Solution Structures and Backbone Dynamics of a Flavodoxin MioC from Escherichia coli in both Apo- and Holo-forms

Flavodoxins play central roles in the electron transfer involving various biological processes in microorganisms. The mioC gene of Escherichia coli encodes a 16-kDa flavodoxin and locates next to the chromosomal replication initiation origin (oriC). Extensive researches have been carried out to investigate the relationship between mioC transcription and replication initiation. Recently, the MioC protein was proposed to be essential for the biotin synthase activity in vitro. Nevertheless, the exact role of MioC in biotin synthesis and its physiological function in vivo remain elusive. In order to understand the molecular basis of the biological functions of MioC and the cofactor-binding mechanisms of flavodoxins, we have determined the solution structures of both the apo- and holo-forms of E. coli MioC protein at high resolution by nuclear magnetic resonance spectroscopy. The overall structures of both forms consist of an α/β sandwich, which highly resembles the classical flavodoxin fold. However, significant diversities are observed between the two forms, especially the stabilization of the FMN-binding loops and the notable extension of secondary structures upon FMN binding. Structural comparison reveals fewer negative charged and aromatic residues near the FMN-binding site of MioC, as compared with that of flavodoxin 1 from E. coli, which may affect both the redox potentials and the redox partner interactions. Furthermore, the backbone dynamics studies reveal the conformational flexibility at different time scales for both apo- and holo-forms of MioC, which may play important roles for cofactor binding and electron transfer.

as pyruvate-formate lyase (13) and (E)-4-hydroxy-3-methylbutyl-2-enyl pyrophosphate synthase (14). It has been established by potentiometry that electrons flow from NADPH to flavodoxin reductase and then to flavodoxin, which subsequently transfers the electron to the downstream targets (15). Eukaryotes and most prokaryotes also contain proteins carrying domains homologous to flavodoxins and flavodoxin reductase in a single polypeptide chain, such as the NADPH-cytochrome P450 oxidoreductase (CPR) 2 (16,17), nitric-oxide synthase (NOS) (18), methionine synthase reductase (19), and sulfite reductase (20,21). These domains function in a similar fashion as the flavodoxins and flavodoxin reductase proteins in prokaryotic cells. Extensive efforts have been devoted to understanding the cofactor binding, redox partner interaction, and electron transfer mechanisms of flavodoxins and flavodoxinlike domains as well as the folding and stability of these proteins (22)(23)(24)(25)(26). However, the detailed mechanisms are not clear yet and remain to be further elucidated.
Flavodoxins are classified into the long-chain and shortchain groups according to the presence or absence of a 20amino acid loop that splits the fifth ␤-strand of the central ␤-sheet (25,26). The Escherichia coli protein MioC is identified as a flavodoxin with 147 amino acid residues and belongs to the short-chain group. The mioC gene is located next to the chromosomal replication initiation origin (oriC) of the E. coli genome, and the transcription of mioC runs through the oriC site (27). Based on its peculiar chromosomal location, extensive efforts were dedicated to establishing the relationship between mioC transcription and DNA replication initiation (28 -37). However, these experimental results are often contradictory, and no ultimate conclusions have been reached thus far. Recently, it was suggested that MioC protein plays an essential role in promoting the biotin synthase (BioB) activity in vitro (11). Two other essential protein components for keeping BioB activity were previously identified to be E. coli flavodoxin 1 (FldA) and flavodoxin reductase (10). It was also shown that both MioC and FldA are needed in the in vitro activity assays, and neither of them can substitute the function of the other (11). Intriguing questions are thus raised. What exact role does MioC play in biotin synthesis? Does it act up-or downstream of FldA in the electron transfer chain, or does it act parallel? Why are two flavodoxins needed in this reaction system, and how do they distinguish their redox partners and their biological roles? Is MioC essential for biotin synthesis in vivo, and what other potential physiological roles may MioC play? In order to elucidate these issues, one needs to answer the following two fundamental questions. Does E. coli MioC exhibit similar or different molecular properties as FldA? What are the structural and dynamic features that might determine the roles of MioC in the biotin synthesis and other redox activities?
Since MioC and FldA share low sequence identity and belong to the short-chain and long-chain flavodoxin groups, respectively, the structural and dynamic diversities may give rise to the different redox activities and functions of the two proteins.
In an effort to obtain insights into the molecular mechanism of the biological functions, we have determined the solution structures of both the apo-and holo-forms of MioC by high resolution NMR spectroscopy. Both structures show the typical flavodoxin fold consisting of an ␣/␤ sandwich. However, the protein structure is considerably stabilized, and the secondary structures are extended upon FMN binding. In addition, backbone dynamics studies of the apo-and holo-forms in conjunction with structural analysis and titration experiments provide further insights into the molecular mechanisms of cofactor binding and electron transfer of flavodoxins.

EXPERIMENTAL PROCEDURES
Expression and Purification of the Recombinant MioC-The mioC gene was cloned into pET21a expression vector and expressed in E. coli BL21(DE3) strain (Novagen). The culture was allowed to grow in LB medium, centrifuged, and resuspended in M9 minimal medium with antibiotics (ampicillin) and 15 NH 4 Cl in the presence or absence of 13 C 6 -glucose for preparations of 13 C/ 15 N-labeled or 15 N-labeled samples, respectively (38). The MioC protein was purified by ion exchange chromatography (DEAE) and subsequently gel filtration (Superdex-75) using an Ä KTA FPLC system (Amersham Biosciences). The purity was determined to be greater than 95% as judged by SDS-PAGE. The apo-form of MioC was prepared by denaturing the purified protein using 5% trichloroacetic acid (39), followed by refolding the protein in a buffer containing 30 mM sodium phosphate (pH 7.0), 30 mM NaCl, and 20 mM dithiothreitol (DTT). The holo-form of MioC was prepared by refolding the protein in a buffer containing 30 mM sodium phosphate (pH 7.0), 30 mM NaCl, and excess fully oxidized FMN (50 mM). 5% D 2 O was added for preparations of the NMR samples, and 2,2-dimethyl-2-silapentanesulfonic acid was added as the internal chemical shift reference.
Structure Calculations-The structure calculations were performed using the program package CYANA (49) and refined by AMBER (50). Distance restraints were derived from interproton nuclear Overhauser effect (NOE). Dihedral angles ( and ) were determined from backbone chemical shifts using TALOS (51). Hydrogen bond restraints were obtained using intermediate range NOEs in combination with the secondary structural information. The initial structures were calculated with the CANDID module of the CYANA program (49,52). The 20 lowest energy structures were selected as models for SANE to extend the NOE assignments (53). Two hundred structures were calculated by CYANA, and the 100 lowest energy structures were used as initial structures and refined using AMBER. Finally, the 20 lowest energy structures for each form were selected to represent the MioC protein in the apoand holo-form, respectively. Hydrogen bond restraints were used in the initial structural determination and refinement but were removed in the final stage of structure refinement. The final structures were analyzed using the program packages MOLMOL (54) and PROCHECK_NMR (55). In order to fix the phosphate group of the FMN cofactor during the structure calculations of the holo-form of MioC, ambiguous restraints were added between the oxygen atoms of the phosphate groups and residues Leu 12 -Gly 14 . These residues showed large chemical shift perturbations upon FMN binding, and their peculiar chemical shifts suggested hydrogen bond formation with the phosphate group (21,56).
Titration Experiments-The FMN titration experiments were performed and monitored by a series of two-dimensional 15 N-edited HSQC experiments. The apo-form of MioC was first exchanged using the NMR buffer without DTT under anaerobic conditions to remove DTT. The two-dimensional HSQC spectra of apo-MioC in the NMR buffer with or without DTT were identical. During the titration, the molar ratio of FMN/protein was gradually increased from 0.1:1 to 2:1. The spectra were collected on a Bruker Avance 800-MHz NMR spectrometer at 25°C.
Backbone 15 N Relaxation Measurements-The backbone 15 N relaxation parameters of the longitudinal relaxation rates (R 1 ), transverse relaxation rates (R 2 ), and steady-state heteronuclear { 1 H}- 15 N NOE values of the apo-and holo-forms of MioC were measured, respectively (57). The experiments were performed on a Bruker Avance 800-MHz NMR spectrometer at 25°C. A spectral width of 11160.7 Hz for the 1 H dimension was used for both the apo-and holo-MioC, whereas for 15 N dimension, spectral widths of 2595.0 and 2676.1 Hz were used for the apo-and holo-MioC, respectively. For the R 1 and R 2 measurements, 1024 ( 1 H) and 100 ( 15 N) complex data points were collected with 24 transients per increment and a recycle delay of 2.7 s. The delays used for the R 1 experiments were 10 (ϫ2), 100, 200, 500, 800, 1100, 1500, 2000, 2500, and 3190 ms, and those used for the R 2 experiments were 8 (ϫ2), 16,32,48,64,88,112,136, and 176 ms. The relaxation rate constants were obtained by fitting the peak intensities to a single exponential function using the nonlinear least squares method as described (58). The { 1 H}-15 N NOE experiments were performed in the presence and absence of a 3-s proton presaturation period prior to the 15 N excitation pulse and using recycle delays of 2 and 5 s, respectively (59). 64 transients were used for each experiment.

NMR Characterizations of the Apo-and Holo-forms of E. coli
MioC-The overexpression and purification procedures of the recombinant MioC were similar to that previously described (11). The elution with bright yellow color from the ion exchange column was further purified by gel filtration chromatography. However, the two-dimensional 15 N-edited HSQC spectrum of the directly purified sample showed an extra set of peaks, indicating a mixture of inhomogeneous compositions.
The trichloroacetic acid precipitation and subsequent refolding procedure were used to prepare the apo-form of MioC (39). The corresponding two-dimensional 15 N-edited HSQC spectrum showed a unique set of cross-peaks with chemical shift distribution corresponding to a reasonably well folded protein (Fig. 1A). However, the backbone amide signals of 21 nonproline residues were missing, including Ala 2 , Gly 9 , Thr 11 , Leu 12 , Gly 14 , Ala 15 , Glu 16   formational exchanges that are most probably coupled to the hinge motions (see below). Other missing residues are mostly located in the long loop (the 120s loop) connecting the fifth ␤-strand and the last ␣-helix, indicating that extensive conformational exchanges occur in this region. The holo-form of MioC was prepared by refolding the protein in the presence of excess FMN, and the corresponding two-dimensional 15 N-edited HSQC spectrum showed a single set of cross-peaks quite different from that of the apo-form (Fig.  1B). Nearly all backbone amide signals of nonproline residues were observed except for residues Ala 2 , Leu 126 , His 128 , Glu 132 , and Asp 133 . Fig. 1C shows an overlay of the HSQC spectra of the apo-and holo-forms of MioC, and the residues significantly perturbed upon binding with FMN are labeled.
FMN titration experiments were performed to characterize the cofactor-binding site of MioC. When the molar ratio of FMN/MioC was less than 1, the two-dimensional HSQC spectra showed two sets of peaks corresponding to the apo-and holo-forms of MioC with varying intensities. The spectrum of the directly purified sample was similar to these spectra, further demonstrating that the protein purified directly from recombinant expression was a mixture of both the apo-and holo-forms of MioC.
After the holo-form of MioC was exchanged using the NMR buffer without free FMN molecules, the bound FMN molecules gradually left off from the protein.
The HSQC spectrum of the sample prepared in this way also showed two sets of peaks, indicating that FMN is noncovalently bound to MioC and the apo-and holo-forms of MioC are in an exchanging process. Taken together, the existence of two sets of peaks instead of one in the HSQC spectra indicates that the exchange of apo-and holo-forms of MioC is in the slow regime. When the molar ratio of FMN/protein reached and went beyond 1, the set of peaks corresponding to the apoform was completely unobservable, and the HSQC spectra were identical to that of the holo-form. These results suggest a 1:1 binding between MioC and FMN with high affinity, which is in agreement with the previously characterized nanomolar range of K d values commonly observed for flavodoxins (60).
The composite chemical shift changes of individual residues in the two forms are plotted in Fig. 1D. Residues exhibiting the largest chemical shift changes are clustered around the three FMN-binding loops: the P-loop (residues Gly 9 -Gly 14 ), the 50s loop (residues Ser 55 -Asp 64 ), and the 90s loop (residues Ser 92 -Gly 100 ). In the HSQC spectra of MioC, the missing residues located in or near the three loops in the apo-form appeared after its binding with FMN.
Solution Structures of the Apo-and Holo-forms of E. coli MioC-To fully characterize the conformational changes upon FMN binding, the solution structures of both the apo-and holo-forms of E. coli MioC protein were determined using the NOE-derived distance restraints in combination with dihedral angle restraints. The lowest energy structures (20 for each form) were selected to represent the apo-and holo-forms of MioC, respectively, and are shown in Fig. 2, together with the ribbon diagrams of the mean structures (energy-minimized using AMBER) of each form. The structural statistics of both forms are summarized in Table 1.
Binding-induced Secondary Structure Formation-Although the core structures of the apo-and holo-forms of MioC are similar, significant differences are observed. An overlay of the ribbon diagrams of the mean structures (energy-minimized using AMBER) of apo-and holo-forms of MioC is shown in Fig.  3A. The backbone r.m.s. deviation between the two forms is 4.63 Å for all residues, and it is 2.30 Å after excluding the four loops near the cofactor-binding sites, indicating the FMNbinding loops are involved in the largest conformational changes. The secondary structures in the holo-form are generally longer than that in the apo-form. The most significant secondary structure extensions are observed for strands ␤3 and ␤4, extending two and three residues further toward the FMNbinding pocket. These two strands are directly connected to the 50s and the 90s loops, which are responsible for coordination of the isoalloxazine ring of the FMN cofactor. The short strand ␤5Ј is flexible and not well defined in the apo-form of MioC, whereas it is fully established and stabilized in the holo-form. The ␣-helices are also extended. Both helices ␣1 and ␣4 extend two residues toward the FMN-binding site. In addition to the binding-induced secondary structure extension and stabilization observed at local structures of the FMN-binding site, a small helix HЈ from residue Leu 40 to Leu 44 is also induced after binding of the FMN molecule. In contrast, this region adopts a looplike structure in the apo-form, and there were not sufficient NOE contacts obtained to establish a helical structure.
The induced formation of secondary structures and the significant stabilization of the overall structure of MioC after FMN binding is mostly responsible for the high affinity toward the FMN molecule and may also be important for maintaining the stable MioC-FMN complex and its biological activity in vivo.
Structural Comparisons of the Apo-form of E. coli MioC with the Crystal Structures of Apo-flavodoxins-For a full understanding of the molecular mechanism of cofactor binding, structures of both the holo-form and apo-form are required. Previous investigations on flavodoxin structures generally employed the crystallographic approach. Crystal structures of flavodoxins in the FMN-bound state from various organisms have been reported (60 -65), whereas much less information is available for apoflavodoxins. Two crystal structures of the longchain apoflavodoxins from Anabaena and Helicobacter pylori were reported (22,66). In both structures, the P-loop was stabilized by the recruitment of an anion (a sulfate ion for Anabaena apoflavodoxin and a chloride ion for H. pylori apoflavodoxin). All of the three FMN-binding loops showed similar rigid conformations as observed in the holoflavodoxins, whereas only the 50s loop appeared to move closer to the gap left by the cofactor (22,66). Two other crystal structures of apo-WrbA proteins, which also belong to the long-chain flavodoxin group, have been determined recently (67). The FMN binding loops in these structures also showed rigid conformations, and no significant conformational changes were observed upon FMN binding. Our structure of the apo-form of MioC is the first solution structure reported for apoflavodoxins and, moreover, the first structure of the apo-form of short-chain flavodoxins thus far. The backbone cross-peaks were missing in the two-dimensional HSQC spectrum and could not be assigned for 10 residues (Gly 9 , Gly 14 , Ala 15 , Glu 16 , Ser 55 , Asp 64 , Arg 93 , Ala 101 , Ile 102 , and Asp 103 ) connecting the FMN-binding loops and the central ␤-sheet, which suggests possible hinge motions. In contrast, most of the residues located in the loops could be observed and assigned (except for residues Thr 11 and Leu 12 in the P-loop and Cys 99 in the 90s loop), although very limited NOE contacts were observed due to the flexibility in these regions. The backbone r.m.s. deviation is 3.7 Ϯ 0.9 Å for the FMN-binding loops, whereas it is only 0.8 Ϯ 0.2 Å for the secondary structures. These results indicate that the FMN-binding loops of the apo-form of MioC are flexible in solution, which are in good agreement with the previous solution NMR characterizations of the structural properties of apo-forms of long-chain flavodoxins (68,69).  Fig. 2A). Since it was also observed that the FMN binding loops in the apo-form of long-chain flavodoxins were flexible in solution (68,69), this may be a common feature for all flavodoxins. It is likely that the structural flexibility is required for the cofactor-binding loops to sample different conformations and thus facilitate the recognition and binding with the FMN molecule.
Structural   NOVEMBER Fig. 3, D-F. The overall structures are very similar for these proteins. The 90s loop of the holo-form of MioC adopts a similar backbone conformation as observed in the other structures. The conformation of the 50s loop is slightly different from the other structures, which is in accordance with previous observations that this loop varies in both length and conformation among flavodoxins (60). Another major structural diversity is observed for the last helix. This helix is at least one turn shorter in the holo-form of MioC than that in the other structures, which is due to the presence of two proline residues that are commonly not found in other flavodoxins. The 120s loop connecting strand ␤5Ј to helix ␣4 is therefore longer in MioC than that in the other structures. In addition, The 120s loop is involved in the intermediate conformational exchanges, which might play specific roles in protein-protein interactions.

Structures and Dynamics of E. coli MioC
The electrostatic potential at the FMN-binding site of MioC is shown in Fig. 4A. For comparison, the cofactor binding site of E. coli flavodoxin FldA and the flavodoxin-like domain of rat CPR are shown in Fig. 4, B and C, respectively. The FMN binding pocket is generally surrounded by many negatively charged residues, which are thought to be important factors for modulating the redox potentials of FMN (72,73). E. coli FldA has a total of five negatively charged residues located in the 50s and 90s loops that are in proximity with the isoalloxazine ring of FMN, whereas E. coli MioC has only three negatively charged residues in these loops. In addition, a positively charged residue Arg 93 locates in the 90s loop next to the negatively charged Glu 94 in MioC, which may neutralize the negative charge to a certain degree. Therefore, the electrostatic charge of MioC is much less negative compared with FldA around the flavin ring.
Similar to other flavodoxins, MioC contains the conserved aromatic residues Tyr 95 and Phe 98 in the 90s loop. In the 50s loop, the aromatic residue Trp 57 in E. coli FldA is substituted by a histidine residue in MioC. This position is also occupied by an aromatic residue tyrosine in both cytochrome P450BM-3 from Bacillus megaterium (Tyr 536 ) and rat CPR (Tyr 140 ). Although not enough NOE contacts between FMN and MioC were obtained and the orientation of the molecule FMN was ill defined, the local backbone conformation of MioC is quite similar to other well characterized flavodoxins, suggesting that MioC may bind FMN in a similar manner as other flavodoxins. Notably, the E. coli FldA has a total of six aromatic residues surrounding the FMN molecule (residues Trp 57 , Tyr 58 , Tyr 59 , Tyr 94 , Tyr 97 , and Phe 98 ). These aromatic residues form a hydrophobic wall that separates the flavin ring of FMN from the solvent (60). In contrast, the E. coli MioC protein does not contain any other aromatic residues except His 57 , Tyr 95 , and Phe 98 , as mentioned above. The positions equivalent to residues Tyr 58 , Tyr 59 , and Tyr 97 in FldA are occupied by the short side-chain residues Gly 58 , Ala 59 , and Thr 97 in the MioC structure. These differences may play a role in determining the different redox partners that MioC and FldA interact with, and thus the biological processes they are involved in. These structural properties may also affect the redox potentials.
Interestingly, the aforementioned structural features at the FMN-binding site of MioC are also shared by the flavodoxinlike domains in cytochrome P450BM-3 from B. megaterium and the flavodoxin-like domains of mammalian CPR. The FMN-binding domains of these multidomain proteins also contain fewer aromatic residues and are less negatively charged. Furthermore, both of these flavodoxin-like domains contain a positively charged lysine residue at the position equivalent to Arg 93 in MioC. The electrostatic potential at the protein surface may act in orienting the protein in protein-protein interactions (74). Therefore, a similar pattern of charge distribution between MioC and these flavodoxin-like domains may suggest similar redox partner-interacting surfaces and thus similar biological functions.
Backbone Relaxation Parameters and Model-free Analysis-It is well understood that the biological function of a protein is strongly dependent on its structure and dynamics, and the changes of both structure and dynamics play critical roles in many biological processes, such as ligand binding, proteinprotein interactions, and enzyme catalysis (75)(76)(77). Proteins mostly undergo submillisecond time scale conformational changes upon ligand binding, and it is of importance to recognize the change of dynamic properties associated with this process. NMR is a well established technique to provide the site-specific motional information with exquisite time resolution (78). Therefore, insights obtained from dynamics may have potential implications in understanding of the biological functions (76).
In order to characterize the dynamic properties of the apoand holo-form of MioC and obtain further insights into the molecular mechanisms of cofactor binding and electron trans-fer, we have determined the 15 N backbone relaxation parameters, including the longitudinal relaxation rates R 1 , transverse relaxation rates R 2 , and heteronuclear Overhauser effect { 1 H}-15 N NOE values for both forms of MioC using freshly prepared samples. During the analysis of the relaxation data, 86 of 147 residues were used for the apo-form, whereas 115 were used for the holo-form of MioC. The unanalyzed residues include the proline residues, the ones that were not assigned and those that overlapped or were too weak to be accurately analyzed. The experimental data of the apo-and holo-form for MioC are shown in Fig. 5A.
Overall, the protein adopts a rigid structure in both forms as reflected by the high { 1 H}- 15  Many residues in the 90s loop were unanalyzed due to weak signals and thus poor data quality for the determination of relaxation rates (R 1 and R 2 ). In addition, many backbone cross- peaks were missing in these regions, especially for residues in the P-loop. This suggests that the intermediate conformational exchanges on submillisecond time scales, which result in the line broadening and the disappearing of the cross-peaks. Upon binding to FMN, the P-loop, the 90s loop, and part of the 50s loop became more rigid, since many new cross-peaks appeared for residues in these regions, which showed close to average { 1 H}-15 N NOE values and relaxation rates. However, motional flexibilities still exist for some parts of the loops, especially the 50s loop and the 120s loop, and will be discussed below.
The precise determination of motional anisotropy is essential for analyzing the relaxation data, particularly for the characterization of chemical or conformational exchanges (79). The motional anisotropy can be described by the rotational diffusion tensor. The ratios of the principle components of the inertia tensors of the apo-and holo-forms of MioC calculated from the solution structures are (1:0.80:0.75) and (1:0.77:0.66), respectively, suggesting motional anisotropy for both forms. The rotational diffusion tensors of apo-and holo-forms of MioC were determined following the general procedures (80). A total of 63 and 95 residues were used for the apo-and holoforms to determine the rotational diffusion tensors, respectively. The results indicated that the diffusion tensors for both forms were best represented the axially symmetric model, demonstrating the motional anisotropy. For the apo-form of MioC, the rotational correlation time is m ϭ 7.57 Ϯ 0.04 ns, and the diffusion anisotropy is D ʈ /D Ќ ϭ 1.09 Ϯ 0.03. For the holo-form of MioC, the rotational correlation time is m ϭ 7.27 Ϯ 0.03 ns, and the diffusion anisotropy is D ʈ /D Ќ ϭ 1.14 Ϯ 0.03. The results suggest that both forms of the MioC protein exist as monomers in solution under our present experimental conditions, which was also observed during the MioC purification procedure using gel filtration (data not shown).
Model-free analysis was performed to extract the dynamic parameters from the experimentally determined relaxation data (81,82), and the axially symmetric diffusion model was used during the calculations. The calculations were carried out by using the experimental data, the uncertainties, and the energy-minimized mean structures for both forms of MioC as input. Five models with increasing complexity (M1, S 2 ; M2, S 2 , e ; M3, S 2 , R ex ; M4, S 2 , e , R ex ; M5, S f 2 , S 2 , e ) were used iteratively to reproduce the experimental data until confidence reached within 95% (83). The confidence level was estimated using 300 Monte Carlo simulations per run in combination with 2 and F-statistic analysis. The amide bond length was fixed at 1.02 Å, and the 15 N chemical shift anisotropy value of Ϫ175 ppm was used during the calculations. The optimized internal mobility parameters (the generalized order parameter S 2 , the fast internal motions on the picosecond to nanosecond time scales e , and the conformational exchanges R ex on the micro-to millisecond time scales) for both apo-and holo-forms of MioC are shown in Fig. 5B.
For the holo-form of MioC, a total of 95 residues were assigned to model M1, with an average S 2 ϭ 0.88 Ϯ 0.03. Twelve residues (Asp 3 , Thr 33 , Ala 46 -Ile 49 , Thr 97 , Gly 113 , Glu 119 , Asp 129 , Trp 142 , and Leu 145 ) were assigned to model M2, with an average S 2 ϭ 0.76 Ϯ 0.03 and internal motions on the picosecond to nanosecond time scales. Seven residues (Tyr 17 , Ala 59 -Ile 62 , Ile 90 , and Ile 125 ) were assigned to model M3, with an average S 2 ϭ 0.84 Ϯ 0.04 and conformational exchanges on the microsecond to millisecond time scales. No residues were assigned to model M4, and one residue (Lys 147 ) was assigned to model M5.

DISCUSSION
Internal Dynamics-Although much information is available on the structures of flavodoxins, there are no systematic dynamic characterizations for either the apo-form or the holoform thus far. For a clear overview, the extracted dynamic parameters are mapped onto the MioC structures of the apoand holo-forms and are shown in Fig. 6. Overall, the core structures of apo-and holo-forms of MioC show a similar rigidity, indicated by the similar average S 2 values for residues assigned to M1. However, as shown in Fig. 7, a closer examination of the differences of the S 2 values between the two forms of MioC demonstrates that the holo-form is less mobile than the apoform, which is also indicated by the fact that many more residues in the holo-form of MioC were assigned to M1 by the model-free analysis. For the apo-form of MioC, significant conformational exchanges on the microsecond to millisecond time scales as well as the internal motions on the picosecond to nanosecond time scales were observed around the FMN binding loops, indicating that these loops may sample multiple conformations in solution in the absence of FMN. In addition, the line broadening for the signals of residues connecting the loops to the central structural core also suggests conformational exchanges on the submillisecond time scales, which may enable the loops to adopt the open and closed conformations, offering the entrance of the FMN molecule into the binding pocket. Upon binding to FMN, the conformational exchanges of the MioC protein are largely diminished, especially those residues in the loops around the binding site. The residues connecting the loops to the structural core are stabilized and mostly are not involved in conformational changes, and residues located in the P-loop and 90s loop regions are greatly stabilized as well. On the other hand, the holo-form of MioC is not completely rigid, and the motional flexibilities at submillisecond time scales still exist for some residues in the loops. Residues Ala 59 -Ile 62 in the 50s loop were assigned but could not be analyzed in the apo-form due to the poor data quality, an indication of conformational exchanges. These residues could be analyzed in the holo-form and were shown also involved in conformational exchanges on the microsecond to millisecond time scales. This fact suggests that the conformational exchanges still exist in this region even after the FMN binding. Interestingly, residues Ala 46 -Ile 49 located in the central ␤-strand show internal motions on the picosecond to nanosecond time scales in both the apo-and holo-forms. Since this strand directly connects to the 50s loop, the internal motions may be coupled to the conformational exchanges occurring in the 50s loop. In addition, residues in the C and D, ribbon diagrams of the apo-(C ) and holo-(D) forms of E. coli MioC representing the internal motions on the picosecond to nanosecond time scales with colors ranging from yellow to red and magenta corresponding to e values from 10 to 100 ps and Ͼ100 ps. E and F, ribbon diagrams of the apo-form (E ) and holo-form (F) of E. coli MioC representing the residues with conformational changes (R ex Ͼ 1 s Ϫ1 ) on the microsecond to millisecond time scales colored in blue. The missing residues caused by the conformational exchanges are colored in black. The figures were generated using MOLMOL (54). NOVEMBER 17, 2006 • VOLUME 281 • NUMBER 46 120s loop, which is near the binding site but are not bound with FMN, also show conformational flexibilities. Residue Asp 129 shows fast internal motion on subnanosecond time scales, whereas residue Ile 125 shows conformational exchanges on the microsecond to millisecond time scales. Notably, the backbone signals of many nonproline residues in this loop (Leu 126 , His 128 , Glu 132 , Asp 133 , and Ala 135 ) were still missing in the holo-form, strongly suggesting intermediate conformational exchanges. Since the FMN-binding site is also involved in the redox partner interaction and subsequently the electron transfer (74,84), the motional flexibility observed in this region of the holo-form of MioC may be required for redox partner recognition.

Structures and Dynamics of E. coli MioC
Functional Implications-Previous studies have established certain structural determinants for the modulation of the FMN redox potentials (72,73). It was argued that the negative charges near the flavin ring destabilize the anionic hydroquinone and thus result in a lower E1 (semiquinone/hydroquinone) potential (60,72,73). In addition, it was proposed for the E. coli FldA that the hydrophobic walls formed by the aromatic rings may also lower the E1 value (60). Since MioC has much less negative charge and fewer aromatic residues around the cofactor-binding site, it may have a higher E1 potential than FldA. However, further investigations are required to clarify this issue.
Our dynamics results showed that the conformational exchanges on the microsecond to millisecond time scales are maintained for some residues in the 50s loop of the holo-form. Since flavodoxins act as electron donors in many biological processes, the protein conformation is expected to change during the oxidation and reduction of the FMN molecule. It was demonstrated by earlier studies that residues 58 -60 have to experience a backbone flip upon the reduction of FMN from the fully oxidized state to the semiquinone state (85,86), and the corresponding redox potential (E2) was shown coupled to local conformational flexibility (87,88). The conformational exchanges on microsecond to millisecond time scales observed for the holo-form of MioC might be important for the protein to sufficiently fine tune to the different redox states of the bound molecule. Moreover, it is highly possible that different dynamic properties of the FMN binding loops may be another fundamental factor affecting the redox potentials besides the structural and electrostatic properties.
In the FMN-bound form, the flavodoxins are able to shuttle electrons from the FAD-containing flavodoxin reductase to diverse downstream targets (15). Several studies were also carried out to investigate the interaction surface of flavodoxins with their redox partners (74,84). It was shown that flavodoxins interact with both the upstream and downstream redox partners using a similar interface, which primarily consists of the FMN-binding loops, and ternary complex formation is excluded (74). Large conformational changes are expected as the flavodoxin dissociates from its upstream redox partner and subsequently interacts with the downstream partner (74). The conformational flexibility on the slow time scales observed for the 50s loop may also be required for the redox partner switching process.
In addition, it has been suggested that residues in the additional loop in the long-chain flavodoxins (25,26) and the acidic residues located in the loop connecting the fifth ␤-strand and the following helix of rat CPR (17,89,90) are not directly involved in FMN binding but may also play a role in redox partner interactions. It is interesting to notice that the backbone signals of residues in the 120s loop are still largely missing in the two-dimensional HSQC spectrum of the holo-form, indicative of the intermediate conformational exchanges. Furthermore, the PEDPAE motif of this loop is present only in the MioC protein subfamily, and the proline and negatively charged residues are highly conserved among different bacterial species. The presence of two proline residues shortens the last helix and makes the 120s loop of MioC significantly longer and more mobile than other flavodoxins previously reported (60 -65). Although it does not bind to FMN directly, we propose that this loop may represent a unique feature of the MioC protein subfamily and thus may play a specific role in redox partner recognition.
Evolutionary Implications-Our current results show that the three-dimensional structure of MioC is similar to that of the FMN-binding domains of multidomain protein CPR and shares similar characteristics around the FMN-binding site. The E. coli genome contains at least four genes predicted to encode flavodoxins: fldA, fldB, mioC, and yqcA. The first two genes, fldA and fldB, encode proteins belonging to the long-chain flavodoxin group, whereas MioC and YqcA proteins belong to the short-chain group. Interestingly, flavodoxin-like domains found in multidomain proteins from mammals are more homologous to the short-chain group in amino acid sequences. When searching for protein sequences from human genome by BLAST (91), the MioC protein sequence shows the highest similarity with the FMN-binding domain from CPR (sequence identity of 31% and similarity of 46% with 5% gaps), whereas YqcA shows the highest similarity with the FMN-binding domain of methionine synthase reductase (sequence identity of 34% and similarity of 48% with 3% gaps). In contrast, FldA shows 27% sequence identity (42% sequence similarity with 25% gaps) with CPR and 25% sequence identity (39% sequence similarity with 19% gaps) with methionine synthase reductase, whereas FldB shows no sequence identity with protein domains from human or other eukaryotic organisms. Although FldA is an essential protein for E. coli growth and it has long been accepted that FldA acts as the electron donor to various enzymes like methionine synthase in E. coli, few investigations have been performed, and there is a lack of insight into the roles of other flavodoxins, especially the short chain flavodoxins MioC and YqcA.
It has been proposed that the long-chain flavodoxins may precede short-chain flavodoxins during evolution (25,26). These short chain flavodoxins may take the place of FldA in specific biological pathways or under special conditions. Future detailed investigations on the short-chain flavodoxins will shed light on the evolution and functions of the flavodoxin family.
Conclusion-We have determined the solution structures of an E. coli short-chain flavodoxin MioC in both the apo-and holo-forms at high resolutions by NMR spectroscopy. Our results show the extensive secondary structure formation and overall stabilization upon FMN binding. The backbone dynamics of apo-and holo-forms of MioC provide detailed descriptions of the backbone conformational mobility of flavodoxins for the first time. Moreover, our results suggest that conformational flexibility is important for both cofactor binding and redox partner interactions. Our current studies provide further insights into the molecular mechanisms of the FMN binding and electron transfer of flavodoxins.