Bacillus amyloliquefaciens orthologue of Bacillus subtilis ... - CiteSeerX

Microbiology (2002), 148, 297–306

Printed in Great Britain

Bacillus amyloliquefaciens orthologue of Bacillus subtilis ywrO encodes a nitroreductase enzyme which activates the prodrug CB 1954 Gill M. Anlezark, Thomas Vaughan,† Elizabeth Fashola-Stone, N. Paul Michael,‡ Heather Murdoch, Meg A. Sims, Simon Stubbs,‡ Stuart Wigley§ and Nigel P. Minton Author for correspondence : Gill M. Anlezark. Tel : j44 1980 612 330. Fax : j44 1980 611 310. e-mail : gill.anlezark!camr.org.uk

Centre for Applied Microbiology and Research (CAMR), Porton Down, Salisbury, Wiltshire SP4 0JG, UK

A nitroreductase with distinct properties that can activate the prodrug 5-aziridinyl-2,4-dinitrobenzamide (CB 1954) was isolated from Bacillus amyloliquefaciens. The encoding gene was identified as a homologue of the ywrO of Bacillus subtilis, and was obtained as a PCR product by reverse genetics, cloned and the entire nucleotide sequence determined. The gene was found to reside between homologues of the B. subtilis alsD and yswB genes ; however, the ywrO and yswB genes of B. amyloliquefaciens were not separated by a fourth gene, ywsA. The B. amyloliquefaciens ywrO gene was overexpressed, the recombinant protein purified and its properties were compared with those of two CB 1954-activating enzymes, Escherichia coli B nitroreductase (NTR) and Walker DT-diaphorase (DTD). In common with these enzymes menadione was an electron acceptor (Km 3 µM) and activity with this substrate was inhibited by the presence of dicoumarol (Ki 10 µM). In contrast, YwrO showed a marked preference for NADPH as a cofactor (Km 40 µM) and therefore could not be classified as a DTD (EC 1.6.99.2). The flavin FMN was an acceptor with high affinity. B. amyloliquefaciens YwrO was shown to be a flavoprotein with a monomeric molecular mass of 215 kDa by calculation and SDS-PAGE. The cytotoxic 4-hydroxylamine derivative was the single CB 1954 reduction product, but B. amyloliquefaciens YwrO was inactive with the bischloroethyl analogue of CB 1954, SN 23862. In both of these properties B. amyloliquefaciens YwrO more closely resembles DTD than NTR. Its Km for CB 1954 was lower than that of NTR (617 µM compared to 862 µM). Enhanced in vitro cytotoxicity of CB 1954 was demonstrated on incubation of V79 cells with prodrug, NADPH and B. amyloliquefaciens YwrO. The work has led to the identification of a previously unknown nitroreductase, B. amyloliquefaciens YwrO, with distinct properties which will aid the rational selection of appropriate genes for applications in directed enzyme prodrug therapy (DEPT).

Keywords : prodrug activation, DEPT, cancer therapy, flavoprotein, in vitro cytotoxicity

.................................................................................................................................................................................................................................................................................................................

† Present address : Department of Biology, University of York, PO Box No. 373, York YO10 5YW, UK. ‡ Present address : Nycomed-Amersham, Cell Based Assay Department, Cardiff Laboratories, Forest Farm, Whitchurch, Cardiff CF4 7YT, UK. § Present address : Perotsystems, 1 Oakwood Court, Sherwood Park, Nottingham NG15 0DR, UK. Abbreviations : ADEPT, antibody-directed enzyme prodrug therapy ; CB 1954, 5-aziridinyl-2,4-dinitrobenzamide ; DEPT, directed enzyme prodrug therapy ; DTD, DT-diaphorase ; GDEPT, gene-directed enzyme prodrug therapy ; NTR, nitroreductase ; SN 23862, bischloroethyl-2,4-dinitrobenzamide ; VDEPT, viraldirected enzyme prodrug therapy ; 2HX, 5-(aziridin-1-yl)-2-hydroxylamino-4-nitrobenzamide ; 4HX, 5-(aziridin-1-yl)-4-hydroxylamino-2-nitrobenzamide. The GenBank accession number for the sequence reported in this paper is AF373598. 0002-4996 # 2002 SGM

297

G. M. A N L E Z A R K a n d O T H E R S

INTRODUCTION

The monofunctional prodrug 5-aziridinyl-2,4-dinitrobenzamide (CB 1954) can be activated to form a cytotoxic hydroxylamine derivative by nitroreductase enzymes (Fig. 1) (Anlezark et al., 1995 ; Knox et al., 1988b). The difunctional alkylating agent formed upon reduction with these enzymes can cross-link DNA via a non-enzymic combination with cellular thiols (Knox et al., 1991). Therefore, CB 1954 has potential applications in cancer therapy. CB 1954 is toxic to Walker rat carcinoma cells due to the presence of a form of NAD(P)H oxidoreductase, DT-diaphorase (DTD ; EC 1;6;99;2) (Knox et al., 1988b), in these cells. The enzyme DTD was originally defined as having the ability to use the cofactors di- and triphosphopyridine nucleotide (NADH and NADPH) equally well. Although other mammalian cells, including human cells, do not possess an equivalent DTD activity (Boland et al., 1991) the conversion of CB 1954 to hydroxylamine can be exploited in cancer therapy using directed enzyme prodrug therapy (DEPT) approaches. DEPT involves directing an exogenous enzyme with the required properties to the target tumour as a fusion protein, as part of an antibody–enzyme conjugate (antibody-directed enzyme prodrug therapy, ADEPT) (Bagshawe, 1987 ; Bagshawe et al., 1988), or by delivering the encoding gene to the target cells. The presence of the functional enzyme will eliminate the targeted cells and, through a bystander effect, will also eliminate cells in the immediate vicinity of the targeted cells on administration of the prodrug (gene-\viraldirected enzyme prodrug therapy, GDEPT\VDEPT, or clostridial-directed enzyme prodrug therapy, CDEPT) (Dachs et al., 1997 ; Minton et al., 1995 ; Lemmon et al., 1997). Bacterial enzymes offer a number of advantages in DEPT strategies in that there is often no mammalian equivalent and their substrate specificity may differ substantially from that of mammalian enzymes so that non-specific prodrug activation, at sites other than the tumour, can be avoided. A ‘ classical ’ nitroreductase (NTR) isolated from Escherichia coli B activates CB 1954, and its bischloroethyl analogues and orthologues, to form cytotoxic derivatives (Anlezark et al., 1992, 1995). The use of NTR in ADEPT is, however, limited by the requirement for a cofactor to be co-administered with the prodrug and by the achievable level of enzyme activity in the tumour when directed there by chemical conjugation with antibody fragments (Anlezark et al., 1996 ; G. M. Anlezark, unpublished data). However, NTR has been used in gene therapy approaches with some success in vitro (Bailey et al., 1996 ; Bridgewater et al., 1997 ; Searle et al., 1998 ; Latham et al., 2000) and in animal models (Clark et al., 1997 ; Drabek et al., 1997 ; McNeish et al., 1998). Nevertheless, in common with Walker DTD, the kinetics of the enzyme–substrate interaction are suboptimal (Knox et al., 1992). Furthermore, in contrast to the single cytotoxic product formed by the Walker enzyme, the reaction catalysed by 298

.................................................................................................................................................

Fig. 1. Structure of CB 1954 and the E. coli B reduction products formed. The cytotoxic 4HX product is converted intracellularly into a DNA cross-linking agent by a non-enzymic mechanism (Knox et al., 1991).

NTR results in the formation of two products, the 4hydroxylamine (4HX) and the relatively non-toxic 2hydroxylamine (2HX) derivatives (Fig. 1 ; Knox et al., 1992 ; Anlezark et al., 1992). This limits the use of NTR in gene therapy, where the efficiency of transfection and the expression of proteins from transfected genes can be low, particularly in vivo. Ideally, a DEPT enzyme should fulfill a number of criteria for its successful use including a high affinity and specificity for the substrate, and a high turnover so that sufficient cytotoxin is generated within the tumour for clinical efficacy even if transfection and expression efficiency is low. We have previously screened a number of bacterial classes for prodrug-activating activity and these studies have indicated that an enzyme with the desired properties is present in some Bacillus spp. (G. M. Anlezark, unpublished data). The present study describes the isolation of a novel nitroreductase from Bacillus amyloliquefaciens, and its purification, characterization and evaluation as a candidate for DEPT strategies. METHODS Bacterial strains, plasmids and culture media. The source of

chromosomal DNA for the cloning of ywrO was B. amyloliquefaciens (CAMR strain 0454). The E. coli host used for cloning experiments was Top10 [F− mcrA ∆(mrr–hsdRMS– mcrBC) φ80lacZ∆M15 ∆lacX74 recA1 deoR araD139 ∆(ara– leu)7697 galU galK rpsL (StrR) endA1 nupG], and the host used for expression studies was E. coli novablue [endA1 hsdR17 (r−K m+K ) supE44 thi-1 recA1 gyrA96 relA1 lac + lacIq"# lacZ∆M15 : : Tn10)]. Plasmids employed (Fh proA+B"# for cloning and expression studies, respectively, were pCR2.1-TOPO (Invitrogen) and pET21b (Novagen). All organisms were cultivated in L-broth or on L-agar at 37 mC and supplemented where necessary with antibiotics (100 µg ampicillin ml−" or 30 µg kanamycin ml−"). For enzyme purification and DNA preparation, 20 ml L-broth was inoculated with a single colony of B. amyloliquefaciens and grown at 37 mC with gentle shaking for 6–8 h. A 1 ml aliquot of the starter culture was thereafter used to inoculate 200 ml L-broth in a 1 l unbaffled flask which was then incubated overnight at 37 mC with shaking at 150 r.p.m. Cells

YwrO : cloning, expression and CB 1954 activation were harvested by centrifugation and washed with 20 mM potassium phosphate buffer, pH 7n0, before use, or stored at k20 mC until required. All chemicals and reagents used throughout this study were purchased from Sigma unless otherwise indicated. General recombinant methods. The pCR2.1-TOPO cloning vector was used in accordance with the manufacturer’s (Invitrogen) instructions. All recombinant methods including isolation of plasmid DNA, cultivation and transformation of E. coli, and all molecular manipulations of DNA were performed as described by Sambrook et al. (1989). All restriction endonuclease and DNA-modifying enzymes were used in accordance with the supplier’s instructions (NBL). For nucleotide sequencing 1–2 µg template DNA was mixed with 4n5 pmol oligonucleotide primer and the sequencing reaction was completed using AmpliTaq FS dye terminator chemistry in a robotic workstation. DNA sequence analysis was performed using an ABI Prism DNA sequencer (Perkin Elmer). Native Pfu DNA polymerase was obtained from Stratagene. Oligonucleotides were synthesized on an ABI synthesizer (Perkin Elmer). Primary data were manipulated using software in the Lasergene program (DNA STAR). Inverse PCR was performed as described by Whelan et al. (1992). Cloning of the ywrO gene. A 230 bp portion of ywrO was amplified from the B. amyloliquefaciens chromosome using the two degenerate primers P1 (5h-GTNCAYCCNGATATGGARAA-3h) and P2 (5h-CCRTANGCCCANCCRTG-3h) ; P1 was capable of encoding an heptapeptide motif from the determined N-terminal sequence of the purified enzyme (VHPDMEN) and P2 could encode an internal sequence motif (HGWAYG) found to be entirely conserved between the hypothetical bacterial proteins YrkL (Bacillus subtilis) and YabF (E. coli). The 230 bp fragment was cloned into pCR2.1TOPO to yield pBAM1 (Fig. 2). The remainder of the gene was isolated by using inverse PCR. B. amyloliquefaciens chromosomal DNA was cleaved with StyI, and the fragments generated were circularized through their subsequent incubation with DNA ligase. The ligated DNA was then used as the template for a PCR employing two divergent primers based on the sequenced 230 bp fragment : primers P3 (5hGCTTATTGACCGCTGAG-3h) and P4 (5h-GTACAGTGCGCCTCCGC-3h). A 2n9 kb fragment was generated and cloned into pCR2.1-TOPO (Invitrogen), and the sequence of the insert of the recombinant plasmid pBAM2 was determined. This allowed the identification of the nucleotide sequence of the remaining parts of the B. amyloliquefaciens gene. Using this information, a contiguous copy of the entire structural gene was amplified from the B. amyloliquefaciens chromosome using primer P5 (5h-GGTGTGATACATATGAAAGTATTG-3h), which encompassed the translational start codon, and a primer residing 3h to the translational stop codon (P6, 5h-CGGGGATTCGAATTCTTTCTCAGG-3h). The contig was then inserted into pCR2.1-TOPO to give pBAM3. The primer at the 5h-end of the gene (P3) was designed such that the sequence immediately 5h to the ATG start codon became CAT. This change created a NdeI site (CATATG), thereby allowing the gene to be cloned into the equivalent site of the expression vector pET21b to yield pBAM4. The primer at the 3h-end of the gene (P4) contained a recognition site for EcoRI. Enzyme analysis. Quantitative enzyme assays using CB 1954

or bischloroethyl-2,4-dinitrobenzamide (SN 23862) (generous gifts from Dr D. Wilman, Institute of Cancer Research, Sutton, Surrey, UK, and Professor W. Denny, University of Auckland, New Zealand, respectively) as substrate were

carried out at 37 mC by HPLC as previously described (Anlezark et al., 1992, 1995). In this system peaks for substrate, cofactor and reduction products are baseline-separated with the following retention times : NAD(P)H 1n1 min, 4HX 2n1 min, 2HX 2n6 min and CB 1954 3n3 min. Reaction components were quantitated by comparison of peak areas with those of standards run on the same column. A typical peak area for 1 mM CB 1954 would be calculated at 6–7i10' milli absorbance units and peaks of 5 % of this area are readily detected. When qualitative assays were used to identify column fractions the standard conditions were as follows : 1 mM prodrug, 2 mM NAD(P)H, 4 % (v\v) DMSO (Aldrich, reagent grade) in 100 mM sodium phosphate buffer, pH 7, at 37 mC. The presence of nitroreductase activity was indicated by a colour change in the reaction mixture from pale yellow to a deep golden yellow for CB 1954, or from bright yellow to brown for SN 23862. Assays using menadione as the substrate were carried out spectrophotometrically as described by Knox et al. (1988a) using cytochrome c as the terminal electron acceptor. Similar procedures were used to assay flavin reductase activity with FMN and FAD as the substrates. Cofactor concentrations in these assays were at least 500 µM to ensure saturation. Protein concentrations were estimated by BCA assay (Pierce). All spectrophotometric analyses were carried out using a Lambda 2 spectrophotometer (Perkin Elmer). Kinetic parameters were calculated using Enzfitter (Biosoft), a non-linear regression data analysis program. Results are expressed as Km or Vmaxp. Each calculation is based on a minimum of at least ten reaction rate determinations. Purification of native B. amyloliquefaciens YwrO. A 54 ml

crude extract was prepared by sonication from a 2n4 l overnight L-broth culture of strain 0454, after resuspending the washed cell pellet (23 g) in 20 mM potassium phosphate buffer, pH 7n0. DNAase and RNAase were added before centrifugation to aid clarification of the extract. Ammonium sulphate precipitation was performed as follows. The extract was brought to 60 % saturation by addition of 19n5 g solid ammonium sulphate and centrifuged (15 000 r.p.m., 15 min) to remove the precipitate. The supernatant (57 ml) was then brought to 90 % saturation by addition of a further 11n5 g solid ammonium sulphate and centrifuged again. The resulting pellet containing the majority of the nitroreductase activity was resuspended in 5 ml 20 mM Tris pH 7n6, desalted using a PD10 column of Sephadex G25 (Pharmacia) and analysed by chromatography using FPLC Mono Q equilibrated in the same buffer. After washing off unbound proteins, elution was by gradient to 0n3 M KCl. Fractions active with CB 1954 eluting between 120 and 150 mM KCl were pooled, concentrated and desalted before reapplying to Mono Q equilibrated in 20 mM Bistris, pH 6. Elution was again by KCl gradient, and the active fractions eluted at 120–140 mM KCl were pooled and concentrated. Electrophoresis. Homogeneity of the purified protein was assessed in SDS- or native-PAGE in 8–25 % gradient Phastgels using a Phastsystem (Amersham), or in 4–12 % gradient NuPage Bistris gels (Novex) according to the manufacturers’ instructions. Visualization of the protein bands was by Coomassie staining (Phastsystem) or silver staining (Bio-Rad) according to the protein concentration. For final identification of the active protein in IEF or the native-PAGE gels, 4–6n5 IEF Phastgels or Novex 4–20 % gradient Tris\glycine gels were used. Activity staining was performed using o-methyl red (200 µM in 20 mM Bistris, pH 7n0, 1 % DMSO) and NAD(P)H (500 µM) as a cofactor. The gels were placed in lidded Petri

299


dishes and flooded with 0n5–1n0 ml NAD(P)H, followed by 1–2 ml o-methyl red solution and left at 37 mC for approximately 30 min. After this time the dye had penetrated the gel and a zone where the dye had been decolourized by the action of the enzyme was clearly visible. Images were captured in an ImageMaster imaging system (Amersham) using a blue filter to enhance contrast. In vitro cytotoxicity assays. Microtitre plates (96-well) were pre-seeded with V79 cells (European Collection of Animal Cell Cultures) at 10 000 cells ml−" in Dulbecco’s modified Eagle’s medium (DMEM ; Sigma) plus 10 % (v\v) fetal calf serum. All treatments were prepared on duplicate plates and transferred to the cells prior to adding enzyme (10 µl) to the appropriate wells. CB 1954 was dissolved in DMSO (Sigma, tissue culture grade) so that the appropriate concentrations could be dispensed by adding 5 µl of the solution to each well. NAD(P)H was dissolved in sterile PBS to give the appropriate final concentration by adding 10 µl of the solution to each well. Enzymes were diluted in sterile PBS. All aqueous solutions were filter-sterilized before use and all operations were carried out in a laminar-flow hood. The cells were exposed for 3 h to CB 1954 (3n9–500 µM in doubling dilutions) alone, or in combination with cofactor (NAD(P)H 125 or 250 µM) and enzyme (4 µg), by removing the growth medium and replacing it, after washing twice with PBS, with 200 µl serum-free medium containing the various reaction components. After exposure, the cells were washed with PBS, fresh medium containing serum was added and the plates were left to incubate at 37 mC, 5 % CO for 4 d. At this time cells in control (untreated) wells had# achieved confluent growth. Cytotoxicity was quantified by the sulphorhodamine B assay (Rubinstein et al., 1990). Briefly, the growth medium was removed and the cells were fixed by addition of cold trichloroacetic acid (TCA) (10 %, w\v ; 100 µl per well) for 30 min. The TCA was removed and the fixed cells were washed with water before the addition of 0n4 % SRB in 1 % (v\v) acetic acid (100 µl per well) and incubation at room temperature for 30 min. Excess dye was removed and the wells were washed four times with 1 % (v\v) acetic acid. After air drying at room temperature the dye was solubilized by adding 100 µl 10 mM Tris to each well and shaking for 15 min. The plates were read at 492 nm in a Titertek plate reader. Cytotoxicity towards treated cells was expressed as a percentage of the A value for the control wells. %*#

RESULTS Purification of a nitroreductase from B. amyloliquefaciens strain 0454

Following the procedures outlined in Methods, SDSPAGE on an 8–25 % gradient Phastgel indicated the presence of one major band and several minor bands in the concentrate. To confirm that the major band was the nitroreductase, the concentrate was applied to an IEF gel and, after focusing, the gel was sliced to allow Coomassie blue staining of one half and an activity stain (omethyl red, which is decolourized by the enzyme) to be applied to the other half. A clear band of decolourized dye corresponding to the position of the major Coomassie-stained band appeared on incubation of the gel with the dye and NADPH. Throughout the purification, qualitative CB 1954 assays only could be carried out since the endogenous enzyme was present at very low levels. 300

Isolation and characterization of the B. amyloliquefaciens ywrO gene

N-terminal amino acid sequencing of the purified B. amyloliquefaciens nitroreductase enzyme resulted in the sequence MKVLVLAVHPDMENSAVN. When this sequence was used to search available protein databases strong similarity (11 out of 18 conserved amino acids) was noted with the hypothetical protein YrkL identified by the B. subtilis genome sequencing project (Kunst et al., 1997). Some similarity was also noted with two hypothetical E. coli proteins, YabF and YheR (28 and 33 % identity, respectively). In view of this observation, a strategy was formulated whereby sequence homology between the identified bacterial proteins, together with the determined N-terminal amino acid sequence of the discovered B. amyloliquefaciens enzyme, was used to amplify a region of the desired encoding gene from the B. amyloliquefaciens genome (see Methods). Nucleotide sequencing of the insert of the plasmid obtained (pBAM1) revealed the presence of an ORF which encoded a polypeptide that shared 66 % sequence similarity with YrkL. To obtain the 5h- and 3h-ends of the gene, inverse PCR was used to clone a 2n9 kb StyI fragment. The insert of the plasmid obtained (pBAM2) was partially sequenced, allowing the subsequent isolation, by PCR, of the entire structural gene as an approximately 550 bp fragment and its insertion into either pCR2.1-TOPO (pBAM3) or pET21b (pBAM4). Comparative analysis of the predicted amino acid sequence of the cloned gene indicated that it shared a high degree of similarity to the hypothetical B. subtilis protein YrkL (50 % identity). However, a higher degree of similarity (69n7 % identity) was evident to a hypothetical protein, YwrO, of a subsequently determined B. subtilis gene. In view of this higher degree of homology, the cloned B. amyloliquefaciens gene was designated ywrO and its encoded protein was designated YwrO. In B. subtilis, the ywrO gene is preceded by a gene (alsD) encoding acetolactate decarboxylase and followed by two hypothetical genes encoding proteins YwsA and YwsB, both of which have unknown function(s). This arrangement of genes is largely conserved in B. amyloliquefaciens (Fig. 2). Thus, an ORF encoding a protein which shares 90n2 % identity with the B. subtilis AlsD protein resides at the 5h-end of the B. amyloliquefaciens ywrO gene, while the incomplete ORF found downstream encodes a protein which exhibits 48n6 % identity with YswB. One obvious difference between the two bacilli, however, is the fact that in B. subtilis a further small gene, yswA, resides between the ywrO and yswB. The equivalent gene does not appear in this position in the B. amyloliquefaciens genome (Fig. 2). Purification and characterization of recombinant B. amyloliquefaciens YwrO

To facilitate subsequent expression studies, a NdeI site was created over the translational start codon of the gene during its amplification by PCR. This NdeI site,

YwrO : cloning, expression and CB 1954 activation 0·15

Absorbance

(a)

0·10

0·05

250

300

350

400

450

500

ì (nm)

(b)

kDa 188 98 62

.................................................................................................................................................

Fig. 2. Molecular cloning of the ywrO gene. The region of the B. amyloliquefaciens genome in the vicinity of ywrO is shown above the four plasmid maps. The arrangement in B. subtilis is identical except that a hypothetical gene, yswA, is present in the indicated position. Only an internal 1n5 kb fragment (indicated by \\ and BB in the map) of the 2n9 kb StyI fragment was sequenced. The region of the ywrO gene originally amplified (cloned into pCR2.1 to generate pBAM1), using the primers P1 and P2, has been shaded black. The position of the primers P3 and P4 used to generate the inverse PCR fragment cloned in pBAM2 is also shown beneath the shaded region of ywrO. Plasmids pBAM3 and pBAM4 were made by inserting a DNA fragment generated by PCR using the primers P5 and P6 into pCR2.1-TOPO and pET21b between the vector NdeI and EcoRI sites (see Methods). A prime (h) before or after the gene indicates either a 5h- or 3h-truncation, respectively. Also marked are the following vector-borne elements : lac, the lacZ promoter ; T7, the T7 polymerase promoter ; ori, the ColE1 origin of replication ; KmR, the kanamycin resistance gene ; ApR, the ampicillin resistance gene.

together with an EcoRI site created at the 3h-end of the gene, was used to subclone ywrO into the equivalent sites of the expression vector pET21b. This placed the translational start codon of ywrO in the resultant plasmid, pBAM4, at the optimal distance from the vector-encoded ribosome binding site. E. coli novablue cells carrying pBAM4 were grown overnight at 37 mC following induction with IPTG. A crude extract was prepared in 20 mM potassium phosphate buffer, pH 7, by sonication with the addition of DNAase and RNAase before centrifugation to aid clarification. Purification was essentially as described above for the native enzyme, omitting the initial ammonium sulphate fractionation step. The supernatant was buffer-exchanged into 20 mM Tris, pH 7n6, for separation on FPLC Mono Q. A second anion-exchange step was performed in 20 mM Bistris, pH 6, and following this the protein appeared essentially homogeneous on SDS-PAGE (not reduced) (Fig. 3b). The yield from the small-scale process was approximately 25 % with a 24-fold increase in specific activity (Table 1). The pure protein was also electrophoresed on native NuPage 4–12 % gradient gels and activity-stained

28 17 1

2

3

4

5

14

(c)

1 2 3 4 5 6 7 8 9 10 11 12 13 14 .................................................................................................................................................

Fig. 3. Absorption spectrum of B. amyloliquefaciens YwrO, SDS-PAGE analysis and activity staining in native-PAGE. (a) Absorption spectrum showing peak maxima at 381 and 451 nm. (b) SDS-PAGE in 4–12 % gradient NuPage Bistris gel, 25 µl YwrO sample or 10 µl molecular mass marker per lane. Lanes : 1, crude extract of E. coli novablue bearing pBAM4 ; 2, partially purified B. amyloliquefaciens YwrO in 20 mM Tris, pH 7n6 ; 3, YwrO purified to homogeneity in 20 mM Bistris, pH 6n0 ; 4, blank ; 5, molecular mass marker (SeeBlue Plus2, Invitrogen). (c) Native PAGE in 4–20 % Novex Tris/glycine gel with samples and markers loaded at 10 µl per lane. Molecular mass markers do not show true mobility in native-PAGE and so are not annotated. Lanes 1–7 were stained with Coomassie blue ; lanes 8–14 were activity-stained with o-methyl red as described in Methods. Lanes : 1, markers (SeeBlue Plus 2) ; 2, blank ; 3, crude extract as in Fig. 3(b) ; 4, partially purified YwrO ; 5, blank ; 6, purified YwrO ; 7 and 8, blank ; 9, purified YwrO ; 10, partially purified YwrO ; 11, blank ; 12, crude extract ; 13, blank ; 14, molecular mass markers.

301


Table 1. Purification scheme for recombinant B. amyloliquefaciens YwrO Fraction

CB 1954 activity (µmol min−1 ml−1)

Protein (mg ml−1)

Specific activity (µmol min−1 ml−1)

Total activity (µmol min−1)

1n19 1n14 0n84 0n70 0n74

9n40  3n19  0n24

0n13  0n26  3n08

5n95  2n08  1n48

Crude extract Crude extract in 20 mM Tris pH 7n6 Mono Q in Tris pH 7n6 Pooled active fractions 20 mM Bistris starting material Mono Q in Bistris pH 6n0 (final product) , Not determined.

Table 2. Substrate and cofactor specificity for recombinant B. amyloliquefaciens YwrO Km (µM)

Substrate CB 1954 SN 23862 Menadione FMN FMN FAD NADPH (substrate 5 µM menadione) Azodyes

618p169 Inactive 3n4p0n5 (Ksi 8n7) 53n0p6n0 104n7p35n6 209n0p27n0 40n0p3n8 Active with o-methyl red

kcat (s−1)

Cofactor

8n2p1n1 * 175n0p9n7

NADPH NAD(P)H NADPH

87n3p1n8 54n2p2n4 78n2p2n0 169n4p3n3 

NADPH NADH NADPH NAD(P)H

, Not determined. * SN 23862 kcat value could not be determined because YwrO is inactive with this substrate.

using NAD(P)H and o-methyl red. Decolourized bands were clearly seen in lanes containing the YwrO protein (Fig. 3b).

0·07

[Menadione] /v

0·06

The estimated molecular mass from comparison with low-molecular-mass standards was 21n5 kDa, which agrees well with the value calculated from the deduced amino acid sequence. By gel filtration (Superdex 75), however, the molecular mass was estimated to be approximately 88 kDa, suggesting that the protein exists as a tetramer in the native state. The spectrum of the pure protein was typical of a flavoprotein with λmax at 381 and 451 nm (Fig. 3a).

0·05 0·04 0·03 0·02 0·01

–15

–10

–5

0

5

10

15

20

25

[Dicoumarol] (íM) .................................................................................................................................................

Fig. 4. The inhibition of menadione reduction by dicoumarol. The concentration of dicoumarol (µM) was plotted against substrate concentration/rate of reaction (i.e. [menadione]/v) for each of the five menadione concentrations tested : $, 1 µM ; #, 1n5 µM ; X, 2 µM ; W, 3 µM ; , 4 µM. Lines of best fit were constructed for each substrate concentration and the point of intersection indicates the inhibition constant (kKi). Converging lines indicate that the inhibition was uncompetitive.

302

Kinetic parameters with respect to CB 1954 were determined by HPLC assay using 10 µl of the final product in a 500 µl assay mix containing 100–1000 µM CB 1954, 500 µM NADPH and incubating the mixture at 37 mC for 10 min. Reduction of CB 1954 was determined by comparison of peak areas at 325 nm in standard and enzyme tubes. Km was calculated to be 618p169 µM with a Vmax of 5n5 µmol min−" mg−", giving a kcat of 8n2 s−". The single product formed was found to be exclusively the cytotoxic 4HX derivative. As described in Methods, reduction product peaks were baseline-separated from each other and from those of

YwrO : cloning, expression and CB 1954 activation

not use nitrofurazone or SN 23862, the bischloroethyl mustard analogue of CB 1954, as substrates with either NADH or NADPH as cofactor. The reduction of menadione was potently inhibited uncompetitively by dicoumarol with a Ki of 1 µM (Fig. 4).

120

A492 as % of control

100 80 60 40

In vitro cytotoxicity tests

20

The sulphorhodamine B assay was used to determine the cytotoxicity of the combination of enzyme, cofactor and CB 1954 on V79 cells of NTR and B. amyloliquefaciens YwrO. Exposure of the cells for 3 h to the various treatments was carried out in serum-free medium to avoid activation of CB 1954 by endogenous enzymes. The results are shown in Fig. 5. Control curves are shown for prodrug alone or in combination with either NADH or NADPH. Other controls were performed concurrently in each experiment (enzyme alone or with prodrug or cofactor, cofactor alone or with prodrug, DMSO alone) but none of the median absorbance values differed significantly from those of untreated cells (P 0n05, Mann–Whitney test).

0 –20

1

10

100

1000

[CB 1954] (íM) .................................................................................................................................................

Fig. 5. Enhanced cytotoxicity in V79 cells of CB 1954 in the presence of B. amyloliquefaciens YwrO and E. coli B NTR. Cytotoxicity was assessed in 96-well plates seeded with V79 cells in DMEMj10 % (v/v) fetal calf serum at 10 000 cells ml−1, by sulphorhodamine B assay 4 d after a 3 h exposure to : $, CB 1954 alone ; #, CB 1954jNADH ; X, CB 1954jNADPH ; W, CB 1954jNADHjNTR ; , CB 1954jNADPHjYwrO. Cells were exposed to CB 1954 alone (doubling dilutions from 500 µM), in combination with CB 1954 plus NAD(P)H (125 or 250 µM), or CB 1954 plus NAD(P)H and enzyme (4 µg) in a total volume of 200 µl DMEM. Results are expressed as a percentage of A492 values for the control cells, meanpSD (untreated, A492 1n6p0n05, n l 24). Each point represents the mean (pstandard deviation, n l 6) of at least six wells in two independent experiments. All treatments were carried out in serum-free DMEM. For wells exposed to prodrug, cofactor and enzyme, five concentrations of CB 1954 in doubling dilutions from 500 µM (B. amyloliquefaciens YwrO) or 50 µM (NTR) were used. Statistical analysis was performed using the Mann–Whitney test comparing the median values for prodrugjcofactor with those for prodrugjcofactorjenzyme at each dose of prodrug. The asterisks denote significant differences between the medians, P 0n05.

the cofactor and substrate, and very small peaks could be detected. The only peaks detected in any sample were for CB 1954, NADPH and the 4HX product. There was no activity with the bischloroethyl analogue of CB 1954, SN 23862, with either NADPH or NADH as the cofactor. B. amyloliquefaciens YwrO could use FMN and FAD as substrates and acted as a menadione reductase (Table 2). With all of these substrates the rate of reaction was 2–10-fold faster with NADPH than with NADH (data shown for FMN and NAD(P)H only). When menadione was used as the substrate the rate of reaction with NADH as cofactor was very low ( 5 µmol min−" ml−" with 10 µM menadione against 40 µmol min−" ml−" with NADPH as cofactor) and kinetic parameters with this cofactor were not determined. The cofactor concentration in all cases was at least 500 µM to ensure saturation. Substrate inhibition was seen with concentrations of menadione greater than 10 µM and therefore the kinetic parameters were determined using the Cornish-Bowden method (Cornish-Bowden, 1979) for substrate inhibition. B. amyloliquefaciens YwrO could

Under the conditions used CB 1954 (3n9–500 µM) was not cytotoxic either alone or in combination with NADH or NADPH. However, in the presence of 4 µg B. amyloliquefaciens YwrO or NTR there was a doserelated cytotoxic effect. The ED values for the two &! analysis and were enzymes were calculated by probit 6n3p0n5 µM (NTR) (95 % confidence intervals 5n4–7n3) and 137n1p8n2 µM (B. amyloliquefaciens YwrO) (95 % confidence intervals 121n9–154n4). DISCUSSION

The aim of this study was to isolate and characterize a novel enzyme for DEPT strategies. For nitroreductases the strategies of choice are likely to be of the GDEPT or VDEPT type because of the necessity and inherent difficulty of providing exogenous cofactor for ADEPT approaches (Knox et al., 1995 ; Anlezark et al., 1996). Whilst the requirements for prodrugs for the alternative approaches may be different (Connors, 1995 ; Denny & Wilson, 1998), the requisite properties of the enzymes are probably substantially alike in that selectivity, a high affinity for the substrate and a high turnover are the principal considerations. Additionally, if CB 1954 is to be used as a prodrug, consideration needs to be given to product formation. The E. coli B NTR reduces CB 1954 to produce a mixture of the two hydroxylamine reduction products (Fig. 1) (Knox et al., 1992), one of which (the 2HX) is substantially less cytotoxic. Therefore, it is considered an advantage for a DEPT enzyme to produce the 4HX derivative only and certainly not solely the 2HX. Alternative prodrugs also merit attention, including analogues of CB 1954 (Palmer et al., 1992 ; Anlezark et al., 1995) and other compounds where nitroreduction is a ‘ trigger ’ for activation (Mauger et al., 1994 ; Siim et al., 1997). An individual enzyme might not possess all the desired properties or substrate specificity, but if a portfolio of enzymes could be 303


identified with specificity for one or more classes of prodrugs, exploitation of nitroreduction for prodrug activation in cancer therapy (Connors, 1995) could find applications in a wide spectrum of disease. The B. amyloliquefaciens YwrO protein described in the present study possesses at least two of the desired attributes for DEPT. Firstly, it produces only 4HX from the reduction of CB 1954 and secondly, it has a higher affinity for the prodrug than either NTR or Walker DTD (Km values of 617, 862 and 826 µM, respectively), with its kcat being comparable to that of NTR (and greater than that of Walker DTD). However it cannot use the mustard analogue of CB 1954 as a substrate. As yet few enzymes with optimal properties have been available, with the only published alternatives to NTR being the Walker rat carcinoma DTD and human NQ02 (Knox et al., 1988a ; Wu et al., 1997). As well as being able to reduce CB 1954, B. amyloliquefaciens YwrO has a number of other properties in common with DTD or NTR. It is a flavin reductase and can use either FAD or FMN as substrates, although the affinity for the latter is greater. NTR has flavin reductase activity, but the flavins are relatively poor substrates and reduction rates are low (G. M. Anlezark, unpublished data), as they are for a highly related protein NfsB, compared with proteins from other related species (Zenno et al., 1996). B. amyloliquefaciens YwrO is also an azoreductase, rapidly reducing the azodye o-methyl red. The B. amyloliquefaciens protein described in this work has been identified with the aid of the complete genome sequence of a highly related organism. The value of sequence homology in these circumstances for identifying and screening enzymes for specific properties, such as prodrug activation, is clear. In the present work we have cloned and overexpressed the previously unknown product of a gene which was identified by sequence homology to be an orthologue of the ywrO of B. subtilis. In Bacillus spp., and amongst members of other genera (Bryant et al., 1981 ; Kinouchi & Ohnishi, 1983 ; Rosenkranz et al., 1982 ; G. M. Anlezark, unpublished data), total nitroreductase activity assayed in crude cell extracts often represents activity of at least three enzymes which can be purified individually only with difficulty (G. M. Anlezark, unpublished data). In the course of this study we have also cloned and overexpressed the B. subtilis gene (G. M. Anlezark, unpublished results), but the protein had relatively little activity with the prodrugs CB 1954 or SN 23862. This demonstrates that sequence homology may not always indicate structural homology or similar enzyme– substrate interactions. Nevertheless, searching genomes and databases for homologous sequences offers a relatively rapid screening method for identification of potentially useful enzymes for DEPT strategies. Using the entire genome sequence of B. subtilis and database mining, we have already identified further genes potentially encoding prodrug-reducing enzymes which will form the basis of further publications. 304

B. amyloliquefaciens YwrO shows no sequence homology with NfnB or other Enterobacteriaceae nitroreductases, but shares a degree of homology with Walker DTD, other mammalian diaphorases and NQ02. Sequence differences may contribute to the greater ability of the bacterial enzyme and NQ02 to turn over CB 1954 compared to other DTDs. In particular, the Walker enzyme and NQ02 (active with CB 1954) have a Tyr residue at position 104, whereas NQ01 (DTD) and the mouse enzymes (low CB 1954 activity) do not. This residue has been shown to have a significant effect on the affinity of the proteins for CB 1954, with the substitution of Tyr for Gln shifting the properties of NQ01 and mouse enzymes towards those of the Walker enzyme and NQ02 (Chen et al., 1997) with respect to CB 1954. B. amyloliquefaciens and B. subtilis YwrO share the Tyr in the equivalent position in their sequences, but these differences may not represent all of the critical residues involved in prodrug reduction because B. subtilis YwrO has very low activity with CB 1954, reducing the prodrug too slowly to be of clinical use, despite its high degree of homology with B. amyloliquefaciens YwrO. Further studies will be necessary to elucidate which parts of the molecules are essential for efficient production of the 4HX cytotoxin. Although YwrO produces only the cytotoxic 4HX derivative of CB 1954, in contrast to the 2HX and 4HX products of the prodrug reduction by NTR, its effects in cytotoxicity assays are less than those of NTR under the same conditions. It is difficult to say how an in vitro experiment will relate to an in vivo situation where the gene product is being expressed in mammalian cells. The kcat value of the novel protein is lower than that of the NTR and it is not yet known which of the kinetic parameters will prove to be crucial in an in vivo setting. Nevertheless, the kcat value of B. amyloliquefaciens YwrO is greater than that of the Walker enzyme, which is able to achieve efficient in vitro cytotoxicity using CB 1954 and NADH (Knox et al., 1988b). It may be that a critical concentration of the toxic product is needed in vitro, where a relatively low initial cell density will allow diffusion of the product away into the surrounding medium. In a dense tumour mass diffusion will be restricted and an enzyme working relatively slowly will not be a disadvantage. Three-dimensional tumour models may provide a better estimate of in vivo efficacy (Hoffman, 1993 ; Santini & Rainaldi, 1999) and it remains to be seen which enzyme will be effective when expressed in in vivo models. ACKNOWLEDGEMENTS This work was supported by the Department of Health, UK.

REFERENCES Anlezark, G. M., Melton, R. G., Sherwood, R. F., Coles, B., Friedlos, F. & Knox, R. J. (1992). The bioactivation of 5-(aziridin-

1-yl)-2,4-dinitrobenzamide (CB 1954) – I. Purification and properties of a nitroreductase enzyme from Escherichia coli – a potential enzyme for antibody-directed enzyme prodrug therapy (ADEPT). Biochem Pharmacol 44, 2289–2295.

YwrO : cloning, expression and CB 1954 activation Anlezark, G. M., Melton, R. G., Sherwood, R. F., Wilson, W. R., Denny, W. A., Palmer, B. D., Knox, R. J., Friedlos, F. & Williams, A. (1995). Bioactivation of dinitrobenzamide mustards by an E. coli

B nitroreductase. Biochem Pharmacol 50, 609–618. Anlezark, G. M., Cook, R., Melton, R. G., Freemantle, P., Pedley, R. B., Boden, J. & Knox, R. J. (1996). Biodistribution of A5B7

F(abh)2 : E. coli B nitroreductase conjugate in nude mice with human xenografts. Br J Cancer 73, 40. Bagshawe, K. D. (1987). Antibody directed enzymes revive anticancer prodrugs concept. Br J Cancer 56, 531–532.

NAD(P)H dehydrogenase (quinone) (EC 1;6;99;2). Biochem Pharmacol 37, 4671–4677. Knox, R. J., Friedlos, F., Jarman, M. & Roberts, J. J. (1988b). A new cytotoxic, DNA interstrand crosslinking agent, 5-(aziridin-1-yl)4-hydroxylamino-2-nitrobenzamide, is formed from 5-(aziridin1-yl)-2,4-dinitrobenzamide (CB 1954) by a nitroreductase enzyme in Walker carcinoma cells. Biochem Pharmacol 37, 4661–4669. Knox, R. J., Friedlos, F., Marchbank , T. & Roberts, J. J. (1991).

Bagshawe, K. D., Springer, C. J., Searle, F., Antoniw, P., Sharma, S. K., Melton, R. G. & Sherwood, R. F. (1988). A cytotoxic agent

Bioactivation of CB 1954 : reaction of the active 4-hydroxylamino derivative with thioesters to form the ultimate DNA–DNA interstrand crosslinking species. Biochem Pharmacol 42, 1691– 1697.

can be generated selectively at cancer sites. Br J Cancer 58, 700–703.

Knox, R. J., Friedlos, F., Sherwood, R. F., Melton, R. G. & Anlezark, G. M. (1992). The bioactivation of 5-(aziridin-1-yl)-2,4-dinitro-

Bailey, S. M., Knox, R. J., Hobbs, S. M., Jenkins, T. C., Mauger, A. B., Melton, R. G., Burke, P. J., Connors, T. A. & Hart, I. R. (1996).

benzamide (CB 1954) – II. A comparison of an Escherichia coli nitroreductase and Walker DT diaphorase. Biochem Pharmacol 44, 2297–2301.

Investigation of alternative prodrugs for use with E. coli nitroreductase in ‘ suicide gene ’ approaches to cancer therapy. Gene Ther 3, 1143–1150. Boland, M. P., Knox, R. J. & Roberts, J. J. (1991). The differences in kinetics of rat and human DT diaphorase result in a differential sensitivity of derived cell lines to CB 1954 (5-(aziridin-1-yl)-2,4dinitrobenzamide). Biochem Pharmacol 41, 867–875.

Knox, R. J., Friedlos, F., Jarman, M., Davies, L. C., Goddard, P., Anlezark, G. M., Melton, R. G. & Sherwood, R. F. (1995). Virtual

cofactors for an Escherichia coli nitroreductase enzyme : relevance to reductively activated prodrugs in antibody directed enzyme prodrug therapy (ADEPT). Biochem Pharmacol 49, 1641–1647. Kunst, F., Ogasawara, N., Moszer, I. & 148 other authors (1997).

Bridgewater, J. A., Knox, R. J., Pitts, J. D., Collins, M. K. & Springer, C. J. (1997). The bystander effect of the nitro-

The complete genome sequence of the gram-positive bacterium Bacillus subtilis. Nature 390, 249–256.

reductase\CB1954 enzyme\prodrug system is due to a cellpermeable metabolite. Hum Gene Ther 8, 709–717.

Latham, J. P., Searle, P. F., Mautner, V. & James, N. D. (2000).

Bryant, D. W., McCalla, D. R., Leeksma, M. & Laneuville, P. (1981).

Type I nitroreductases of Escherichia coli. Can J Microbiol 27, 81–86. Chen, S., Knox, R., Wu, K., Deng, P. S., Zhou, D., Bianchet, M. A. & Amzel, L. M. (1997). Molecular basis of the catalytic differences

among DT-diaphorase of human, rat and mouse. J Biol Chem 272, 1437–1439. Clark, A. J., Iwobi, M., Cui, W., Crompton, M., Harold, G., Hobbs, S., Kamalati, T., Knox, R., Neil, C., Yull, F. & Gusterson, B. (1997).

Prostate-specific antigen promoter\enhancer driven gene therapy for prostate cancer : construction and testing of a tissue-specific adenovirus vector. Cancer Res 60, 334–341. Lemmon, M. J., van Zijl, P., Fox, M. E., Mauchline, M. L., Giaccia, A. J., Minton, N. P. & Brown, J. M. (1997). Anaerobic bacteria as

a gene delivery system that is controlled by the tumour microenvironment. Gene Ther 4, 791–796. McNeish, I. A., Green, N. K., Gilligan, M. G., Ford, M. J., Mautner, V., Young, L. S., Kerr, D. J. & Searle, P. F. (1998). Virus directed

enzyme prodrug therapy for ovarian and pancreatic cancer using retrovirally delivered E. coli nitroreductase and CB1954. Gene Ther 5, 1061–1069.

Selective cell ablation in transgenic mice expression E. coli nitroreductase. Gene Ther 4, 101–110. Connors, T. A. (1995). The choice of prodrugs for gene directed enzyme prodrug therapy of cancer. Gene Ther 2, 702–709. Cornish-Bowden, A. (1979). Fundamentals of Enzyme Kinetics, pp. 73–96. Boston : Butterworths.

Mauger, A. B., Burke, P. J., Somani, H. H., Friedlos, F. & Knox, R. J. (1994). Self-immolative prodrugs : candidates for antibody-

Dachs, G. U., Dougherty, G. J., Stratford, I. J. & Chaplin, D. J. (1997). Targeting gene therapy to cancer : a review. Oncol Res 9,

Minton, N. P., Mauchline, M. L., Lemmon, M. J., Brehm, J. K., Fox, M., Michael, N. P., Giaccia, A. & Brown, J. M. (1995). Chemo-

313–325.

therapeutic tumour targeting using clostridial spores. FEMS Microbiol Rev 17, 357–364.

Denny, W. A. & Wilson, W. R. (1998). The design of selectively-

directed enzyme prodrug therapy in conjunction with a nitroreductase enzyme. J Med Chem 37, 3452–3458.

activated anti-cancer prodrugs for use in antibody-directed and gene-directed enzyme-prodrug therapies. J Pharm Pharmacol 50, 387–394. Drabek, D., Guy, J., Craig, R. & Grosveld, F. (1997). The expression of bacterial nitroreductase in transgenic mice results in specific cell killing by the prodrug CB1954. Gene Ther 4, 93–100. Hoffman, R. (1993). To do tissue culture in two or three dimensions? That is the question. Stem Cells (Miamisburg) 11, 105–111. Kinouchi, T. & Ohnishi, Y. (1983). Purification and characterization of 1-nitropyrene nitroreductases from Bacteroides fragilis. Appl Environ Microbiol 46, 596–604.

Palmer, B. D., Wilson, W. R., Cliffe, S. & Denny, W. A. (1992).

Knox, R. J., Boland, M. P., Friedlos, F., Coles, B., Southan, C. & Roberts, J. J. (1988a). The nitroreductase enzyme in Walker cells

Rubinstein, L. V., Shoemaker, R. H., Paull, K. D., Simon, R. M., Tosini, S., Skehan, P., Scudiero, D. A., Monks, A. & Boyd, M. R. (1990). Comparison of in vitro anti-cancer-drug-screening data

that activates 5-(aziridin-1-yl)-2,4-dinitrobenzamide (CB 1954) to 5-(aziridin-1-yl)-4-hydroxylamino-2-nitrobenzamide is a form of

Hypoxia-selective antitumour agents. 5. Synthesis of watersoluble nitroaniline mustards with selective cytotoxicity for hypoxic mammalian cells. J Med Chem 35, 3214–3222. Roberts, J. J., Freidlos, F. & Knox, R. J. (1986). CB 1954 (2,4dinitro-5-aziridinyl benzamide) becomes a DNA interstrand crosslinking agent in Walker tumour cells. Biochem Biophys Res Commun 140, 1073–1078. Rosenkranz, E. J., McCoy, E. C., Mermelstein, R. & Rosenkranz, H. S. (1982). Evidence for the existence of distinct nitroreductases

in Salmonella typhimurium : roles in mutagenesis. Carcinogenesis 3, 121–123.

generated with a tetrazolium assay versus a protein assay against 305


a diverse panel of human tumour cell lines. J Natl Cancer Inst 82, 113–118. Sambrook, J., Fritsch, E. F. & Maniatis, T. (1989). Molecular Cloning : a Laboratory Manual, 2nd edn. Cold Spring Harbor, NY : Cold Spring Harbor Laboratory. Santini, M. T. & Rainaldi, G. (1999). Three-dimensional spheroid model in tumour biology. Pathobiology 67, 148–157. Searle, P. F., Weedon, S. J., McNeish, I. A., Gilligan, M. G., Ford, M. J., Friedlos, F., Springer, C. J., Young, L. S. & Kerr, D. J. (1998).

Sensitisation of human ovarian cancer cells to killing by the prodrug CB1954 following retroviral or adenoviral transfer of the E. coli nitroreductase gene. Adv Exp Med Biol 451, 107–113. Siim, B. G., Denny, W. A. & Wilson, W. R. (1997). Nitro reduction as an electronic switch for bioreductive drug activation. Oncol Res 9, 357–369. Whelan, S. M., Elmore, M. J., Bodsworth, N. J., Atkinson, T. & Minton, N. P. (1992). The complete amino acid sequence of the

306

Clostridium botulinum type-E neurotoxin, derived by nucleotidesequence analysis of the encoding gene. Eur J Biochem 204, 657–667. Wu, K. B., Knox, R., Sun, X. Z., Joseph, P., Jaiswal, A. K., Zhang, D., Deng, P. S. & Chen, S. (1997). Catalytic properties of

NAD(P)H : quinone oxidoreductase-2 (NQ02), a dihydronicotinamide riboside dependent oxidoreductase. Arch Biochem Biophys 347, 221–228. Zenno, S., Koike, H., Tanokura, M. & Saigo, K. (1996). Gene cloning, purification, and characterization of NfsB, a minor oxygen-insensitive nitroreductase from Escherichia coli, similar in biochemical properties to FRase I, the major flavin reductase in Vibrio fischeri. J Biochem 120, 736–744. .................................................................................................................................................

Received 22 May 2001 ; revised 20 August 2001 ; accepted 29 August 2001.