Purification and Characterization of Wild-type and Mutant “Classical” Nitroreductases of Salmonella typhimurium

“Classical” nitroreductase ofSalmonella typhimurium is a flavoprotein that catalyzes the reduction of nitroaromatics to metabolites that are toxic, mutagenic, or carcinogenic. This enzyme represents a new class of flavin-dependent enzymes, which includes nitroreductases ofEnterobacter cloacae and Escherichia coli,flavin oxidoreductase of Vibrio fischeri, and NADH oxidase of Thermus thermophilus. To investigate the structure-function relation of this class of enzymes, the gene encoding a mutant nitroreductase was cloned from S. typhimuriumstrain TA1538NR, and the enzymatic properties were compared with those of the wild-type. DNA sequence analysis revealed a T to G mutation in the mutant nitroreductase gene, predicting a replacement of leucine 33 with arginine. In contrast to the wild-type enzyme, the purified protein with a mutation of leucine 33 to arginine has no detectable nitroreductase activities in the standard assay conditions and easily lost FMN by dialysis or ultrafiltration. In the presence of an excess amount of FMN, however, the mutant protein exhibited a weak but measurable enzyme activity, and the substrate specificity was similar to that of the wild-type enzyme. Possible mechanisms by which the mutation greatly diminishes binding of FMN to the nitroreductase are discussed.

Nitroreduction is an initial step in the metabolism of a variety of structurally diverse nitroaromatic compounds, including nitrofurans, nitropyrenes, and nitrobenzenes (1)(2)(3)(4)(5). Enzymes that catalyze this process are termed nitroreductases and are classified into two groups: oxygen-sensitive and oxygen-insensitive (6). The former enzymes, such as NADPH-cytochrome P-450 oxidoreductase (EC 1.6.2.4) and NADPH-b 5 oxidoreductase (EC 1.6.2.2), catalyze the one-electron reduction of nitro moiety in which case the anion free radicals are formed (7)(8)(9). These enzymes are termed oxygen-sensitive, because the resulting radicals are easily reoxidized to the parent compounds by O 2 in a futile redox cycle, which generates superoxide. Thus, these enzymes can mediate the reduction of nitroaromatics only under anaerobic conditions (3). The latter enzymes, such as NAD(P)H-quinone oxidoreductase (formerly called DT-diaphorase, EC 1.6.99.2) and nitroreductases of enteric bacteria, catalyze the two-electron reduction of the nitro moiety through nitroso and hydroxylamine intermediates to the fully reduced amino compounds (10,11). Although this process does not produce superoxide, some of the hydroxylamine intermediates are mutagenic and carcinogenic (12,13).
In the strains of Salmonella typhimurium used in the Ames mutagenicity test, a "classical" nitroreductase plays an important role in the reductive metabolic activation process (2). In fact, S. typhimurium TA98NR, a nitroreductase-deficient strain, is resistant to both the killing and mutagenic effects of nitroarenes, whereas S. typhimurium YG1021, a nitroreductase-overproducing strain, is extremely sensitive to the effects (14,15). The latter strain was constructed in this laboratory by introducing a multicopy number plasmid carrying the gene encoding the nitroreductase of S. typhimurium into an Ames tester strain TA98 (16). The enzyme is termed classical because other nitroreductases were identified in S. typhimurium later (17). The NfsB protein of Escherichia coli, which is about 90% homologous to the nitroreductase of S. typhimurium, has been used in antibody-directed enzyme prodrug cancer therapy, because it can activate a nitroaromatic monofunctional prodrug CB1954 (18 -20).
In 1990, we identified the nucleotide sequence of the gene encoding the nitroreductase of S. typhimurium and estimated that the enzyme is composed of 217 amino acids with a calculated M r of 23,955 (21). Since then, several enzymes have been reported to share similarities with the deduced amino acid sequence of the nitroreductase of S. typhimurium. Such enzymes include the oxygen-insensitive, flavin-dependent nitroreductases of Enterobacter cloacae (22) and E. coli (18,20) and flavin reductases of Vibrio fischeri (23). In addition, significant levels of similarities to the Salmonella enzyme have been observed in NADH oxidase of Thermus thermophilus (24), DrgA of Synechocystis sp. (GenBank TM accession numbers L29426 and D90910) (25), a putative flavin reductase of Hemophilus influenzae (National Center for Biotechnology Information ID B64116) (26) and a putative nitroreductase of a Mycoplasma-like organism (GenBank TM accession number L22217). The NADH oxidase of T. thermophilus reduces a number of nitro compounds and contains flavin as a cofactor (27). DrgA controls resistance to metronidazole, a nitroaromatic compound, in Cyanobacterium, suggesting that DrgA also participates in nitroreduction. These bacterial enzymes are functionally similar to mammalian NAD(P)H-quinone oxidoreductase, in that they are oxygen-insensitive and flavin-dependent enzymes (28). However, no sequence similarities are observed between the bacterial enzymes and NAD(P)H-qui-none oxidoreductase (29 -31). In addition, the bacterial enzymes tightly associate with FMN (11,20,31,32), whereas FAD is a prosthetic group in the mammalian enzyme (33,34). Thus, it is suggested that the bacterial enzymes represented by the nitroreductase of S. typhimurium constitute a separate class of flavoproteins.
To investigate structure-function relation of this class of enzymes, we cloned the gene encoding a mutant nitroreductase from S. typhimurium strain TA1538NR and compared the enzymatic properties between the mutant and wild-type nitroreductases. The parent strain of TA1538NR, i.e. S. typhimurium TA98NR, was isolated as a mutant resistant to the killing effects of nitrofuran (35). The purified mutant nitroreductase showed a reduced affinity for flavin, suggesting that the replacement of leucine 33 with arginine, which was found in the mutant nitroreductase gene, causes destabilization of folding FMN in the enzyme. Based on the amino acid sequence similarities between the nitroreductase of S. typhimurium and NADH oxidase of T. thermophilus, we discuss the possible mechanisms by which the mutation reduces the affinity of the enzyme for FMN.
Cloning of the Mutant Nitroreductase Gene from S. typhimurium Strain TA1538NR-Genomic DNA of S. typhimurium strain TA1538NR was isolated and digested completely with restriction endonucleases EcoRI and PstI. A 5.65-kilobase pair DNA fragment was recovered from agarose gel and ligated into EcoRI-and PstI-digested plasmid pHSG399. E. coli XL1-Blue (Stratagene, La Jolla, CA) was transformed by the resulting plasmids. Plasmid DNAs were prepared from the transformants, and the restriction patterns were analyzed by digesting them with EcoRI plus PstI, followed by agarose gel electrophoresis. Plasmids having the identical restriction pattern with that of the wild-type nitroreductase gene (15,21) were further selected by the digestion with EcoRV and the plasmid carrying the mutant nitroreductase gene was termed pYG143.
DNA Sequencing Analysis of the Mutant Nitroreductase Gene-DNA sequence of the cloned gene in pYG143 was determined by ALFred DNA sequencer using AutoRead sequencing kit (Amersham Pharmacia Biotech, Little Chalfont, Great Britain). Primers used for sequencing were 24-mer oligonucleotides corresponding to nucleotides 134 -157 (AAT-GACTCATGGAATCTGGTCGTA) in the coding strand and 1027-1004 (TTCGCGCCATTGATCATTGAGGAA) in the complementary strand shown in the Fig. 1 of Ref. 21. The coding region of nitroreductase is nucleotides 299 -949. Cy5-dATP (Amersham Pharmacia Biotech) was used for internal labeling. Plasmid pYG220, a pBluescript-based plasmid carrying the wild-type nitroreductase gene, was used as a control. Direct DNA sequencing of the mutant nitroreductase gene of strain TA1538NR was also carried out using polymerase chain reaction techniques described previously (13). In this case, strain TA1538, which carries the wild-type nitroreductase gene, was used as a control. The 24-mer oligonucleotides described above were used for amplifications.
Purification of the Wild-type Nitroreductase of S. typhimurium-S. typhimurium strain YG1021 was grown in 16 liters of LB broth (1% Bacto-tryptone, 0.5% Bacto-yeast extract, and 1% NaCl) supplemented with 10 g/ml tetracycline for 16 h at 37°C with vigorous shaking. The cells (101 g) were washed with 50 mM Tris-HCl buffer, pH 7.5 (buffer A). All the following steps were performed on ice or at 4°C. The washed cells were suspended in 3 volumes of buffer A containing 1 mM dithiothreitol (buffer B) and disrupted by sonication. The cytosol fraction was prepared, and streptomycin sulfate was added at a final concentration of 1% to precipitate nucleic acids. After centrifugation, nitroreductase was precipitated between 40 and 60% saturation of ammonium sulfate. The precipitate was dissolved in 60 ml of buffer B plus 1 M FMN (buffer C), followed by dialysis against buffer C. The dialyzed sample was applied to a DE52 anion-exchange DEAE-cellulose (Whatman) column (5 ϫ 15 cm) that was preequilibrated with buffer A. Nitroreductase was eluted with a 3,000-ml linear gradient of 50 -300 mM NaCl in buffer C at a flow rate of 2.5 ml/min. Fractions were subjected for enzyme assays and the analysis by SDS-polyacrylamide gel electrophoresis. The activity was eluted with about 130 mM NaCl in buffer C. The active fractions (160.5 ml) were dialyzed against buffer C, and nitroreductase was precipitated by the addition of ammonium sulfate to 35% saturation (buffer C ϩ 35%). Then hydrophobic interaction chromatography was performed using a phenyl-Sepharose CL-4B (Amersham Pharmacia Biotech) column (1.5 ϫ 30 cm). After the column was washed with 30 ml of buffer C ϩ 20% ammonium sulfate saturation solution, nitroreductase was eluted with a 500-ml linear gradient of 20 to 0% saturation of ammonium sulfate in buffer C at a flow rate of 0.4 ml/min. The nitroreductase activity was eluted in about 5% saturation of ammonium sulfate in buffer C. The active fractions (115 ml) were dialyzed against 10 mM sodium phosphate buffer, pH 6.8, and applied to a Bio-Gel HT hydroxyapatite (Bio-Rad) column (1 ϫ 6 cm). The nitroreductase activity was recovered in the pass-through fractions. The fractions (170.5 ml) were concentrated using Centriprep 10 concentrators (Amicon, Beverly, MA) to 4 ml. After adjustments of the concentrate to 50 mM Tris-HCl, pH 7.5, and 50 mM NaCl, the sample was loaded onto a Sephadex G-100 (Amersham Pharmacia Biotech) column (2.5 ϫ 100 cm). The elution buffer was buffer C containing 50 mM NaCl, and the flow rate was 0.25 ml/min. The active fractions (30 ml) were pooled and concentrated to 2 ml by Centriprep 10 concentrators. The final preparation was used as purified nitroreductase.
Amino-terminal Sequence Analysis-Purified enzyme solution (1.5 l) was desalted and diluted to 30 l. The sample was loaded on to a glass fiber disc and was analyzed by Applied Biosystems 477 sequencer for the first 19 amino acids.
Purification of the Mutant Nitroreductase-Plasmid pYG143 was digested with NotI and SacII. A 1.0-kilobase pair DNA fragment, which contains the mutant nitroreductase gene, was isolated from agarose gel and inserted into the NotI-and SacII-digested pBluescript KSϩ (Stratagene). The resulting plasmid was designated as pYG149 and was introduced into the nitroreductase-deficient strain S. typhimurium TA1538NR. LB broth (1 liter) supplemented with 50 g/ml ampicillin was inoculated with the transformant and incubated for 16 h at 37°C. Purification protocols were similar to those for the purification of the wild-type enzyme, but several modifications were made because of different chromatographic behavior of the mutant enzyme. After streptomycin sulfate precipitation, the mutant nitroreductase was precipitated between 33 and 45% saturation of ammonium sulfate. The mutant nitroreductase was eluted from a DEAE-cellulose column DE52 in about 80 mM NaCl solution, which was lower than that needed for elution of the wild-type enzyme. The mutant nitroreductase tightly bound to a phenyl-Sepharose CL-4B column and could not be eluted from the column without using organic solvent. Thus, this step was omitted. Contrary to the wild-type enzyme, the mutant enzyme bound to a hydroxyapatite column and was eluted with about 70 mM sodium phosphate buffer, pH 6.8. A fraction that eluted from a hydroxyapatite column showed a single band when analyzed by SDS-polyacrylamide gel electrophoresis followed by visualization with Coomassie Brilliant Blue. Thus, this fraction was used as a source of the purified mutant nitroreductase.
Western Blot Analysis-Samples were separated by SDS-polyacrylamide gel electrophoresis and blotted on nitrocellulose sheets (BA83, Schleicher & Schuell, Dassel, Germany) by semidry electrophoretic transfer. The rabbit antiserum raised against the wild-type nitroreductase was prepared by Takara-shuzo Co. (Otsu, Japan). The nitroreductase antiserum and horseradish peroxidase-linked anti-rabbit antibody (Amersham Pharmacia Biotech) combined with diaminobenzidinestaining technique were used to visualize the wild-type and mutant nitroreductases (36).
Enzyme Assays-Nitrofurazone reductase activity was determined by the method of McCalla et al. with modifications (37). Reaction was performed in 0.6 ml of buffer B (50 mM Tris-HCl buffer, pH 7.5, 1 mM dithiothreitol) containing 0.05 mM nitrofurazone, NADPH-generating system (6.7 mM glucose 6-phosphate, 0.04 mM NADP ϩ , 0.8 unit of glucose-6-phosphate dehydrogenase), and an enzyme preparation. The reaction was carried out at 37°C, and the initial reaction velocity was determined by monitoring the decrease in absorption at 375 nm of nitrofurazone. The extinction coefficient of 1.5 ϫ 10 4 M Ϫ1 cm Ϫ1 was used to calculate the amount of nitrofurazone reduced per min per mg of protein. Because the absorbance of NAD(P)H overlaps that of nitrofurazone at 375 nm, the NADPH-generating system was employed to maintain constant concentrations of NADPH. When a fixed concentration of NADH or NADPH was used, the decrease in absorption at 400 or 420 nm was monitored. Flavin-reductase activity was determined by the method described by Zenno et al. with modifications (38). The reaction mixture (0.6 ml) consisted of enzyme preparation, flavin (FMN, FAD, or riboflavin, 0.1 mM), and electron donor (NADH or NADPH, 0.1 mM) in buffer B. Before the addition of NADH, the reaction mixture was preincubated at 23°C for 5 min. Reaction was carried out at 23°C, and the initial reaction rate was determined by measuring the decrease in absorption of NADH at 340 nm. All assays were performed in Shimadzu double beam spectrophotometer UV-200S (Shimadzu, Kyoto, Japan). Protein concentration was determined using a Bio-Rad Bradford protein assay kit. Bovine serum albumin (Boehringer Mannheim number 711454) was used as a standard.
Flavin Cofactor Analysis-Apoenzyme of the wild-type nitroreductase was prepared using potassium bromide (39). Briefly, the purified enzyme preparation was combined with 3 volumes of buffer B containing 4 M potassium bromide. The solution was applied to Centricon 10 concentrators (Amicon). Once concentrated, the solution was diluted with buffer B containing 3 M potassium bromide. This concentrationdilution cycle was repeated several times, and the final preparation was used as the apoenzyme. Then, 1 nmol of the apoenzyme was combined with 2 or 100 nmol of FMN, FAD, or riboflavin in a final volume of 20 l in buffer B and incubated overnight at 4°C. Riboflavin was not completely dissolved when the final concentration was 5 mM. The incubated enzyme preparations were analyzed for their nitrofurazone-reductase activities.

RESULTS
Mutant Nitroreductase Gene in S. typhimurium TA1538NR-S. typhimurium TA98NR and its pKM101-removed derivative, i.e. TA1538NR, are deficient in nitroreductase and are insensitive to mutagenic and cytotoxic effects of nitroaromatic compounds (15,35). To identify the mutation that inactivates the enzyme activity, the nitroreductase gene of the deficient TA1538NR strain was cloned, and the entire coding region plus the 5Ј-flanking region was sequenced. A base change mutation of T:A to G:C transversion at nucleotide 396 shown in the Fig. 1 of Ref. 21, which leads to a replacement of leucine 33 with arginine, was identified (Fig. 1). Direct amplification and sequence analysis of the genomic DNA of TA1538NR also detected the same mutation. These results suggest that the base change mutation at the nitroreductase gene confers the phenotypes upon the strains TA1538NR and TA98NR.
Purification of the Wild-type and Mutant Nitroreductases of S. typhimurium-From 101 g of cell pellet, we obtained 24 mg of the purified wild-type nitroreductase. The NH 2 -terminal amino acid sequence of the purified protein was MDIVS-VALQRYSTKAFDPS, which was exactly the same as that deduced from the nucleotide sequence of nitroreductase gene (21). To purify the mutant protein without any contamination of the wild-type enzyme, the strain TA1538NR deficient in nitroreductase was used as a host strain for the overproduction of the mutant enzyme. The mutant protein showed different behavior from the wild-type enzyme during the protein purification: it bound tightly to phenyl-Sepharose CL-4B and hydroxyapatite columns. We have obtained 2.3 mg of the purified mutant protein (Table I). The final preparations were more than 99 and 95% pure in the wild-type and the mutant enzymes, respectively, as judged by SDS-polyacrylamide gel electrophoresis followed by staining with Coomassie Brilliant Blue (Fig. 2, A  and B). The mutant and wild-type nitroreductases could react with antiserum raised against the wild-type nitroreductase with similar efficiencies (Fig. 2C).
The Nitroreductase Is a Flavin-dependent Oxygen-insensitive Nitroreductase-The absorption spectrum of the purified wildtype nitroreductase, the color of which was yellow, showed two peaks at 370 and 455 nm (Fig. 3). The absorption spectrum suggests that it is a flavoprotein containing an oxidized form of flavin (11). The enzyme was judged as an oxygen-insensitive nitroreductase, because the product generated from nitrofurazone by this enzyme showed an absorption maximum at 276 nm (data not shown). This indicates the formation of an open chain nitrile, which is a two-electron-reduced product of nitrofurazone (6). We compared the abilities of NADH and NADPH as electron donors and those of FMN, FAD, and riboflavin as cofactors for supporting nitroreductase activities. Both NADH and NADPH were effective as the electron donors, although the K m for NADH was more than four times lower than that of NADPH (Table II). FMN was most effective when the nitroreductase holoenzyme was reconstituted from the apoenzyme plus FMN, FAD, or riboflavin (Fig. 4). Although it was less effective, FAD also acted as a cofactor for the nitroreductase. A higher concentration of FAD (100 ϫ FAD) yielded a higher activity than did a lower concentration (2 ϫ FAD). There was no activity when riboflavin was used as a cofactor. We also compared the reduction rates of FMN, FAD, and riboflavin by the nitroreductase holoenzyme (Table III). In this case, riboflavin was most efficiently reduced. When the apoenzyme was used, the order of reduction rates was FMN Ͼ FAD Ͼ Ͼ riboflavin. Riboflavin was not reduced, probably because riboflavin did not act as a cofactor for supporting flavin reductase activity. Both NADH and NADPH acted as electron donors for the  a Nitrofurazone reductase activity was determined in a reaction mixture containing 0.05 mM nitrofurazone and the NADPH-generating system. To measure the mutant nitroreductase activity, 0.1 mM of FMN was supplemented to the reaction mixture. Activity was calculated based on the reduction of nitrofurazone.
b -Fold increase of the specific activity of each fraction relative to that of crude extract was presented in parentheses.
reduction of FMN, as in the case of the reduction of nitrofurazone (data not shown).
The Mutant Nitroreductase Has a Reduced Affinity for FMN-The wild-type nitroreductase showed the characteristic yellow color of flavin, and the color could hardly be removed by simple dialysis or ultrafiltration. However, the mutant nitroreductase was colorless. In an attempt to reconstitute the holoenzyme, the colorless preparation was incubated with excess FMN, overnight at 4°C, followed by dialysis or ultrafiltration to remove free FMN. Nevertheless, we could only recover colorless, inactive enzyme after this procedure. To examine the possibility whether the mutant nitroreductase has a reduced affinity for FMN, the enzyme activities were assayed in the presence of FMN (0.015 and 0.1 mM). Under the conditions, the mutant enzyme exhibited a weak but measurable nitrofurazone reductase activity: in the presence of 0.015 and 0.1 mM FMN, the mutant enzyme exhibited 0.02 and 0.14% nitrofurazone reductase activity, respectively, as compared with the wild-type enzyme. The values represent the results of three independent experiments. To exclude the possibility that the activity was due to different nitroreductases contaminated in the preparation of the purified enzyme, we compared the ac-tivities of crude extracts of strain TA1538NR harboring plasmid pYG143 carrying the mutant nitroreductase gene with those of crude extracts of strain TA1538NR without plasmids. Only the crude extract of strain TA1538NR harboring pYG143 exhibited an elevated nitroreductase activity in the presence of 0.1 mM FMN (data not shown), suggesting that the mutant nitroreductase itself has an FMN-dependent nitroreductase activity. Another formal possibility that the activity was due to trace amounts of the wild-type nitroreductase mistranslated from the mutant nitroreductase gene seems unlikely. Even if the wild-type enzyme were contaminated in the crude extracts, it could be efficiently removed from the L33R enzyme during the purification, in particular, in the step of hydroxyapatite chromatography. From these results, we concluded that the major deficit of the mutant nitroreductase is its reduced affinity for FMN, which is an essential cofactor for nitroreductase activities.
Comparison of Enzymatic Properties between the Wild-type and Mutant Enzymes-If the major deficit of the mutant nitroreductase is its reduced affinity for FMN, it may exhibit weak but similar enzymatic properties to the wild-type enzyme in the presence of FMN. To examine this possibility, we characterized the mutant enzyme for its NAD(P)H requirement and for its substrate specificity in the presence of 0.1 mM FMN (Table II). The mutant enzyme could use both NADH and NADPH as electron donors. Like the wild-type enzyme, the K m for NADH was smaller than that for NADPH. The mutant enzyme exhibited p-nitroacetophenone-and menadione-reducing activities (Table III). Relative activities of the mutant enzyme to the wild-type were about 0.25% for the substrates. In addition, the mutant enzyme exhibited a very low but significant flavin reductase activity (0.006 mol/min/mg of protein), which is 0.4% of the corresponding activity of the wild-type enzyme. These results suggested that the mutant enzyme has weak but similar enzymatic properties to the wild-type nitroreductase. In other words, the mutant enzyme could be an oxygen-insensitive NAD(P)H-nitroreductase associated with quinone reductase and flavin reductase activities, with a reduced affinity for FMN.

DISCUSSION
It has been suggested that there is an oxygen-insensitive NAD(P)H nitroreductase/flavin reductase family among bacteria, because oxygen-insensitive NAD(P)H nitroreductase of E. cloacae, nitroreductase of E. coli, i.e. NfsB, NAD(P)H flavin reductase of V. fischeri, NADH oxidase from T. thermophilus, and the nitroreductase of S. typhimurium share sequence similarities at the amino acid level (18, 20 -22, 38, 40). These enzymes require flavin as a cofactor, and the most effective flavin is FMN. The amino acid sequences of the bacterial enzymes do not share significant similarity with other proteins such as FMN-binding protein in bacteria (41) or NAD(P)Hquinone oxidoreductase in higher organisms (29,31,42). Thus, we suggested that the enzymes belonging to the bacterial nitroreductase family may have common structural characteristics necessary to bind FMN tightly.
In this study, we identified a transversion mutation at nucleotide 396 of the gene encoding the nitroreductase of S. typhimurium TA1538NR, a pKM101-removed derivative of TA98NR (Fig. 1), and characterized the purified mutant enzyme (Tables I, II, and III). TA98NR is a strain deficient in the nitroreductase and was isolated as a mutant resistant to the killing effects of nitrofuran (14,35). The purified mutant protein, in which leucine 33 is replaced with arginine (L33R), showed no detectable level of nitrofurazone nitroreductase ac-tivity under the standard assay conditions. In addition, even the crude extract of the strain S. typhimurium TA1538NR harboring plasmid pYG143 highly expressing the mutant protein exhibited an extremely low activity. From these results, we concluded that the mutation of L33R is responsible for the nitroreductase deficiency of S. typhimurium TA98NR.
The nitroreductase protein having the mutation of L33R exhibited low but significant nitroreductase activities when FMN (0.1 mM) was present in the assay mixture. The mutant nitroreductase easily lost FMN by dialysis or ultrafiltration. These findings suggest that the replacement of leucine 33 with arginine reduces the affinity for FMN. Why is Leu 33 essential for FMN binding? One possibility is that this residue forms part of the cofactor binding site itself. Alternatively, Leu 33 may be distant from the binding site, but critical for determining the overall native conformation of the wild-type enzyme. It appeared that the mutation slightly affected the protein structure, because the mutant protein exhibited different chromatographic behavior from the wild-type enzyme. However, the mutant enzyme as well as the wild-type could use both NADPH and NADH as electron donors and exhibited quinone reductase and flavin reductase activities (Tables II and III). These results raised the possibility that the mutant protein has a structure similar to that of the wild-type enzyme but with a reduced affinity for FMN.
Fine structures of nitroreductase of S. typhimurium or closely related enzymes such as NfsB of E. coli are not available at present (43,44). However, x-ray crystallographic analysis of TABLE II Kinetic parameters for the wild-type and mutant nitroreductases of S. typhimurium Michaelis constant (K m ) and maximal velocity (V max ) were determined from double-reciprocal plots of initial velocity of nitrofurazone reduction versus NADH, NADPH, or nitrofurazone concentrations. When the kinetic parameters for NADH and NADPH were determined, the initial velocity was determined in a reaction mixture containing nitrofurazone (0.15 mM), an electron donor (NADH or NADPH), and a purified enzyme preparation. When the L33R mutant was used, FMN (0.1 mM) was also added to the reaction mixture. When the kinetic parameters for nitrofurazone were determined, the fixed concentration of NADH (0.3 mM) was used. ND, not determined.

TABLE III
Substrate specificity of the wild-type and mutant nitroreductases of S. typhimurium Reaction velocities were determined based on the oxidation of NADH by monitoring the decrease of absorption at 340 nm (E ϭ 6.3 ϫ 10 3 M Ϫ1 cm Ϫ1 ). Nitroaromatics and quinone reductase activities were determined using a reaction mixture containing substrate (0.1 mM), NADH (0.1 mM), and a purified enzyme. When the mutant L33R protein was used, FMN (0.1 mM) was also added to the reaction mixture. Apoenzyme was prepared from the purified wild-type nitroreductase using potassium bromide as described under "Experimental Procedures." Flavin reductase activities were determined in a reaction mixture containing substrate (0.1 mM FMN, FAD, and riboflavin), NADH (0.1 mM), and a purified enzyme preparation. ND, not determined. NADH oxidase from T. thermophilus (45) shows that the enzyme consists of two identical subunits associated with symmetry. This association creates two deep clefts between the subunits. Each cleft contains an active site where an FMN molecule is bound. Ser 45 and Pro 156 of NADH oxidase from T. thermophilus are directly involved in holding of FMN. These residues and the neighboring region, i.e. Ala-Pro-Ser 45 -Ala and Pro 156 -Met-Leu-Gly-Phe, are relatively conserved (Fig. 5). Arg 21 involved in holding of FMN is also conserved as positively charged arginine or lysine throughout the family. The sequence conservation at the regions, together with the observed sequence similarity throughout the Salmonella and Thermus enzymes, suggest that the conserved regions are involved in FMN binding in nitroreductase of S. typhimurium as well. In addition, the NfsB protein, a nitroreductase of E. coli, which shares about 90% amino acid identity with nitroreductase of S. typhimurium, has been reported to be a dimeric enzyme (18,46). This suggests that nitroreductase of S. typhimurium could make a dimeric structure containing two FMN-binding sites in the clefts between the subunits. Sequence similarity also suggests that Ala 39 of the Thermus enzyme may correspond to residue Leu 33 of the Salmonella enzyme, where the mutation was located in the mutant enzyme of S. typhimurium. Ala 39 of the Thermus enzyme is located almost at the end of helix "B" (45). The site is followed by a polypeptide loop interacting with FMN at the conserved region APS 45 A. Therefore, it seems plausible that the amino acid change at Ala 39 leads to a local conformational change of the conserved region, where FMN is bound. Another interesting possibility is that the mutation of L33R may disrupt the hydrophobic interaction between Leu 33 of one subunit and Ala 7 of the other. In the Thermus enzyme, there seems to be hydrophobic interaction between Ala 39 of one subunit and Leu 15 of the other. Leu 15 is close to Arg 21 , which is directly involved in the binding to FMN. Thus, L33R may affect the interaction between Arg 21 and FMN by changing the microenvironment between two subunits, thereby diminishing binding of FMN to the enzyme. Alternatively, the change would affect the structure of the open cleft, where FMN is bound, because Ala 39 of the Thermus enzyme is located near the helix "A" and "G," which form the cleft. This possible structural change at the clefts may increase the distance between two conserved regions, APS 45 A of one subunit and P 156 MLG of the other, both are directly associated with FMN and cause reduction of affinity for the cofactor. Although Leu 33 of the Salmonella enzyme and Ala 39 of the Thermus enzyme are not conserved among the family, no enzymes except for the mutant Salmonella nitroreductase have a positively charged residue, such as Arg 33 (Fig. 5). Thus, the mutation arginine for Leu 33 of the Salmonella enzyme may be critical for affinity for FMN at the possible binding site. Further structural analysis of the wild-type and the L33R enzyme of Salmonella will shed light on the structure-function relation of this FMN-dependent nitroreductase family.