Introduction

Top Abstract Introduction Materials & Methods Results Discussion References

Tirilazad mesylate (Freedox) is the first in a series of novel compounds that have been demonstrated to be potent inhibitors of membrane lipid peroxidation in vitro (Braughler et al., 1989). Tirilazad has been demonstrated to be efficacious in animal models of neuronal damage subsequent to head trauma (Hall et al., 1988a), subarachnoid hemorrhage (Kanamura et al., 1990), spinal cord injury (Anderson et al., 1988) and ischemic stroke (Hall et al., 1988b). As a consequence, tirilazad mesylate has the potential to improve the clinical prognosis in conditions where lipid peroxidation occurs secondarily to traumatic tissue injury. Moreover, tirilazad mesylate has been shown to decrease mortality in male subarachnoid hemorrhage patients (Schmiedek, 1994).

Studies conducted to elucidate the basic pharmacokinetics and disposition characteristics of tirilazad have indicated that tirilazad is highly bound to protein in serum (Bombardt et al., 1994) and eliminated by hepatic clearance, where clearance approaches liver blood flow (Stryd et al., 1992). In Phase I studies, 12% of the dose of tirilazad was excreted in urine, whereas the remainder of the dose was observed in bile and feces, primarily as metabolites (Fleishaker et al., 1993). Previous in vitro metabolism studies have demonstrated a significant role for cytochrome P450 3A4 in the metabolic disposition of tirilazad (fig. 1) (Wienkers et al., 1996). These observations are consistent with extensive hepatic extraction of tirilazad and elimination of metabolites via biliary excretion. In addition to oxidative pathways associated with tirilazad clearance, a major circulating metabolite of tirilazad in humans arises by metabolic reduction of the steroid A-ring (Fleishaker et al., 1994). A similar pathway was demonstrated to be the primary means of tirilazad elimination in rats and was mediated by microsomal 5-reductase (Wienkers et al., 1995).

View larger version (16K): [in this window] [in a new window]   Fig. 1.   Metabolism of tirilazad and U-89678 in human liver microsomal incubations. Heavy arrows denote major metabolites, light arrows represent minor metabolites.

The enzyme 3-oxo-5-steroid-⁴-oxidoreductase (EC 1.3.99.5; 5-reductase) is an NADPH-dependent membrane bound protein that catalyzes the reduction of ^4,5 double bonds in a variety of steroid substrates (Wilson, 1975). In humans, 5-reductase exists as two isoforms, type 1 and type 2, which share ~49% sequence homology (Thigpen et al., 1992; Russell et al., 1994). The objective of the current investigation was to characterize the non-P450 enzymatic pathways responsible for tirilazad metabolism in humans using liver microsomal preparations. In these experiments, we have demonstrated that type 1 and type 2 5-reductase contribute to the metabolism of tirilazad in vitro. Furthermore, additional in vitro studies were undertaken to establish the metabolic fate of the reduced tirilazad metabolite in humans. In a companion manuscript, the effect of finasteride, a potent inhibitor of type 1 and type 2 5-reductase, on the pharmacokinetics of tirilazad and reduced metabolites has been evaluated (Fleishaker et al., 1998).

	Materials and Methods

Top Abstract Introduction Materials & Methods Results Discussion References

Reagents

Tirilazad mesylate, and analogs of tirilazad mesylate specifically labeled with carbon-14, viz. 16-[¹⁴C-methyl]tirilazad mesylate ([¹⁴C]steroid label; specific activity 24.1 mCi/mg), and stable isotopically labeled analogs of tirilazad mesylate, [2,4,6-¹³C₃,1,3-¹⁵N₂]tirilazad mesylate (+5 label) and viz. 16-[trideuteromethyl]tirilazad mesylate (+3 label) were supplied by Drs. R. S. P. Hsi and J. A. Easter (Pharmacia & Upjohn, Kalamazoo, MI). The radiochemical purity of 16-[¹⁴C-methyl]tirilazad was >98% by HPLC analysis. Authentic standard of 5-reduced tirilazad (U-89678) was supplied by Drs. E. J. Jacobson and L. S. Stelzer (Pharmacia & Upjohn). Finasteride was a gift from Dr. AE Buhl (Pharmacia & Upjohn). Ketoconazole and NADPH were obtained from Sigma Chemical (St. Louis, MO). Human livers were acquired from the Arizona Organ Bank and the International Institute for the Advancement of Medicine (IIAM). Liver microsomal protein isolation and the specific catalytic activity of individual isoforms of P450 were determined as previously described (Wienkers et al., 1996).

Formation of Tirilazad-Reduced Metabolite

Incubations conducted for mass spectral analysis contained microsomal protein (0.3 mg), NADPH (1 mM) and 200 µM tirilazad mesylate (100 µM 16-[trideuteromethyl]tirilazad mesylate (+3 label), and 100 µM nonlabeled), in a final volume of 0.5 ml of Chelex-treated Tris buffer (100 mM, pH 7.4). Incubations were carried out for 20-min periods as described below, before analysis by PB/LC-MS.

Identification of Tirilazad Metabolite by PB/LC-MS

Analytical separation of tirilazad metabolites was accomplished using an HPLC system equipped with a PE410 pump (Perkin Elmer, Norwalk, CT), a Waters 490 MS UV detector (Waters, Milford, MA), a PE ISS-200 autosampler (Perkin Elmer, Norwalk, CT) and a Kromasil 5 mm C18 column (250 × 4.6 mm; Bodman, Aston, PA). The mobile phase consisted of solvent A (100 mM ammonium acetate) and solvent B (100% acetonitrile). A linear gradient (0.5 ml/min) was programmed from 10% to 100% B in 60 min, the final conditions were held for 40 min. PB/LC-MS (chemical ionization) was performed on a Finnigan 4021 quadrupole mass spectrometer (Finnigan, San Jose, CA) equipped with a thermal pneumatic nebulizer coupled with a momentum separator (Thermabeam, Extrel Corp., Pittsburgh, PA). The instrument was operated in positive ion chemical ionization (CI) mode using ammonia as the moderating gas. Particle beam parameters were tip temperature 180°C, expansion region 80°C, and the helium flow rate was restricted by a 360 × 75-mm fused silica capillary in the 0.02-inch nebulizer assembly, a 150 psi helium head pressure on the nebulizer produced the appropriate helium flow rate. The instrument was scanned at a rate of 2.0 sec per scan over the range 60 to 800 amu under control of a Teknivent Vector II data system (Teknivent, Maryland Heights, MO). PB/LC-MS (electron ionization) was performed on a VG AutoSpecQ tandem hybrid mass spectrometer (VG Analytical, Division of Fisons, Manchester, UK), equipped with a thermal pneumatic nebulizer coupled with a momentum separator (Thermabeam, Extrel Corp., Pittsburgh, PA). The particle beam interface conditions were identical to those described above. The electron energy was 70 eV. The accelerating voltage was 8 kV. The resolution was set to 1500 for low resolution scans. Data were acquired using VG Analytical's Opus software package version 2.0 FX.

Human Microsomal Reduction of Tirilazad

Kinetic studies. Experiments to characterize the kinetics of tirilazad metabolism to U-89678 were performed using human hepatic microsomes. Incubations contained microsomal protein (0.3 mg), NADPH (1 mM), ketoconazole (10 µM) and tirilazad (0.5, 1.0, 2.5, 5.0, 10.0, 50.0 and 100 µM) in a final volume of 0.5 ml of Chelex treated Tris buffer (100 mM, pH 7.4). Samples were preincubated for 4 min at 37°C, and reactions were initiated by the addition of NADPH. Incubation reactions were allowed to proceed for 15 min and then terminated by addition of acetonitrile (500 µl) and internal standard (50 µl), tubes were vortex mixed for 20 sec and then centrifuged at 2100 × g for 25 min at 4°C. The incubation supernatant (800 µl) was transferred to HPLC autoinjector vials, capped under an atmosphere of nitrogen and placed in a chilled autosampler tray (4°C). Metabolite formation was quantitated by HPLC/APCI-MS/MS using 50-µl aliquots.

Kinetic analysis. Initial kinetic characterization of tirilazad microsomal reduction was determined by Eadie-Hofstee plot. The kinetic parameters of tirilazad microsomal reductase were determined by modeling the formation of U-89678, using a (1/v²) weighted nonlinear regression method of least squares with the statistical analysis program SYSTAT (SYSTAT Inc., Evanston, IL) modeling to both single and multienzyme systems.

pH dependency studies. Experiments to characterize the kinetics of tirilazad metabolism to form U-89678 were performed using human hepatic microsomes. Incubations contained microsomal protein (0.3 mg), NADPH (1 mM), ketoconazole (10 µM) and tirilazad (100 µM) in a final volume of 0.5 ml of phosphate buffer (100 mM) at various pH (5.0, 6.0, 7.0, 8.0, 9.0). Samples were preincubated for 4 min at 37°C, and reactions were initiated by the addition of NADPH. Incubation reactions were allowed to proceed for 15 min and then terminated by addition of acetonitrile (500 µl). Sample workup and analysis were performed as described above.

Finasteride inhibition study. A study to characterize the inhibition kinetics of tirilazad metabolism to U-89678 was performed using human hepatic microsomes in the presence of the potent 5-reductase inhibitor finasteride. Incubations contained microsomal protein (0.3 mg), NADPH (1 mM), ketoconazole (27 µM), tirilazad (50.0 µM) and finasteride (0.0, 0.001, 0.003, 0.01, 0.03, 0.1, 0.3, 1.0, 3.0, 10.0 and 30.0 µM) in a final volume of 0.5 ml of Chelex-treated Tris buffer (100 mM, pH 7.4). Samples were preincubated for 4 min at 37°C, and reactions were initiated by the addition of NADPH. Incubation reactions were allowed to proceed for 15 min and then terminated by addition of acetonitrile (500 µl). Incubation reactions were allowed to proceed for 15 min and then terminated by addition of acetonitrile (500 µl). Sample workup and analysis was performed as described above.

Quantification of U-89678 by HPLC/APCI-MS/MS. Quantification of U-89678 was accomplished using a HPLC system designed to facilitate elimination of residual analyte from the HPLC autosampler following injection of high concentration samples. Installation of a second HPLC pump and a custom valve switching device permitted the autosampler to be switched off-line during the analytical run. During this time, protein precipitated blank plasma samples were injected by the autosampler; mobile phase flow to the analytical HPLC column was supplied by the second HPLC pump at the time of column bypass. The HPLC system consisted of a Hewlett Packard (Naperville, IL) 1050 Series pump and autosampler, a Perkin Elmer (Norwalk, CT) Series 410 pump and a custom valve switching device (The Upjohn Co., Kalamazoo, MI) containing two pneumatically assisted Valco six position valves. The first valve was used to toggle the fluid flow to the analytical column from the autosampler to the second HPLC pump, whereas the second valve was used to toggle the fluid flow from the analytical column to either the APCI source or waste. Analytical separations were performed on a Kromasil 100-5C18 column (15 cm × 4.6 mm, 5 mm) using a mobile phase consisting of methanol/isopropanol/1 M ammonium acetate pH 4.5 (90:5:5) at a flow rate of 1.0 ml/min. Under these conditions the retention time of the analyte and internal standard was 8 min. Mass spectrometric analysis was performed on a Finnigan-MAT (San Jose, CA) TSQ 700 triple quadrupole mass spectrometer directly coupled to the HPLC system via a Finnigan API source operated in the positive ion APCI mode. The protonated molecular ion of U-89678 at m/z 627 underwent collisionally activated dissociation to afford a product ion at m/z 260. In a similar manner, the protonated molecular ion of [¹³C₃,¹⁵N₂]U-89678 (IS) at m/z 632 afforded a product ion at m/z 265. Selected reaction monitoring of these transitions was used for quantification. The APCI inlet, ion optics, and MS/MS conditions were optimized by infusing a tuning solution containing 1 mg/ml U-89678 at a flow rate of 5 µl/min into the mobile phase flow path post column via a T connector. An enhancement in sensitivity was achieved by deresolving the first and third quadrupoles. However, the selectivity of the assay was not compromised by selecting the product ions at m/z 260 and 265 for the analyte and IS, respectively. Product ion spectra were obtained using argon (99.999% pure, AGA Maumee, OH) as the collision gas at a pressure of 2 mm Hg and a collision offset of 48 V. The conversion dynode and electron multiplier were set to 15 kV and 1400 V, respectively. The APCI vaporizer and capillary temperatures were 450 and 260°C, respectively. The corona was set to 5 mA and nitrogen was employed as a drying gas at a sheath pressure of 70 psi. The instrument was scanned at a rate of 0.5 sec per scan for each transition. Data were collected on a Personal DEC station 5000 computer running Ultrix version 4.2 as the operating system. The mass spectrometer was controlled using Instrument Control Language (ICL) version 7.2 and the data were processed using Interactive Chemical Information System (ICIS) version 7.0 software.

In Vitro Metabolism of U-89678

Correlation of U-89678 metabolism across the human liver bank. The relative rates of metabolism of U-89678 (60 µM) were determined across nine different human liver microsomal preparations, previously characterized for specific P450 substrate activities. Each incubation contained 0.3 mg microsomal protein, NADPH (1 mM), U-89678 (60 µM) in a final volume of 0.5 ml of Chelex-treated Tris buffer (100 mM, pH 7.4). A preincubation equilibrium period was performed for 4 min at 37°C, metabolic reactions were initiated by the addition of NADPH. Incubation reactions were allowed to proceed for 5 min and then terminated by addition of acetonitrile (500 µl), tubes were vortex mixed for 20 sec then centrifuged at 2100 × g for 25 min at 4°C. The incubation supernatant (800 µl) was transferred to HPLC autoinjector vials, capped under an atmosphere of nitrogen and placed in a chilled autosampler tray (4°C). Loss of U-89678 was quantitated by HPLC with UV detection. The consumption of starting material was defined as U-89678 hydroxylase activity.

HPLC-UV analysis of U-89678. Quantification of U-89678 was achieved using a HPLC system equipped with a PE410 pump (Perkin Elmer, Norwalk, CT), a PE ISS-200 autosampler (Perkin Elmer, Norwalk, CT) equipped with a chilled sample tray maintained at 4°C, a PE LC-235 diode array detector (Perkin Elmer, Norwalk, CT). Analytical separations were performed on a Kromasil 5 mm C18 column (250 × 4.6 mm; Bodman, Aston, PA). The mobile phase consisted of solvent A (100 mM ammonium acetate) and solvent B (acetonitrile). A linear gradient (0.5 ml/min, to mimic PB/LC-MS conditions) was programmed from 10% to 100% B in 60 min, the final conditions were held for 40 min.

Identification of U-89678 metabolite, DHT-1. Incubations conducted for mass spectral analysis contained either cDNA expressed CYP3A4 or human liver microsomes (0.3 mg), NADPH (1 mM) and (200 µM) U-89678 mesylate ([100 µM] "pyrimidine-ring"-[2,4,6-¹³C₃,1,3-¹⁵N₂]U-89678 mesylate (+5 labeled) and 100 µM (nonlabeled), in a final volume of 0.5 ml of Chelex-treated Tris buffer (100 mM, pH 7.4). Incubations were carried out for 20-min periods as described above, analysis of ensuing metabolite(s) was accomplished using the PB/LC-MS method previously described for identification of U-89678.

	Results

Top Abstract Introduction Materials & Methods Results Discussion References

Identification of U-89678

To study the reductive component of tirilazad metabolism in vitro, a human liver microsomal system devoid of CYP3A4 activity was artificially generated by coincubation of tirilazad with the potent noncompetitive inhibitor of CYP3A4, ketoconazole (Wrighton and Ring, 1994). In preliminary studies, it was determined that while ketoconazole abolished nearly all of the tirilazad hydroxylase (CYP3A4) activity it did not decrease the catalytic activity of tirilazad reductase in human microsomal incubations (fig. 2). The tirilazad reduced metabolite, U-89678, was characterized by PB/LC-MS and mass spectra were compared with spectra of tirilazad, and a stable-labeled analog obtained under EI and CI conditions. The EI mass spectrum of U-89678 displayed ion-pairs for the M^+. species at m/z 626/629, a gain of 2 amu from tirilazad (M^+. = 624; fig. 3); in addition, structurally informative ions observed at m/z 246 and m/z 315 were unchanged from the analogous fragment ions in the EI spectrum of tirilazad (Wienkers et al., 1996). This indicates that the gain of 2 amu in U-89678 occurred by reduction of the steroid moiety of tirilazad. LC-MS analysis of an authentic standard of U-89678 afforded an HPLC retention time and an EI mass spectrum that were directly comparable to the reduced metabolite (data not shown). The stereochemistry of reduction of the A-ring of tirilazad in rats had previously been characterized by NMR in our laboratory (Wienkers et al., 1995). Therefore, the structure of U-89678 was assigned as the 5-reduced tirilazad, ^4,5-dihydrotirilazad (fig. 1).

View larger version (20K): [in this window] [in a new window]   Fig. 2.   Typical HPLC radiochromatograms of metabolites formed from incubations of [14C]tirilazad mesylate (50 µM) with human liver microsomes; control (top) and coincubation with ketoconazole (27 µM) (bottom).

View larger version (20K): [in this window] [in a new window]   Fig. 3.   Particle beam LC/MS electron ionization mass spectrum obtained for U-89678. Electron ionization mass spectrum was obtained following coincubation of tirilazad mesylate with an equimolar ratio of 16-[trideuteromethyl]tirilazad in human liver microsomal preparations.

Characterization of 5-Reductase Activity

Kinetic analysis. In vitro kinetic experiments revealed that reduction of tirilazad was mediated by two enzymes as suggested by a biphasic Eadie-Hofstee plot (fig. 4). Determination of the kinetic parameters (V_max and K_M) associated with the multienzyme reduction of tirilazad in two different human liver microsomal preparations was accomplished using nonlinear least-squares regression analysis (table 1).

View larger version (16K): [in this window] [in a new window]   Fig. 4.   Effect of substrate concentration on the reduction of tirilazad by human liver microsomes. Tirilazad (0.5-100 µM) was incubated with human liver microsomes, as described in Materials and Methods. Presented are the Eadie-Hofstee and Michaelis-Menten (inset) plots for formation of U-89678 by human liver (HL17) microsomes. Each point represents the mean of duplicate determinations from a single human liver.

pH dependency and inhibition studies. Two types of experiments were performed to establish which 5-reductase enzyme(s) were responsible for the formation of U-89678. The first experiment yielded indirect evidence and involved the use of the potent 5-reductase inhibitor finasteride. Literature values for finasteride inhibitory constants (K_i) toward 5-reductase type 1 and type 2 are ~325 and ~12 nM, respectively (Weisser et al., 1994; Delos et al., 1994; Houston et al., 1987). Coincubation of varying concentrations of finasteride with tirilazad resulted in an IC₅₀ for the formation of the reduced metabolite of ~0.01 mM. These data were then compared with theoretical IC₅₀ plots, assuming the competitive interaction between finasteride and the two isoforms of 5-reductase (fig. 5). These results suggest a principle role for 5-reductase type 2 in formation of U-89678 under saturating substrate conditions. Further support for the identity of the low affinity/high capacity 5-reductase isoform responsible for tirilazad reduction was accomplished by establishing the optimal pH for tirilazad reductase activity under saturating substrate conditions. Results of the pH study revealed that greater U-89678 formation was obtained at an acidic pH (pH = 5) (fig. 6) consistent with a role for 5-reductase type 2 as the low affinity/high capacity enzyme.

View larger version (16K): [in this window] [in a new window]   Fig. 5.   IC50 plot for finasteride inhibition of tirilazad reductase activity in human liver (HL17) microsomes. The shaded circles represent inhibition curve generated for finasteride inhibition of tirilazad (60 µM) microsomal reduction. The dashed line represents a theoretical curve for the inhibition of tirilazad reduction by finasteride based on literature Ki values for the competitive 5-reductase type 2/finasteride interaction. The solid line represents a theoretical curve for the inhibition of tirilazad reduction by finasteride based on literature Ki values for the competitive 5-reductase type 1/finasteride interaction. Each experimental data point represents the mean of duplicate incubations.

View larger version (44K): [in this window] [in a new window]   Fig. 6.   pH-dependence of 5-reductase activity in human liver microsomes. Enzyme activity was measured as indicated in Materials and Methods. Each column represents the mean of quadruplicate incubations.

Correlation of U-89678 metabolism with CYP activity. U-89678 consumption was determined in microsomal preparations from nine human livers. The rates of U-89678 consumption were plotted against metabolic activities determined in microsomal preparations for the following cytochrome P450 isoforms: CYP1A2 (caffeine N-demethylation), CYP2A6 (coumarin 7-hydroxylation), CYP2C9 (tolbutamide hydroxylation), CYP2D6 (dextromethorphan O-demethylation), CYP3A4 (testosterone 6-hydroxylation). The rates of U-89678 metabolism at saturating concentrations (60 µM) displayed an 11-fold interindividual variation across nine human liver microsomal preparations. A comparison of U-89678 consumption with microsomal P450 activities revealed a strong linear correlation between tirilazad hydroxylation and 6-testosterone hydroxylation (r² = 0.869) which intercepted the y-axis close to the origin (fig. 7). In addition, metabolism of U-89678 did not correlate with the activities of the other cytochrome P450 enzymes within the human liver microsome bank (table 2).

View larger version (20K): [in this window] [in a new window]   Fig. 7.   Correlation of U-89678 (60 µM) metabolism vs. CYP3A4 activity across a bank of nine different human liver microsomal preparations. P450 3A4 activity was characterized by testosterone 6-hydroxylase activity. Each point represents the mean of duplicate determinations from a single human liver.

Characterization of U-89678 metabolism. The effect of incubating U-89678 in the presence of cDNA expressed CYP3A4 microsomes or human liver microsomes was investigated. In both enzyme systems the major microsomal metabolite of U-89678, DHT-1, appeared to arise via hydroxylation of one of the U-89678 pyrrolidine rings as indicated by mass shifts in mass spectral fragment ions that are characteristic for the heterocyclic domain of U-89678. The EI mass spectrum of DHT-1 was characterized by an odd-electron molecular ion (M^+.) pair at m/z = 642/647 from the [¹³C₃,¹⁵N₂]U-89678: U-89678 (1:1) and ion pairs at 624/629 that arose from the loss of one molecule of water from the molecular ion pair (fig. 8). Structurally informative ion pairs were observed at m/z 313/318 and 244/249 that were derived from cleavage between C-20 and C-21 and cleavage across the piperazine ring of the heterocyclic domain, respectively. The corresponding fragments for U-89678 observed at m/z 315 and 246 indicate that fragment ions derived from the heterocyclic domain have decreased by 2 amu, a mass shift consistent with dehydration of a carbinolamine (M^+. = 642) in the mass spectrometer to afford a pyrroline ring (fig. 8).

View larger version (19K): [in this window] [in a new window]   Fig. 8.   Particle beam LC/MS electron ionization mass spectrum obtained for the oxidized U-89678 metabolite, DHT-1. Electron ionization mass spectrum was obtained following coincubation of U-89678 with an equimolar ratio of [13C3,15N2]U-89678 mesylate in human liver microsomal preparations.

	Discussion

Top Abstract Introduction Materials & Methods Results Discussion References

In humans, tirilazad undergoes extensive biotransformation leading to the formation of a large number of multigenerational drug-related products that are observed in vitro and in vivo. As a consequence, in vitro metabolism studies have focused on the factors which modulate the metabolic clearance of the parent drug (i.e., formation of primary metabolites). The fate of tirilazad in human liver microsomes is characterized by formation of four hydroxylated metabolites and one reduced metabolite (Wienkers et al., 1996). The primary mechanism of tirilazad metabolism in human is mediated by CYP3A4 oxidation. As a consequence, metabolic reduction of tirilazad to U-89678 represents a relatively minor component of the overall metabolic clearance of tirilazad in vivo. However, U-89678 is a pharmacolgically active metabolite of tirilazad that is present in human plasma at levels ~25% (single dose) to 50% (multiple dose) of the AUC_0- of the parent drug following intravenous administration (Fleishaker et al., 1996).

The intent of the present in vitro experiments were 2-fold. First, we sought to characterize the enzymatic and kinetic basis for formation of U-89768 in vitro. Second, we wished to prospectively address potential clinical drug-drug interactions between tirilazad and modulators of activity of the reductase responsible for tirilazad clearance. Because microsomal 5-reductase generation of U-89678 is relatively minor compared with the CYP3A4 derived metabolites it was necessary to artificially produce a microsomal preparation which was functionally devoid of CYP3A4 activity. The suppression of CYP3A4 activity was accomplished by co-incubating the reactions with the potent noncompetitive CYP3A4 inhibitor ketoconazole (fig. 2). This strategy allowed the direct inspection of microsomal 5-reductase activity without the constraint of significant depletion of starting substrate.

The enzyme 3-oxo-5-steroid-⁴-oxidoreductase (EC 1.3.99.5; 5-reductase) is an NADPH-dependent membrane bound enzyme that catalyzes the reduction of ^4,5 double bonds in a variety of steroid substrates. In humans, 5-reductase exists as two isoforms designated as type 1 and type 2 (Delos et al., 1994; Andersson and Russell, 1990), is present in most tissues and catalyzes the conversion of testosterone to the more potent androgen, 5-dihydrotestosterone (Wilson, 1975; Anderson and Liao, 1968). The type 1 enzyme is characterized by a neutral to basic pH-optima and is relatively insensitive to finasteride inhibition (Thigpen et al., 1992; Jenkins et al., 1992). In contrast, the type 2 enzyme has an acidic pH-optima and is sensitive to inhibition by finasteride (Delos et al., 1994; Andersson et al., 1990). The reduction of tirilazad to U-89678 in human liver microsomes was modulated by two enzymes as evidenced by a biphasic Eadie-Hofstee plot (fig. 4) and characterized as a low affinity/high capacity enzyme and the other a high affinity/low capacity enzyme (table 1). Comparison of the theoretical contribution of each enzyme towards the formation of U-89768, based upon calculated in vitro kinetic parameters suggests that at physiologically relevant tirilazad concentrations the high affinity type 1 5-reductase (see below) is responsible for the formation of U-89678 in vivo.

Two isoforms of 5-reductase are expressed in the human liver (Eicheler et al., 1994) and appear to contribute to varying degrees to the reduction of tirilazad to U-89678 in vitro (table 1). To characterize the role of the the two kinetically distinct isoforms in tirilazad reduction and to assess the potential for pharmacokinetically based drug-drug interactions in vivo, two experimental approaches were designed to identify the high K_M 5-reductase. The first involved the determination of the IC₅₀ value for finasteride inhibition of tirilazad reductase activity using tirilazad concentrations which were saturating for both enzymes. Under saturating substrate conditions, the low affinity/high capacity enzyme, owing to its large catalytic capacity, was predicted to be theoretically responsible for >90% of the reduced metabolite formed. Therefore, comparison of observed finasteride inhibition of tirilazad reduction with the theoretical IC₅₀ plots, based on literature K_i values (assuming competitive inhibition), would indirectly suggest which of the 5-reductase isoforms was the low affinity/high capacity enzyme. This strategy resulted in an IC₅₀ plot for experimental inhibition which closely resembled the theoretical plot generated for the finasteride/5-reductase type 2 interaction (fig. 5). A small contribution of the low K_M enzyme introduces a slight shift of the IC₅₀ plot toward the right as finasteride will not as effectively compete with tirilazad for the 5-reductase type 1 active site (i.e., K_i/K_M ratio equals 0.0009 and 0.66, for 5-reductase type 2 and type 1, respectively). The second experiment to determine the identity of the low affinity/high capacity tirilazad 5-reductase exploited differences in pH-optima associated for each isoform; 5-reductase type 1 is characterized by a slightly basic pH-optima of ~8.0, while 5-reductase type 2 has been demonstrated to have an acidic pH-optima of about 5.0 (Andersson et al., 1990; Jenkins et al., 1992). Results of the pH-dependence study clearly demonstrated an acidic pH optima for U-89678 formation in human liver microsomes, consistent with a primary involvement of type 2 enzyme under saturating substrate conditions (fig. 6). This information coupled with the indirect evidence of the finasteride inhibition study suggests that 5-reductase type 2 is the low affinity/high capacity isoform that contributes to the in vitro microsomal reduction of tirilazad.

U-89678 is a pharmacologically active metabolite of tirilazad that has been detected in human plasma at levels that may represent >50% of the area under the tirilazad plasma concentration time curve on multiple dosing (Fleishaker et al., 1994). Our interest in the metabolic fate of U-89678 stems in part from an earlier clinical investigation where administration of ketoconazole resulted in a doubling of tirilazad plasma AUC and a six-fold increase in the AUC of U-89678 (Fleishaker et al., 1996). From these observations, it appeared that the clearance of both tirilazad and U-89678 were mediated by CYP3A4. Results of the present investigation indicates that U-89678 is a substrate for CYP3A4 and that oxidation occurs at one of its two pyrrolidine rings. Thus CYP3A pathway appears to be the primary in vitro means of U-89678 metabolism and is consistent with a major route of metabolism established for tirilazad (Wienkers et al., 1996). Interestingly, a major CYP3A4 mediated biotransformation of tirilazad, i.e., 6-hydroxylation, was not observed for U-89678. Lack of oxidation at this position, while maintaining susceptibility to hydroxylation in the heterocyclic domain, may reflect a change in A-ring conformation upon reduction of the 4,5 double bond (Wienkers et al., 1995), or may reflect an increased energy requirement for proton abstraction at the 6-position in the reduced metabolite as compared with the allylic proton of tirilazad. In this case, the structural/energy considerations do not dramatically affect the heterocyclic domain of the tirilazad, which may orient itself in the CYP3A4 active site independently of the steroid portion.

In conclusion, metabolic reduction of tirilazad to U-89678 is catalyzed by two kinetically distinct 5-reductase isoforms. The low affinity/high capacity/low clearance (V/K) enzyme was identified as the type 2 enzyme. Based on kinetic calculations, only one enzyme, the high affinity/low capacity type 1 5-reductase enzyme is predicted to contribute significantly to tirilazad reduction in vivo. In addition, a role for CYP3A in the metabolic clearance of U-89678 was established. Collectively, these data confirm previous in vivo observations that 5-reductase is a relatively minor pathway in the total metabolic elimination of tirilazad and that clearance of tirilazad and U-89678 is mediated primarily via a CYP3A4 dependent pathway (Fleishaker et al., 1998).