Oxygen-Insensitive Nitroreductases: Analysis of the Roles of nfsA and nfsB in Development of Resistance to 5-Nitrofuran Derivatives in Escherichia coli

Abstract

Nitroheterocyclic and nitroaromatic compounds constitute an enormous range of chemicals whose potent biological activity has significant human health and environmental implications. The biological activity of nitro-substituted compounds is derived from reductive metabolism of the nitro moiety, a process catalyzed by a variety of nitroreductase activities. Resistance of bacteria to nitro-substituted compounds is believed to result primarily from mutations in genes encoding oxygen-insensitive nitroreductases. We have characterized the nfsA and nfsB genes of a large number of nitrofuran-resistant mutants of Escherichia coli and have correlated mutation with cell extract nitroreductase activity. Our studies demonstrate that first-step resistance to furazolidone or nitrofurazone results from an nfsA mutation, while the increased resistance associated with second-step mutants is a consequence of an nfsB mutation. Inferences made from mutation about the structure-function relationships of NfsA and NfsB are discussed, especially with regard to the identification of flavin mononucleotide binding sites. We show that expression of plasmid-carried nfsA and nfsB genes in resistant mutants restores sensitivity to nitrofurans. Among the 20 first-step and 53 second-step mutants isolated in this study, 65 and 49%, respectively, contained insertion sequence elements in nfsA and nfsB. IS1 integrated in both genes, while IS30 and IS186 were found only in nfsA and IS2 and IS5 were observed only in nfsB. Insertion hot spots for IS30 and IS186 are indicated in nfsA, and a hot spot for IS5 insertion is evident in nfsB. We discuss potential regional and sequence-specific determinants for insertion sequence element integration in nfsA and nfsB.

Nitroheterocyclic and nitroaromatic compounds constitute an enormous range of chemicals that are characterized by the presence of one or more nitro groups on a heterocyclic or aromatic nucleus. For many years nitro-substituted compounds have been used in manufacturing processes (23, 67) and as antimicrobial agents (49, 70). More recently, nitropolyaromatic compounds have been identified as by-products of a variety of combustion processes (20, 21, 52, 62, 63, 71). Human health concern in regard to all of these compounds has resulted from the identification of several nitro derivatives as potent bacterial mutagens, mammalian cell mutagens and clastogens, and rodent carcinogens (4, 37, 71).

The biological activity of nitro-substituted compounds is derived from reductive metabolism of the parent compound’s nitro moiety, a process catalyzed by a variety of nitroreductase activities (12, 24, 49, 59). Metabolism of nitroaromatic compounds proceeding through two-electron reduction of the nitro moiety yields nitroso and hydroxylamine intermediates and an amino-substituted product. Metabolism of 5-nitrofuran compounds produces an open-chain nitrile isomer in addition to the aminofuran derivative (49). The end products of metabolism are biologically inactive; the biological activities of these chemicals are conferred by the reactivities of short-lived intermediates with protein and DNA. Nitroreduction proceeding through single-electron increments yields an additional product, the nitro-anion free radical, at the first step. Reoxidation of this radical intermediate in the presence of oxygen accounts for the oxygen sensitivity of some biological nitroreductases (43, 49, 59).

In Escherichia coli two classes of nitroreductase can be distinguished based on their sensitivities to oxygen: oxygen-insensitive (type I) nitroreductases and oxygen-sensitive (type II) nitroreductases (2, 22, 46, 59). The nfsA gene (mdaA [8]), located at 19.2 min on the E. coli chromosome, has recently been cloned (79) and encodes the protein previously described as nitroreductase IA (14). NfsA is a flavin mononucleotide (FMN)-containing protein with a monomeric molecular mass of 26.8 kDa that uses NADPH as an electron donor (14, 79). The nitroreductase gene nfsB (nfnB [51]), located at 13 min on the E. coli chromosome (8), appears to encode the minor nitroreductase component previously referred to as IB₁ (14). The NfsB protein is an FMN-containing flavoprotein with a monomeric molecular mass of approximately 24 kDa; NfsB can use either NADH or NADPH as a source of reducing equivalents (14, 51). The genes encoding type I nitroreductase IB₂ (14) and type II nitroreductases have not yet been identified.

Bacterial resistance to antimicrobial nitroheterocyclic agents occurs in a stepwise manner. Early biochemical studies of nitroreductase mutants (14, 45, 48) suggested that increased resistance is associated with the sequential mutagenic inactivation of multiple nitroreductases (14, 48, 64). Although E. coli genes encoding the NfsA and NfsB nitroreductases have now been identified, the direct link between mutation in these genes and the development of bacterial resistance has yet to be established. To address this point, we have returned to several nitrofurazone-resistant mutants that were well characterized in earlier studies (14, 48) and have isolated a large number of additional first- and second-step furazolidone-resistant mutants. We have analyzed the cross-resistance of the mutants, determined cellular nitroreductase activities, and characterized the DNA sequence alterations in the nfsA and nfsB genes.

MATERIALS AND METHODS

Bacterial strains, plasmids, and growth conditions.

E. coli K-12 strains AB1157, NFR402, NFR502, and SIL41 were provided by D. Bryant, McMaster University, Hamilton, Canada. AB1157 (F⁻ thr-1 leu-6 thi-1 supE44 lacY1 galK2 ara-14 xyl-5 mtl-1 proA2 his-4 argE3 str-31 tsx-33 λ⁻ sup-37) (3) is the parent strain of the putative nitroreductase mutants NFR402, and NFR502 (48). Strain SIL41 was derived from a cross between AB1157 and an HfrH strain lacking NfsA and NfsB activities (14). DH5α [supE44 lacU169 (φ80lacZΔM15) hsdR17 recA1 endA1 gyrA96 thi-1 relA1] is a standard laboratory strain used for most plasmid manipulations. INVαF′ [F′ endA1 recA1 hsdR17 (r_K⁻ m_K⁺) supE44 thi-1 gyrA96 relA1 φ80lacZΔM15 Δ(lacZYA-argF)U169 λ⁻] was included in the Original TA Cloning Kit. The vectors used in this study were pUC118 (76), pKK232-8 from Pharmacia Biotech Inc. (11), and pCR 2.1 from the Original TA Cloning Kit. Bacterial strains were routinely grown at 37°C in Luria-Bertani (LB) medium or on LBA (LB medium with 1.5% agar), obtained as a premixed powder from Canadian Life Technologies Inc. (Burlington, Ontario, Canada). When appropriate, the medium was supplemented with antibiotics at the following concentrations: 100 μg of ampicillin per ml, 30 μg of chloramphenicol per ml, and 50 μg of kanamycin per ml. Nitrofurazone (5-nitro-2-furaldehyde semicarbazone) and furazolidone were kind gifts from D. Bryant, McMaster University; stock solutions were prepared in dimethyl sulfoxide.

Isolation of furazolidone-resistant strains.

Furazolidone-resistant mutants were obtained as described previously but with furazolidone instead of nitrofurazone (48). Briefly, furazolidone gradient plates were prepared as previously described (68); 24 h was allowed for diffusion of the drug prior to plating. Multiple cultures of AB1157 were grown to exponential growth phase and spread on the furazolidone gradient plates (0 to 1.25 μg/ml), which were incubated overnight. Several colonies representing independent first-step mutants, were selected, subcultured, and grown on gradient plates containing higher concentrations of the drug (0 to 2.5 or 0 to 4 μg/ml). Isolated colonies on these plates represented second-step mutants.

Determination of nitrofurazone sensitivity.

Overnight cultures of each mutant strain were streaked with an absorbent applicator onto a gradient plate containing 0 to 50 μg of nitrofurazone per ml. This provided a standard measure of resistance and allowed the new mutants to be compared with previously isolated strains (48). Absolute concentrations of nitrofurazone to which strains were tolerant were determined as described by McCalla’s group (48): 10 μl of overnight cultures was inoculated in a grid pattern onto plates containing 5 to 45 μg of nitrofurazone per ml.

Preparation of cell extracts.

Overnight cultures (6 ml) were harvested by centrifugation. The cells were washed and resuspended in 0.25 ml of Tris-HCl buffer (pH 7.0). The suspensions were chilled on ice, sonicated with a Kontes Micro-Ultrasonic cell disrupter, and centrifuged at 14,000 × g for 30 min at 4°C. The protein concentration was determined as described for the Bio-Rad (Richmond, Calif.) protein assay with a bovine serum albumin standard curve.

Nitroreductase assay.

Nitroreductase activity was measured as nitrofurazone reductase. The reduction of nitrofurazone was determined by a decrease in the absorbance of nitrofurazone at 400 nm (molar extinction coefficient, 12,960 M⁻¹ cm⁻¹ [79]) at 22°C. The reaction mixture (1 ml) contained 50 mM Tris-HCl (pH 7.0), 0.1 mM nitrofurazone, and 0.1 mM NADPH.

Manipulation of DNA.

Genomic DNA was isolated as described previously (61); DNA was collected by centrifugation (12,000 × g). Plasmid DNA was isolated by the boiling miniprep method described previously (36). If further purification was necessary, it was accomplished by using a GeneClean kit (Bio 101, Inc.). Standard procedures were used for molecular cloning.

Amplification of the nfsA and nfsB genes.

Amplification of the nfsA (Fig. 1) and nfsB (Fig. 2) fragments from genomic DNA was accomplished by PCR with the Expand High Fidelity PCR System (Boehringer Mannheim Canada). Prior to amplification of nfsA, genomic DNA was denatured with 0.2 N NaOH. Prior to amplification of nfsB, genomic DNA was digested with HindIII (New England Biolabs). Reaction mixtures (100 μl) included the Expand High Fidelity buffer with 15 mM MgCl₂, 0.2 mM each deoxynucleoside triphosphate, 300 nM appropriate primers (Fig. 1 and 2), 1.7 U of Expand High Fidelity enzyme mix, and 20 to 750 ng of template DNA. Amplification was performed through 30 repetitions of a cycle of denaturation for 15 s at 94°C, annealing for 30 s at 55°C, and elongation for 1 min at 72°C, with the addition of a 4-min preincubation at 94°C and a 7-min postincubation at 72°C. Prior to further analysis, the products were purified by using a High Pure PCR Product Purification Kit (Boehringer Mannheim Canada). Amplification product and restriction endonuclease analyses were done by agarose gel electrophoresis (1%) in the presence of ethidium bromide.

FIG. 1.

Nucleotide sequence of the PCR-generated fragment containing the wild-type nfsA gene from E. coli AB1157, with putative −35 and −10 promoter elements, ribosomal binding site, and transcription terminator indicated. NFR402 refers to the mutants selected by McCalla and coworkers (47). Mutations for the 20 mutants characterized in this study are indicated. Altered bases and amino acids are underlined in the wild-type sequence, with substituted bases indicated above the sequence. IS elements are shown as boxes above the sequence, with arrows indicating the insertion point. Deletions and frameshifts are indicated below the sequence. PCR primers used for amplification of the gene are shown as horizontal arrows, and the lowercase letters indicate synthetic restriction endonuclease sites. S.D., Shine-Dalgarno sequence.

FIG. 2.

Nucleotide sequence of the PCR-generated fragment containing the nfsB gene from E. coli AB1157 with putative −35 and −10 promoter elements, ribosomal binding site, and transcription terminator indicated. NFR502 and SIL41 refer to the mutants selected by McCalla and coworkers (14, 47). Mutations for the 53 mutants characterized in this study are indicated. Altered bases and amino acids are underlined in the wild-type sequence, with substituted bases indicated above the sequence. IS elements are shown as boxes above the sequence, with arrows indicating the insertion point. Deletions and frameshifts are indicated below the sequence. The horizontal arrows represent the PCR primers. Lowercase letters indicate the synthetic restriction endonuclease sites added to the PCR primers. S.D., Shine-Dalgarno sequence.

Cloning of nfsA and nfsB.

Cloning of amplified nfsA and nfsB into pCR 2.1 was carried out according to the Original TA Cloning Kit instructions with the following changes. The Expand High Fidelity PCR System was used to amplify nfsA from AB1157, followed by a 10-min incubation at 72°C with 1 U of Taq polymerase (Canadian Life Technologies Inc.) to add a 3′ A overhang. Successful clones, pTAA and pTAB, and sequence fidelity were verified by sequencing with universal primers for pUC vectors as well as with specific internal primers for nfsA and nfsB. The nfsB PCR fragment was inserted into the EcoRI and BamHI sites of pUC118 to create pJAB. The primers used for amplification of the 850-bp nfsB fragment contained these 5′ restriction endonuclease recognition sites (Fig. 2) to facilitate cloning. pJAC was constructed by subcloning the 469-bp EcoRI (filled)-BglII fragment of pJAB into the SmaI and BamHI sites of pKK232-8. The 3′ termini of EcoRI were filled in by adding 2 μl each of dATP and dTTP (1 mM stocks) and 2 U of Klenow polymerase to the pJAB EcoRI digest.

Sequencing of clones and PCR products.

Sequencing of the nfsA and nfsB PCR products from each mutant was achieved by using the Sequenase version 2.0 T7 DNA Sequencing Kit and α-³⁵S-dATP (Amersham Canada Ltd.). The double-stranded template was denatured by boiling with 15 pmol of primer and quickly chilled on ice. The reaction products were analyzed on an 8% acrylamide-bisacrylamide gel.

Insertion sequence (IS) elements in nfsA and nfsB were identified by DNA sequencing of the gene or IS element endpoints followed by a nonredundant similarity search of sequence databases and the E. coli databank (Nara Institute of Science and Technology, Nara, Japan) with BLASTX (1).

Nitrofurazone sensitivity assay.

Overnight cultures (100 μl) of each strain were added to 4 ml of LB top agar (LB medium solidified with 0.7% agar) to create a lawn of bacteria on which 10-μl amounts of nitrofurazone at various concentrations in dimethyl sulfoxide were applied in triplicate to paper disks arranged on the lawn. Overnight incubation at 37°C yielded cleared zones in the bacterial lawn, the diameters of which were measured.

CAT assay.

The specific activity of chloramphenicol acetyltransferase (CAT) was determined by a previously described spectrophotometric assay (61). This method measures free 5-thio-2-nitrobenzoate at λ = 412 nm, which has a direct molar relationship to the acetylation of chloramphenicol in the presence of 5,5′-dithiobis-2-nitrobenzoic acid and acetyl coenzyme A (66).

Protein sequence analysis.

A similarity search of protein sequence databases was conducted by using the BLAST server program (1) with default parameters. Protein alignments were performed by the ClustalV program (34) with default parameters.

RESULTS

Isolation of mutant strains.

Putative nitroreductase mutants derived from AB1157 were identified based on their resistance to furazolidone after stepwise selection on gradient plates. Resistance to nitrofurazone as measured by growth on a gradient plate was used to standardize the level of nitrofuran resistance exhibited by our mutants and allow comparison to mutants selected in previous studies (14, 48). The first and second selection steps yielded mutants that were resistant to nitrofurazone concentrations of approximately 15 and 35 μg/ml, respectively. Resistance occurred incrementally; there were no instances in which a single selection step yielded a mutant that exhibited the resistance to the elevated concentrations of furazolidone that is characteristic of second-step mutants. Subsequent selection steps did yield third-step mutants which exhibited resistance to still-higher concentrations of furazolidone (>4 μg/ml). We have focused on the genetic analysis of first- and second-step mutants.

Genetic analysis.

DNA sequence analysis was performed on the PCR-generated amplified products of nfsA and nfsB obtained from all of the furazolidone-resistant mutants isolated in this study, as well as from the nitrofurazone-resistant mutants NFR402, NFR502, and SIL41 obtained previously by other workers (48). Following sequence analysis, some mutants derived through a single experiment had identical mutations. To ensure that all mutations were of independent origin, only one identical mutant from each experiment was considered in our analysis. NFR402 is a first-step mutant with IS186 integrated within the coding region of nfsA at position G³⁸⁸ (Fig. 1); NFR402 contains a wild-type nfsB gene. NFR502, a second-step mutant derived from NFR402, contains the identically integrated IS186 in nfsA, as well as a missense mutation in nfsB at position C¹³⁴ resulting in a Pro-to-Leu substitution at position 45 (Fig. 2). SIL41 was derived from a genetic cross between an Hfr nfsA nfsB double mutant and AB1157 and exhibits wild-type levels of resistance to nitrofurazone. However, SIL41 can be mutated to high (two-step) levels of nitrofuran resistance through a single mutational step. SIL41 has a wild-type nfsA gene and contains an amber mutation, Gln-142 (CAG) to amber (TAG), at position C⁴²⁴ in nfsB (Fig. 2).

All 20 of the first-step nitrofuran-resistant mutants isolated in this study contain a mutation in the nfsA gene (Table 1). No mutations in the nfsB gene were observed in any of the first-step mutants, while all but 4 of the 53 second-step mutants had a mutation in the nfsB gene (Table 1).

TABLE 1.

Mutation distribution in nfsA and nfsB of nitrofuran-resistant mutants

Mutation	No. in:
	nfsA of first-step mutantsa	nfsB of second-step mutantsb
Base substitutions
Total	5	21
G · C→T · A	1	7
G · C→C · G	0	1
A · T→T · A	2	4
A · T→C · G	1	3
G · C→A · T	1	5d
A · T→G · C	0	1
Frameshifts and insertions
Total	1	3
−G	1	0
−A	0	1
+C	0	1
+(TCT)	0	1
Deletions	1 (177 bp)	2 (∼80 bp)
Insertion elements
Total	14	26
IS1	9c	13
IS2	0	3
IS5	0	10
IS30	2	0
IS186	3	0
None	0	3
Total no. of mutants	20 (21c)	53 (55d)

^aThe same mutations, and no others, are in nfsA of second-step mutants derived from first-step mutants. 

^bThe nfsB gene of first-step mutants is wild type. 

^cThe mutant NFR402 isolated by McCalla et al. (48) is included. 

^dThe mutants NFR502 and SIL41 isolated by McCalla et al. (48) are included. 

Among the first-step mutants selected in the current study, all have a single mutation affecting nfsA. All five base substitution mutations are located within the first third of the gene’s coding region (Fig. 1). One mutant, JVR1, has a −1 frameshift resulting from a −G event at position G^626–628, thereby altering the remaining 31 amino acids and adding 15 more residues on the carboxyl terminus of the NfsA protein. One nfsA mutant, JVQ1, has a 177-base deletion, including the final 88 bases of the nfsA gene’s coding region. The other 13 first-step mutants, comprising 65% of all the nfsA mutations, have IS elements integrated within and upstream of the coding region of nfsA. IS1, IS30, and IS186 were identified in nfsA. Eight of the mutants contain IS1: four with integrations in the putative promoter region upstream of the coding sequence and two near each terminus of the gene. IS30 integrated near the 5′ end of nfsA in two mutants, both at position C⁶⁷⁵. IS186 is the only IS element that integrated near the middle of nfsA, and it is found in all three mutants at the identical position.

The second-step mutants isolated in this study were confirmed to have only the parental mutation in nfsA. Only single mutations were found in the nfsB genes of second-step mutants. Eighteen of the 53 second-step mutants have a base substitution in nfsB; 1 results in an ochre mutation, and 17 are missense mutations. Of the 17 missense mutations, 9 effected changes in a 17-amino-acid sequence, from Leu-33 to Ile-49, of the protein NfsB (Fig. 2). One mutant, JVZ2, has a three-base insert (TCT) at position T⁴⁷⁷ that results in the addition of a leucine residue between Leu-159 and Asp-160 (Fig. 2). The two mutants JVAE6 and JVI3 contain frameshift mutations in nfsB resulting from the insertion of a cytosine at position C³⁶⁹ and the deletion of A¹²¹, events causing premature termination of translation after 2 and 53 amino acids, respectively. In two mutants, JVQ3 and JVR4, nearly identical deletions of approximately 80 bp are situated in the middle of the gene (Fig. 2). The most frequent type of mutation identified in the nfsB gene of the furazolidone-resistant mutants is IS element integration. Three IS elements, IS1, IS2, and IS5, account for 49% of the second-step mutants isolated in this study. With the one exception of IS1 inserting near the Shine-Dalgarno sequence of JVP3, all IS elements integrated within the coding region of nfsB. As in nfsA, IS elements demonstrated regional specificity for insertion in the regions proximate to either terminus of nfsB. This is particularly apparent in the first 165 bp of the gene’s coding region, where IS element integration occurred in 19 mutants. IS5 integrated at the identical sequence, after G⁴³ (Fig. 2), in all 10 mutants with this element; IS5 inserted in the identical orientation in each mutant except JVD2, in which it was reversed.

In four second-step mutants, no mutations were observed within the coding region of nfsB, although multiple sequencing reactions were carried out on both strands of the PCR products. Of these, one mutant, JVAE4, contained a transversion in the putative transcription terminator of nfsB. The resistance of the mutants JVA6, JVN3, and JVO2 to a concentration of nitrofurazone (35 μg/ml) consistent with a second-step mutant was confirmed. To ensure that there is not a mutation in or near the putative −35 region of nfsB, an upstream primer, EBUP1 (Fig. 2), was used to amplify and sequence this region. No mutations were found. We confirmed the presence of a functional promoter in the PCR-amplified product of nfsB by using the construct pJAC. pJAC contains the promoterless CAT gene of plasmid pKK232-8 directly downstream from the truncated fragment of the wild-type nfsB, containing the putative promoter region, so that any expression of CAT, and hence resistance to chloramphenicol, was due to the promoter activity of nfsB. Growth of transformants derived from pJAC on chloramphenicol medium confirmed the presence of a functional nfsB promoter in the PCR-amplified product. This contruct enabled the measurement of promoter activity by determination of the specific activity of CAT (126 nmol min⁻¹ mg⁻¹). The specific activity of CAT in the negative control containing only pKK232-8 was zero.

Nitroreductase activity.

Nitroreductase activities in cell extracts of AB1157 (wild type), first-step mutants, and second-step mutants were spectrophotometrically measured as the reduction of nitrofurazone by using NADPH as a source of reducing equivalents. The average NADPH-dependent nitroreductase activities of AB1157, first-step mutants, and second-step mutants were 7,900, 290, and 5.5 nmol min⁻¹ mg⁻¹, respectively (Table 2). Four second-step mutants have significantly greater nitroreductase activity than the average: JVA6, JVN3, and JVO2, the three second-step mutants that do not have a mutation in NfsB; and JVA5, which has a missense mutation in NfsB. Of particular interest is JVA6, which has nitroreductase activity (261 ± 50 nmol min⁻¹ mg⁻¹) comparable to that of a first-step mutant. JVN3 has activity (41 ± 4.3 nmol min⁻¹ mg⁻¹) that is about twice that of JVA5 and JVO2. The second-step mutant JVAE4, which has a mutation in the transcription terminator but not in the coding region of nfsB, has nitroreductase activity consistent with second-step mutants.

TABLE 2.

Nitroreductase activity in cell extracts of AB1157 and first- and second-step nitrofuran-resistant mutants

Strain	Mutation	Nitroreductase activitya (nmol min−1 mg of protein−1)
AB1157 (wild type)		7,900 ± 760
First-step mutants
All		290 ± 80b
NfsA missense
JVC1	Gln67→Pro	441 ± 6
JVK1	Arg15→Cys	233 ± 4
JVO1	Ile89→Asn	284 ± 6
JVY1	His11→Leu	253 ± 44
Second-step mutants
All		5.5 ± 4.0c
NfsB missense
JVA5	Ser40→Cys	22 ± 1.9
JVB4	Asp71→Lys	4.0 ± 0.3
JVE2	Ser39→Cys	7.9 ± 0.1
JVE4	Gly148→Ser	3.1 ± 0.5
JVG2	Asp105→Tyr	10 ± 9.4
JVH2	Leu33→Pro	4.1 ± 0.7
JVI2	Val83→Asp	6.2 ± 7.0
JVL6	Val187→Glu	8.9 ± 4.6
JVP2	Ser37→Arg	9.4 ± 0.6
JVQ4	Pro45→Gln	7.8 ± 6.5
JVR3	Gln44→Pro	11 ± 2.1
JVY2	Arg207→His	6.8 ± 5.0
JVAE5	Pro45→Gln	2.1 ± 1.8
JVAG3	Tyr183→Asp	5.7 ± 0.8
JVAG6	Leu34→Arg	0.7 ± 0.8
No mutation in NfsB
JVA6		261 ± 50
JVN3		41 ± 4.3
JVO2		19 ± 1.1
JVAE4	Transcription terminator	9.4 ± 4.2

^aAverages and standard deviations from assays done at least twice. 

^bAverage for all 20 first-step mutants. 

^cAverage for all 50 second-step mutants for which a mutation has been characterized. The cultures of five second-step mutants were not viable after storage, and enzyme activity was not determined. Genetic characterization of these mutants had revealed the following mutations in nfsB: two missense (Ala¹⁵⁶→Pro and Ile⁴⁹→Asn), one stop (Glu¹⁷⁵→ochre) and two IS elements (IS1⁶² and IS5⁴³). 

Restoration of nitrofuran sensitivity by cloned NfsA and NfsB proteins.

The influence of plasmid-encoded NfsA and NfsB in vivo was evaluated for the wild-type E. coli strain AB1157, a first-step mutant containing a 177-bp deletion in nfsA (JVQ1), and a second-step mutant derived from JVQ1 with IS5 integrated after position 43 in nfsB (JVQ2) by comparing the levels of nitrofurazone sensitivity of the strains transformed with pTAA (nfsA) or pTAB (nfsB) to that of the strains transformed with vector alone (pTA). Sensitivities were compared by measuring the diameters of cleared zones around a nitrofurazone-infused paper disk on a lawn of bacteria (Table 3). The greatest differences in sensitivity were noted for nitrofurazone concentrations between 5 and 10 mg/ml. Further increases in sensitivity were not observed at concentrations above 10 mg/ml, and concentrations of 1 mg/ml or lower did not yield cleared zones at all.

TABLE 3.

Diameters of cleared zones in a lawn of each strain with and without vector-encoded NfsA (pTAA) and NfsB (pTAB) after the application of 5 mg of nitrofurazone per ml

Strain	Diam (mm)a for strain with:
	Vector alone (pTA)	pTAA	pTAB
AB1157	18.4 (0.4)	20.6 (0.5)	20.9 (0.5)
JVQ1	15.5 (0.5)	18.8 (1.3)	18.9 (0.7)
JVQ2	12.5 (0.5)	20 (1)	18.8 (0.8)

^aResults are means (standard deviations) calculated from five determinations. 

The cleared-zone diameters calculated from five determinations with each strain are indicated in Table 3. From this experiment, four observations were made. (i) AB1157, JVQ1, and JVQ2 had significantly different sensitivities to nitrofurazone; AB1157 demonstrated the greatest sensitivity, and JVQ2 had the least. The sensitivity correlates with nitroreductase levels in cell extracts (Table 2). (ii) Upon the introduction of either plasmid-encoded nitroreductase, an increase in sensitivity over that exhibited by the strain transformed with vector alone was observed. The increases in AB1157 were presumably due to the multiple copies of the plasmid-encoded gene. (iii) Between strains, no significant difference between the sensitivities conferred by the nfsA- and the nfsB-containing clones was apparent. While the percent increase in sensitivity was far greater in the mutants than in the wild type, the absolute sensitivities of the transformants, as indicated by the diameter of the cleared zone in the lawn, were similar to that of AB1157 transformants. This maximum diameter could represent a limit either for the chemical’s diffusion or for cellular sensitivity to nitrofurazone. (iv) Transformation with either nfsA or nfsB increases sensitivity of all strains to the same level.

DISCUSSION

The genetic and biochemical characterizations in this study corroborate previous studies that suggested that NfsA and NfsB nitroreductase activities are necessary for the majority of nitrofuran biological activity under aerobic conditions (1, 14). Furthermore, by restoring nitrofuran sensitivity through the introduction of cloned nfsA and nfsB in AB1157 (wild type) and its nitroreductase-deficient derivatives (Table 3), we have shown that NfsA and NfsB are sufficient for maximal nitroreductase activity. Selection of E. coli K-12 nitrofuran-resistant strains on medium supplemented with furazolidone (this study), nitrofurazone (45, 48), or nitrofurantoin (10, 64) consistently yields two discrete steps of increased resistance. In this study we show through DNA sequence analysis of a large number of nitrofuran-resistant mutants that the observed pattern of two-step resistance is a consequence of an obligatory first-step event, mutation in nfsA, followed by mutational inactivation in nfsB. Each step of increased resistance correlates with a decrease in nitroreductase activity in cell extracts (with the exception of JVA6, which exhibits second-step resistance but a first-step level of nitroreductase activity). The observation that SIL41 (nfsA⁺ nfsB [amber]) exhibits wild-type nitrofurazone resistance, despite having a reduction in oxygen-sensitive nitroreductase activity of approximately 25% (14), shows that inactivation of nfsB does not produce a resistance phenotype in the presence of a wild-type nfsA gene. However, in contrast to wild-type bacteria, SIL41 can attain a high (second-step) level of nitrofurazone resistance in a single step that presumably corresponds to inactivation of nfsA. Since the majority of aerobic nitroreductase activity is attributable to NfsA (Table 2), the simplest explanation for these results is that the loss of only the minor NfsB component from nfsA⁺ strains is insufficient to reduce the total amount of nitroreductase activity to the extent that altered resistance will be conferred on the bacterium.

Of the mutants selected in this study, three second-step mutants did not have a mutation either in the coding region of nfsB or in the 100 bases upstream of the coding region, and one had a base substitution downstream of the coding region in the putative transcription terminator (Fig. 2). Confirmation that this nfsB fragment contained a functional promoter was achieved by demonstrating CAT activity in the plasmid pKK232-8, which contains a promoterless CAT gene, upon insertion of the putative promoter region of nfsB. We have also shown that NfsB is normally sufficient for biological activation of nitrofurans: sensitivity to nitrofurazone is increased upon the introduction of vector-encoded NfsB in both wild-type and mutant E. coli cells. In each of the three mutants resistance to elevated concentrations of nitrofurazone and furazolidone has been confirmed, eliminating the possibility of mutant reversion. Interestingly, each mutant has extract nitroreductase levels that are atypical of second-step mutants, with JVA6 exhibiting nitroreductase levels similar to those of first-step mutants and JVN3 and JFO2 exhibiting intermediate nitroreductase levels. This suggests that additional factors may modulate nitrofuran resistance. Possible explanations include (i) a nitroreductase enzyme distinct from NfsA or NfsB (for example, nitroreductase component IB₂, described by Bryant and coworkers [14]), (ii) a trans-acting factor with which interaction is essential for full nitroreductase activity; or (iii) a factor mediating transport of the nitrofuran into the cell. The last possibility seems particularly applicable to JVA6.

Beyond the initial two levels of resistance, nitrofurantoin- and furazolidone-isolated mutants appear to have nitrofuran tolerance properties not shared by nitrofurazone-resistant mutants. Third-step furazolidone-resistant mutants displayed inconsistent and nonuniform resistance when plated on nitrofurazone medium. Similar characteristics were observed in putative third- and fourth-step nitrofurantoin-resistant mutants (10), which did not display cross-resistance to nitrofurazone but instead were characterized by a second-step level of nitrofurazone resistance. It is possible that the principal mode of toxic action of the reactive metabolites derived from nitroreduction differs between nitrofuran derivatives. For example, nitrofurans have a variety of toxic actions, including genotoxicity, metabolic effects, effects on nucleotide pools, and formation of active oxygen species (47, 73). Alternatively, the participation of other, as-yet-uncharacterized, nitroreductase activities may contribute to the reduction of a subset of nitrofuran derivatives (e.g., furazolidone and nitrofuranoin, but not nitrofurazone) in the absence of NfsA and NfsB activity. A third possibility is that the high levels of unreduced nitrofurazone in higher-order nitroreductase mutants may have an inhibitory effect on bacterial metabolism when mutants are plated on nitrofurazone medium (41).

The mutational distribution in the nfsA and nfsB genes of the nitrofuran-resistant mutants (Fig. 1 and 2) implicates particular residues as having functional significance in the NfsA and NfsB proteins. NfsA nitroreductase has 51 and 24% amino acid identity with NADH oxidoreductase (Frp) from Vibrio harveyi and NADH oxidase (Nox) from Thermus thermophilus, respectively (Fig. 3). The three-dimensional structures of both Frp (69) and Nox (33) have been determined experimentally, and the similarities between the two structures suggest that both Frp and Nox belong to the same group of flavoproteins. It is quite likely that NfsA constitutes another member of this group, since both Frp and Nox exist in the physiological state as homodimers, with gel exclusion chromatography data suggesting the same for NfsA (79). All three oxidoreductases use FMN as a cofactor (although Nox can also use flavin adenine dinucleotide [FAD]) and display kinetic profiles characteristic of a ping-pong bi-bi reaction mechanism (33, 69, 79). Particularly interesting is the conservation of the amino acid region corresponding to residues 11 to 15 of NfsA (Fig. 3) in the three flavoproteins, which has been shown to be involved in cofactor binding in both Frp and Nox. Two of four amino acid substitutions recovered from nfsA mutants characterized in this study are confined to this region (His-11 to Lys and Arg-15 to Cys); the change of a conserved residue, Gln-67, to His in another nfsA mutant could also be localized to the putative cofactor binding site, since glutamine residues in Frp and in Nox, corresponding to Gln-67 of NfsA, form two hydrogen bonds with the isoalloxazine ring of FMN (33, 69). These data suggest conservation of the basic cofactor binding site among the three flavoproteins.

FIG. 3.

ClustalV alignment of NfsA with T. thermophilis (T.t.) Nox and V. harveyi (V.h.) FrpI proteins. Exclamation points indicate completely conserved residues. Asterisks indicate residues, either identical or chemically similar, that are conserved between NfsA and one of the other proteins. The arrowheads above the sequences indicate residues altered in NfsA as a result of a mutation in nfsA of a furazolidone-resistant mutant. The five residues in the box comprise the putative FMN binding site (see text for details).

NfsB has 85% amino acid sequence identity with homologous nitroreductases isolated from Salmonella typhimurium (78) and Enterobacter cloacae (13) and 32% identity with FRase I, the major flavin oxidoreductase of Vibrio fischeri (80, 81). The sequences of RdxA from Helicobacter pylori (29, 72), Nox from T. thermophilus (33), and a putative protein from Bacillus subtilis (5) also have some amino acid sequence identity with NfsB, particularly within a 14-residue stretch corresponding to Ser-37 to Val-50, near the amino terminus (Fig. 4). Although the overall similarity of these sequences is low, the identity between the proteins in this 14-residue region is high: it is 100% for the E. coli, S. typhimurium, and E. cloacae proteins, and the V. fischeri, H. pylori, B. subtilis, and T. thermophilus proteins have 78, 71, 57, and 50% identity, respectively, with the NfsB sequence. It is interesting that of the 17 nfsB mutants with missense mutations, 7 of the mutations (41%) are located in this region. The other 10 missense mutations also affected conserved residues: Arg-207 is conserved in all seven proteins, while the other amino acids altered by mutation in NfsB were conserved in at least four of the proteins. No consensus sequences for NAD(P)H or FMN binding have been found in NfsB, either by searching the protein databases (PROSITE, PIR, and Swiss-Prot) with MotifFinder (28) or by comparisons with known FMN binding sites in other proteins (17, 35, 40, 42, 54–56, 58, 77). However, the amino acid sequence YSxSS (similar to the conserved sequence identified in this study at residues 36 to 40 of NfsB) was identified as part of the possible FAD-isoalloxazine ring binding region in two oxygenase enzymes of Acinetobacter calcoaceticus (55). The sulfite reductases of E. coli (17) and S. typhimurium (55) and the cytochrome P-450 oxidoreductases of pig and rat also contain a similar sequence, YSIAS, that aligns with the oxygenase enzymes and corresponds to FAD binding domains (55). Phe-124 appears to be spatially situated near the isoalloxazine ring of a bound flavin in NfsB, as manipulation of this residue can alter the affinity of NfsB for binding different flavins, and deletion of the residue substantially increases flavin reductase activity (80).

FIG. 4.

ClustalV alignment of NfsB with the NfsB-like proteins from S. typhimurium (S.t.) and E. cloacae (E.c.), the major flavin reductase (FRase I) of V. fischeri (V.f.), the RdxA protein of H. pylori (H.p.), Nox of T. thermophilus (T.t.), and a hypothetical protein of B. subtilis (B.s.). Exclamation points indicate residues conserved in all seven proteins. Asterisks indicate residues, either identical or chemically similar, that are conserved in NfsB and at least three of the other proteins. The 14 residues indicated by the box constitute the most conserved region in the proteins and may include the FMN binding site (see text for details).

Among the 20 mutations isolated in the nfsA genes and the 50 mutations characterized for the nfsB genes of the furazolidone-resistant mutants isolated in this study, 13 (65%) and 26 (52%), respectively, were derived from the insertion of one of six IS elements (Table 1). Transposition is a rare event, occurring at frequencies of 10⁻⁷ to 10⁻⁵ per generation, with only 5 to 15% of all spontaneous mutations being attributed to IS elements (26). Although there is a high degree of variability in the copy number and presence of IS elements in E. coli strains (9), all IS element types identified in this study have been mapped to fixed locations on the chromosome of E. coli K-12 strains, as well as having additional strain-specific locations (6, 7, 9, 75). Clustering of particular IS families to certain chromosomal locations is believed to result from local hopping, a process in which local transposition events give rise to daughter IS elements inserted in close proximity (<10 kb) to the parent element (9, 18). The nfsA gene is located at 19.2 min on the E. coli K-12 chromosome (8): no IS elements have previously been mapped within 10 kb (approximately 0.2 min) of nfsA. The nfsB gene is located at 13 min on the E. coli chromosome (8). IS5, at 13.1 min, is the only element identified in this study to be located near nfsB in the three mapped E. coli K-12 strains (6). Proximity of IS5 to nfsB on the chromosome, in addition to the probable IS5 hot spot in nfsB (discussed below), and the high transposition rate observed for IS5 (53) could account for the large number of nfsB mutants (19%) containing IS5 in this study.

There are two putative IS element hot spots in nfsA and one in nfsB. In nfsA a hot spot for IS186 is indicated after G³⁸⁸, as all three IS186 mutants independently isolated in this study, as well as the previously isolated NFR402, had IS186 inserted at this site (Fig. 1). The hot spot for IS186 in nfsA is similar to the G+C-rich insertion sites observed in other studies (16, 39, 65), as it is directly downstream from two G₃ runs and one G₄ run. The target duplications, GC or GT, generated by IS186 in the nfsA genes of mutants in this study and in NFR402 (48) were of identical length. In contrast, other studies noted that even at an identical integration site, the target duplications were of variable lengths (39, 65). Specific target sequences are often associated with the transposition of IS30 (15, 26, 57). A putative hot spot for IS30 insertion after C⁶⁷⁵ in nfsA is suggested by our data. The target site has similarities with the IS30 hot spots observed in the prophage P1 and the R plasmid NR1-Basel (15): the sequence GCC immediately precedes the target duplication, and the sequence ANNNCCTTTNTTA, suggested to be involved in target site selection of IS30 in P1 (15), is present, reversed and with one mismatch (A⁶⁸⁹ ACACGTTTATTA⁷⁰¹), immediately upstream of the IS30 insertion site in nfsA. A hot spot for IS5 after G⁴³ is evident in nfsB: 10 independent mutants that contained IS5 at the same site were isolated in this study. The target sequence for IS5 in nfsB, AAGG⁴³, does not conform to the previously described IS5 target site consensus sequence C(A/T)A(G/A) (25). However, of the 12 sequences within the coding region of nfsB that did match this consensus sequence, two, CTAA⁴¹ and TTTG⁴⁹, flanked the insertion point of IS5.

IS element target site sequence specificity may sometimes be attributable to protein consensus sequences. In particular, E. coli integration host factor (IHF), a protein that induces DNA bending, has been implicated in some instances of IS1 site recognition (27, 60). While the nfsA gene does not have any obvious IHF binding sites, the nfsB gene contains one site at bases 640 to 652 that conforms to an IHF binding site consensus sequence, WWWCAWNANNTTA (W is A or T) (28, 31). Notably, all IS elements, five IS1 and one IS2, that integrated near the 3′ end of nfsB were inserted within 100 bp of this consensus sequence, and two integration sites were exactly 9 bases on either side. IS1 is often associated with regional preferences for approximately 100-bp segments of DNA in which they insert throughout (27, 50, 82).

With the exception of IS186, all IS elements demonstrated specificity for insertion in the regions proximate to either terminus of nfsA and nfsB, suggesting that despite the variable target sequence specificity of IS elements, there may be some common determinants in target site selection. Several studies have shown that IS1 has a preference for insertion into stretches of A+T-rich sequences (44, 50, 82). It has been suggested that IS1, like RNA polymerase, may require regions high in A+T for their low helix stability (19, 44, 50), so A+T-rich promoter regions could be expected to be targets for IS1. Among the mutants described in the current study, four mutants had IS1 inserted within the promoter region of nfsA, while eight nfsB mutants had IS1 inserted within the first 165 bases of the coding region, which has a high density of homopolymeric runs of adenine and thymine which, according to the wedge model of DNA conformation, could, like IHF (26, 27, 60), promote bending of DNA (30, 32, 38, 74).