An Analysis of the Binding of Repressor Protein ModE tomodABCD (Molybdate Transport) Operator/Promoter DNA ofEscherichia coli *

Expression of the modABCDoperon in Escherichia coli, which codes for a molybdate-specific transporter, is repressed by ModE in vivo in a molybdate-dependent fashion. In vitro DNase I-footprinting experiments identified three distinct regions of protection by ModE-molybdate on the modAoperator/promoter DNA, GTTATATT (−15 to −8; region 1), GCCTACAT (−4 to +4; region 2), and GTTACAT (+8 to +14; region 3). Within the three regions of the protected DNA, a pentamer sequence, TAYAT (Y = C or T), can be identified. DNA-electrophoretic mobility experiments showed that the protected regions 1 and 2 are essential for binding of ModE-molybdate to DNA, whereas the protected region 3 increases the affinity of the DNA to the repressor. The stoichiometry of this interaction was found to be two ModE-molybdate per modA operator DNA. ModE-molybdate at 5 nm completely protected themodABCD operator/promoter DNA from DNase I-catalyzed hydrolysis, whereas ModE alone failed to protect the DNA even at 100 nm. The apparent K d for the interaction between the modA operator DNA and ModE-molybdate was 0.3 nm, and the K d increased to 8 nm in the absence of molybdate. Among the various oxyanions tested, only tungstate replaced molybdate in the repression ofmodA by ModE, but the affinity of ModE-tungstate formodABCD operator DNA was 6 times lower than with ModE-molybdate. A mutant ModE(T125I) protein, which repressedmodA-lac even in the absence of molybdate, protected the same region of modA operator DNA in the absence of molybdate. The apparent K d for the interaction between modA operator DNA and ModE(T125I) was 3 nm in the presence of molybdate and 4 nmwithout molybdate. The binding of molybdate to ModE resulted in a decrease in fluorescence emission, indicating a conformational change of the protein upon molybdate binding. The fluorescence emission spectra of mutant ModE proteins, ModE(T125I) and ModE(Q216*), were unaffected by molybdate. The molybdate-independent mutant ModE proteins apparently mimic in its conformation the native ModE-molybdate complex, which binds to a DNA sequence motif of TATAT-7bp-TAYAT.

Molybdenum is necessary for the activity of several enzymes found in animals, plants, and bacteria, such as sulfite oxidase, xanthine dehydrogenase, nitrate reductase, formate dehydrogenase, and nitrogenase (1)(2)(3). From structural studies, it has been determined that molybdenum exists in molybdoenzymes in the form of a pterin-containing molybdenum cofactor, with the exception of dinitrogenase, which has an iron-molybdenum cofactor (1,3,4).
For Escherichia coli, the production of its complement of molybdoenzymes requires the efficient uptake of molybdate via the molybdate-specific transporter encoded by the modABCD operon (5)(6)(7)(8). Analysis of the proteins encoded by the mod-ABCD operon suggests that ModA functions as a molybdatespecific periplasmic binding protein (9), ModB as an integral membrane channel-forming protein, and ModC as an ATPbinding energizer protein. Mutations in genes coding for the molybdate-specific transporter in E. coli result in a loss of active molybdoenzymes, such as nitrate reductase and formate dehydrogenase (8). However, when sufficient levels of molybdate are added to the culture medium of these mutant strains, activity of the molybdoenzymes is restored, suggesting that in molybdate-supplemented media, molybdate enters the cell through alternate transport systems (10). It was determined that expression of the modABCD operon is regulated by the intracellular concentration of molybdate. High levels of intracellular molybdate resulted in reduction of transcription of the modABCD operon (10 -12), which implies that E. coli has a molybdate-dependent repressor that regulates expression of the modABCD operon.
Isolation of E. coli mutants that did not exhibit the classical repression of the modABCD operon upon molybdate-supplementation of the growth medium enabled us to identify a putative molybdate-dependent repressor (ModE) of the mod-ABCD operon (13). Similar genetic analysis also identified modE product as a potential repressor of the modABCD operon (7,14). Purified ModE protein prevented transcription of modABЈ template DNA in an in vitro coupled transcriptiontranslation experiment (13). Molybdate is an essential component of this repression, and ModE-molybdate is the active form of the repressor. Molybdate-independent mutant forms of the repressor have been described (13,14). Some of these mutant forms carried point mutations within the amino acid sequence TSARNQXXG (positions 125-133), which is conserved in ModE-homologs from several organisms and in Clostridium pasteurianum molybdopterin-binding proteins (8,13). Other molybdate-independent mutants were missing different lengths of the C terminus of the ModE protein (13,14). However, the kinetics of interaction of the mutant proteins with the modA operator DNA are yet to be established.
Confirmation of the binding of ModE to E. coli modA operator/promoter DNA was provided by electrophoretic mobility shift and by DNase I-footprinting of this DNA in the presence of ModE (15)(16)(17). McNicholas et al. (17), using the modABCD coding strand, determined that ModE protected a 28-bp 1 stretch of modA DNA spanning from Ϫ18 to ϩ10. It was proposed that two inverted repeat sequences located within the ModE-protected region serve as possible ModE binding sites. This inverted repeat sequence is similar to a consensus sequence, CGTTATATAN 4 -12 TATATAACG, identified for molybdenum-regulated genes based on sequence similarity (18). However, McNicholas et al. (17) reported that ModE protected the modA operator DNA even in the absence of added molybdate, which is in contrast to the in vivo observations of strict molybdate dependence. A K d value of about 25 nM was reported for the interaction between ModE-molybdate and the operator DNA (15,17). This value of 25 nM is significantly higher than the subnanomolar K d values reported for binding of other repressors to their cognate operator DNA (19 -21). We reevaluated the kinetics of interaction between ModE and modA operator DNA, as well as the role of molybdate in this interaction, and the results are presented in this communication. Additional information on the interaction of various mutant forms of ModE with modA operator DNA is also presented.

EXPERIMENTAL PROCEDURES
Reagents-Biochemicals were purchased from Sigma. Inorganic and organic chemicals were obtained from Fisher Scientific Co. and were analytical or molecular biological grade. Restriction endonucleases and DNA modifying enzymes were purchased from New England Biolabs, Inc., or Promega. Sequenase 2.0 was obtained from United States Biochemical Corp. All oligonucleotides used in this study were synthesized by National Biosciences, Inc., or Genosys.
Media and Growth Conditions-L broth (LB), which served as rich medium, was supplemented with glucose (0.3%) or sodium molybdate (1 mM). Antibiotics, as needed, were included in media at the following concentrations; ampicillin, 100 g/ml; kanamycin, 50 g/ml. Genetic and molecular biological experiments were performed essentially as described previously (6,10,13).
DNA Sequencing Experiments-All DNA sequences generated in the course of this study were obtained using the Sanger dideoxy method and appropriate primers and plasmids isolated by alkaline lysis procedure and purified by cesium chloride density-gradient centrifugation (22,23). The DNA sequences were analyzed using computer software programs GCG (24) and Genepro (Riverside Scientific).
Construction of ModE Mutant Proteins-ModE(T125I) and ModE-(Q216*) were described previously (13). For the construction of plasmid pJKR12, which carries a different length of C-terminal deletion, an AseI fragment carrying part of modE, modA, and part of modB was removed from plasmid pSE1004 (25) and cloned into AseI-digested plasmid vector pBR322. This plasmid produced a ModE C-terminal hybrid protein of 227 amino acids in which ModE was 220 amino acids long. The truncated C terminus of the protein was extended by an additional 7 amino acids from the vector sequence (IVAGKLE). This hybrid protein is designated as ModE(220hyb). For construction of other mutant ModE proteins (Table II), two primers that flank the modE gene in plasmid pAG1 (primer 1, GTGTGGAATTGTGAGCGG; primer 2, GTCAGGAT-GGCCTTCTGC) were used in combination with two internal primers carrying the designed mutation to amplify the modE gene in plasmid pAG1. The two polymerase chain reaction products were hydrolyzed with the designed restriction enzyme and ligated to generate the fulllength modE gene. After further digestion with EcoRI and BsaI, the mutated modE was cloned into plasmid pAG1, which was also hydrolyzed with EcoRI and BsaI. This method replaced the modE ϩ DNA in plasmid pAG1 with mutated modE DNA. The specific mutations in six such modE mutants used in this study are listed in Table II. The modE DNA in these plasmids was sequenced to confirm the presence of the specific mutation(s).
Construction of modA Operator DNA Plasmids-Plasmid pAM4, which provided the native modA operator/promoter DNA used in DNase I protection experiments, was produced by ligating a KpnI-BstEII fragment from plasmid pSE1009 (25) (the BstEII end was modified using DNA polymerase I-Klenow fragment; Klenow) that covered the Ϫ247 to ϩ25 sequence of the modA operator/promoter region into the KpnI-HincII sites of plasmid pUC19. Plasmid pAM6, was obtained by Bal-31 nuclease treatment of KpnI-digested plasmid pAM4 followed by ligation. In the construction of plasmids pAM17 and pAM18, complementary oligomers that contained various portions of modA operator/promoter sequence (Table I) were synthesized, annealed, phosphorylated with polynucleotide kinase, and digested with BamHI prior to ligating into the BamHI-SmaI sites of plasmid vector pUC19. All mutations were confirmed by sequencing the DNA. Throughout this paper, the modABCD transcription start site was assigned ϩ1 in reference to the numbering of the bases.
ModE Protein Purification-ModE protein purified from E. coli as described previously (13) was used in this study. Although the native ModE protein remained in solution upon expression from the T7-expression plasmid, the mutant ModE proteins were localized in inclusion bodies. The method described below overcame this problem and allowed expression of mutant ModE into the soluble fraction and purification to homogeneity using nickel affinity chromatography.
Polymerase chain reaction was utilized to amplify the wild type modE as well as mutant modE alleles from plasmid pAG1 (and its mutant derivatives) (13) using two primers, 5ЈGGACATTCATATGCA-GGCCGAAATC-3Ј and 5Ј-GCGGATCCTTAGCACAGCGTGGCGA-3Ј, for insertion into the expression vector pET15b (Novagen) at the NdeI/ BamHI sites. The N termini of the ModE proteins produced from these plasmids also contain six histidines, which are separated from the ModE protein by a linker containing thrombin cleavage site. The mutation in each plasmid containing the mutant modE allele was confirmed by sequencing the entire modE gene in the pET15b-derived plasmids.
Each of the ModE proteins was purified as follows. Strain BL21(DE3) carrying pET15b-modE derivative was grown in 250 ml of L broth with shaking (225 rpm) at 37°C. When the absorbance of the culture reached 0.7 (420 nm; Spectronic 710 spectrophotometer), isopropyl-1-thio-␤-Dgalactopyranoside was added to a final concentration of 0.1 mM. The temperature of incubation was reduced to 23°C, and the culture was incubated for an additional 3 h. The cells were harvested, washed once with Tris buffer (50 mM Tris, pH 8.0), and resuspended in 25 ml of Tris buffer. Cells were broken by passage through a French pressure cell operating at 20,000 psi, and the extract was clarified by centrifugation at 27,000 ϫ g for 30 min at 4°C. The supernatant was filtered through a 0.2 m filter (Gelman Sciences) and then loaded onto a HiTrap chelating column (Amersham Pharmacia Biotech). Before loading the sample, the column was washed with NiCl 2 (40 mM in Tris buffer), and free Ni was removed with Tris buffer. After loading the sample, the column was washed with Tris buffer followed by the same buffer containing 50 mM imidazole to remove the proteins bound nonspecifically to the matrix. The ModE protein was eluted with 0.3 M imidazole in Tris buffer. After adding CaCl 2 to a final concentration of 2.5 mM, thrombin protease (25 units; Amersham Pharmacia Biotech) was added to the ModE preparation to remove the N-terminal His tag sequence. The thrombin-ModE mixture was incubated for 16 h at 4°C. After cleavage was complete, based on SDS-polyacrylamide gel electrophoresis analysis, the sample was loaded on a HiTrap Heparin column (5 ml; Amersham Pharmacia Biotech) to separate ModE from thrombin. After washing the column with 5 volumes of Tris buffer, ModE was eluted with 0.3 M NaCl and was determined to be pure by SDS-polyacrylamide gel electrophoresis. ModE protein was dialyzed overnight in 50 mM Tris (pH 8.0) containing 0.5 mM dithiothreitol and stored on ice until use. This protein was found to be stable for several months. ModE proteins purified by this nickel affinity method, contained an extra three amino acids (Gly, Ser, and His) in the N termini and are designated *ModE to denote this alteration. Protein concentration was determined using either the BCA assay or Coomassie Blue (Bradford reagent) with bovine serum albumin as standard (26,27).
Kinetic Properties of ModE-DNA Interaction-The kinetic characteristics of the interaction between ModE and the modA operator DNA were determined using an IAsys optical evanescent wave biosensor (Affinity Sensors). The principles of this biosensor and kinetic analysis of the data were described previously (28 -30). In this particular experiment, a cuvette containing biotin on the sensor surface was utilized. Biotinylated DNA was bound to the sensor surface using streptavidin (Calbiochem) as an intermediate. Two 43-base complimentary oligonu- 1 The abbreviation used is: bp, base pair(s). cleotides covering the modA operator region between Ϫ18 and ϩ25, were synthesized (National Biosciences, Inc.), and biotin was incorporated into the 5Ј-end of one of the oligonucleotides that upon annealing would be located in the 5Ј-end of the DNA. All cuvettes were prepared by applying the same amount of modA operator DNA for generation of the immobilized ligand. The quantity of DNA present in the cuvette was further verified by monitoring the response of the instrument to a standard amount of ModE-molybdate.
The biotin cuvette in the optical sensor was rinsed with phosphatebuffered saline containing 0.1% Tween 20 (PBST). Streptavidin (100 l; 1 mg/ml) in 10% PBST was added to the cuvette (200-l reaction volume) and incubated for 6.5 min with mixing. After removing the unbound streptavidin, DNA was added and incubated until no further response was noted. Excess DNA was removed from the cuvette by 3 washes with PBST. ModE protein with or without molybdate (1 mM) was added to the cuvette, and the rate of increase in refractive index (response units in arc⅐s) was recorded. After the maximum response was attained, free ModE in the cuvette was removed by a wash with PBST, and the rate of dissociation of ModE from the ModE-DNA complex was monitored. The experiment was repeated with various ModE concentrations, as well as with various mutant forms of ModE and mutated modA operator DNA.
In Vitro DNase I-Footprinting Experiments-DNase I protection experiments were performed as described previously (31). For these experiments, a 446-bp FspI-HindIII fragment from plasmid pAM4, which carries the modA operator/promoter DNA spanning from Ϫ247 to ϩ25, was used after purification using a 10 -30% sucrose gradient (32).
DNA Mobility Shift Experiments-DNA mobility shift experiments were performed as described by Fried and Crothers (33), with modification (31,34). These experiments used a binding reaction buffer (10 mM Tris-HCl, pH 7.9, 10 mM MgCl 2 , 50 mM NaCl, 1 mM dithiothreitol, and 5% glycerol), DNA (0.1 pmol; Ϫ18 to ϩ25 region of the modA operator/promoter DNA or mutant modA operator derivative), and protein in a final volume of 10 l. Sodium molybdate or other oxyanions at a final concentration of 1 or 10 mM were also included in the reaction mixtures, gels, and electrophoresis running buffer, as needed.
Determination of the Molecular Weight of the ModE-DNA Complex-In an effort to determine the stoichiometry of the ModE-DNA association, binding reactions containing 0.1 pmol of a 43-bp DNA (Ϫ18 to ϩ25 region) and 25 nmol of ModE were prepared as described for the DNA mobility shift experiments. The binding reaction samples were then subjected to electrophoresis through a 5.0, 6.0, 7.0, or 8.0% polyacrylamide-Tris-borate-EDTA-nondenaturing gel as described for DNA mobility shift experiments (31). The distance of migration of ModEmolybdate-DNA complex was compared with values obtained with protein standards. This information was used to produce a Ferguson plot (35), and the apparent molecular weight of the ModE-DNA complex was determined by extrapolation from the Ferguson plot. For determination of the R f values of the protein standards, 15 g of lactalbumin (14,200), 20 g of carbonic anhydrase (29,000), 20 g of chicken egg albumin (45,000), 15 g of bovine serum albumin (monomer, 66,000; dimer, 132,000) and 6 g of urease (dimer, 240,000; tetramer, 480,000) were subjected to electrophoresis along with the ModE-DNA complex.
Fluorescence Spectroscopy Measurements-Emission spectra were collected using an Aminco Bowman Series 2 spectrofluorometer (Spectronic) at a bandpass width of 4 nM. The detector voltage was set at 610V. ModE proteins were diluted in Tris buffer (pH 8.0) with 0.5 mM dithiothreitol for measurement using a 1-cm path length. Three separate concentrations of ModE and mutant ModE proteins (1, 2.4, and 4.8 M) were analyzed for intrinsic fluorescence with and without added molybdate.

RESULTS AND DISCUSSION
Identification of ModE Binding Sites in modA Operator/ Promoter DNA Using DNase I-Footprint Analysis-ModE has previously been shown to repress transcription of the mod-ABCD operon only in a molybdate-dependent fashion (7,13,14). However, the purified ModE was reported to bind to modA operator DNA even in the absence of added molybdate, and it protected the DNA from hydrolysis by DNase I (17). This raised the possibility that a fraction of ModE used in these experiments is contaminated with molybdate and the analytical sensitivity of molybdate determination methods is not high enough to detect its presence. In order to evaluate this possibility, ModE purified as described previously (13) was compared with ModE purified using the nickel affinity method (*ModE). The nickel affinity purification method was chosen because it involves only a limited number of steps, thereby decreasing the exposure of ModE protein to large volumes of buffer, which could be a possible source of molybdate contamination. The E. coli host used for expression of modE in our experiments, strain BL21(DE3), was found to carry an unidentified mod operon mutation and thus is incapable of transporting molybdate when grown in rich medium not supplemented with molybdate. 2 ModE binding to DNA was used as the assay for molybdate contamination of ModE, and to increase the sensitivity, DNase I footprinting was used as the analytical tool.
The results presented in Fig. 1 (top panel) show that ModE alone partially protected modA DNA from DNase I hydrolysis at a concentration of 50 nM (lanes 2-4). A hypersensitive A at position ϩ5 can also be seen at this concentration of ModE. Addition of molybdate to the binding reaction reduced the amount of ModE-molybdate required for complete protection to less than 5 nM (Fig. 1, lane 6). In the presence of *ModE protein purified using the nickel affinity procedure, protection of modA operator/promoter DNA from DNase I hydrolysis was dependent on molybdate (Fig. 1, lanes 11-18). Even at 100 nM *ModE, the modA operator DNA was not protected from DNase I (Fig.  1, lane 14). These results suggest that greater than 95% of the *ModE in this preparation is free of molybdate and that the active form of ModE binding to modA operator DNA is ModEmolybdate.
Analysis of the ModE-molybdate protected region of the modA operator/promoter DNA after DNase I cleavage of the coding strand (Fig. 1, lanes 6 -9 and 15-18) revealed that there are three areas of protection detectable at a ModE-molybdate concentration as low as 5 nM. The first protected region, GT-TATATT, spans from Ϫ15 to Ϫ8 and overlaps the modA Ϫ10 sequence (Fig. 1, bottom panel). The second ModE-protected region, GCCTACAT spans from Ϫ4 to ϩ4 of the modA operator/ promoter DNA, whereas the third protected region, GTTACAT, is located at bases ϩ8 to ϩ14. Each of the three DNase I-cleavage protected regions contains either a TATAT (protected region 1) or a TACAT (protected regions 2 and 3) sequence, which suggests that it is these sequences that are recognized by ModE as the target binding site. The independent protected sites are separated by three bases, and the pentameric sequences in regions 1 and 3 are preceded by the same two bases, GT. Aside from the three protected regions, the base A at position ϩ5, located next to the 3Ј end of the second protected region, is hypersensitive to DNase I-cleavage. An inverted repeat of bases (Ϫ16 to ϩ8) also exists within the ModE-protected region of the modA operator/promoter DNA. McNicholas et al. (17), based on DNase I footprinting, reported that ModE protected the entire region of DNA spanning from Ϫ18 to ϩ10. The hypersensitive A at position ϩ5 was not detectable in the reported ModE-footprint (17). Although the hypersensitive site is recognizable in the ModE-modA DNase I footprint reported by Anderson et al. (15), the unprotected bases between the regions 1 and 2 were not discernable. The fact that the DNase I footprint presented in Fig. 1 resolves the modA operator/ promoter region into discrete ModE-binding regions with a distinct DNase I hypersensitive site delineating protected regions 2 and 3 suggests that ModE binds and protects the three indicated regions in vivo also. The *ModE also protected the same bases with the hypersensitive adenine in the modA operator DNA, but the amount of protein required for complete protection was slightly higher (10 nM versus 5 nM for ModEmolybdate). Similar DNase I-footprinting of the noncoding strand of the modA operator/promoter DNA also was observed (data not presented).
Because the DNase I protection experiments revealed that ModE-molybdate protected three regions of modA operator DNA, this DNA was mutated to establish the requirement for each of the three regions for ModE-molybdate binding. In a DNA electrophoretic mobility shift experiment, modA operator region from plasmids pAM6 and pAM18, lacking either the first or second pentamer sequence, respectively, (Table I; Fig. 1), failed to bind ModE-molybdate (data not presented). The modA operator DNA from plasmid pAM17 that lacks the third pentamer sequence did bind ModE-molybdate, and the DNA-protein complex migrated at a lower rate than the DNA alone during electrophoresis (data not shown). The apparent K d for the interaction between the DNA lacking the region three and ModE-molybdate in an optical sensor experiment was about three times higher (1 nM) than the native DNA with all three sites (0.3 nM). Based on these results, the DNA in the first and second protected regions is essential for binding ModE-molybdate, whereas the third pentamer enhances ModE-molybdate binding.
The GTTA sequence found in the ModE-protected regions 1 and 3 repeats itself nine times within the presented 81-bp region of the modA operator/promoter DNA (Fig. 1, upper panel). The function of this tetramer sequence in modA DNA is not known. However, the presence of this tetramer sequence in the ModE-molybdate-protected region in both the hyc and narXL promoter DNA (31) suggests that ModE-molybdate does recognize the sequence GTTA.
Determination of the Stoichiometry of the Association of ModE with modA Operator/Promoter DNA-The DNA mobility shift experiments with native and mutant forms of modA operator DNA indicated that a stable ModE-molybdate-DNA complex may involve contact with only two of the three ModEprotected regions. If this is the case, and given the size of the ModE protein (28, 271 Da) and the target binding sites (TATAT and TACAT), it is likely that ModE would bind to each of the two ModE-protected regions as a monomer. To determine the stoichiometry of the ModE-DNA association, ModE was bound to a 43-bp modA operator/promoter DNA that contains all three ModE-protected regions in a standard DNA mobility shift binding reaction. The calculated apparent molecular weight of ModE-molybdate-DNA complex, based on its migration through a set of different percentage polyacrylamide gels and the resulting Ferguson plot (35) of these data, was 81,247. This molecular weight compares favorably with an apparent molecular weight of 83,026 expected for the association of a ModEdimer (56, 424) with the 43-bp DNA (26, 602) (Fig. 2). Thus, the ModE-DNA complex consists of a ModE dimer associated with the 43-bp-long DNA. For this stoichiometry determination, the axial ratio and electrophoretic mobility of the protein-DNA complex was considered to be similar to the characteristics of the protein itself, because the DNA used in this experiment was only 43 bp. Sedimentation equilibrium analysis of ModE also revealed that the protein exists as a homodimer in solution (15). The ModE-molybdate apparently binds to the modA operator/promoter DNA with 2-fold symmetry as a dimer using the palindromic consensus operator with the proteins centered around the pentamer sequence TATAT or TACAT.
Molybdate  previously described several ModE mutants that are either inactive (A76V) or molybdate-independent (T125I and Q216*) in modA repression. These mutant proteins were purified using His tag, nickel affinity chromatography. In agreement with the in vivo results, the *ModE(A76V) did not protect modA operator/promoter DNA from DNase I hydrolysis (Fig. 3, lanes 9 -11).
Inclusion of molybdate in the reaction mixture did not increase the affinity of *ModE(A76V) for the DNA (data not presented). Although the *ModE(T125I) mutant protein protected the same region of DNA protected by native ModE, the amount of protein required for complete protection in the absence of molybdate was close to 100 nM (Fig. 3, lanes 6 -8). At this concentration, the level of protection by the *ModE(T125I) protein was about 2 times better than the *ModE protein. The higher affinity of *ModE(T125I) in the absence of molybdate is in agreement with the observed in vivo results on repression of modABCD operon by the ModE mutant protein.

Kinetics of Interaction between ModE and modA
Operator DNA-The apparent K d for binding of ModE-molybdate to modA operator DNA was previously reported to be about 25 nM (15,17). This K d value is significantly higher than the Ͻ5 nM of ModE-molybdate required for complete protection of modA operator/promoter DNA from DNase I hydrolysis (Fig. 1, lane 6). To resolve this difference, the interaction between ModE and modA operator/promoter DNA was determined using an evanescent wave biosensor. Upon addition of 13.5 nM *ModEmolybdate, a rapid response signifying binding was observed, and the maximum binding was achieved within 80 s (Fig. 4A). In the absence of added molybdate, ModE bound to the same DNA at a lower rate, and the total response signifying the amount of protein bound was also lower. However, both forms of protein showed a concentration response to association with DNA in the cuvette (Fig. 4, B and C). Upon removal of excess ModE, the ModE without molybdate dissociated from DNA at a higher rate than the sample with molybdate. ModE-molybdate did not bind to a mutant modA operator DNA (TACAT between Ϫ1 and ϩ4 changed to CTTGG; plasmid pAM18) (Fig. 4A), demonstrating the specificity of ModE-Mo for native modA operator DNA. The small increase in response observed with the mutant DNA and ModE-molybdate was seen with the native DNA also and is apparently due to a change in refractive index of the buffer associated with ModE addition. However, nonspecific adsorption of the protein to the DNA cannot be ruled out.
The apparent K d for this interaction between *ModE and modA operator/promoter DNA (Ϫ18 to ϩ25) was calculated to be 0.3 nM. An apparent K d of 0.4 nM was obtained when native ModE was used in these experiments. This apparent K d value is similar to the subnanomolar apparent K d values reported for other repressor/operator DNA interactions (19 -21). In the absence of molybdate, this apparent K d value for the interaction between ModE and modA operator/promoter DNA increased to 8 nM. When *ModE(T125I) mutant protein was used in these experiments, no significant difference was observed either in the rate of association or dissociation of the protein with the DNA both in the presence and absence of molybdate (Fig. 5). The apparent K d for the interaction between *ModE(T125I) and modA operator DNA was 3 nM in the presence of molybdate and 4 nM in the absence of molybdate. These values are about 10 times higher than the values obtained with native ModE-molybdate but about 1 ⁄2 of the K d value obtained with ModE alone. The *ModE(A76V) protein did not bind to the DNA at a protein concentration as high as 54 nM, either with or without molybdate (Fig. 5A). This is in agreement with the observed inability of this mutant protein to repress modA-lac in vivo or to protect the modA DNA from DNase I hydrolysis (Fig. 3) (13). These results also show that the interaction between ModE-Mo and modA operator DNA is specific. The kinetics of *ModE(Q216*) interaction with DNA was found to be complex, and the apparent K d value for this interaction was not determined.
It is interesting to note that the calculated apparent K d for *ModE(T125I) in the absence of molybdate is about 3 nM, whereas that of native ModE is 8 nM. However, in vivo, ModE(T125I) repressed modA-lac expression in the absence of molybdate, whereas native ModE failed to repress it in the absence of molybdate. The rate of dissociation of the two pro- teins from the DNA was found to be significantly different (Figs. 4 and 5). Thus, the apparent discrepancy between the in vivo activity of ModE(T125I) and its in vitro kinetic parameters compared with the native ModE characteristics can possibly be reconciled by the observation that the ModE(T125I) dissociates from the protein-DNA complex at an appreciably lower rate than does the native ModE. This phenomenon of a lower rate of dissociation of a mutant protein has been previously described for the interaction of trp superrepressor with its operator DNA (36).
Fluorescence Characteristics of ModE Mutant Proteins-Anderson et al. (15) reported that the intrinsic fluorescence of ModE protein decreased upon binding molybdate and this could be a result of conformational change of the protein. Because the ModE mutant proteins are molybdate-independent, it is possible that the mutant proteins structurally mimic the ModE-molybdate complex. If this is indeed the case, the intrinsic fluorescence of the mutant proteins should not be altered by addition of molybdate. When the *ModE protein was excited by irradiation at 290 nm, the emission spectrum had a peak at 347 nm and a shoulder at 370 nm (Fig. 6). The peak of emission shifted down to 343 nm in the presence of molybdate, and the relative fluorescence was also reduced by about 60%. This reduction in fluorescence was dependent on the molybdate concentration. Relative fluorescence of ModE decreased linearly between the molybdate concentrations of 0.2 and 2.0 M, and maximum response was observed at about 10 M. The amount of molybdate required for a reduction of 1 ⁄2 of the total extent of fluorescence change was about 0.75 M, and this value is similar to the 0.8 M reported by Anderson et al. (15) for an equivalent change. ModE has three tryptophans (positions 49, 131, and 186) (Fig. 7), and the tryptophan at 131 is located within a sequence motif ( 125 TSARNQWFG 133 ) that is conserved in several ModE homologs from other organisms ((T/ S)SARNQXXG) and also in a molybdopterin-binding protein from C. pasteurianum (8). It is possible that molybdate binding to ModE buried the tryptophan at 131, and its fluorescence is not readily observable. Because a mutation that changed the Thr (to Ile at 125) or Gly (to Asp at 133) in E. coli ModE converted the mutant proteins to molybdate-independent forms (13), it is possible that the conformation of the mutant protein mimics ModE-molybdate.
In agreement with the possible conformational change, the relative fluorescence of ModE(T125I) was not altered by molybdate (Fig. 6). The peak of emission was 339 nm, which is closer to the peak of emission of ModE-molybdate (343 nm) than that of ModE (347 nm). The intrinsic fluorescence of the C-terminal deletion protein ModE(Q216*) was also not affected by the addition of molybdate (Fig. 6). These results suggest that the molybdate-independent ModE mutant proteins resemble ModE-molybdate in its conformation. This "locked-on" conformation may have allowed the protein to bind to DNA, even in the absence of added molybdate. However, this conformation is apparently not the same as that of ModE-molybdate because the apparent K d for ModE(T125I) was about 10 times higher than that of the native ModE-molybdate (Figs. 4 and 5).
Other ModE Mutant Proteins-The C-terminal 47 amino acids (217-262), which contain all three cysteines in the protein (positions 217, 230, and 262) (Fig. 7) are apparently essential for normal function of ModE, because ModE(Q216*), which lacks the amino acids from 216 to the C terminus, is molybdateindependent for repression of modABCD operon. The conformation of ModE(Q216*), based on intrinsic fluorescence emission characteristics, is also significantly different from that of native ModE. It is possible that one or more of these cysteines play a role in the conformation change of ModE in response to molybdate. In order to evaluate this possibility, two different sets of ModE mutants were constructed. In the first set, two additional deletion derivatives lacking one or two cysteines were constructed (ModE-220hyb and ModE-N252*). In ModE(220hyb), the cysteines at positions 230 and 262 were removed, and in ModE(N252*), only the C-terminal cysteine was removed. Both of these ModE mutants were found to be molybdate-independent for repression of modABCD operon (Table II). In an alternate experiment, the three cysteines were changed individually by mutagenesis, and the length of the protein was maintained at the full length of 262 amino acids. The mutant proteins repressed modA-lacZ expression only in the presence of molybdate (Table II), suggesting that the cysteines are not directly involved in the molybdate-dependent conformational change. However, removal of the C-terminal 11 amino acids, which are mostly hydrophobic, shifted the protein into a molybdate-independent repressor (ModE-N252*) and apparently altered the conformation of the protein significantly to expose the DNA-binding helix-turn-helix region located in the N-terminal part of the protein. The hydrophobic amino acids ILLTL in the N-terminal part of the protein (positions 5-9) are also conserved in Hemophilus influenzae ModE protein (37), and the H. influenzae ModE protein can functionally replace the E. coli ModE in vivo (16). Deleting these amino acids eliminated the biological activity of the ModE protein (Table  II), suggesting that the N-terminal part of the protein is critical for repression of modABCD expression.
Changing the amino acids in the SARNQ region (amino acids 126 -130; Fig. 7), especially the basic amino acids RN, to PG would be expected to cause a structural change, and such a mutant protein was found to be inactive (Table II). McNicholas et al. (14) replaced some of the amino acids in the SARNQ region individually and found that the repression ratio was altered (␤-galactosidase activity from mod-lac; ϪMo/ϩMo). Taken together, these results indicate that the amino acids in the TSARNQXXG region, as well as the C-terminal 11 amino acids, are critical for molybdate-dependent conformational  change of the ModE protein.
Oxyanion Specificity of ModE-The possibility that other oxyanion analogs of molybdate could functionally substitute for molybdate in promoting the association of ModE with modA operator/promoter DNA was investigated under both in vivo and in vitro conditions. It was found that molybdate and tungstate can support ModE-mediated repression in vivo, because addition of 1 mM sodium molybdate or sodium tungstate to strain SE2069 culture medium resulted in extremely low modAЈ-ЈlacZ expression (30 and 40 units of ␤-galactosidase activity, respectively, compared with 1300 units of activity produced by strain SE2069 grown without molybdate or tungstate). Neither sodium sulfate nor sodium vanadate could substitute for molybdate in vivo, as evidenced by the resultant high or nonrepressed ␤-galactosidase activity levels. These results show that tungstate is the only other oxyanion tested that can substitute for molybdate in modABCD regulation. This was also borne out by direct ModE-binding experiments with the modA operator/promoter DNA in a DNA electrophoretic mobility experiment. The apparent K d for the interaction between ModE-tungstate and modA operator/promoter DNA was about 6-fold higher than with ModE-molybdate. With sulfate, the apparent K d increased by about 21-fold. Vanadate did not support DNA binding. These results are in agreement with those of Anderson et al. (15), in which tungstate was able to replace molybdate in a DNA electrophoretic mobility experiment involving native ModE and modA operator/promoter DNA. Therefore, the successful activation of ModE relies solely on the binding of an appropriately sized and shaped moiety to affect the change in the conformation of ModE that allows for efficient binding of DNA.
The ability of sodium tungstate to substitute for molybdate in this regulation is not surprising because it has been shown previously that another molybdate-binding protein, periplasmic binding protein (ModA), also binds tungstate with an apparent K d of 7 M (9). The apparent K d for molybdate and ModA interaction was reported to be 3 M. In some thermophiles, the "molybdoenzymes" contain tungsten rather than molybdenum (38). E. coli trimethylamine N-oxide reductase, a periplasmic molybdoenzyme, was reported to function with either molybdate or tungstate (39). However, other molybdoenzymes in E. coli, such as nitrate reductase and formate dehydrogenase-H, are inactive when the cells are grown with tungstate.
Conclusion-The results presented and discussed in this communication show that the ModE-molybdate binds to modA operator DNA with an apparent K d of about 0.3 nM; in the absence of molybdate, the affinity for modA operator DNA decreased by about 25-fold, and the apparent K d increased to 8 nM. The subnanomolar K d value for the ModE-molybdate binding to modABCD operator/promoter DNA is in concert with its apparent role as a repressor and in this regard is similar to other repressors (19 -21). ModE mutant proteins that are molybdate-independent for modA repression have similar intrinsic fluorescence characteristics of ModE-molybdate complex. However, the apparent K d for interaction between these mutant proteins and modA operator DNA is significantly higher than that of the native ModE-molybdate complex with its cognate operator DNA. Three regions in the modA operator DNA can be identified as ModE-binding areas (Ϫ15 to ϩ15), and only the segment between Ϫ15 and ϩ10, which also includes the sequence motif TATAT-7bp-TACAT, appears to be critical for binding of ModE-molybdate as a dimer.