Cloning and Characterization of a Novel Human Dual Flavin Reductase*

Flavoprotein reductases play a key role in electron transfer in many physiological processes. We have isolated a cDNA with strong sequence similarities to cytochrome P-450 reductase and nitric-oxide synthase. The cDNA encodes a protein of 597 amino acid residues with a predicted molecular mass of 67 kDa. Northern blot analysis identified a predicted transcript of 3.0 kilobase pairs as well as a larger transcript at 6.0 kilobase pairs, and the gene was mapped to chromosome 9q34.3 by fluorescencein situ hybridization analysis. The amino acid sequence of the protein contained distinct FMN-, FAD-, and NADPH-binding domains, and in order to establish whether the protein contained these cofactors, the coding sequence was expressed in insect cells and purified. Recombinant protein bound FMN, FAD, and NADPH cofactors and exhibited a UV-visible spectrum with absorbance maxima at 380, 460, and 626 nm. The purified enzyme reduced cytochromec, with apparent K m andk cat values of 21 μm and 1.3 s−1, respectively, and metabolized the one-electron acceptors doxorubicin, menadione, and potassium ferricyanide. Immunoblot analysis of fractionated MCF7 cells with antibodies to recombinant NR1 showed that the enzyme is cytoplasmic and highly expressed in a panel of human cancer cell lines, thus indicating that this novel reductase may play a role in the metabolic activation of bioreductive anticancer drugs and other chemicals activated by one-electron reduction.

Flavin-containing enzymes catalyze a broad spectrum of biochemical reactions ranging from oxidase, dehydrogenase, and mono-oxygenase reactions. Most flavoproteins contain either FMN or FAD as prosthetic groups; however, a small number of enzymes contain both cofactors. In mammalian systems, NADPH cytochrome P-450 oxidoreductase (cytochrome P-450 reductase) was the first such enzyme isolated (1,2), followed by several other dual flavin enzymes including nitric-oxide synthases (NOS) 1 in higher organisms (3,4), and CYP102 (5) and sulfite reductase (6) in bacteria. More recently, the cDNA sequence encoding a putative FMN-and FAD-binding enzyme, methionine synthase reductase, has been described (7).
Cytochrome P-450 reductase, the most extensively characterized of these enzymes (8 -10), is found in the endoplasmic reticulum of most eukaryote cells and is an integral component of the monooxygenase system transferring electrons from NADPH to cytochromes P-450 via FMN and FAD co-factors. Cytochrome P-450 reductase may also donate electrons to heme oxygenase (11), cytochrome b 5 (12), and the fatty acid elongation system (13), and can reduce cytochrome c (14). Both the crystal and NMR structure of the FMN domain of human cytochrome P-450 reductase (15,16) and the crystal structure of the NH 2 -terminally truncated form of the rat enzyme (17) have been resolved, providing high resolution structural information on this enzyme class. The amino-terminal region of cytochrome P-450 reductase bears striking amino acid homology with FMN-containing flavodoxins, while the carboxyl-terminal region shows similarities with the FAD-containing ferredoxin-NADP ϩ reductases, thus leading to the hypothesis that cytochrome P-450 reductase has evolved as a fusion of these two ancestral proteins (18,19). A carboxyl-terminal cytochrome P-450 reductase-like domain is also a component of the NOS family of enzymes, where it is fused to an amino-terminal heme domain. The NOS reductase domain shuttles electrons from NADPH to the active site iron where the amino acid, L-arginine, is metabolized to nitric oxide (NO) (20).
In addition to its normal physiological functions, cytochrome P-450 reductase plays a role in the reduction of one-electron acceptors such as the therapeutically important anticancer agents mitomycin c (22), adriamycin (23), and the benzotriazine di-N-oxide, tirapazamine (24). Evidence is also emerging that NOS can transfer electrons to these compounds via its reductase domain (25,26). The expression of these dual flavin reductases will therefore influence the outcome of cancer therapy.
In this study we report the cloning of a novel member of the FNR family containing FMN and FAD as cofactors, which supports the NADPH-dependent metabolism of cytochrome c, the quinone anti-neoplastic agent doxorubicin, and menadione. Interestingly, the enzyme, which we have called NR1 (novel reductase 1) appears widely expressed in human cancer cell lines and, therefore, could play a potential role in the activation (or deactivation) of drugs used in cancer therapy.

MATERIALS AND METHODS
Chemicals and Reagents-All chemicals were purchased from Sigma (Poole, Dorset, United Kingdom (UK)) and all enzymes and cell culture * This work was supported by SmithKline Beecham and United Kingdom Medical Research Council Grant G9203175. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBank TM /EBI Data Bank with accession number(s) AF199509.
ʈ To whom correspondence should be addressed. Tel.: 44-1382-632-621; Fax: 44-1382-668278; E-mail: rooney@dundee.ac.uk. 1 The abbreviations used are: NOS, nitric-oxide synthase; PBS, phosphate-buffered saline; FNR, flavodoxin-NADP ϩ reductase; FISH, fluorescent in situ hybridization; PCR, polymerase chain reaction; HPLC, high performance liquid chromatography; EST, expressed sequence tag; TBST, Tris-buffered saline with Tween 20; RB, resuspension buffer. media were from Life Technologies, Inc. (Paisley, UK), except where stated. All solvents were of HPLC grade (Rathburn Chemicals Ltd., UK). Novel Reductase Constructs-The EST data base was screened for putative FAD-or FMN-containing proteins using the human P-450 reductase cDNA as the probe sequence (accession no. S90469). Two novel cDNAs were identified, one was subsequently reported to be methionine synthase reductase (7) and the other (pNR1-SPORT) was used in these studies. This clone contained a 2506-nucleotide sequence and contained putative FAD-and FMN-binding domains. This cDNA in pSPORT (Life Technologies, Inc.) was used as a template for PCR amplification of the sequence for expression in baculovirus and Escherichia coli systems. Oligonucleotides 5Ј-GAGAATTCCATATGCCGAG-CCCGCAGCTTCTG-3Ј and 5Ј-GGAATTCCTCGAGTCAGGCCCACGT-CTCTGTCTGGAA-3Ј corresponding to putative 5Ј and 3Ј ends of the NR1 coding sequence were synthesized. The 5Ј oligonucleotide contained overhanging NdeI and EcoRI restriction sites, while the 3Ј oligonucleotide contained XhoI and EcoRI sites. Following 25 cycles of amplification using Pfu polymerase (Stratagene), the cDNA was ligated into pCR SCRIPT (Stratagene) to produce the plasmid pNR1-SCRIPT and sequenced to confirm that no PCR errors had been introduced. For baculoviral expression, the coding sequence was removed from pNR1-SCRIPT by EcoRI digestion and cloned into the unique EcoRI site of pFastBac Hta (Life Technologies, Inc.) and in frame with an aminoterminal 6-histidine linker and rTEV protease cleavage site. Recombinant baculovirus (vNR1) was produced following transposition of the cDNA sequence downstream of the polyhedrin promoter in Bacmid DNA and transfection into insect Sf9 cells using the Bac-to-Bac baculovirus expression system (Life Technologies, Inc.), following manufacturer's instructions.
The FAD domain constructs were generated by PCR amplification and cloning of nucleotides 579 -1860 (encoding amino acid residues 194 -597). This region was cloned into pCRSCRIPTto produce pFAD-SCRIPT using a Stratagene kit system and sequenced to verify clone integrity. Forward (5Ј-GGGAATTCCATATGGTAGCTCACCCCGGCT-CTCAGG-3Ј) and reverse (5Ј-GGAATTCCTCGAGTCAGGCCCACGTC-TCTGTCTGCAA-3Ј) primers were then used, which contained NdeI and XhoI sites for subcloning into the unique NdeI/XhoI sites of pET15b (Novagen) downstream of a 6-histidine linker and thrombin cleavage site. The resulting plasmid pFAD-PET was used for expression of the FAD domain.
Recombinant Protein Expression-For baculoviral expression of NR1, Sf9 cells were maintained at 27°C in SF900 II media (Life Technologies, Inc.) according to standard procedures (27). For expression, a 300-ml suspension culture (ϳ2.0 ϫ 10 6 cells/ml) was infected with virus at a multiplicity of approximately 2 plaque-forming units/ cell. Cells were harvested 72 h after infection and resuspended in 10 ml of PBS, 0.1% Tween 20. Protein purification steps were carried out at 4°C. The suspension was sonicated (MSE probe, several short bursts at highest power) and centrifuged at 100,000 ϫ g for 1 h (Sorvall Ultra Pro 80 with A641 rotor). The supernatant was loaded onto a 5-ml Hi-Trap nickel-agarose column (Amersham Pharmacia Biotech) and washed sequentially with 20 ml of 20 mM Tris-HCl, pH 7.5, 100 mM NaCl, 10% glycerol; 20 ml Tris-HCl, pH 7.5, 500 mM NaCl, 60 mM imidazole, 10% glycerol; and then with 25 ml of PBS. Approximately 3 ml of yellow protein was then eluted with PBS,10% glycerol, 0.3 M imidazole. The protein was diluted with 10 ml of affinity buffer (20 mM Tris-HCl, pH 7.5, 150 mM NaCl, 1 mM EDTA, 0.1 mM dithiothreitol, 10% glycerol), and loaded onto a 2-ml 2Ј,5Ј-ADP-Sepharose column. The column was washed sequentially with 15 ml of affinity buffer and affinity buffer with 0.5 M NaCl. The protein was eluted in 2.5 ml of affinity buffer containing 0.5 M NaCl and 10 mM 2Ј-AMP. FMN was added to 40 M to replace any cofactor lost during purification, and protein exchanged into PBS, 10% glycerol using a PD-10 (Amersham Pharmacia Biotech) gel filtration column. Protein concentrations were determined by Bradford analysis using Bio-Rad reagents and bovine serum albumin as a protein standard.
For E. coli expression of the FAD domain, BL21 (plys S) strains containing the domain expression plasmid pFAD-PET were grown overnight at 37°C in LB broth containing ampicillin (50 g/ml) and chloramphenicol (34 g/ml). One-liter cultures of fresh LB broth were inoculated with 10 ml of the overnight culture and the bacteria grown at 37°C to an optical density of 0.6 -1.0. Isopropyl-1-thio-␤-D-galactopyranoside was then added (0.5 mM) to initiate expression and the culture transferred to 30°C and grown overnight. Cells were harvested at 5,000 ϫ g for 15 min and resuspended in 20 ml of binding buffer (5 mM imidazole, 500 mM NaCl, 20 mM Tris-HCl, pH 8.0). The recombinant protein was extracted and purified over a 5-ml Hi-Trap nickel-agarose column (Amersham Pharmacia Biotech) as described (19). The FAD domain was washed with binding buffer containing 60 mM imidazole and eluted with binding buffer containing 1 M imidazole. The protein eluted off the column was then exchanged into affinity buffer using PD-10 gel filtration columns and purified over 2Ј,5Ј-ADP-Sepharose as described above. Final yields of pure protein were between 2 and 5 mg/liter of culture. The final purified protein was stored in PBS, 10% glycerol at Ϫ20°C.
Antibodies and Immunoblots-Antibodies against NR1 were generated in sheep using 1 mg of purified recombinant NR1-FAD domain by Scottish Antibodies Production Unit (Carluke, UK). Antibodies to fulllength cytochrome P-450 reductase have been described previously (19). SDS-polyacrylamide gel electrophoresis and immunoblots were carried out using a Mini-PROTEAN II (Bio-Rad) electrophoresis system. Except where indicated, proteins were separated using 10% SDS-polyacrylamide gels and electroblotted onto nitrocellulose (Schleicher & Schuell) according to manufacturer's instructions. For immunodetection, the blots were blocked in TBST (20 mM Tris-HCl, pH 7.5, 0.5 M NaCl, 0.05% v/v Tween 20) with 5% w/v milk powder (Marvel) overnight at 4°C with shaking. After washing several times with TBST, blots were incubated with appropriate antibody diluted in TBST, 5% milk powder at room temperature for 1-2 h. The binding of the primary antibodies was detected using a chemiluminescence detection system (Amersham Pharmacia Biotech, ECL). The secondary antibodies used were anti-rabbit IgG and anti-sheep IgG (Scottish Antibody Production Unit).
To prepare whole cell extracts, approximately 2 ϫ 10 7 cells were harvested by trypsinization, washed once with PBS, and frozen at Ϫ70°C. Cells were resuspended in 1 ml of 100 mM PBS, 0.25 M sucrose and sonicated on ice using several pulses with an MSE probe. Following centrifugation at 4°C, 12,000 ϫ g for 1 h, the supernatant was aspirated and stored at Ϫ70°C. For subcellular fractionation studies, approximately 2 ϫ 10 8 MCF-7 cells were harvested, washed twice in PBS, and homogenized using a 20-ml glass homogenizer in 10 ml of resuspension buffer (RB) consisting of 0.25 M sucrose, 50 mM HEPES, 1 mM EDTA, 0.2 mM dithiothreitol, 2 g/ml aprotonin, 2 g/ml leupeptin, 0.2 mM phenylmethylsulfonyl fluoride. Nuclear material and particulate cell matter were pelleted at 10,000 ϫ g for 10 min and resuspended in 5 ml of RB. The supernatant material was centrifuged at 100,000 ϫ g for 1 h, and the resultant pellet was resuspended in 1 ml of RB. The supernatant, containing the cytoplasmic fraction, and the pellet, containing the membrane fractions, were stored at Ϫ70°C.
Flavin Determination and Spectral Analysis-FMN and FAD content was determined by HPLC (28) using a Hewlett Packard 1050 HPLC chromatograph and fluorescence detector. Flavins were released from NR1 by boiling for 5 min, and denatured protein removed by 20,000 ϫ g centrifugation for 10 min. FMN and FAD were detected by fluorescence (excitation, 450 nm; emission, 520 nm) following isocratic separation (10 mM sodium acetate, pH 6.0, methanol; v/v ratio 78:22) over a 25-cm Spherisorb ODS-2 5-m column. Authentic FMN and FAD standards purchased from Sigma were used as control. Both were over 95% pure, as judged by HPLC analysis. Absorption spectra were obtained using a Shimadzu 160 UV spectrophotometer.
Northern Blot-Human Multiple Tissue Northern blots (CLON-TECH) were hybridized with a 521-base pair cDNA fragment generated by SacI/SmaI restriction endonuclease digestion of pNR1. This probe was radiolabeled by incorporation of [ 32 P]dCTP (RadPrime DNA labeling system, Life Technologies, Inc.) and purified using a Chroma Spin ϩ TE-30 column (CLONTECH). After a 1-h prehybridization, hybridization was carried out for 1 h using ExpressHyb buffer (CLONTECH) at 68°C. The membrane was washed twice in 2ϫ SSC, 0.05% sodium dodecyl sulfate for 20 min, twice in 0.1ϫ SSC, 0.05% v/v sodium dodecyl sulfate at 50°C for 20 min, and exposed to x-ray film at Ϫ70°C with two intensifying screens.
Chromosomal Location-The full-length 2.6-kilobase NR1 cDNA segment was digested from pSPORT with EcoRI and HindIII and used as a probe in fluorescence in situ hybridization mapping (FISH). Standard cytogenetic techniques were used to prepare fixed normal lymphocyte slides. The probe was labeled with Spectrum Red dUTP using a nick translation kit (both Vysis, Downers Grove, IL) using the following kit protocol modifications. Slides were pretreated four times for 2 min in 2ϫ SSC, pH 7.0, and dehydrated through 70%, 85%, and 100% ethanol (2 min each) prior to air drying. DNA was dried at 37°C for 15 min, and an extra posthybridization wash of 1ϫ SSC, 0.3% v/v Nonidet P-40 at 73°C for 2 min was added between the other washes. Chromosomes were visualized using an Olympus BX60 fluorescent microscope fitted with a cooled CCD camera and using Vysis QUIPS image analysis software.
Enzyme Assays-Potassium ferricyanide and cytochrome c reduction was measured as described for cytochrome P-450 reductase (29). The reduction of doxorubicin and menadione was carried out in 50 mM Tris-HCl (pH 7.5), 1 mM NADPH, and various substrate concentrations at 37°C. The total incubation volume was 500 l. Reactions were initiated by the addition of 10 g of enzyme. The oxidation of NADPH was then monitored at 340 nm using a Shimadzu UV 2000 spectrophotometer. Final doxorubicin concentrations ranged from 20 to100 M and menadione from 10 to 22.5 M. Control reactions were carried out in the absence of active enzyme.

Molecular Cloning of a cDNA Encoding a Dual Flavin
Reductase-A DNA fragment was identified, which contained an open reading frame with significant homology to human cytochrome P-450 reductase, following an extensive data base search of EST data base libraries. The cDNA insert for the EST was 2452 nucleotides in length (excluding the poly(A) tail) and contained the complete coding sequence for a putative cytochrome P-450 reductase-like enzyme (Fig. 1). The initiation codon is predicted to be the first in-frame methionine residue based on sequence alignment with human cytochrome P-450 reductase and is preceded by several nucleotides bearing homology to the Kozak initiation sequence consensus (30). There are also in-frame protein stop codons upstream of the predicted start site, which place this ATG codon in good context for the initiation of translation. A 1791-nucleotide sequence encodes a 597-amino acid residue polypeptide, NR1 (Fig. 1), with a predicted molecular mass of 66,700 Da.
Comparative alignment of the identified amino acid sequence with human P-450 reductase and other FMN-and FAD-containing human flavoproteins shows sequence similarities ranging from 41% for NOS II and methionine synthase transferase to 44% for cytochrome P-450 reductase (Fig. 2). The cytochrome P-450 reductase subfamily of enzymes contain distinctive amino-terminal FMN-binding and carboxyl-terminal FAD-and NADPH-binding domains, which are aligned for efficient electron transfer by a connecting domain (17). As shown in Fig. 2, a similar domain arrangement is found in NR1 with particularly strong sequence conservation in the regions shown to be involved in FMN, FAD, and NADPH cofactor binding. The major difference in domain organization is associated with the extreme amino-terminal region. Cytochrome P-450 reductase contains a hydrophobic 60-amino acid aminoterminal anchor domain, which is involved in tethering the molecule to the endoplasmic reticulum (9,17). This domain is absent in NR1, implying a different cellular location for the enzyme.
Chromosomal Location-The 2.5-kb NR1 cDNA was used to probe a spread of human metaphase chromosomes. The gene was localized close to the telomere on the short arm of chromosome 9q34.3 by FISH analysis (Fig. 3). An 80-nucleotide gene sequence identical to NR1 cDNA has been identified in random mapping studies of chromosome 9 (31), thus confirming the chromosomal localization.
Production of Active NR1 in Insect Cells-In order to determine whether the NR1 cDNA codes for a biologically active enzyme, we expressed the cloned cDNA using a baculovirus system. The full-length NR1 coding sequence was subcloned into pFastBac downstream of the very late polyhedron promoter, and the cDNA fused with a 6-histidine-tagged sequence at the amino terminus to facilitate affinity purification by nickel-agarose chromatography. Recombinant baculovirus vNR1 was generated by homologous recombination with Bacmid DNA. NR1 was detectable by polyacrylamide gel electrophoresis at around 3 days after infection, at which time cells were harvested for protein purification. Approximately 20% of recombinant enzyme remained in the soluble fraction following cell lysis and 100 ϫ g centrifugation. Recombinant NR1 was purified by affinity purification over nickel-agarose and ADP-Sepharose to a final purity of over 90% as judged by SDSpolyacrylamide gel electrophoresis (Fig. 4A). The final yield of purified NR1 was approximately 1.5 mg/liter suspension culture.
Flavin Analysis and UV-visible Spectral Analysis of NR1-Purified NR1 was yellow, indicating the presence of flavin, and bound to 2Ј,5Ј-ADP-Sepharose, indicating the presence of an NADPH-binding domain. HPLC analysis of heat-denatured enzyme determined that it released two fluorophores whose retention times exactly matched those of authentic FMN and FAD (Fig. 4B). There were 1.2 and 1.1 mol each, respectively, of FMN and FAD bound per mole of enzyme. NR1 exhibited a UV-visible spectrum similar to cytochrome P-450 reductase (32), possessing absorbance maxima at 380, 460, and 626 nm (Fig. 4C). Like cytochrome P-450 reductase, the addition of NADPH under aerobic conditions caused a decrease in the absorbance at 380 and 460 nm, and an absorbance increase at 580 nm. Furthermore, the UV-visible spectrum of NR1 reduced with NADPH was stable over a 24-h period, which is consistent with the reduction of the flavin co-factors by NADPH and the production of an air-stable semiquinone form (10,32,46).
The above data showed that NR1 is a flavoenzyme that binds both FMN and FAD cofactors as predicted from the amino acid sequence. Furthermore, the spectral changes associated with the addition of NADPH indicate that electrons are transferred from NADPH to FAD and FMN, which indicates that NR1 follows the same pattern of electron transfer as in other dual flavin enzymes.
Biological Activity-The cytochrome P-450 reductase family of flavoenzymes are generally capable of reducing the hemoprotein cytochrome c, which thus serves as a model substrate for the comparative analysis of enzyme activity and electron transfer. Cytochrome c reducing activity was maximal when it occurred using potassium phosphate concentrations of between 300 and 400 mM and the enzyme had a pH optimum of around 8.0 (data not shown). There was also no detectable enzyme activity using NADH as a reducing cofactor. The conditions for optimal enzyme activity were thus similar to those observed for cytochrome P-450 reductase (9, 10). The kinetic parameters of cytochrome c reduction were compared with human cytochrome P-450 reductase. As shown in Table I, the apparent K m value of NR1 for cytochrome c was 21 M, which was similar to cytochrome P-450 reductase (15 M). Reported K m values of mammalian P-450 reductase for cytochrome c range between 5 and 21 M (14, 49, 50). The apparent k cat value was calculated as 1.3 s Ϫ1 , which was approximately 100-fold lower than cytochrome P-450 reductase. NR1 also metabolized a range of oneelectron acceptors, including the quinone-containing compounds doxorubicin and menadione (Table II). Although all the activities measured were significantly lower than cytochrome P-450 reductase (in the range 1-4%), they were reasonably similar to activities previously measured (25) for the reductase domain of NOS III (Table II). Taken together, these results indicate that the cloned cDNA encodes an authentic NADPHdependent reductase enzyme, which is capable of catalyzing the reduction of cytochrome c and one-electron acceptors.
Expression of NR1 in Human Tissue and Cancer Cell Lines-Northern blot analysis of mRNA from human tissue and cancer cell lines indicate two main species of approximately 3 and 6 kilobases in length (Fig. 5). The 3-kilobase mRNA corresponds to the size expected for the full-length NR1 transcript. In human tissues, levels of expression were generally low, with highest levels seen in the placenta (Fig. 5A). In cancer cell lines, highest levels were found in HeLa and colonic adenoma cells followed by myeloid leukemia cells and melanoma cells ( 5B). The more prominent high molecular weight band followed a similar profile of expression but appeared to be expressed at higher levels in the tumor cell lines. The nature of the larger mRNA species is unclear. It may represent a partially spliced variant NR1 mRNA, a homologous gene sequence, or possibly a fusion protein between an NR1-related protein and another protein. This is currently under investigation through the characterization of the NR1 gene and its intron/exon organization.
Antibodies were generated against the carboxyl-terminal FAD domain of NR1 in order to identify the native form of the enzyme and to investigate its expression in cancer cell lines. These antibodies were used for Western blot analysis of whole cell extracts from a range of human cancer cell lines derived from different tissues, including ovary (PEO1), breast (MCF7), bladder (EJ9), lung (NCIH322), colon (HT29), liver (Hep G2), and cervical carcinoma (HeLa). The murine fibroblast NIH3T3 cell line was also analyzed. As shown in Fig. 6A, a ϳ62-kDa protein of similar size to that predicted from the NR1 primary sequence was detectable at similar levels in all cell lines apart from the murine NIH3T3 cells. The expression profile was significantly different to cytochrome P-450 reductase (Fig. 6B), which showed high levels of expression in HepG2 cells, low levels in MCF7 and HT-29 cell lines, and undetectable levels in the other cell lines. In the murine-derived fibroblast NIH3T3 cell line, NR1 antibodies detected two different sized polypeptides of ϳ80 and 30 kDa, respectively. Since NR1 genes have so far not been found in mice, it is unclear what relationship these polypeptides have with the human form of the enzyme, but the higher molecular mass protein could indicate the presence of the fusion protein suggested in the Northern blot analysis above. Interestingly, in the HeLa cell line, a high molecular weight protein is also observed in the Western blot consistent with the presence of the 6-kilobase transcript. Subcellular Localization-From the primary amino acid sequence data, a major difference between NR1 and cytochrome P-450 reductase is the lack of a membrane anchor at the aminoterminal end of NR1. To compare the subcellular distribution of these enzymes, crude subcellular fractionation of MCF-7 cells was carried out by differential centrifugation and NR1 and cytochrome P-450 reductase identified in different fractions by Western blotting (Fig. 7). NR1 was detectable primarily in the 10,000 ϫ g and 100,000 ϫ g supernatant fractions, indicating that the enzyme is associated with the cytoplasmic fraction. There was some signal detectable in the nuclear pellet fraction; thus, possible targeting of NR1 to the nucleus cannot be ruled out. By contrast, cytochrome P-450 reductase was found predominantly in nuclear pellet and microsomal membrane fractions, consistent with its localization to the endoplasmic reticulum. These results indicate that the subcellular localization of NR1 differs from microsomal cytochrome P-450 reductase and is found associated with the cytosolic fraction.

DISCUSSION
We have cloned and characterized a novel dual flavin reductase, NR1, which represents a new member of the FNR family of flavoenzymes. NR1 binds FMN, FAD, and NADPH co factors and shares about 44% similarity with human cytochrome P-450 reductase. Analysis of the prototypical rat cytochrome P-450 reductase crystal structure (17) highlights amino acids of potential functional significance and indicates that there may be close structural similarities. With respect to FMN binding, the isoalloxazine ring of FMN in rat cytochrome P-450 reductase is covered by the phenolic ring of Tyr-140 at the re-side and Tyr-178 on the si-side (17). In NR1 the equivalent residues are Thr-61 and Tyr-102, respectively. Thus, the aromatic residue in the Tyr-178 position, which is essential for FMN binding, is conserved while, interestingly, there is a non-aromatic substitution of Tyr-140. Site-directed mutagenesis studies have shown that such a substitution does not necessarily affect FMN binding but may reduce electron transfer efficiency (33). In cytochrome P-450 reductase, the FAD ring is stacked by the indole ring of Trp-677 while the aromatic residue Tyr-456 lies on the si-side. NR1 contains corresponding aromatic residues in Phe-384 and Trp-596. Strong similarities also exist with the rat cytochrome P-450 reductase peptide fragments 455 YYSIAS and ICVAV, which encompass FAD (17) and with other FADcontaining proteins, including ferredoxin-NADP reductase (FNR) (34).
Although there is no crystal structure yet available of an FNR family member with bound NADPH, amino acids that are involved in NADPH binding have been identified (17,34). In human cytochrome P-450 reductase, a Gly-Thr-Gly-Tyr-His-Pro sequence similar to the pyrophosphate binding consensus sequence Gly-X-Gly-X-X-(Gly/Ala) common to NADPH binding reductases (34,35) is found between residues 534 and 539 (Fig.  2). A similar sequence is found in NR1 between amino acid residues 459 and 464, and there is also strong overall homology in the peptides considered to be involved in pyrophosphate and NADPH adenine binding. In addition to co-factor binding motifs, enzymes that contain both FMN and FAD contain an extra 70 -80-amino acid insertion sequence in the FAD domain (residues 253-377 in rat reductase) relative to FNR (17). A similar insertion is present in NR1, which may be responsible for controlling electron transfer between the two flavins (17).  Abbreviations are: w.c., whole cell homogenate; 10K Sup., 10,000 ϫ g supernatant; 10K Pel., 10,000 ϫ g nuclear pellet; 100K Pel., 100,000 ϫ g microsomal membrane pellet; 100K Sup., 100,000 ϫ g cytoplasmic fraction.
The overall conservation in the elements required for cofactor binding and their sequence arrangement indicate that NR1 may be structurally similar to cytochrome P-450 reductase. Like cytochrome P-450 reductase, recombinant NR1 catalyzed the reduction of cytochrome c and various one-electron accepting compounds. Overall, however, the apparent enzymatic activity was significantly lower than human cytochrome P-450 reductase. It is possible that this may be related to the non-aromatic substitution of Thr-72 in the Tyr-140 position of the FMN domain, as described above. However, since reduction of potassium ferricyanide by NR1, which occurs via the FAD redox center, was also slower than cytochrome P-450 reductase, it is also possible that amino acid sequence differences in the FAD/NADPH domain may be responsible for different rates of electron transfer. Recent studies have shown that Ser-457, Asp-675, and Cys-630 in rat cytochrome P-450 reductase interact to form a catalytic site for hydride transfer from NADPH to FAD (36). It is notable that, in NR1, Ala-549 corresponds to Cys-630 in the rat enzyme and Cys-629 in the human P-450 reductase sequence shown in Fig. 2. A similar non-conservative amino acid substitution in cytochrome P-450 reductase significantly reduced catalytic activity (36) in this enzyme, and may possibly do so in NR1 as well. A more detailed structural analysis, for example the independent expression of the FAD/ NADH domain, will provide more definitive information on the functional relationship between NR1 and cytochrome P-450 reductase.
It has recently been shown that the NOS family of enzymes play an important role in the bioactivation of anti cancer drugs via the reductase domain (25,26). Rates of reduction of the quinone-containing compounds including the anticancer drugs doxorubicin and menadione were comparable with NOS, with k cat values of the NOS III reductase domain measured at 1.45 and 2.07 s Ϫ1 , respectively (26). Thus, under appropriate physiological circumstances NR1 may also affect the metabolism of one-electron accepting compounds. In this respect, it is interesting that Western blot analysis indicated that NR1 expression was detectable at high levels in a wide range of cancer cell lines. The enzymatic factors involved in the metabolic activation of bioreductive drugs are complex and not fully understood. In solid tumor tissue, bioreductive enzyme activity is located at different subcellular locations throughout the cell (37). Key bioreductive enzymes are thought to include the cytosolic enzyme DT-diaphorase and the microsomal cytochrome P-450 reductase (37,47). However, other enzymes with novel activities may well be involved.
The biological role of NR1 is unknown. Clues as to the natural function of genes frequently come from analysis of genetic abnormalities or recurrent chromosomal breakpoints in cancer. We have mapped the gene for NR1 to the telomeric region of the long arm of chromosome 9. There are, however, comparatively few reports of constitutional chromosomal abnormalities, recurrent cancer breakpoints or single gene disorders for this region. Two cases of infants with deletion to 9q34.3 have been recently reported (38,39), which show that this deletion may be associated with a recognizable pattern of malformation associated with severe developmental delay and respiratory problems (39). Three diseases have also been localized to the region between 9q32 and 9q34. These include limb-girdle muscular dystrophy, characterized by muscle weakness and wasting (40); lethal congenital contracture syndrome, characterized by the fetal akinesia phenotype, with highly focused degeneration of motor neurons in the spinal cord (41); and amytophic lateral sclerosis, characterized by slow progressive, distal limb amyotrophy, and severe loss of motor neurons in the brain stem and spinal cord (42).
Further studies investigating tissue-specific expression and interaction with other cellular proteins will help to elucidate the normal function of the gene. Similarities with cytochrome P-450 reductase enzymes indicate that it is likely to transfer electrons from NADPH to the heme-or transition metal-containing center of an appropriate redox partner. Possible redox partners include heme-binding enzymes such as cytochrome P-450 enzymes, or possibly the cobalamin-dependent methionine synthase. Microsomal cytochrome P-450 is unlikely to be the physiological partner, since NR1 lacks an apparent membrane anchor sequence, which is an important requirement for efficient coupling with cytochrome P-450 reductase at the membrane surface (9). Furthermore, we have found that NR1 is unable to reconstitute ethoxyresorufin hydroxylase activity with CYP1A2 in vitro, or CYP 2D6 bufaralol hydroxylase activity when co-expressed in insect Sf9 cells. 2 It seems more likely, therefore, that NR1 may be involved in some other function. One possibility is in methionine synthesis. Methionine synthase is a cobalamin-dependent enzyme that catalyzes the transfer of a methyl group from CH 3 -H 4 folate to homocysteine. During catalysis, accidental build-up of the inactive cob(II)alamin state is prevented by reduction by oxidoreductases (43). Interestingly, the identity of the mammalian proteins that regulate reductive activation of methionine synthase have not been established. However, NADPH-dependent auxiliary redox proteins are known to be involved (44), and the most recent evidence suggests that cytochrome P-450 reductase and cytochrome b 5 may reactivate methionine synthase (45). Thus, NR1 may represent an alternative pathway for methionine synthase reactivation.