Introduction

Top Abstract Introduction Materials and Methods Results Discussion References

Dexrazoxane [ICRF-187; (+)-1,2-bis(3,5-dioxopiperazin-1-yl) propane; Zinecard; Fig. 1) is clinically used to reduce doxorubicin-induced cardiotoxicity (Hasinoff, 1998; Hasinoff et al., 1998). There is now considerable evidence to indicate that this toxicity may be due to iron-dependent oxygen free radical formation (Malisza and Hasinoff, 1995; Meyers, 1998) on the relatively unprotected cardiac muscle (Halliwell and Gutteridge, 1989). Dexrazoxane can be considered a prodrug analog of EDTA that is activated upon hydrolysis to its one-ring open intermediates B and C, and then to its fully rings-opened form ADR-925 (Hasinoff, 1990, 1994a,b, 1998; Hasinoff et al., 1998). Neutral dexrazoxane is permeable to cells (Dawson, 1975). ADR-925 (Fig. 1) may either remove iron from the iron-doxorubicin complex (Buss and Hasinoff, 1993) or bind free iron, thus preventing oxygen radical formation.

View larger version (17K): [in this window] [in a new window]   Fig. 1.   a, reaction scheme for the hydrolysis of dexrazoxane to metabolites B and C and its strongly metal ion-chelating form ADR-925; b, DHOase-catalyzed reversible conversion of L-dihydroorotate into N-carbamoyl-L-aspartate (carbamyl aspartate); c, structure of the DHOase sulfonamide inhibitor furosemide.

Our previous spectrophotometric and HPLC¹ studies (Hasinoff, 1994a,b) showed that under physiological conditions dexrazoxane is only slowly hydrolyzed to B and C (t_1/2 of 9.3 h at 37°C and pH 7.4) and the final hydrolysis product ADR-925 (t_1/2 of 23 h), according to the kinetic scheme shown in Fig. 1a. Given the slow rate at which hydrolysis-activation occurs in vitro, it is thus unclear how sufficient amounts of ADR-925 could be present in heart tissue to chelate iron and prevent oxygen radical damage before dexrazoxane was eliminated (t_1/2 of 4.2 ± 2.9 h in humans) (Hochster et al., 1992).

We have previously shown that B and C are rapidly formed from dexrazoxane in a primary rat hepatocyte suspension (Hasinoff et al., 1994) and in vivo using a rat model (Hasinoff and Aoyama, 1999). These results were both consistent with dexrazoxane being metabolized by dihydropyrimidine amidohydrolase EC 3.5.2.2 (DHPase). Furthermore, we have shown that pure DHPase enzymatically hydrolyzed dexrazoxane to B and C but did not enzymatically hydrolyze B and C to ADR-925 (Hasinoff et al., 1991; Hasinoff, 1993). The fact that dexrazoxane was not stoichiometrically converted into B and C in our hepatocyte suspension study (Hasinoff et al., 1994) suggested that there was a second unidentified enzyme that acted on B and C. Our previous pharmacokinetic study in the rat showed that B and C only reached relatively low steady-state levels (Hasinoff and Aoyama, 1999). Our preliminary pharmacokinetic studies in the rat (Schroeder and Hasinoff, 2001) and humans (Schroeder et al., 2002) that showed that dexrazoxane is rapidly metabolized to ADR-925 in vivo also suggested that B and C were being converted to ADR-925 enzymatically.

The first three steps of de novo synthesis of pyrimidines, uracil, and thymine are carried out by the multifunctional protein called CAD (Simmer et al., 1990; Davidson et al., 1993). In mammals, CAD is a trifunctional protein that contains glutamine-dependent carbamyl phosphate synthetase (EC 6.3.5.5), aspartate transcarbamylase (EC 2.1.3.2), and the zinc hydrolase DHOase (EC 3.5.2.3) (Davidson et al., 1981; Kelly et al., 1986; Simmer et al., 1990). DHOase catalyzes the reversible cyclization of N-carbamyl-L-aspartate to form L-5,6-dihydroorotate (Fig. 1b), which is the third reaction in the de novo pyrimidine biosynthetic pathway (Carrey, 1993; Evans et al., 1993).

In our previous studies, DHPase was identified as an enzyme that would likely hydrolyze the ring opening of dexrazoxane to B and C, given the structural similarity of dexrazoxane to dihydrouracil and dihydrothymine (Hasinoff et al., 1991). Likewise, given the structural similarity of B and C to dihydroorotate (Fig. 1b), the endogenous substrate of DHOase, it was decided to test the hypothesis that DHOase hydrolyzed B and C to ADR-925, and thus, that DHOase might be responsible for ultimately converting the dexrazoxane metabolites to ADR-925.

	Materials and Methods

Top Abstract Introduction Materials and Methods Results Discussion References

Materials. Dexrazoxane hydrochloride and ADR-925 were gifts from Adria Laboratories (Columbus, OH) and were used as supplied. The dexrazoxane was assayed by HPLC and was found to contain 0.98, 0.41, and 0.08 mol% B, C, and ADR-925, respectively. HPLC-grade methanol was from Fisher Scientific (Nepean, ON, Canada). Tris, Chelex resin, 5-aminoorotic acid, furosemide, L-dihydroorotic acid, and 1-octanesulfonic acid were from Sigma-Aldrich (St. Louis, MO). 4-Chlorobenzenesulfonamide was from Aldrich Chemical Co. (Milwaukee, WI).

Preparation of Purified Recombinant DHOase. Six-histidine-tagged CAD protein (DHOase) was purified from transfected hamster cells as described previously (Qiu and Davidson, 2000).

Preparation and Separation of B and C. Microgram quantities of B and C were prepared by hydrolyzing 5 mg/ml dexrazoxane with NaOH (40 µl/ml of 1 M NaOH) at 25°C for 40 min and quenching the reaction with HCl (45 µl/ml of 1 M HCl) to pH 3 as described previously (Hasinoff, 1994a). Under these conditions a mixture of dexrazoxane, B, C, and ADR-925 is produced. Dexrazoxane was efficiently removed from the reaction mixture by loading 500 µl of the mixture on a Sep-Pak Plus C₁₈ cartridge (Waters, Mississauga, ON, Canada) and eluting with 2% (v/v) methanol at a flow rate of 1 ml/min. Although dexrazoxane was highly retained on the cartridge, B, C, and ADR-925 eluted together and were collected at elution volumes between 1.5 and 2.5 ml. HPLC analysis confirmed that dexrazoxane was not detectable in this fraction. This 1-ml fraction, pH 6, was loaded on three Sep-Pak Accell Plus QMA (Waters) ion exchange cartridges connected in series and eluted with 2% (v/v) methanol at a flow rate of 5 ml/min. Fractions containing B were collected at elution volumes between 3 and 4.5 ml, and those containing C between 5 and 9 ml. The B fraction contained less than 0.1 mol% and 0.01 mol% of C and ADR-925, respectively. The C fraction contained less than 0.1 mol% B and 0.05 mol% of ADR-925, respectively. These fractions were brought to pH 2 with 5 M HCl and evaporated to dryness under a stream of nitrogen, stored at 80°C, and reconstituted in water just before use. Neither of these fractions contained detectable amounts of dexrazoxane (<0.001 mol%). Typical yields of B and C were 10 and 6 µg, respectively, which represent about 1% of the starting dexrazoxane.

Kinetics of DHOase-Catalyzed Hydrolysis of B, C, and Dihydroorotate. The DHOase-catalyzed hydrolysis of B, C, and dihydroorotate was generally determined by measuring the decrease of substrate concentration as a function of time to obtain the initial velocities (v). The much lower molar absorptivity of ADR-925 (Hasinoff, 1990) compared with B and C prevented its routine use in the enzyme kinetic assays. The 60-µl reaction mixture contained Chelex-treated 10 mM Tris buffer, pH 7.4; 2 µg/ml DHOase; and B, C, or dihydroorotic acid at 15°C. DHOase [198 µg/ml in 30% (v/v) dimethyl sulfoxide, 5% (w/v) glycerol, 12 mM Tris buffer, pH 7.9, 0.3 M NaCl, 0.6 M imidazole, and 1 mM 1,4-dithiothreitol] was thawed for 1 min at 37°C and added to the reaction mixture to give a final DHOase concentration of 2 µg/ml. A reaction temperature of 15°C was found to greatly minimize the background hydrolysis of B and C while retaining good DHOase activity. Under these conditions, nonenzymatic hydrolysis of either B or C was not detectable at 45 min, and the substrate-concentration plots were linear to at least 1 h. When inhibitors were used either 1 mM 5-aminoorotic acid, furosemide, or 4-chlorobenzenesulfonamide was incubated with DHOase in the reaction buffer for 1 min before addition of C. After incubation periods of 0, 10, 20, 30, and 45 min post-DHOase addition, 10-µl aliquots were removed and added to 25 µl of 3 mM HCl, pH 2, and stored at 80°C to stop the reaction and prevent further hydrolysis of B or C (Hasinoff, 1994a). The initial velocities for the first 10% (or less) of the reaction for the decrease of B, C, or dihydroorotate were calculated from a linear least-squares fit of five substrate concentration-time data points.

HPLC Analysis of B, C, ADR-925, and Dihydroorotate. The HPLC analysis of B and C using an ion-pair reagent with the reversed phase C₁₈ column (detection wavelength 205 nm) has been described previously (Hasinoff, 1993, 1994a,b; Hasinoff et al., 1994; Hasinoff and Aoyama, 1999). Duplicate determinations were carried out on each sample. The calibration plots for ADR-925 (20-140 µM, n = 7), B, and C (50 µM-10 mM, n = 8) were linear (r² = 0.995, 0.999, and 0.998, respectively). ADR-925 was determined separately (n = 2) under isocratic conditions (500 µM Na₂EDTA/2 mM octanesulfonic acid, pH 3.5, 1 ml/min). After ADR-925 eluted (t_r of 3.5 min) the column was washed with 500 µM Na₂EDTA/methanol (20:80 v/v) for 20 min followed by reequilibration with the initial mobile phase for 25 min.

Molecular Modeling. Molecular modeling, based on the MM2 Allinger algorithm (Burkert and Allinger, 1982) was carried out with PCModel version 6 (Serena Software, Bloomington, IN) on a PC-compatible computer.

	Results

Top Abstract Introduction Materials and Methods Results Discussion References

Kinetics of DHOase-Catalyzed Hydrolysis of Dihydroorotate, B, and C. The dependence of the initial velocity on the concentration of dihydroorotate, B, and C are shown in Fig. 2 and indicate that DHOase catalyzes the ring-opening reaction of B and C. The initial velocities were fit to Michaelis-Menten kinetics by nonlinear least-squares analysis (SigmaPlot; Jandel Scientific, San Rafael, CA) of the data in the following equation:

(1)

The specific V_max and K_m values are given in Table 1.

View larger version (15K): [in this window] [in a new window]   Fig. 2.   a, initial velocities for the rate of loss of dihydroorotate by the action of DHOase (); b, initial velocities for the rate of loss of B () and C () by the action of DHOase. The reactions were carried out in Tris buffer (10 mM, pH 7.4, 15°C) in the presence of 2 µg/ml DHOase and followed by HPLC. The solid lines are obtained from nonlinear least-squares fits of the data to the Michaelis-Menten equation. The best fit Vmax values for dihydroorotate, B, and C were 0.82 ± 0.13, 8.7 ± 1.4, and 22 ± 8 µmol·min1, respectively. The best fit Km values for dihydroorotate, B, and C, respectively, were 20.0 ± 5.8, 4,800 ± 1,700, and 11,000 ± 6,300 µM, respectively.

Effect of Inhibitors on DHOase-Catalyzed Hydrolysis of C. Given that DHOase is only one of the three activities of the multifunctional CAD enzyme, the kinetics of DHOase-catalyzed hydrolysis of C was determined in the presence of specific inhibitors of DHOase to determine whether the hydrolysis of B and C was due to the DHOase domain. 5-Aminoorotic acid is an inhibitor (K_i of 6 µM) of mammalian DHOase-catalyzed hydrolysis of dihydroorotate, although 4-nitrobenzenesulfonamide is not (Christopherson and Jones, 1980). 4-Chlorobenzenesulfonamide and 4-nitrobenzenesulfonamide are, however, noncompetitive inhibitors (K_i of 200 and 1100 µM, respectively) of bacterial dihydroorotase (Pradhan and Sander, 1973). Furosemide (Fig. 1) was included because it is a sulfonamide with a carboxylate group. As shown in Fig. 3, 1 mM 4-chlorobenzenesulfonamide did not inhibit DHOase. However, 1 mM 5-aminoorotic acid and furosemide inhibited DHOase by 91 and 80%, respectively.

View larger version (21K): [in this window] [in a new window]   Fig. 3.   Initial velocity plots for the DHOase-catalyzed hydrolysis of metabolite C without DHOase () and with DHOase in the absence () and the presence of 1 mM 5-aminoorotic acid (), furosemide (), and 4-chlorobenzenesulfonamide (). The reaction was carried out in Tris buffer (10 mM, pH 7.4, 15°C) in the presence of 2 µg/ml DHOase and followed by HPLC. The straight lines are least-squares calculated. 5-Aminoorotic acid and furosemide inhibited DHOase by 91 and 80%, respectively, whereas 4-chlorobenzenesulfonamide was not inhibitory.

Molecular Modeling of C in the DHOase Active Site. Metabolite C was modeled into the X-ray crystal structure-determined active site of Escherichia coli DHOase (Thoden et al., 2001) to determine what movement of the Arg 20 side chain would be required to fit C into the active site. The Arg 20 forms a salt bridge with the carboxylate of dihydroorotate (Thoden et al., 2001). The modeling was done with a number of distance constraints. The distances from the carbonyl carbon that undergoes hydroxide attack to the binuclear zinc centers and its bridging hydroxide were fixed at those determined in the X-ray structure of dihydroorotate bound to DHOase (Thoden et al., 2001). Also the distance between the carboxylate group of C and the guanidinium group and the two carbon atoms at each extreme of the Arg 20 side chain were also fixed to that found in the X-ray structure. The minimized structure of C in the DHOase active site is shown in Fig. 4. The distance from the carbonyl carbon in C undergoing hydroxide attack to the CZ carbon of the guanidinium group increased by 2.2 Å from 7.6 Å determined in the X-ray structure of dihydroorotate bound to DHOase (Thoden et al., 2001).

View larger version (16K): [in this window] [in a new window]   Fig. 4.   Ball and stick structure of dihydroorotate in the active site of DHOase (top). The structure was obtained from the X-ray crystal determination of E. coli DHOase (Thoden et al., 2001). Only dihydroorotate, Arg 20, the binuclear zinc atoms, and their bridging hydroxide are shown. The carboxylate group of dihydroorotate forms a salt bridge with the guanidinium group of Arg 20. Ball and stick structure of C in the active site of DHOase obtained by molecular modeling (bottom). Only the heteroatoms are labeled.

	Discussion

Top Abstract Introduction Materials and Methods Results Discussion References

This report is the first that DHOase might be involved in the metabolism of any drug. The specific V_max value of 0.41 µmol·min¹·mg¹ (10 mM Tris buffer, pH 7.4, 15°C) for the hydrolysis of dihydroorotate by the recombinant 6-histidine-tagged hamster DHOase (Qiu and Davidson, 2000) used in this study is smaller than the recombinant hamster DHOase specific V_max values of 1.2 to 2.1 µmol·min¹·mg¹ (50 mM HEPES buffer, pH 7.4, 37°C, 100 µM ZnCl₂) (Williams et al., 1995; Huang et al., 1999) and 1.2 µmol·min¹·mg¹ (100 mM Tris buffer, pH 7.4, 37°C, 25 mM MgCl₂, 0.1 M KCl, 3.3 mM glutamine, 15 mM L-aspartate, 1,4-dithiothreitol, and 7.5% dimethyl sulfoxide) (Kelly et al., 1986). However, given the different reaction conditions (buffer, presence of activating ZnCl₂, and temperature) and that a histidine-tagged DHOase was used in this study, the agreement is reasonable. The K_m of 20 µM for dihydroorotate determined in this study is in the range of K_m values of 4 to 20 µM as observed previously (Kelly et al., 1986; Christopherson et al., 1989; Williams et al., 1995; Huang et al., 1999).

The results of Table 1 show that both B and C are good substrates for DHOase. The specific V_max values for B and C are 11- and 27-fold higher than the endogenous substrate dihydroorotate. The reason for the high V_max for B and C probably lies with the fact that the imide bond in B and C is much more easily hydrolyzed than is the imide bond in dihydroorotate. We previously showed that B and C underwent nonenzymatic hydrolysis (pH 7.4, 37°C) with t_1/2 values of 17 and 8.4 h, respectively (Hasinoff, 1994a), whereas dihydroorotate is very stable under these conditions. The K_m values for both B and C are, however, 240- and 550-fold larger than for dihydroorotate. This result suggests that neither B nor C is a good fit in the active site of DHOase. A comparison of the V_max/K_m values of Table 1 indicates that under nonsaturating conditions ([S] < K_m) dihydroorotate is still a 23-fold better substrate than either B or C. Given that the V_max/K_m values for B and C are similar, under nonsaturating conditions DHOase would act on B and C at about the same rate. The high K_m values for B and C may be partially compensated for by the high V_max values, thus providing a role for DHOase in the metabolism of dexrazoxane.

The X-ray structure of homodimeric bacterial E. coli DHOase with dihydroorotate bound has recently been determined (Thoden et al., 2001). Each subunit contains a binuclear zinc center, and dihydroorotate is bound to one subunit with the negatively charged carboxylate group of dihydroorotate, forming a salt bridge with the positively charged guanidinium group of Arg 20. Assuming mammalian DHOase has a similar structure, the positively charged active site is the reason that B and C with their negatively charged carboxylate groups are substrates for DHOase, but neutral dexrazoxane is not. A mechanism is proposed in which a bridging hydroxide group attacks the re-face of dihydroorotate (Thoden et al., 2001). Given the structural similarity of dihydroorotate to B and C, a similar mechanism is likely. Although the sequence identity homology of the hamster DHOase domain of CAD with that of E. coli DHOase is only 20%, there are clusters of highly conserved amino acids (Simmer et al., 1990). The Arg 20 is, however, conserved in human DHOase (Thoden et al., 2001). The presence of the guanidinium salt bridge had been predicted previously based on the homology of DHOase with carbonic anhydrase II (Williams et al., 1995).

The results of the molecular modeling shown in Fig. 4 indicated that C can be accommodated in the DHOase active site with a movement of the Arg 20 side chain of 2.2 Å. Although this distance is not large, this and other interactions in the binding pocket are, nonetheless, sufficient to increase the K_m value for C to 11,000 µM from 20 µM for dihydroorotate (Table 1). It should be noted the modeling was done with bacterial DHOase and may not be valid for the DHOase domain of the CAD enzyme.

The ability of furosemide, but not 4-chlorobenzenesulfonamide, to inhibit DHOase is probably due to the presence of a negatively charged carboxylate group on furosemide (Fig. 1) allowing for stronger binding to the positively charged active site of DHOase (Williams et al., 1995; Thoden et al., 2001). Furosemide is a widely used diuretic in the treatment of congestive heart failure. Given that only micromolar peak plasma levels of furosemide are achieved after a typical 40-mg dose (Straughn et al., 1986), it is unlikely that furosemide inhibition of DHOase has any significant pharmacological effects in vivo.

DHOase is present in a variety of tissues, including the heart, liver, and kidney (Kennedy, 1974) and in erythrocytes and leukocytes (Smith and Baker, 1959). The level of DHOase activity in heart homogenate is 23% of that found in the liver (Kennedy, 1974). The presence of DHOase in the heart suggests that B and C may be enzymatically hydrolyzed by DHOase to ADR-925 in the heart tissue. Hydrolysis would likely be occurring in other tissues and in the blood as well. The presence of DHOase in the heart, in particular, may explain the cardioprotective effect of dexrazoxane. The ADR-925 thus formed may either remove iron from the iron-doxorubicin complex (Buss and Hasinoff, 1993), or bind free iron, thus preventing oxygen radical formation.

The rapid appearance of ADR-925 in our preliminary pharmacokinetic studies in the rat (Schroeder and Hasinoff, 2001) and humans (Schroeder et al., 2002) is consistent with a DHOase-catalyzed conversion of B and C to ADR-925. We previously showed that B and C were present at relatively low steady-state levels in a rat pharmacokinetic study (Hasinoff and Aoyama, 1999). This result is consistent with the subsequent metabolism of B and C to ADR-925 by DHOase. We also previously showed that the zinc hydrolase DHPase, which is present in the liver and kidneys, but not in the heart (Hasinoff et al., 1991), converted dexrazoxane to B and C, but did not convert these intermediates to ADR-925 (Hasinoff, 1993). In preliminary studies we also showed that B and C were permeable enough to be taken up by attached neonatal rat myocytes and dequench an intracellularly trapped iron-calcein complex (Hasinoff et al., 2002). Thus, DHPase and DHOase acting in succession on the parent drug and its metabolites B and C provide a mechanism by which dexrazoxane may exert its cardioprotective effects.