Identification of a Bis-molybdopterin Intermediate in Molybdenum Cofactor Biosynthesis in Escherichia coli*

Background: Some molybdoenzymes in prokaryotes contain the bis-molybdopterin guanine dinucleotide cofactor. Results: The bis-Mo-MPT cofactor is a novel intermediate in Moco biosynthesis in E. coli. Conclusion: Bis-MGD formed by MobA is fully functional and restores the catalytic activity in apoTorA. Significance: Bis-Mo-MPT assembles spontaneously on MobA prior to forming bis-MGD. The molybdenum cofactor is an important cofactor, and its biosynthesis is essential for many organisms, including humans. Its basic form comprises a single molybdopterin (MPT) unit, which binds a molybdenum ion bearing three oxygen ligands via a dithiolene function, thus forming Mo-MPT. In bacteria, this form is modified to form the bis-MPT guanine dinucleotide cofactor with two MPT units coordinated at one molybdenum atom, which additionally contains GMPs bound to the terminal phosphate group of the MPTs (bis-MGD). The MobA protein catalyzes the nucleotide addition to MPT, but the mechanism of the biosynthesis of the bis-MGD cofactor has remained enigmatic. We have established an in vitro system for studying bis-MGD assembly using purified compounds. Quantification of the MPT/molybdenum and molybdenum/phosphorus ratios, time-dependent assays for MPT and MGD detection, and determination of the numbers and lengths of Mo–S and Mo–O bonds by X-ray absorption spectroscopy enabled identification of a novel bis-Mo-MPT intermediate on MobA prior to nucleotide attachment. The addition of Mg-GTP to MobA loaded with bis-Mo-MPT resulted in formation and release of the final bis-MGD product. This cofactor was fully functional and reconstituted the catalytic activity of apo-TMAO reductase (TorA). We propose a reaction sequence for bis-MGD formation, which involves 1) the formation of bis-Mo-MPT, 2) the addition of two GMP units to form bis-MGD on MobA, and 3) the release and transfer of the mature cofactor to the target protein TorA, in a reaction that is supported by the specific chaperone TorD, resulting in an active molybdoenzyme.

The molybdenum cofactor is an important cofactor, and its biosynthesis is essential for many organisms, including humans.

Its basic form comprises a single molybdopterin (MPT) unit, which binds a molybdenum ion bearing three oxygen ligands via a dithiolene function, thus forming Mo-MPT. In bacteria, this form is modified to form the bis-MPT guanine dinucleotide cofactor with two MPT units coordinated at one molybdenum atom, which additionally contains GMPs bound to the terminal phosphate group of the MPTs (bis-MGD). The
MobA protein catalyzes the nucleotide addition to MPT, but the mechanism of the biosynthesis of the bis-MGD cofactor has remained enigmatic. We have established an in vitro system for studying bis-MGD assembly using purified compounds. Quantification of the MPT/molybdenum and molybdenum/phosphorus ratios, time-dependent assays for MPT and MGD detection, and determination of the numbers and lengths of Mo-S and Mo-O bonds by X-ray absorption spectroscopy enabled identification of a novel bis-Mo-MPT intermediate on MobA prior to nucleotide attachment. The addition of Mg-GTP to MobA loaded with bis-Mo-MPT resulted in formation and release of the final bis-MGD product. This cofactor was fully functional and reconstituted the catalytic activity of apo-TMAO reductase (TorA). We propose a reaction sequence for bis-MGD formation, which involves 1) the formation of bis-Mo-MPT, 2) the addition of two GMP units to form bis-MGD on MobA, and 3) the release and transfer of the mature cofactor to the target protein TorA, in a reaction that is supported by the specific chaperone TorD, resulting in an active molybdoenzyme.
The biosynthesis of the molybdenum cofactor (Moco) 4 is an ancient, ubiquitous, and highly conserved pathway leading to the biochemical activation of molybdenum (1). In Moco, the molybdenum atom is coordinated to the dithiolene group of the 6-alkyl side chain of a tricyclic pyranopterin called molybdopterin (MPT) (2). Moco biosynthesis has been studied in detail in Escherichia coli by using a combination of biochemical, genetic, and structural approaches (3,4) and has been divided into four major steps: 1) formation of cyclic pyranopterin monophosphate from 5Ј-GTP (5,6), 2) formation of MPT from cyclic pyranopterin monophosphate by insertion of two sulfur atoms (7)(8)(9)(10), 3) insertion of molybdenum to form Mo-MPT via an adenylated MPT intermediate (11)(12)(13), and 4) additional modification by the covalent addition of GMP or CMP to the C4Ј phosphate of MPT via a pyrophosphate bond to form the MPTguanine or MPT-cytosine dinucleotide cofactors (MGD (14) or MCD (15)), respectively.
After the synthesis of MCD or MGD in E. coli, the cofactor can be further modified. In MCD-containing enzymes, like the periplasmic aldehyde oxidoreductase PaoABC (16), the Moco contains an equatorial sulfido ligand at the molybdenum atom, which is essential for the catalytic activity of this class of enzymes (17). For the final step of MGD biosynthesis, two cofactor molecules are ligated to one molybdenum atom, forming the bis-MGD cofactor (18). In E. coli, GMP attachment to Mo-MPT is catalyzed by the MobA and MobB proteins, thereby forming MGD (19). MobA is crucial for this reaction and acts as a GTP:molybdopterin guanylyltransferase (14), whereas MobB is not essential (20). The type of Moco and ligand composition at the molybdenum atom divides the molybdoenzymes of E. coli into three families with the following coordination environment: the sulfite oxidase family (dioxo Mo-MPT with a protein cysteinate ligand), the xanthine oxidase family (mono-oxo MCD with a terminal sulfur ligand), and the dimethyl sulfoxide (DMSO) reductase family (bis-MGD with one oxo and one amino acid ligand) (1,3). Most E. coli molybdoenzymes, like the TMAO reductase TorA, belong to the DMSO reductase family and utilize the bis-MGD form of Moco (21). However, it has remained unclear at which stage of Moco biosynthesis the bis-form of the MGD cofactor is built and whether this occurs on MobA or at the respective target enzyme during the insertion process.
It has been shown that MGD was only formed by MobA when the molybdenum atom was already ligated to MPT (22,23). The crystal structure of MobA has revealed two conserved binding sites, one of which was predicted to bind MPT and the other of which was proposed to bind GTP (24). The MobA enzyme has an overall ␣␤ architecture, in which the N-terminal domain of the molecule adopts a Rossmann fold (24). From the crystal structure and from previous studies, it is not known so far whether MobA, in addition to MGD formation, also catalyzes the step of the bis-MGD assembly (21).
The last steps of Moco modification, including the formation of bis-MGD, prepare the cofactor for insertion into the specific apoenzymes. The insertion step is catalyzed by Moco-binding molecular chaperones, which bind the respective molybdenum cofactor and insert it into the target molybdoenzyme (25). With a few exceptions, most of the molybdoenzymes have a specific chaperone for Moco insertion. One well studied example is the TorD/TorA system for TMAO reductase in E. coli. TorD was shown to be the specific chaperone for TorA (26) and plays a direct role in the insertion of Moco into apoTorA (27). During this reaction, TorD interacts with both MobA and apoTorA and further stabilizes apoTorA for Moco insertion to avoid a proteolytic attack of the latter. This is consistent with its role as "facilitator" of the bis-MGD insertion and maturation of the apoenzyme (21,25,28).
In this work, we have established an in vitro system for specifically addressing the mechanism for bis-MGD formation. By studies quantifying the metal and cofactor content, in addition to determination of the structure of the molybdenum center by X-ray absorption spectroscopy (XAS), we show that bis-Mo-MPT formation precedes nucleotide addition in the bis-MGD synthesis and that these steps are solely catalyzed by MobA. The detection of the bis-Mo-MPT intermediate is a novel finding for Moco biosynthesis in E. coli.
Moco Binding Experiments-MogA, MoeA, and MobA were incubated with Mo-MPT extracted from hSO (29,32) under anaerobic conditions. Mo-MPT was extracted from aliquots of hSO (240 -250 M) by incubation at 95°C for 4 min. The supernatant was filtered through Centricon ultrafiltration devices (10 kDa cut-off). The filtered Mo-MPT was added to either 30 -40 M MogA, MoeA, or MobA and incubated for 2 h in a total volume of 7-8 ml. After separation of the protein fractions using G25 columns (PD10, GE Healthcare), the proteins were concentrated to molybdenum contents of 0.5-1 mM for XAS measurements. The binding experiments were performed in the presence or absence of 2.5 mM MgCl 2 and 380 M GTP.
Metal and Cofactor Content Quantification-Molybdenum analysis and quantification of further components in protein samples were performed using inductively coupled plasma optical emission spectroscopy on a PerkinElmer Life Sciences Optima 2100DV instrument as described previously (33) or total reflection X-ray fluorescence analysis (34) on a PicoFox spectrometer (Bruker). By total reflection X-ray fluorescence analysis, molybdenum and phosphorus contents per protein were determined in samples, to which sodium phosphate and/or molybdenum acetate had been added as standards in a concentration series. The respective element concentrations in the protein samples were determined from linear fits to the magnitudes of the elemental K ␣ X-ray fluorescence of the sample series and extrapolation to the point of zero molybdenum or phosphorus addition (not shown), relative to cesium acetate and gallium elemental standards (Sigma) and relative to the respective protein concentration. MPT contents were determined fluorometrically after conversion of the molecule to form A, as described previously (35). The formation of MGD was assayed fluorometrically after its conversion to form A-GMP (36). The fluorescence of form A and form A-GMP was monitored by an Agilent 1260-series detector using excitation at 383 nm and emission detection at 450 nm.
Time-dependent MGD Formation by MobA-MGD formation as a function of the incubation time was assayed at room temperature in a total sample volume of 200 l, containing 5 M MobA, 1 mM MgCl 2 , 1 mM GTP, 50 l of supernatant of 300 M heat-denatured hSO (incubated at 95°C for 2 min), and 120 l of 100 mM Tris buffer (pH 7.2).
Bis-MGD Insertion into apoTorA-An in vitro assay was used for the insertion of in vitro synthesized bis-MGD into apoTorA (26). The assay consisted of apoTorA (1 M), MobA (0.955 M), GTP (1 mM), MgCl 2 (1 mM), and 200 l of Mo-MPT filtrate from heat-denatured hSO (250 M) (29) in a total volume of 300 l of 100 mM phosphate buffer (pH 6.5) and was incubated at 37°C under anaerobic conditions. The assay was performed in the presence or absence of TorD (3.74 M). TorA activity was measured in 3944 l of 100 mM phosphate buffer (pH 6.5), containing 20 l of 1.5 M TMAO, 16 l of 100 mM benzyl viologen, 30 l of the incubation mixture and was adjusted with sodium dithionite to an A 600 of 1.0. The oxidation of reduced benzyl viologen was monitored at 600 nm. The specific TorA activity was defined as the oxidation of 1 mol of benzyl viologen/min/mg of protein.
XAS-XAS at the molybdenum K-edge was performed at SOLEIL (Paris, France) at the SAMBA beamline as described previously (37), using an Si[220] double-crystal monochromator. The synchrotron was operated at a current of 400 mA in top-up mode. The incident energy axis was calibrated (accuracy Ϯ0.15 eV) using the first inflection point at 20,003.9 eV in the simultaneously measured absorption spectrum of a molybdenum foil as a standard. Fluorescence-detected XAS spectra were measured using energy-resolving 7-or 36-element solidstate germanium detectors (Canberra), which were shielded by 10-m zirconium foil against scattered incident X-rays. Samples were held in a liquid helium cryostat at 20 K. Dead timecorrected XAS spectra (1-2 scans/sample spot) were averaged (5-10 scans/sample) and normalized, and EXAFS oscillations were extracted as described previously (38). k 3 -weighted EXAFS spectra were simulated (S 0 2 ϭ 1.0) using phase functions calculated with FEFF7 (39). Fourier transforms of EXAFS spectra were calculated using in-house software and cos 2 windows extending over 10% at both k-range ends (k ϭ 2-14 Å Ϫ1 ). E 0 was refined to 20,014 Ϯ 2 eV in the fit procedure. The fit quality was judged by calculation of the Fourier-filtered R-factor (R F ) (38). The pre-edge structure of XANES spectra was isolated by subtracting a polynomial spline from the main K-edge rise using the software XANDA (XANES Dactyloscope for Windows, K. V. Klementiev; available on the World Wide Web). K-edge energies reflect values at 50% of normalized edge absorption (edge half-height).
Bond valence sum (BVS) calculations were performed using Equation 1 (41) and N (coordination number) and R (interatomic distance) values derived from EXAFS analysis (the sum is over all Mo-S and Mo-O bonds). The used B value was 0.37. For R 0 values, see the legend to Table 4.

Characterization of MGD Cofactor Formation by
MobA-To analyze the reaction catalyzed by MobA, we made use of an in vitro system consisting of purified MobA, Mo-MPT from hSO, MgCl 2 , and GTP (22). On the basis of this system, MGD formation was quantified fluorimetrically after conversion to its stable fluorescent degradation product form A-GMP (36). Fig. 1 shows that MobA catalyzed the step of MGD formation continuously in the reaction mixture. A saturation was reached after 2 h, due to the limitation of intact Mo-MPT in the incubation mixture. In addition, the reaction was dependent on the presence of MgCl 2 , as suggested by the Rossman fold of the protein, showing that GTP is only bound and converted as a Mg-GTP complex. However, the results also showed that MGD was produced by MobA under the experimental conditions and that the incubation mixture was suitable to analyze the reaction product of MobA further.
We therefore analyzed whether a portion of MGD remained bound to MobA after the nucleotide addition. MobA was separated from the small molecular weight fraction of the reaction mixture by gel filtration on a desalting column, and the total MGD content was related to MGD that remained bound to MobA or was released to the solution. MGD was quantified after its conversion to form A-GMP ( Fig. 2A). The purified MobA fraction contained about 19.3 Ϯ 7.9% of the total MGD formed, whereas the majority of MGD (90.9 Ϯ 15.5%) was  This result was further corroborated by determination of the relative molybdenum contents by assaying the molybdenum X-ray fluorescence intensity (not shown) in the MobA samples used in the XAS experiments described further below. After the addition of Mg-GTP to MobA, the molybdenum content was decreased to ϳ50% for a 9-fold excess of GTP and to ϳ20% for an 80-fold excess of GTP, in comparison with a MobA sample containing only bis-Mo-MPT (Fig. 2B). These observations consistently suggest the formation of bis-MGD on MobA in the presence of Mg-GTP and subsequent release of the product from the protein. Additionally, we determined the oligomerization state of MobA during the experiments by dynamic light scattering. The results showed that during the course of the reaction, MobA did not change its oligomerization state and existed as a monomer in solution, also in the incubation mixtures containing Mg-GTP and Mo-MPT (data not shown).
Reconstitution of ApoTorA Using Mo-MPT, Mg-GTP, and MobA-The molybdoenzyme TorA (TMAO reductase from E. coli) was used to investigate whether the cofactor produced by MobA was able to reconstitute enzyme activity in the purified apoprotein isolated from a Moco-deficient strain. Purified apoTorA was incubated with MobA, Mo-MPT, GTP, and MgCl 2 at 37°C, and TMAO reductase activity was determined after increasing incubation times of the reaction mixture (Fig.  3). The data showed that the presence of only MobA was sufficient for pronounced activation of apoTorA. No activation was observed in the absence of MobA, showing that other components besides MobA in the reaction mixture were not active in apoTorA activation and that MobA is essential to provide the matured cofactor for TorA, which probably is the bis-MGD cofactor. However, in the presence of the specific Moco-binding chaperone TorD (26), an about 2-fold increase of the maximal TorA activity was observed (Fig. 3). Accordingly, TorD either stabilized the released bis-MGD cofactor synthesized by MobA or facilitated its insertion into apoTorA, thus leading to the higher activity of TorA.  (Table 1). An MPT/molybdenum ratio close to 1:1 was found in hSO, consistent with the quantitative presence of the Mo-MPT cofactor in the enzyme. A similar ratio was determined for MoeA incubated with Mo-MPT, supporting previous suggestions that this protein is able to bind exogenously added Mo-MPT (42,43). These controls emphasized the accuracy of the used method for detection of the MPT/molybdenum ratio.
The MobA protein, which was incubated with Mo-MPT and purified thereafter from the reaction mixture, revealed an MPT/molybdenum ratio close to 2:1, irrespective of the presence or absence of GTP and MgCl 2 ( Table 1). These results show that, most likely, two MPT units were bound to a single molybdenum ion on MobA, and apparently GTP and MgCl 2 were not necessary for the formation of this novel bis-Mo-MPT cofactor precursor. To obtain further proof for the bis-Mo-MPT intermediate, total reflection X-ray fluorescence analysis was used to determine the molybdenum/phosphorus ratios in MobA samples in comparison with the bis-MGD-containing TorA used as a control (Table 1). In samples containing the MobA protein to which Mo-MPT had been previously added, but GTP was absent, a phosphorus/molybdenum ratio close to 2:1 was determined, consistent with one molybdenum and two MPT molecules, which carry one phosphate group each. After the addition of Mo-MPT and Mg-GTP, the phosphorus/molybdenum ratio for MobA was close to 4:1 and is thus similar to the value derived for TorA, which contains the bis-MGD cofactor. This suggested the attachment of two GMP units to the bis-Mo-MPT cofactor on MobA in the presence of Mg-GTP and hence formation of the bis-MGD cofactor (containing four phosphate groups in total).   addition, XAS experiments were performed to determine the valence state and first-sphere coordination of the molybdenum atom. The coordination of the molybdenum atom was analyzed in samples of MogA, MoeA, or MobA, which were incubated with Mo-MPT in the presence or absence of Mg-GTP. In addition, protein spectra were compared with further reference compounds with known molybdenum coordination. In the hSO enzyme, for example, the molybdenum in the Moco is coordinated by the two sulfurs of the dithiolene moiety of the MPT, two oxygen ligands, and the sulfur from the thiol group of a cysteine residue (MoS 3 O 2 ).

Determination of Molybdenum Oxidation States and Site
The molybdenum K-edge (XANES) spectrum reflects electronic transitions from the 1s core level to unoccupied localized states with mainly metal-d/p characters (Fig. 4). The main differences in the XANES spectra of the proteins and reference compounds were observed with respect to the amplitude of the pre-edge peak (Fig. 4, asterisk). In the case of coordination of molybdenum by oxygen and/or sulfur ligands, this feature is attributable to formally dipole-forbidden 1s34d transitions into * orbitals oriented along Mo-O bond vectors and thus gains intensity for an increasing number of oxygen ligands (44). The pre-edge peak magnitude therefore was particularly large for the molybdate ion (MoO 4 coordination), decreased for MoS 4 O coordination in a synthetic bis-dithiolene model complex (45), and almost absent for MoS 6 coordination in molybdenum disulfide (Fig. 4), revealing a direct dependence of the pre-edge area on the number of oxygen ligands at molybdenum (Fig. 5).
The pre-edge peak areas as determined from the XANES spectra of the protein samples (Fig. 4, inset) were compared with the correlation between the pre-edge area and the number of oxygen ligands (Fig. 5). The determined ϳ3. The position of the K-edge on the incident energy axis is indicative of the metal oxidation state. For various molybdenum reference compounds, the edge energy increased by ϳ1.2 eV per oxidation state in the range of Mo(IV) to Mo(VI) for varied sulfur/oxygen ligand configuration, but the absolute energies depended on the relative numbers of sulfur and oxygen ligands and were higher by ϳ5.5 eV for oxygen-only compared with sulfur-only coordination of molybdenum (data not shown). The K-edge energies for the protein samples were determined from the XANES spectra (Fig. 4) and compared with the edge energies of the references ( Table 2 (Table 2).

GTP (Table 2). This suggests Mo(V)S 4 O coordination in MobA
EXAFS analysis was performed to determine the bond lengths and numbers of oxygen and sulfur ligands in the first molybdenum coordination sphere in the protein samples in comparison with selected reference compounds (Fig. 6). Visual inspection of the Fourier transforms in Fig. 6A calculated from the EXAFS oscillations in Fig. 6B revealed two main Fourier transform peaks for the protein spectra, which are attributable to molybdenum-oxygen (shorter distances) and molybdenum-sulfur (longer distances) bonds, in comparison with the references.
Simulations of the EXAFS spectra yielded the fit parameters listed in Table 3. We show the results of two fit approaches, the first one including only variable fit parameters and the second one using best fit rounded values for the coordination numbers (N). The fit results may be summarized as follows, in particular emphasizing the sulfur/oxygen ligand ratios (Table 3). For the Mo-MPT sample, the N values for oxygen ligands were higher and for sulfur ligands the values were lower than for a pure MoS 2 O 3 coordination. The EXAFS spectra of MogA ϩ Mo-MPT and MoeA ϩ MPT were well described by N values, which were close to a pure MoS 2 O 3 coordination. However, MoeA ϩ Mo-MPT showed two longer Mo-O Ϫ bonds (ϳ1.8 Å) and one shorter MoϭO bond (ϳ1.7 Å), whereas for MogA ϩ MPT, this was reversed. This is further evidence for the presence of more reduced molybdenum in MoeA. For the MobA ϩ Mo-MPT samples, irrespective of the presence or absence of Mg-GTP, resulting coordination numbers, in comparison with a MoS 4 O model complex (45), consistently revealed only one short MoϭO bond and four Mo-S bonds. This clearly indicates the presence of two MPT units bound to molybdenum in MobA. The differences in Mo-S bond lengths of ϳ0.08 Å suggest that each dithiolene moiety of the two MPT units contributes one longer and one shorter Mo-S bond to the asymmetric ligation at the molybdenum in MobA.
The BVS, as calculated from the molybdenum ligand distances, is a measure of the molybdenum oxidation state (41). BVS values for the protein samples, which were calculated from the distances in Table 3, are summarized in Table 4 In summary, the XANES and EXAFS analyses provided a consistent picture of the molybdenum oxidation states and coordination environments. Mo-MPT extracted from hSO, which was used for the binding experiments, revealed only a ϳ50% fraction of intact cofactor. Nonspecific binding of the cofactor to MogA could be used for purification and stabilization of the intact cofactor, in terms of (MPT)S 2 Mo(VI)(ϭO) 2 Table 3 (fits 2, 4, 6, 8, 10, 12, and 14). Spectra were vertically displaced for comparison.  Only after the addition of Mg-GTP to the bis-Mo-MPT structure on MobA was the final product bis-MGD formed, and the cofactor was released from the protein thereafter. The mature cofactor could be readily inserted into apoTorA, resulting in reconstitution of TMAO reductase activity. This shows that the bis-MGD cofactor formed under our in vitro conditions is fully functional and that only MobA is involved in this step. The activity of reconstituted TorA was 2-fold increased in the presence of TorD, which is the specific chaperone for TorA, suggesting either stabilization of bis-MGD in the mixture or facilitation of the insertion process by TorD. The results show that TorD itself is not involved in the bis-MGD formation.
In our study, a Mo-MPT preparation extracted from heatdenatured hSO (29) was used as an effective in vitro source for the production of functional bis-MGD by MobA. In vitro sys-tems have been used before to insert the cofactor produced by MobA into molybdoenzymes. Moco produced by E. coli MobA has been inserted into Rhodobacter sphaeroides DMSO reductase (22). Alternatively, E. coli TMAO reductase was used to test the activity of MobA in conjunction with TorD using the total extract from E. coli cells as a rather undefined Moco source (26). These assay systems thus were intrinsically inhomogeneous, because proteins and/or (non-purified) cofactors from different organisms were used. In our present system, only enzymes from E. coli were employed. In addition, a defined Mo-MPT source from hSO was used, which proved to be more effective, because reconstitution of TMAO reductase required less incubation time as compared with the system using DMSO reductase. Inclusion of TorD in the incubation mixture resulted in a 2-fold higher rate of reconstitution and of the maximum activity; thus, TorD accelerates bis-MGD insertion 2-fold and additionally acts as a stabilizing protein for TorA and bis-MGD in this reaction, as proposed before (26).
The molecular mechanism of bis-Mo-MPT formation and its binding mode to MobA, however, need further consideration. Presumably, one molecule of molybdate is released during the combination of two Mo-MPT molecules to form the bis-Mo-MPT cofactor, but the underlying chemistry remains elusive. It also remains possible that MobA binds one Mo-MPT molecule and one MPT molecule from which the bis-Mo-MPT could be formed. Crystal structures have shown either a monomeric (48) or an octameric (24) organization of MobA in the crystals. We have studied the oligomerization state of MobA using analytical gel filtration or dynamic light scattering techniques, which revealed that MobA was present as a monomer in solution under all tested conditions, including the presence of Mo-MPT and Mg-GTP and high protein concentrations. Accordingly, bis-Mo-MPT formation from two Mo-MPT/MPT molecules in solution may occur on monomeric MobA using the MPT and predicted GTP binding sites.

TABLE 3 Simulation parameters for EXAFS spectra
Simulation parameters describe EXAFS spectra in Fig. 6. N, coordination number; R, interatomic distance; 2 2 , Debye-Waller parameter; R F , error sum as defined in Ref. 38 and calculated for a reduced distance range of 1-2.5 Å.

Sample
Fit no.   Table 3 for best fits of EXAFS spectra with rounded coordination numbers, using R 0 values that were the average of values for Mo(IV,V,VI) species (i.e. R 0 (Mo-O) ϭ 1.878 Å and R 0 (Mo-S) ϭ 2.285 Å (40,41). Here we observed facilitated release of bis-MGD from MobA in the presence of GTP, which could suggest release of the product by competition of GTP with the same site occupied by an MPT unit. Binding of GTP and MPT to the same site may indeed occur, because MPT is derived from GTP in a reaction catalyzed by the MoaA protein (6). Based on these results, we propose that two MPT moieties bind to both the predicted GTP-binding site and the predicted MPT binding site, thus enabling bis-Mo-MPT and bis-MGD synthesis by monomeric MobA. The favorable release only of the final product may then be induced by a different binding mode of bis-MGD compared with bis-Mo-MPT to MobA.

Mo
In our in vitro system, the formation of bis-MGD readily occurred after the addition of Mg-GTP to MobA loaded with bis-Mo-MPT. In vivo, however, MobB, a GTP-binding protein interacting with MobA (20,49), may assist the GTP binding step. A docking model of MobA and MobB has suggested that GTP is bound to a shared binding site at the interface between both proteins (20). However, MobB did not enhance the activity of MGD formation under our assay conditions (data not shown). In the cell, MobB may deliver GTP to MobA with a high specific affinity in a reaction, which was not required in our in vitro assay due to the higher concentrations of GTP. The mature bis-MGD cofactor after its release from MobA is captured by Moco-binding chaperones like TorD, TorZ, NarJ, DmsD, FdhD, or NarW (21). These chaperones assist bis-MGD insertion into the respective target proteins. This has been studied in detail for the TorA/TorD system (25,28), revealing that a complex comprising the TorA, MobA, and TorD proteins is involved (28).
Based on our present findings and the earlier results, we propose the following sequence of events during bis-MGD biosynthesis for TorA activation in E. coli (Fig. 7). 1) MogA forms an MPT-AMP intermediate from MPT and ATP. 2) MoeA takes over the product, inserts the molybdenum ion derived from molybdate in a Zn 2ϩ -dependent reaction, and detaches the AMP to form Mo-MPT (43). The MoeA reaction may involve reduction of molybdenum to the Mo(V) level, and the short molybdenum-oxygen bond lengths of the (MPT)MoS 2 O 3 site suggest that no amino acid-derived metal ligands are involved in the Mo-MPT binding to MoeA. 3) Mo-MPT is then captured by MobA, which first forms the bis-Mo-MPT cofactor and thereafter attaches two GMP molecules in a GTP-and MgCl 2dependent reaction, producing bis-MGD. MobA-bound bis-Mo-MPT and bis-MGD seemingly contained Mo(V), and the MoS 4 O coordination suggests that amino acids are not ligating the molybdenum of either cofactor. 4) TorD then is involved in channeling bis-MGD to apoTorA, and MobA and TorD are released from the complex, resulting in cofactor-loaded enzyme (pre-TorA) (25). 5) Pre-TorA is translocated to the periplasm, where active TorA enzyme is finally generated (21).