Article pubs.acs.org/IC
Sulfido and Cysteine Ligation Changes at the Molybdenum Cofactor during Substrate Conversion by Formate Dehydrogenase (FDH) from Rhodobacter capsulatus Peer Schrapers,†,∇ Tobias Hartmann,‡,∇ Ramona Kositzki,† Holger Dau,† Stefan Reschke,‡ Carola Schulzke,§ Silke Leimkühler,*,‡ and Michael Haumann*,† †
Institut für Experimentalphysik, Freie Universität Berlin, 14195 Berlin, Germany Institut für Biochemie und Biologie, Molekulare Enzymologie, Universität Potsdam, 14476 Potsdam, Germany § Institut für Biochemie, Bioanorganische Chemie, Ernst-Moritz-Arndt-Universität Greifswald, 17487 Greifswald, Germany ‡
S Supporting Information *
ABSTRACT: Formate dehydrogenase (FDH) enzymes are attractive catalysts for potential carbon dioxide conversion applications. The FDH from Rhodobacter capsulatus (RcFDH) binds a bis-molybdopterin-guanine-dinucleotide (bis-MGD) cofactor, facilitating reversible formate (HCOO−) to CO2 oxidation. We characterized the molecular structure of the active site of wildtype RcFDH and protein variants using X-ray absorption spectroscopy (XAS) at the Mo K-edge. This approach has revealed concomitant binding of a sulfido ligand (Mo=S) and a conserved cysteine residue (S(Cys386)) to Mo(VI) in the active oxidized molybdenum cofactor (Moco), retention of such a coordination motif at Mo(V) in a chemically reduced enzyme, and replacement of only the S(Cys386) ligand by an oxygen of formate upon Mo(IV) formation. The lack of a Mo=S bond in RcFDH expressed in the absence of FdsC implies specific metal sulfuration by this bis-MGD binding chaperone. This process still functioned in the Cys386Ser variant, showing no Mo−S(Cys386) ligand, but retaining a Mo=S bond. The C386S variant and the protein expressed without FdsC were inactive in formate oxidation, supporting that both Mo− ligands are essential for catalysis. Low-pH inhibition of RcFDH was attributed to protonation at the conserved His387, supported by the enhanced activity of the His387Met variant at low pH, whereas inactive cofactor species showed sulfido-to-oxo group exchange at the Mo ion. Our results support that the sulfido and S(Cys386) ligands at Mo and a hydrogen-bonded network including His387 are crucial for positioning, deprotonation, and oxidation of formate during the reaction cycle of RcFDH.
■
INTRODUCTION
Protein crystallography has revealed that the metal-dependent FDHs from prokaryotic organisms belong to the so-called dimethyl sulfoxide (DMSO)-reductase family of enzymes,9,13,14 as characterized by a complex active-site cofactor, in which two pyranopterin molecules, each carrying a terminal guaninenucleotide moiety, are coordinated via their dithiolene-sulfur functionalities to the redox-active metal center in an often trigonal-prismatic geometry, resulting in the prototypic bisMGD structure (Figure 1). The two apical ligands complementing the metal coordination sphere have been proposed to be of crucial functional relevance for the proton and electron transfer reactions during the enzyme’s formate conversion chemistry.15−17 However, both their chemical identity and exact coordination configuration have remained ambiguous in many cases. In the Mo-FDH from Escherichia coli and in the W-FDH from Desulfovibrio gigas, the Se atom of a selenocysteine
The enzymatic conversion of carbon dioxide (CO2) has gained increasing interest in recent years, because such processes may be valuable for the sequestration of CO2 from the atmosphere, to combat the greenhouse effect, for providing important feedstock to the chemical industry (such as carbon monoxide), and for the generation of products such as formate (HCOO−) to serve in renewable energy applications.1−7 The diverse class of enzymes, which catalyzes the conversion between HCOO− and CO2, is denoted as formate dehydrogenase (FDH).8 Particularly interesting among the FDHs are metalloenzymes, bearing functionalized second- or even third-row transition metals, namely molybdenum (Mo) or tungsten (W), at their active sites (for recent reviews, see refs 5 and 9−11). Important discriminators between the metal-containing FDHs are the degree of preservation of their catalytic activity in the presence of O2 and the reversibility of the conversion chemistry to facilitate not only substrate oxidation, but also CO2 reduction to generate the formate product.12 © 2015 American Chemical Society
Received: December 2, 2014 Published: March 24, 2015 3260
DOI: 10.1021/ic502880y Inorg. Chem. 2015, 54, 3260−3271
Article
Inorganic Chemistry
conversion have been formulated using computational approaches: one involving ligation of the Se/S(Cys) to the metal throughout the cycle, and the other relying on a transient exchange of the Se/S(Cys) ligand by the substrate.15−17 In the present study, we have focused on characterization of the metal site structure in the FDH protein from Rhodobacter capsulatus (RcFDH), using X-ray absorption spectroscopy (XAS). The RcFDH is a NAD+-dependent enzyme and possesses a molybdenum cofactor (Moco),28,29 four [4Fe4S] clusters and one [2Fe2S] cluster in the α-subunit, one [4Fe4S] cluster and one FMN in the β-subunit, and one [2Fe2S] cluster in the γ-subunit.28 Previous biochemical and functional studies have revealed a bis-MGD cofactor in the RcFDH and a cysteine (Cys386) homologous to the metal-binding SeCys in the E. coli FDH. 28 Significant O2-tolerance was demonstrated for RcFDH,28 similar to other Mo-FDHs containing a putative cysteine ligand.30−33 RcFDH further shows reversible substrate conversion with increased CO2 reduction activity compared to the E. coli FDH.28 The RcFDH thus exhibits versatile catalytic properties, but a crystal structure of this type of enzymes is not available. The biosynthesis and insertion of the bis-MGD cofactor is a complex process34−36 involving in the case of RcFDH two system-specific chaperones, namely, FdsC and FdsD.28,37 FdsC was shown to bind the bis-MGD cofactor and a role of FdsC in bis-MGD insertion into RcFDH was suggested.28,37 However, when FdsC is absent during RcFDH expression, the enzyme was shown to be inactive, whereas, in the purified protein, bisMGD was present. This suggested a further role of FdsC in bisMGD modification/activation before the insertion into the FdsA subunit of RcFDH.28,37 FdsC can be substituted functionally by the homologous FdhD protein from E. coli, which has been proposed to add a sulfur ligand to the molybdenum,22 suggesting a similar role for RcFdsC.37 Accordingly, the Cys386 and a sulfido ligand at the molybdenum were suggested to be involved in the RcFDH activity.28,37 However, direct structural evidence for such a coordination has not been obtained so far. XAS is well-suited for the determination of metal-site fine structures, compared to protein crystallography, with respect to its significantly higher resolution of interatomic distances (a metal−ligand bond length change of ∼0.02 Å may be discriminated in high-quality EXAFS data), enhanced discrimination between oxygen and sulfur ligands, and facile access to metal oxidation states, as well as more-reliable preservation of high-valent species such as Mo(VI) by avoiding X-ray photoreduction due to a usually lower specific photon flux.38−44 This technique has significantly contributed to unravel the Mo site structures in various bis-MGD enzymes also in the post-structural era (see, for example, refs 34, 41, and 45−53). Discrimination of oxo (Mo=O) vs sulfido (Mo=S) ligation at the molybdenum by XAS can be expected, because of their significantly differing bond lengths (∼1.7 Å vs ∼2.2 Å) observed in synthetic model compounds.54−58 In an XAS study on the SeCys containing E. coli FDH, a close interaction between the Se atom and a dithiolene sulfur has been proposed,59 which may require reinterpretation, with regard to the crystal structures involving a sulfur ligand. Herein, we report the first Mo K-edge XAS study on RcFDH wildtype (WT) and protein variants (Cys386Ser and His387Met) expressed in the presence or absence of the bisMGD binding chaperone FdsC. The results imply that a Mo=S ligand, inserted by the FdsC chaperone, and Cys386
Figure 1. Metal ligation in bis-MGD cofactors in protein crystal structures. The shown structure represents the molybdenum cofactor in an FDH from Escherichia coli (PDB code 2IV2, 2.3 Å resolution, reduced state),18 exemplifying the overall bis-MGD cofactor geometry. Se(Cys140) (cys) and His141 (his) in E. coli correspond to S(Cys386) and His387 in RcFDH. The ligand X (cyan) has initially been modeled as an O atom18,18,21 in the structure shown and in the structures of the oxidized E. coli enzyme (PDB codes 1KQF, 1.6 Å resolution, and 1FDO, 2.8 Å resolution) and was later reinterpreted as an S atom.21 The structure of the tungsten FDH from Desulfovibrio gigas shows an Se(Cys) ligand and a vacant X-site (PDB code 1HOH, 1.8 Å resolution).19 The structure of molybdenum polysulfide reductase from Thermus thermophilus shows an S(Cys) ligand and X is an O atom (PDB code 2VPW, 3.1 Å resolution).25 The structure of molybdenum nitrate reductase from Cupriavidus necator shows an S(Cys) ligand and X is an S atom (PDB code 3ML1, 1.6 Å resolution).23 Mo/W−S(MGD) bond lengths range between 2.3 Å and 2.5 Å, Mo/W−Se/S(Cys) bond lengths are 2.1−2.6 Å, and Mo/ W−X bond lengths are 2.2−2.4 Å in these structures. Color code: Mo/W, magenta; S, yellow; Se/S(Cys), orange; O, red; N, blue; P, green; C, gray; protons were not resolved and therefore were omitted in the drawing.
(SeCys) residue has been shown to be a metal ligand in crystal structures.18−20 Reinterpretation of the crystal structure for the reduced E. coli enzyme, however, favored positioning of the SeCys at a larger distance to the molybdenum under such conditions.18,21 A second non-pyranopterin ligand was initially modeled as an oxygen species, probably reflecting a molybdenum-oxo (Mo=O) bond, in the Mo-FDH structures.18,20 The presence of a respective ligand was less clear in the W-FDH.19 In later revisions of the structures, binding of a sulfur species to molybdenum was suggested.9,19,21 Involvement of a sulfur has gained confidence from biochemical studies, because the system-specific chaperone FdhD has been shown to insert a sulfur ligand in the E. coli Mo-FDHs.22 A coordination motif with six respective sulfur ligands (two dithiolenes, a cysteine, and a terminal sulfur) to molybdenum has been assigned in the crystal structures of two periplasmic nitrate reductases.23,24 By comparison, a terminal oxygen ligand was suggested for the Mo active site of a polysulfide reductase.25 So far, the experimental data has not yet established an unambiguous relationship between the fundamentally different reaction types of bis-MGD enzymes,9 the varying O2-tolerance and catalytic bias in FDHs, as well as the Mo/W metalation, Se/S(Cys) ligation, and oxo/sulfido coordination. In addition, conserved amino acid residues in the vicinity of the cofactor in FDHs may be important for proton transfer (Figure 1).26−28 Based on the crystal data of FDHs, two different reaction mechanisms of formate 3261
DOI: 10.1021/ic502880y Inorg. Chem. 2015, 54, 3260−3271
Article
Inorganic Chemistry
value was 0.37 Å, and R0 represented mean values for Mo(VI) species (R0(Mo−O) = 1.87 Å, R0(Mo−S) = 2.33 Å),64−67
coordination at the Mo center are essential for catalytic activity. Mo-ligation changes in response to pH variation, metal site reduction, and amino acid exchange, as well as Cys386 substitution by formate at the reduced Mo center, support a reaction scheme of formate oxidation involving Mo(VI) to Mo(IV) redox cycling, substrate binding at molybdenum, and viable roles of Cys386 and His387 in proton handling.
■
BVS =
■
⎛ R0 − Ri ⎞ ⎟ B ⎠
∑ Ni exp⎜⎝
(1)
RESULTS Activity of RcFDH Protein Preparations. XAS at the Mo K-edge was carried out on purified wildtype (WT) RcFDH protein, which was prepared in the oxidized and reduced states at pH values of 7.0, 8.0, and 9.0. The protein was expressed in the presence or absence of the bis-MGD binding chaperone FdsC, treated with the substrate formate (HCOO−), or variants with amino acid exchanges (Cys386Ser and His387Met) at the Moco active site were analyzed. The relative catalytic activities of formate oxidation of the WT, ΔFdsC, C386S, and H387M RcFDH variants at pH values of 7.0, 8.0, and 9.0 are summarized in Table 1. RcFDHWT was almost inactive at pH
MATERIALS AND METHODS
Protein Sample Preparation and Activity Assay. Rhodobacter capsulatus formate dehydrogenase (RcFDH) was expressed and purified as described previously.28 The variants of RcFDH, Cys386Ser and His387Met, were obtained by PCR mutagenesis. Redox changes in RcFDH samples were induced by adding either 5 mM NAD+ as an oxidant or 10 mM NaDT or 50 mM sodium formate as reductants. For pH conditions of 8.0 and 9.0, 75 mM potassium phospate was used as a buffer, while 50 mM Tris/HCl buffer was used for pH conditions of 7.0. Samples were transferred to ultracentrifugation devices (VIVASPIN 20, 50-kDa cutoff; Sartorius AG, Göttingen, Germany) and sealed with parafilm inside an anaerobic chamber and concentrated to ∼1 mM of enzyme-bound Mo (4 °C, 3000g overnight centrifugation). Samples were finally transferred to sample holders (30 μL) for XAS and stored in liquid nitrogen. The activities of RcFDH WT and protein variants were analyzed with formate as the substrate, as described elsewhere.28 Measurements over a wide pH range (5−11) were carried out using overlapping buffer systems (100 mM Britton/ Robinson, 100 mM potassium phosphate, 100 mM Tris/HCl, 100 mM CHES, or 100 mM CAPS). X-ray Absorption Spectroscopy. XAS at the Mo K-edge was performed at the synchrotron SOLEIL (Paris, France) at the SAMBA bending-magnet beamline, with the storage ring operated in top-up mode (430 mA), as previously described.34,50,51 A double-crystal Si(220) monochromator was used for scanning of the excitation energy, and two palladium-coated mirrors in grazing incidence mode were used for focusing of the X-ray beam and for harmonics rejection. The X-ray spot size on the sample was set by slits to ∼3 × 1 mm2. The energy axis was calibrated (accuracy ±0.15 eV) using the first inflection point at 20003.9 eV in the simultaneously measured absorption spectrum of a Mo foil as a standard. Fluorescence-detected XAS spectra were measured using an energy-resolving 36-element monolithic planar germanium pixel array detector (Canberra), which was shielded by 10-μm-thick zirconium foil against scattered incident X-rays. The total incoming count rate was kept below 30 000 s−1 per element, to avoid detector saturation. Samples were held in a laboratory-built liquid-helium cryostat at 20 K. XAS spectra (scan range of 19 850−20 770 eV, approximately, up to k = 14.2 Å−1; 1−2 scans per sample spot) were corrected for detector deadtime. Averaging (6−10 scans per sample, 1−2 scans per spot), normalization, and extraction of EXAFS oscillations and conversion of the energy scale to the wave-vector (k) scale were performed as previously described.34,39 k3-weighted EXAFS spectra were simulated (S02 = 1.0) by a least-squares procedure using phase functions calculated with FEFF7,60,61 and Fourier transforms (FTs) were calculated using the in-house software SimX39,62 (k-range of 1.8−14.2 Å−1, cos2 windows extending over 10% at both k-range ends). E0 was refined to 20 014 ± 2 eV in the fit procedure. The fit quality was judged by calculation of the Fourier-filtered R-factor (RF).39 We note that, because of the statistical coupling between coordination numbers (N) and Debye− Waller parameters (2σ2) in the EXAFS fits, the precision in these values is in the range of 15%−25%.41 K-edge energies were determined at the 50% level of normalized XANES spectra (edge half-height). Areas of the pre-edge peak feature in the XANES were determined within an energy range of ∼20 001−20 015 eV after subtraction of a polynomial spline to remove the main edge rise from the spectra using the program XANDA.63 Bond valence sum (BVS) calculations were performed using eq 1 and Ni (coordination number) and Ri (interatomic distance) values derived from EXAFS or crystallographic data (the sum is over all molybdenum−ligand bonds). The used B-
Table 1. Relative Activities for Formate Oxidation of RcFDH WT and Variants Relative Activity (%)a RcFDH variant
pH 7.0
pH 8.0
pH 9.0
pH optimumb
WT ΔFdsC C386S H387M
6 0 0 48
88 0 0 56
100 0 0 21
9.0
7.5
a
Activities were normalized to the maximal WT enzyme activity at pH 9.0, which was set to 100%. bMeasured in a pH range of 5.0−11.0, using overlapping pH-buffer systems.
7.0 and most active at pH 9.0, and the RcFDHΔFdsC and RcFDHC386S variants were inactive in the entire pH range. In the RcFDHH387M variant, the pH optimum was shifted to a value close to 8.0, where it showed ∼56% WT activity, and lower activities were observed at pH 7.0 and 9.0. XANES and EXAFS spectra at the Mo K-edge of the RcFDH protein variants were collected and the spectral analysis is presented in the following. A Sulfido Ligand at the Mo Site in RcFDH. Mo-XANES spectra of oxidized RcFDH WT protein at pH values ranging from 7.0 to 9.0 are shown in Figure 2. The spectra of all RcFDHWTox samples were relatively featureless, showed only a small pre-edge absorption peak, and exhibited rather minor spectral differences at the varying pH values. Comparison of the mean pre-edge area (Apre‑edge) and K-edge energy (Eedge) of RcFDHWTox with the respective parameters of other molybdoenzymes and model compounds with known Mo coordination and redox level (Table 2) revealed that RcFDHWTox contained mostly Mo(VI) and suggested on average one or fewer short Mo−O bonds at the Moco (such as Mo=O or Mo− O−; we note that the assignment of the longer bonds of ∼1.8 Å to Mo−O− is formal and reflects the partial double bond and O− character of the ligand; the expected Mo(VI)−OH bond length is ∼1.9−2.0 Å (see ref 68)) (Figure 2, inset). The presumably noninteger number of Mo=O/−O− bonds and relatively low edge energy suggested admixtures of molybdenum species in other oxidation states and/or of cofactor sites with varying numbers of oxygen ligands at molybdenum and, therefore, heterogeneity in the metal coordination in RcFDHWTox. 3262
DOI: 10.1021/ic502880y Inorg. Chem. 2015, 54, 3260−3271
Article
Inorganic Chemistry Table 2. Mo K-Edge Energies and Pre-edge Areasa sample WT(ox) pH 7 WT(ox) pH 8 WT(ox) pH 9 WT(red) pH 9 WT(formate) pH 7 WT(formate) pH 9 C386S(ox) pH 7 H387M(ox) pH 7 H387M(ox) pH 9 ΔFdsC(ox) pH 7 ΔFdsC(ox) pH 9 b
(O4=)MoVI (O3=)MoVIS2 d (O2=)MoVIS3 e (O2=)(S=)MoVIS3 f (O=)MoVIS4 g (O=)MoIVS4 h (S=)MoIVS4 i Mo(IV)S6 c
E (K-edge) ± 0.1 [eV] RcFDHa 20013.0 20013.1 20013.2 20012.5 20012.8 20012.4 20013.3 20012.7 20012.9 20012.9 20012.7 References 20019.3 20017.5 20014.7 20014.3 20013.8 20011.3 20011.0 20010.7
A (pre-edge) ± 0.05 [r.u.] 0.35 0.38 0.40 0.43 0.35 0.36 0.31 0.43 0.45 0.19 0.13 2.13 1.30 0.80 0.92 0.49 0.43 0.40 0.05
a
RcFDH was oxidized (ox) by NAD+, reduced (red) by NaDT, or formate-reduced at the given pH. bMolybdate in aqueous solution (pH 7.5). cThe isolated mono-pyranopterin Moco.34 dThe monopyranopterin Moco in sulfite oxidase with an S(Cys) ligand at Mo.81 e The mono-pyranopterin Moco in xanthine oxidase with an Mo=S bond (Figure S3 in the Supporting Information). fThe bis-MPT Moco in MobA protein.34 gBisdithiolene Mo=O and hMo=S model complexes (Figure S2 in the Supporting Information) shifted by 2 eV to higher energies. iSolid molybdenum sulfide (MoS2) shifted by 2 eV to higher energies.
Figure 2. Mo XANES spectra of RcFDH samples. Spectra were vertically displaced for comparison. Spectra (solid lines) represent averaged data for RcFDH samples (WT, wildtype; ox, NAD+-oxidized; red, NaDT-reduced; ΔFdsC, expressed in the absence of FdsC; formate, reduced with formate; C386S and H387M, mutated variants) in a pH-range of 7.0−9.0; bluish lines, five WT(ox) spectra at pH 7.0, 8.0, or 9.0; broken lines, spectra of reference samples (a)−(h) (spectrum (d) is shown in Figure S2B in the Supporting Information, and spectra (e) and (g) are shown in Figure S3 in the Supporting Information) (Table 2); the spectrum of MoS2 (h) was up-shifted by 2 eV for comparison). The asterisk (*) marks the pre-edge feature and the horizontal bar (−) the edge half-height. Inset: correlation of preedge areas with edge energies. Open circles (○) denote references (Table 2 and Figures S2 and S3 in the Supporting Information) with varying Mo coordination (Table 2); solid symbols represent the mean values for two RcFDH variants (Table 2; bars mark the value ranges for pH 7.0−9.0). The linear fit (line) to the reference data was drawn to guide the eye.
for all RcFDH preparations (Table 3) suggested a bond length spread (1σ) of ∼0.1 Å, which was in agreement with bond lengths differences observed in crystal structures of bis-pterin ligated Mo sites in enzymes and model complexes.41,51 Noninteger numbers of Mo−O bonds (∼1.7 Å and ∼1.8 Å) were observed. The addition of further Mo−X distances (X = C, N, O, or S) at ∼2.65 and 2.85 Å further improved the fit quality. In particular, the inclusion of a short Mo−S bond strongly diminished RF by a factor of ∼2 (Table 3). A contour plot of the RF value revealed a minimum in a coordination number range of ∼0.4−0.6 bond per Mo ion, meaning that only about every second Mo ion carried such a sulfur ligand. A bond length of ∼2.17 Å was determined for this interaction (Figure 3). Comparative EXAFS analysis of two model compounds with either S=MoIVS4 or O=MoIVS4 motifs, using the same set of EXAFS phase functions and 2σ2 values for Mo=O/S bonds as for the RcFDH, yielded a Mo=O distance of 1.68 Å, a Mo=S distance of 2.16 Å, and coordination numbers close to the expected unity values (Figure S2 and Table S1 in the Supporting Information). EXAFS analysis of the monopyranopterin cofactor containing Mo(VI) in xanthine dehydrogenase (XDH) at pH 9.5 revealed the expected single Mo=S bond (2.18 Å)68,71 plus one shorter Mo−O bond (1.69 Å) and one longer Mo−O bond (1.79 Å), besides the two Mo−S bonds (2.4−2.5 Å) (Figure S3 and Table S1 in the Supporting Information). Protonation of an oxygen ligand at Mo(VI), for example, in XDH at acidic pH has been reported to elongate the Mo−OH bond to ∼2.0 Å.68 According to these results, we attribute the shortest distances to Mo=O (∼1.7 Å) and Mo−
The EXAFS spectra of RcFDHWTox samples are shown in Figure 3. The averaged spectrum obtained for samples at more alkaline pH was subjected to a detailed EXAFS simulation analysis (Table 3). Gradual refinement of the curve-fitting approach, starting from the inclusion of only two Mo−O/S distances, by the addition of further metal−ligand interactions, as judged by the decreasing error sum (RF) finally yielded an excellent fit quality (RF < 5%) (Table 3, fits 1−7). Bond valence sum (BVS) analysis has contributed to determination of the metal site structure in various molybdenum enzymes.59,69,70 The BVS that was calculated on the basis of the coordination numbers (N) and first-sphere molybdenum−ligand distances (R) from EXAFS in parallel increased to a value close to six, in agreement with predominant Mo(VI) species. The detection of close to four Mo−S bonds in the range of ∼2.3−2.4 Å indicated a bis-MGD Moco structure in the RcFDH. Notably, the Debye−Waller parameters (2σ2) of the Mo−S(pterin) bonds 3263
DOI: 10.1021/ic502880y Inorg. Chem. 2015, 54, 3260−3271
Article
Inorganic Chemistry
The EXAFS best-fit for RcFDHWTox yielded a total firstsphere coordination number of bonds shorter than ∼2.7 Å (N) and a respective BVS that were considerably larger than 6 and therefore exceeded the values expected for a hexa-coordinated Mo(VI) ion (Table 3, fit 7). The Mo=O/−O− bonds contributed particularly strongly to the N and BVS values. Background contributions to the low-frequency EXAFS oscillations of the short Mo−O distances (Figure 3) might cause an overestimation of their coordination numbers. However, the analysis of EXAFS spectra obtained using further enhanced background suppression revealed an optimal fit error for the processing procedure used for the data in Figure 3 and a simultaneous decrease of the N values of the Mo=O/−O− and Mo=S bonds for excess background suppression (Figure S3 in the Supporting Information). Selective diminishing of the N value of the Mo=O/−O− bonds therefore was not feasible. The relative abundance of Mo=O/−O− bonds, compared to the Mo−S(MGD) bonds, may further be enhanced by contributions from incomplete Mo centers, in which, for example, only one pyranopterin and three oxygen ligands were bound to molybdenum and, therefore, the used N value for the Mo− S(MGD) bonds (4) therefore was too large (Table 3). The comparable N values for the Mo=S (∼2.17 Å) and Mo−X (∼2.65 Å) bonds suggested that they belong to the same metal centers. This could be explained if the thiol sulfur of Cys386 was a molybdenum ligand in those centers that also contained a Mo=S bond. The longer bonds were assigned to C/N/O/S atoms in the second metal coordination sphere. Testing of these assumptions still yielded a very good EXAFS fit result (Table 3, fit 8), and even restricting the total first-sphere coordination number to six led to an RF value of
