies (19). X-ray absorption spectra at the iron K-edge were recorded at 10 K at the Stanford Synchrotron Radiation Laboratory, beamline 7-3, with the SPEAR2 storage ring operating at 3.0 GeV and 60 -100 mA. A Si (220) double crystal monochromator (with one crystal detuned to 50% reflected intensity for harmonic rejection) and an energy-resolving 30-element Ge solid state array detector (provided by the National Institutes of Health/ Biological and Environmental Research Structural Molecular Biology Research Resource) were employed for data collection. Energies were calibrated using an internal iron foil standard, assigning the first inflection point to 7111.3 eV. All other data collection parameters were as reported previously (25,26).
EXAFS data analysis was performed with the EXAFSPAK software. Calibration and background removal were performed according to established procedures (25,27). The theoretical amplitude and phase shift functions were calculated using the ab initio code FEFF 8.2 (28). A scale factor (S 0 2 ) of 0.9 was used to fit experimental EXAFS amplitudes with those calculated by FEFF 8.2.

RESULTS AND DISCUSSION
Rational Design of an Archetypal Metal Site-Replacement of the two histidine ligands to the [2Fe-2S] cluster of S. solfataricus ARF by cysteine residues (in the double mutant) largely impaired the cluster assembly in the recombinant variant protein (Fig. 2, a and d). The remaining [2Fe-2S] ferredoxin-type cluster in the purified double mutant (H44C/H64C) underwent irreversible decomposition of the cluster upon dithionite reduction (data not shown) and was unstable to heat treatment at 65°C for 15 min. Similarly, replacement of His-44 (H44C, and H44I/K45C) resulted in impaired cluster assembly and iron binding, in contrast to H64C (20) (Figs. 1 and 2, a, b, e, and f). The significantly higher stability of the wild-type ARF (20) compared with the double mutant suggests natural optimization of the surrounding polypeptide for efficient [2Fe-2S] cluster insertion/assembly with the "two histidines-plus-two cysteines" coordination pattern in the Rieske-type protein scaffold, rather than complete cysteinyl coordination.
In light of the structural similarity of the biological Riesketype [2Fe-2S] clusters to the mononuclear iron center in Rd (17), which plays a central electron transfer function in the molecular oxygen-scavenging system in the hyperthermophiles (29,30), we replaced three residues (His-44, Lys-45, and His-64) in S. solfataricus ARF with cysteines and isoleucine (H44I/ K45C/H64C) to mimic the mononuclear iron site in the Rd from the hyperthermophile Pyrococcus furiosus (31) (Fig. 1). These mutations resulted in heterologous overproduction of a rubycolored recombinant protein in E. coli, in stark contrast to the dark reddish purple color of the wild-type ARF (Fig. 2, a and c). The resultant triple mutant (H44I/K45C/H64C) was resistant to heat treatment at 65°C for 20 min like the wild-type ARF but unlike the double mutant.
The iron content in purified protein varies among preparations, and, in several cases, purified protein is enriched with zinc (50 -70% zinc) when it is overproduced with a smaller amount of supplementary iron (ϳ50 M or less) in the LB growth medium. Similar observations have been reported for recombinant Rds heterologously overproduced in E. coli in which zinc occupies the iron site (32-35). 2 We used the ironenriched triple variant (prepared by supplying 150 -200 M FeCl 3 in the growth medium during overexpression) in the following analyses.
Mononuclear Iron Site Designed into the Rieske-type Protein Scaffold-The visible absorption and EPR spectra of the triple 2 A zinc-enriched sample of the triple variant (ϳ90% zinc), obtained from host cells grown in the presence of 50 M ZnSO 4 , tends to aggregate at high protein concentrations presumably because of apoprotein formation (and/or possible changes in the geometry upon metal substitution that may potentially modify the protein structure). Therefore, this was not pursued in this study. variant showed the presence of a high spin ferric iron site (Fig.  2, c and g). It displays a resonance at g ϭ 9.22 associated with one principal direction of the lowest Kramers doublet (Ϯ1/2 or Ϯ5/2) at 4.3 K and a number of features in the g ϭ 3.9 -4.7 range associated with the three principal directions of the Ϯ3/2 doublet (Fig. 2g). These features are very similar to those reported for archaeal and bacterial Rds (34,36). In accordance, the dithionite-reduced triple variant showed the expected bleaching of the visible absorption spectrum, as reported for Rds (data not shown).

FIG. 2. Comparative visible near-UV absorption spectra of the purified wild-type ARF (a) and the H64C (b), triple (H44I/K45C/H64C) (c), double (H44C/H64C) (d), H44C (e), H44I/K45C (f) variants, X-band EPR spectra of the oxidized triple variant (g) at 4.3 (solid) and 12 K (dashed), and comparative resonance
The resonance Raman spectrum at 77 K of the oxidized triple variant differs significantly from that of the wild-type protein (20,21) or other biological [2Fe-2S] 2ϩ clusters with complete cysteinyl ligations (37,38) and clearly indicates the Rd-like tetrahedral coordination geometry of the iron site (Fig. 2, h and  i). The dominant bands at 310 and 323 cm Ϫ1 are assigned to skeletal deformational vibrations having totally symmetric Fe-S t stretching displacements confined to within this frequency region, and two other bands at 361 and 381 cm Ϫ1 are assigned to skeletal deformational modes consisting of nontotally symmetric Fe-S t stretching vibrations (39).
Detailed structures of the iron site in the triple variant were investigated by iron K-edge x-ray absorption spectra, and the results were compared with that of C. pasteurianum Rd (23, 34) ( Fig. 3 and Table I). In agreement with the EPR and resonance Raman results shown in Fig. 2, the EXAFS and FT for the iron-enriched triple variant are more similar to bacterial Rd than to the previously reported ARF spectra (19,20) (Fig. 3, b  and c). There is a significant increase in both the intensity and distance of the first shell Fourier transform peak, indicative of a substitution of sulfur ligands for histidine nitrogens and a more ordered first shell coordination sphere. The iron edge spectrum (Fig. 3a) for the oxidized triple variant exhibits the 1s-3d feature at 7113 eV expected for approximately tetrahedral environment. Upon dithionite reduction, the edge shifts to lower energy (Fig. 3a) and the Fe-S distance increases (Fig. 3c and Table I), as observed for bacterial Rds. For comparison, the artificial [2Fe-2S] sites engineered into the Rd (23) and ARF (20) scaffolds undergo irreversible cluster degradation and iron release upon dithionite reduction, presumably because of an unfavorable coordination geometry of the cluster (40); this was not observed in the present case.
The average Fe-S bond lengths of the iron-enriched triple variant (2.27 Å in the oxidized state and 2.32 Å in the dithionite-reduced state) are comparable with those of archaeal and bacterial Rds (2.28 and 2.33 Å, respectively) (23,34,41) (Table  I). Furthermore, the Fe-Fe interaction around 2.7 Å observed in the Fourier transform spectrum of the wild-type ARF (19) is not observed in the triple variant (Fig. 3), and the inclusion of an iron shell around 2.7 Å to the ARF triple variant EXAFS data fitting was unsuccessful, resulting in an unreasonably large Debye-Waller factor (data not shown). These results collectively demonstrate the assembly of a redox-active, Rd-type mononuclear iron site with complete cysteinyl ligation in the triple variant. This case is strikingly different from the best characterized instance of the [4Fe-4S] u [3Fe-4S] interconversion in bacterial-type ferredoxins and aconitase, in which the overall geometry of a cubane core does not change significantly (1, 2, 25, 42).  b For Rieske-type ͓2Fe-2S͔ proteins, the EXAFS is a measure of the average coordination environment of both iron atoms in the cluster. One iron has a coordination sphere including four sulfur atoms and a distant iron, whereas the other iron has a sphere that includes two sulfur atoms, two nitrogen atoms, and a distant iron. Thus, the average coordination environment in the wild-type ARF is three sulfur atoms, one nitrogen atom, and a distant iron (Fits 4,5 ). The iron K-edge EXAFS data for the wild-type ARF are from Ref. 19 and for the Fe-Rd are from Ref. 23. General Discussion-Our successful rational design of the thermostable Rd-like, mononuclear iron site with complete cysteinyl ligation in the recombinant triple mutant protein (Figs. 1-3) experimentally shows that the in vivo assembly of a [2Fe-2S] cluster in the Rieske protein scaffold is determined by the nature and spacing of the ligands at the cluster-binding motif ( Figs. 1 and 2). It seems plausible to postulate that a "nativelike" semiordered structure of the cluster binding site in a folding intermediate, wherein the spacing and types of ligands near the protein surface should play a decisive role, may behave as a substrate in the enzyme-assisted dinuclear cluster assembly/maturation steps (7)(8)(9)(10). This is in accord with the previous report by Meyer et al. (23), clearly showing the (unexpected) assembly of an oxidized [2Fe-2S] cluster into a recombinant, single ligand-substituted (C42A) variant of C. pasteurianum Rd, whose polypeptide chain normally accommodates a mononuclear Fe(Cys) 4 (20)), we suggest that His-44 in the cluster-binding loop (Fig. 1) plays a crucial role in the cluster assembly and metal-type recognition in the archaeal Rieske-type protein scaffold.
The "two cysteines plus two histidines"-type coordination pattern commonly observed in the Rieske-type proteins/domains (17,43,44) might have been preferred in the modular evolution process of an early Rieske-type protein scaffold, as a consequence of the geometric tolerance of the metal-binding loops for the [2Fe-2S] cluster insertion and/or assembly. The ligand exchange and metal-and/or cluster-type conversion might have been key evolutionary events from an archetypal mononuclear metalloprotein module toward a modern, high potential Rieske protein subunit of the respiratory complexes III, in which one of two histidyl ligands plays crucial electron/ proton transfer roles in the quinol-oxidizing Q o -site catalysis (16,18). Likewise, the dinuclear copper center in the Cu A site of mitochondrial and bacterial respiratory complexes IV, which is the primary electron acceptor site from mobile cytochrome c, is known to be structurally and evolutionarily related to the mononuclear copper proteins such as azurin and plastocyanin in the photosynthetic electron transfer chain (45)(46)(47)(48)(49)(50)(51). Possible common prototypal evolutionary patterns in both cases are (i) the mono-to dinuclear metal center conversion with a few residue substitutions in the immediate metal-binding site and (ii) the modular evolution from water-soluble (mononucleartype) ancestral protein modules, with minimal changes of the native protein scaffolds (Fig. 4). This evolutionary pattern can explain the complicated quaternary structural features of a wide range of redox metalloenzyme complexes, yet consists of a relatively limited number of protein scaffolds within a variety of biological electron transfer pathways. It is envisaged that the evolutionary origins of some primordial redox protein modules may have preceded the divergence of the three different domains of life (52).
A likely biological and evolutionary benefit of having a polynuclear cluster site in a complex metalloenzyme would be that the cluster synthesis/assembly can be more strictly controlled by one or more specific synthetic and assembly apparatuses (7-10) thereby facilitating a unique redox chemistry for specific cellular needs; the simple binding of a mononuclear transient metal site in a primordial metalloprotein might have been more severely influenced by the in vivo availability of environmental metal ions to the common ancestral organisms (because of the simpler metal binding equilibrium). We suggest that prototypal polynuclear cluster formations, followed by early modular evolutionary events, might have afforded a "stepwise" development of new catalytic and electron transfer functions of primordial complex metalloenzymes, consisting of ensembles of redox protein modules of convergent/divergent evolutionary origins using a limited number of basic protein scaffolds to meet the versatile requirements of early metabolisms and environmental conditions.