Molecular basis of the catalytic differences among DT-diaphorase of human, rat, and mouse.

DT-diaphorase (EC 1.6.99.2), also referred to as NAD(P)H:(quinone-acceptor) oxidoreductase, is involved in the reductive activation process of several cytotoxic antitumor quinones and nitrobenzenes. It has been observed in our and other laboratories that the rat enzyme is significantly more effective in activating these drugs than the human and mouse enzymes. These results indicate that the available cytotoxic drugs are better substrates for the rat enzyme and are not the most ideal prodrugs for activation by DT-diaphorase in human tumors. In this study, using site-directed mutagenesis to replace residues in the rat enzyme with the human sequences and residues in the human enzyme with the rat sequences, we have found that residue 104 (Tyr in the rat enzyme and Gln in the human and mouse enzymes) is an important residue responsible for the catalytic differences between the rat and the human (and mouse) enzymes. With an exchange of a single amino acid, the rat mutant Y104Q behaved like the wild-type human enzyme, and the human mutant Q104Y behaved like the wild-type rat enzyme in their ability to reductively activate the cytotoxic drug CB 1954 (5-(aziridin-1-yl)-2,4-dinitrobenzamide). The study also confirms the conclusion of the x-ray structural analysis of rat enzyme that residue 130 (Thr in the rat enzyme and Ala in the human and mouse enzymes) is positioned near the binding region of the nicotinamide portion of NAD(P)H. This structural information is very important for designing suitable drugs and approaches for human cancer chemotherapy mediated by DT-diaphorase.

lyzes the obligatory two-electron reduction of quinones (such as menadione, for structure, see Fig. 1) and quinonoid compounds to hydroquinones, using either NADH or NADPH as the electron donor (1,2).
During the last few years, evidence has accumulated that DT-diaphorase is involved in the reductive activation process of several cytotoxic antitumor quinones, such as mitomycins, anthracyclines, and aziridinylbenzoquinones, in cells (3)(4)(5). The enzyme can also act as a nitroreductase in that it reduces nitrobenzenes, such as CB 1954 (5-(aziridin-1-yl)-2,4-dinitrobenzamide) (6) (for structure, see Fig. 1). The enzymatic reduction of these compounds gives rise to reactive intermediates that can then undergo nucleophilic additions with DNA and other macromolecules, suggesting a possible mechanism for their cytotoxicity (7). Cells expressing a higher level of DTdiaphorase have been shown to be more sensitive to the drug treatment (e.g. Refs. 8 -10), strong support for the role of DTdiaphorase in drug activation.
While there is experimental evidence that DT-diaphorase is responsible for the reductive activation of antitumor quinones and related compounds, the molecular basis for the activation of these compounds by this enzyme is not yet clearly understood. Particularly, it has been recognized that a major difference exists in the capability of the human and rat DT-diaphorase to activate cytotoxic antitumor compounds. Human DT-diaphorase is not as active as rat DT-diaphorase in activating most, if not all, of the available antitumor quinones and nitrobenzenes. For example, the rate of the reductive activation of the indoloquinone antitumor agent EO9 (3-hydroxymethyl-5-aziridinyl-1-methyl-2-(H-indole-4,7-indione)-propenol) by the rat enzyme was found to be significantly higher than that of the human enzyme (5). Similarly, the rat enzyme reduces CB 1954 at a faster rate than the human enzyme (ratio of K cat ϭ 6.4) (6), resulting in a large difference in the effectiveness of this prodrug. Therefore, CB 1954 is a very potent antitumor agent in the rat, but is not effective against human tumors in vivo. It has also been shown that the rat enzyme is more effective than the human enzyme in reducing diaziquone and mitomycin C, two clinically used anticancer agents (3,4) and also 2,5-dimethyl-3,6-diaziridinyl-1,4-benzoquinone, streptonigrin, and mitomycin A (11). These results indicate that the available cytotoxic drugs are better substrates for the rat enzyme and are not the most ideal prodrugs for activation by DT-diaphorase in human tumors.
The mouse DT-diaphorase cDNA was cloned 3 years ago, and the mouse enzyme was successfully expressed in Escherichia coli (12). As expected, the amino acid sequence of mouse DTdiaphorase was found to be more homologous to the rat enzyme than the human enzyme. There are 17 and 37 differences, respectively, between the mouse sequence and the rat and human sequences (see Fig. 2). Thus, it was thought that the mouse enzyme would be similar to the rat enzyme in reducing the above prodrugs in a more efficient manner than the human enzyme. Unexpectedly, the mouse enzyme was found to reduce two prodrugs, CB 1954 and EO9, at a rate similar to the human, but not the rat, enzyme (13).
In order to determine the region of DT-diaphorase that is responsible for the catalytic differences, two mouse-rat chimeric enzymes were generated utilizing a conserved PstI site in both rat and mouse DT-diaphorase cDNA (13) (see Fig. 2). MR-P, a chimeric enzyme that has mouse amino-terminal and rat carboxyl-terminal segments of DT-diaphorase, has been shown to have catalytic properties resembling the rat enzyme, and RM-P, a chimeric enzyme that has rat amino-terminal and mouse carboxyl-terminal segments of DT-diaphorase, has been shown to have catalytic properties resembling the mouse enzyme (13). Based on these results, we propose that the carboxyl-terminal portion of the enzyme plays an important role in the reduction of cytotoxic drugs.
Therefore, it was thought that the catalytic differences among the DT-diaphorase of the three species is due to the amino acid residues that are within the carboxyl-terminal region (after the conserved PstI site) and identical between the human and mouse enzyme, but different from that of the rat enzyme (see Fig. 2). In this study, a series of rat and human DT-diaphorase mutants with changes in the carboxyl-terminal region were generated and analyzed. The study has revealed that residue 104 (Tyr in rat DT-diaphorase and Gln in human and mouse DT-diaphorase) is an important residue responsible for the catalytic differences between the rat, the human (and mouse) enzymes.

MATERIALS AND METHODS
Site-directed Mutagenesis-A PCR 1 -based mutagenesis method described by Nelson and Long (14) was used to generate DT-diaphorase mutant cDNAs. Desired PCR product was resolved over a 1% agarose gel and then extracted using the QIAquick Gel Extraction Kit (Qiagen Inc., Chatsworth, CA). The gel-purified PCR product was cloned into PCRII vector from the TA cloning kit (Invitrogen Co., San Diego, CA). Mutant clones were then selected by dideoxy sequencing. The resulting mutant constructs were religated into the E. coli expression vector, pKK233-2 vector (Pharmacia Biotech Inc.), through the NcoI and Hin-dIII restriction sites. The expressed mutants were purified by Affi-Gel Blue affinity chromatography following the previous published procedures (15).The purity of mutant preparations were examined by SDSpolyacrylamide gel electrophoresis by the method of Laemmli (16).
Other-NADH-menadione reductase activity was determined spectrophotometrically following the published procedure (17). In the assay, menadione was used as the electron acceptor and 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium was included to continuously reoxidize the menadiol formed. The reduction of CB 1954 by DT-diaphorase was analyzed by HPLC. DT-diaphorase was incubated with NADH (500 M) and CB 1954 at different concentrations (0.1 to 2 mM), in sodium phosphate buffer (10 mM, pH 7) at 37°C. At various times, aliquots (10 l) were injected onto a Partisphere SCX (250 ϫ 4.7 mm) HPLC column and eluted isocratically (1.5 ml/min) with 50 mM sodium phosphate containing 1% methanol. The elute was continuously monitored for absorption at 340 and 260 nm, and the spectra of the eluting components were recorded using a diode array detector (ABI 1000S). This separation system could resolve all the expected reduction products (6) and reduction of CB 1954 monitored by either the decrease in its peak area or an increase in the area of the peak corresponding to the reduction product, 5-(aziridin-1-yl)-4-hydroxylamino-2-nitrobenzamide.

RESULTS AND DISCUSSION
In this study, five rat mutants, in which the mutated residues are after the conserved PstI site, were generated. These mutants were converted to the human (mouse) sequence, i.e. Y104Q(r), T130A(r), V204I(r), S218T(r), and L238M(r). In order to prevent confusion, (r) and (h) are included in the name of each mutant to indicate whether it is a rat or human DTdiaphorase mutant, respectively. The mutants were purified to homogeneity, and the purity of the preparations was examined by SDS-polyacrylamide gel electrophoresis (results not shown). Among the five mutants, Y104Q(r) was found to reduce menadione and CB 1954 at a rate similar to that of the human (and mouse) DT-diaphorase, i.e. similar V max values (Table I). In addition, the K m value of T130A for NADH resembles that of the human (and mouse) enzyme. In order to further evaluate the role of residue 104, we prepared the human mutant Q104Y(h). While the mutant Q104Y(h) had a lower menadione reductase activity than the wild-type human DT-diaphorase, its CB 1954 reductase activity was 7-fold that of the wild-type human enzyme, identical to that of the wild-type rat enzyme (Table I). Furthermore, a human double mutant Q104Y/ A130T(h) was prepared and found to have a K m value for NADH similar to the wild-type rat enzyme. Therefore, our mutagenesis experiments have provided definitive results indicating that residue 104 of DT-diaphorase plays a critical role in reductive activation of drug CB 1954, and residue 130 is involved in NADH binding. The x-ray structure of rat DT-diaphorase reveals that the phenolic ring of Tyr-104, together with the main chain carbonyl of Trp-103, provides the main interactions with the "bottom face" of the isoalloxazine ring (the face opposite to the one that interacts with substrate and nicotinamide) of FAD (18). In addition, a water molecule hydrogen-bonded to the OH of Tyr-104 is hydrogen-bonded with the O3Ј and with a phosphate oxygen of FAD. Replacing Tyr-104 by a Gln manifests itself predominantly through effects in the positioning of the isoalloxazine ring. 2 The change in size of the side chain allows the flavin to move deeper into the protein (by approximately 0.7 Å). Such a change may modify the rate of electron transfer between FAD and the substrate menadione. It is not yet known how the drug CB 1954 binds to the active site of the enzyme. However, the results from this study indicate that the positioning of the isoalloxazine ring will have a major effect on the reduction of the prodrug CB 1954. Thr-130 of rat DT-diaphorase is on the outside of the protein in a tight turn of a long loop. Two residues immediately following this turn, Tyr-126 and Tyr-128, are involved in interactions with the nicotinamide or with the substrate. All of the kinetic effects of substituting Thr-130 with Ala are probably a result of small local modifications of the main chain conformation that result in FIG. 1. Structure of menadione and CB 1954.   FIG. 2. Amino acid sequence of DT-diaphorase from rat liver, human liver, and mouse liver. For the amino acid sequence of the human and mouse enzyme, only the residues that are different from those of the rat enzyme are shown. The residues (after the conserved PstI restriction site) that are identical between the human and mouse enzymes, but different from the rat enzyme, are boxed.
slightly altered positions of the side chains of Tyr-126 and Tyr-128. The observed effects of the mutations agree, in general, with the structural observations: change of Tyr-104 affects the V max (by affecting the relation between the flavin and the nicotinamide and the substrate), and change of Thr-130 affects NADH binding.
In addition to the mutations at positions 104 and 130 which led to major changes in the catalytic properties, V204I(r) was found to have a K m value for menadione similar to that of the wild-type mouse enzyme (Table I). The x-ray structural analysis of the rat enzyme has revealed that the adenine ring of FAD interacts strongly with Arg-201 (18). It is thought that the mutation may modify the interaction of menadione with the enzyme by a slight change in the FAD binding environment. The human double mutant Q104Y/I204V(h) was found to have a K m value for menadione similar to the wild-type rat enzyme and a K m value for NADH similar to the wild-type human enzyme. Residue 238 is situated in the carboxyl-terminal domain of the enzyme (18,19). Both the rat mutant L238M(r) and the human mutant Q104Y/M238L(h) were found to have large K m values for NADH (Table I). In addition, the V max value of L238M(r) for CB 1954 was slightly lower than that of the wild-type rat enzyme. It is thought that mutations at this position may affect the catalytic function in an indirect fashion.
Flavones were found to be potent inhibitors of DT-diaphorase (20), and the human and rat forms of the enzyme were found to be more sensitive to flavones than the mouse enzyme (12). Studies involving rat-mouse chimeric enzymes have generated results indicating that the carboxyl-terminal region of DTdiaphorase plays a critical role in the binding differences for flavones between the rat and mouse enzyme (13). Since none of the mutations in this study modified the binding affinities for flavones, it is thought that the mutated amino acid residues are not involved in the binding of flavones (results not shown).
Perhaps the most important implication of the results from our evaluation of various rat and human DT-diaphorase mutants is its potential therapeutic utility. It is now known that by affecting the position of FAD, residue 104 plays a critical role in the reductive activation of cytotoxic drugs such as CB 1954. Suitable drugs that are effectively activated by human DT-diaphorase can be developed by proper modification of the structure of existing drugs, allowing them to interact better with FAD in the human enzyme. Structure-activity studies of a series of the derivatives of CB 1954 have revealed that a more hydrophobic side chain at the C1 position improves the interaction with human DT-diaphorase. 3 In addition, the human mutant Q104Y(h) will be a good candidate for gene therapy involving DT-diaphorase. 9 and 40% of the population have been found to be homozygous and heterozygous for a P187S mutation of DT-diaphorase, respectively (21). The P187S mutation results in an enzyme with one-twentieth of the FAD binding affinity and one-eighth the activity of the wild-type human enzyme. 4 It is thought that the population with the P187S mutation might have a reduced ability to activate cytotoxic quinone drugs, and, therefore, the mutation should have a significant impact for treatment of cancer patients with the P187S mutation. It might be possible to introduce the human Q104Y(h) mutant into the cancerous tissues of these patients to increase the response to cytotoxic drug treatment. Such a possibility is currently being evaluated in our laboratory.