Introduction

Top Abstract Introduction Experimental Procedures Results Discussion References

Flavopiridol (FLAP¹; ()-cis-5,7-dihydroxy-2-(2-chlorphenyl)-8[4-(3-hydroxy-1-methyl) piperidinyl]-4H-benzopyran-4-on, NSC 649890, L86-8275, HMR 1275) is a semisynthetic flavone presently in clinical trials, which as a competitive ATP site antagonist effectively inhibits cyclin-dependent kinases (cdk1, cdk2, cdk4, cdk7) with an IC₅₀ of 0.1 to 0.3 µM causing cell cycle arrest at G₁ or G₂ (Kaur et al., 1992; Losiewicz et al., 1994; Carlson et al., 1996; De-Azevedo et al., 1996; Sedlacek et al., 1996; Sausville et al., 1999; Senderowicz and Sausville, 2000). At higher concentrations (IC₅₀ values of 10 µmol or greater), FLAP also inhibits other protein kinases, which include epidermal growth factor receptor tyrosine kinase, protein kinase C, and cAMP-dependent protein kinase A (Sedlacek et al., 1996). FLAP appears potently antiproliferative and cytotoxic in a variety of cell types, e.g., breast cancer cells (Worland et al., 1993; Carlson et al., 1996), prostate carcinoma cells (Drees et al., 1997), human lung, head, and neck cancer cells (Bible and Kaufman, 1996; Patel et al., 1998), and human tumor xenografts, respectively. FLAP also induces apoptosis via p53-independent pathways in neuronal (Park et al., 1997), lymphoid (Arguello et al., 1998; Parker et al., 1998), and esophageal cells (Schrump et al., 1998). Furthermore, in combination with paclitaxel, cytarabine, doxorubicin, topotecan, and cisplatin in A549 nonsmall cell lung carcinoma cells (Bible and Kaufman, 1997), and with mitomycin-C in MKN-74 gastric and MDA-MB-468 cells, a synergistic toxicity was observed before FLAP exposure (Schwartz et al., 1997).

Clinical phase I studies (Thomas et al., 1997; Senderowicz et al., 1998) revealed that FLAP has a half-life for the terminal phase of 11.6 h, a total clearance of 14.6 l/h/m², and an apparent volume of distribution at a steady state of 13.2 l/m². Toxicities noted at FLAP doses less than 35 mg/m²/day × 3 were severe secretory diarrhea treatable with loperamide and cholestyramine.

Because at least 30% of the patients showed a postinfusional increase in FLAP plasma concentrations 3 to 24 h after the end of the infusion, reabsorption of FLAP in its conjugated form from the gastrointestinal tract was noted (Lush et al., 1997). Studies in our laboratory indeed showed that FLAP undergoes biotransformation in rat liver to two monoglucuronides mainly excreted into bile (Jäger et al., 1998). A very recent clinical phase II trial (Innocenti et al., 2000) also confirmed glucuronidation of FLAP in 22 metastatic renal cancer patients treated with 50 mg/m²/day of FLAP administered every 2 weeks as a 72-h continuous infusion. FLAP and one glucuronide reached plasma concentrations after 47 h of 297 to 566 and 196 to 553 nM, respectively, indicating extensive biotransformation of FLAP. Furthermore, 13 patients experiencing diarrhea had a lower metabolic ratio (FLAP glucuronide:FLAP) than patients without diarrhea, indicating that systemic glucuronidation of FLAP is inversely associated with the risk of developing diarrhea. Glucuronidation of FLAP may also explain the increased level of bilirubin in 22% of patients receiving FLAP at doses of 78 mg/m²/day × 3 (Senderowicz et al., 1998). This suggests a possible interaction on the level of the enzymatic pathway between FLAP and the extensively glucuronidated bilirubin. Therefore, it was the aim of the present study to investigate the metabolism of FLAP in rat and human liver microsomes and to identify the chemical structures of the isolated glucuronides by liquid chromatography/mass spectrometry (LC/MS) and NMR. Furthermore, the UDP-glucuronosyltransferases (UGTs) responsible for FLAP glucuronidation should be identified and a possible interaction with bilirubin determined.

	Experimental Procedures

Top Abstract Introduction Experimental Procedures Results Discussion References

Materials. FLAP was a gift from Dr. Sherman Stinson, National Cancer Institute, Bethesda, MD. Propofol was kindly provided by the National Institute for Quality Control of Drugs, Vienna, Austria. UGT1A1, 1A3, 1A6, 1A7, 1A10, and 2B7 were obtained from Panvera (Madison, WI). UGT1A4, 1A9, and 2B15 were purchased from GENTEST (Woburn, MA). Bilirubin, uridine 5'-diphosphoglucuronic acid (UDPGA) and -glucuronidase were obtained from Sigma (Munich, Germany). All other chemicals and solvents used were commercially available and of analytical grade.

Microsomal Preparations. Human liver samples (HL11, HL18, and HL19) were obtained from patients undergoing partial hepatectomy with their written informed consent (Department of General Surgery, University of Innsbruck, Innsbruck, Austria). The rat liver samples were from male Wistar rats (235-265 g) raised at the Institut für Versuchstierzucht und-haltung, University of Vienna, Himberg, Austria. Livers were homogenized in 0.1 M phosphate buffer (pH 7.4), and liver microsomes were prepared using standard procedures (Jäger et al., 1999). Protein concentration was determined by the method of Lowry et al. (1951), using bovine serum albumin as the standard. UGT activity (rat, 19.41 ± 4.66 mU/mg of protein; human, 10.76 ± 3.81 mU/mg of protein) was determined according to a standard protocol used in our lab (Jäger et al., 1997).

Metabolism of FLAP by Rat and Human Liver Microsomes. Rat and human liver microsomes (0.2 and 2 mg/ml, respectively) and digitonin (0.5 mg/mg of protein) were preincubated on ice for 30 min. Afterward, magnesium chloride (10 mM), D-saccharic acid-1,4-lactone (5 mM), and FLAP (final concentration, 0.1-3 mM) in DMSO were added to a 100 mM potassium phosphate buffer (pH 7.4) in a total volume of 250 µl. The reaction was started by the addition of 4 mM UDPGA, and the metabolism of FLAP was assessed by the formation of metabolite M1 and M2 after a 120-min incubation at 37°C. Control experiments in the absence of UDPGA were run in parallel. High-performance liquid chromatography (HPLC) analysis was performed with minor modifications as previously described (Jäger et al., 1998). Briefly, after the reaction was stopped by the addition of 500 µl of methanol, the samples were centrifuged (5000g for 5 min at 4°C). 100 µl of the supernatant was injected onto the HPLC column. Chromatographic separations were performed on a Hypersil BDS-C₁₈ column (5 µm, 250- × 3.6-mm i.d.) preceded by a Hypersil BDS-C₁₈ precolumn (5 µm, 10- × 3.6-mm i.d.), at a column temperature of 35°C using a mobile phase consisting of a continuous gradient mixed from methanol and 10 mM ammonium acetate/acetic acid buffer, pH 5.0 (Hypersil, Astmoor, England). As standards were not available, quantification of metabolite concentrations were monitored at 264 nm based on the assumption that the unknown metabolites had a similar molar extinction coefficient to FLAP. Previous experiments in our lab indeed showed only slight differences in the absorption spectra of the metabolites. While M2 exhibits the max at 262.4 nm, almost identical with that of FLAP (max, 263.6 nm), the max for M1 was determined to be at 256.4 nm (Jäger et al., 1998). Linear calibration curves were performed from the peak areas of FLAP and its metabolites to the external standard FLAP by spiking drug-free rat and human liver preparations with standard solutions of FLAP to give a concentration range of 0.2 to 30 µg/ml. As metabolite formation, mainly that of M1, was significantly lower in human liver than in rat liver, incubation mixtures of human liver microsomes were concentrated by solid phase extraction before HPLC. Briefly, after the reaction was stopped by methanol and centrifugated, the supernatant was diluted with 2.0 ml of phosphate buffer (0.1 M, pH 3.0) and passed through a C₁₈ column (Isolute, ICT, Vienna, Austria) equilibrated with 1 ml of methanol and water (pH 3.0), respectively. The column was washed with water (pH 3.0), and FLAP and its metabolites were eluted with methanol (100%; 1 ml) and dried under a stream of nitrogen. The residue was reconstituted with 250 µl of the mobile phase before injection (200 µl) onto the HPLC column. The recovery of extractions of FLAP and its metabolites generally exceeded 90%.

Effect of Detergents on FLAP Glucuronidation. Human liver microsomes (2 mg/ml) were preincubated on ice for 30 min with 0.1, 0.25, 0.5, and 1 mg of Triton X-100, sodium cholate, Brij 35, Brij 58, and digitonin/mg of microsomal protein, respectively. Then magnesium chloride (10 mM), D-saccharic acid-1,4-lactone (5 mM), and FLAP (final concentration, 0.1-3 mM) in DMSO were added to 100 mM potassium phosphate buffer (pH 7.4) in a total volume of 250 µl. The glucuronidation assay was then initiated by adding 4 mM UDPGA to the incubation mixture and the metabolism of FLAP assessed by the formation of metabolite M2 after a 120-min incubation at 37°C as described above. Control experiments in the absence of detergent were run parallel.

Glucuronidation of FLAP by Human Recombinant UGT Isoenzymes. To assay FLAP M1 and M2 formation, microsomes (1 mg/ml) prepared from lymphoblasts or Sf9 cells containing the cDNA for human UGT1A1, 1A3, 1A4, 1A6, 1A7, 1A9, 1A10, and 2B7, respectively, were incubated at 37°C with FLAP (100 µM), UDPGA (4 mM), and digitonin (0.5 mg/mg of protein). After 120 min, the incubations were stopped by the addition of methanol and the clear supernatant analyzed by HPLC, as described above. Glucuronidation of FLAP was omitted when control microsomes from cells containing only the vector were used.

Chemical Inhibition of FLAP Metabolite Formation. Human liver microsomes (2 mg/ml) and digitonin (0.5 mg/mg of protein) were preincubated at 4°C for 30 min. Then bilirubin (10 µl of a 0, 10, 20, 40, 60, 80, or 100 µM solution in ethanol), propofol (10 µl of a 0, 10, 20, 40, 60, 80, 100, 200, 400, or 800 µM solution in ethanol), magnesium chloride (10 mM), D-saccharic acid-1,4-lactone (5 mM), and FLAP (final concentration, 60 µM) in DMSO were added to 100 mM potassium phosphate buffer (pH 7.4) in a total volume of 250 µl. Each reaction was initiated by the addition of 4 mM UDPGA at t = 0 min. Control experiments included the same amount of solvent without inhibitor. Ethanol concentrations 2% (v/v) did not affect the rates of FLAP metabolite formation. After 60 min at 37°C, reactions were stopped and analyzed by HPLC as described above.

Isolation of FLAP Glucuronides from Rat Liver Microsomes. Biochemical synthesis of the FLAP metabolites M1 and M2 was performed by the incubation of rat liver microsomes (total protein content, 100 mg) in 100 mM potassium phosphate buffer (pH 7.4) in the presence of digitonin (0.5 mg/mg of protein), magnesium chloride (10 mM), D-saccharic acid-1,4-lactone (5 mM), FLAP (final concentration, 3 mM), and UDPGA (4 mM) in a total volume of 15 ml as described above. The reaction was stopped by the addition of 30 ml of methanol, and the incubation mixture was centrifuged (5000g for 5 min at 4°C). Supernatant (700 µl) was repeatedly injected onto a preparative HPLC column using a flow rate of 5 ml/min. Isolation of M1 and M2 was performed on a Hypersil BDS-C₁₈ column (5 µm, 250- × 10-mm i.d.) using the identical mobile phase and gradient as described above for the analytical HPLC assay. The isolated fractions were rotary evaporated and subsequently freeze dried to remove the remaining water.

Structural Identification of FLAP Metabolites. After a 120-min incubation of rat liver microsomes, 2 µl of -glucuronidase (200 U) was added to 100 µl of microsomal samples and further incubated at 37°C for 60 min. The reaction was stopped by the addition of methanol and the supernatant analyzed as mentioned above. To discriminate between mono- and diglucuronidation of FLAP, LC/MS measurements were performed using an HPLC system fitted with an LC200 pump, a 235C DAD detector, and a 200 autosampler (PerkinElmer Sciex Instruments, Wellesley, MA). The system was coupled in line to an API 150Ex mass selective detector (PerkinElmer Sciex) fitted with an atmospheric pressure ionization source for electrospray ionization in the positive mode. The operating conditions were as follows: capillary voltage, 5.00 kV; orifice voltage, 36 V; and gas temperature, 400°C. Column, mobile phase, gradient, flow rate, and injection volume were identical to those used in the analytical HPLC assay (see above). For structural identification of the flavopiridol metabolites M1 (~0.6 mg) and M2 (~1.3 mg), 1D ¹H, 2D ¹H, ¹H double quantum-filtered correlation spectroscopy (DQF-COSY), ¹H,¹H total correlation spectroscopy (TOCSY), ¹H,¹H rotation frame Overhauser experiment spectroscopy (ROESY), ¹H,¹³C heteronuclear single quantum correlation (HSQC), and heteronuclear multiple bond correlation (HMBC) spectra were acquired using a 600-MHz Varian UnityInova spectrometer equipped with a 5-mm triple resonance probe with actively shielded z-gradient (Varian, Palo Alto, CA). The ¹³C spectra of these compounds were acquired using a Varian UnityInova 400 spectrometer equipped with a 5-mm switchable probe. Typical values for the experimental parameters of the NMR experiments can be found in Debella et al. (1999). All spectra were recorded in DMSO-d₆ at 303 K, and tetramethylsilane was used as an internal standard. Using UV spectroscopy, we previously showed that M2 but not M1 exhibited a bathochromic shift of the electron absorption bands of the chromophore indicating a free hydroxyl-group in position 5 and 7, respectively (Jäger et al., 1998). To confirm the present NMR data we repeated the "UV shift" analysis for both metabolites with AlCl₃ and sodium acetate according to Markham (1982) to unambiguously determine the point of attachment of the saccharide part. Two milliliters of methanol was added to 20 µl of the NMR sample of the metabolites and the MeOH spectrum was recorded. After addition of six drops of 0.4 M AlCl₃ and an additional three drops of concentrated HCl, the AlCl₃ and AlCl₃/HCl spectra were recorded. Powdered sodium acetate was added to a new sample of the metabolites until a 1-mm layer settled to the bottom of the cell and a NaOAc spectrum was recorded.

Data Analyses. For characterization of M1 and M2 formation, FLAP concentrations were varied up to 3 mM, and kinetic parameters were estimated by nonlinear regression analysis (Prism 3.0; GraphPad Software Inc., San Diego, CA). Bilirubin and propofol inhibition was characterized by graphical analyses with Lineweaver-Burk and Dixon plots. The enzymatic efficacy, which is defined as the ratio V_max/K_M, quantifies the glucuronidation capacity and corresponds to the intrinsic clearance. If not otherwise indicated, values are expressed as mean ± S.D. of three individual experiments. Statistical differences from control values were evaluated using the Student's paired t test, at a significance level of p < 0.05.

	Results

Top Abstract Introduction Experimental Procedures Results Discussion References

FLAP Metabolism in Rat and Human Liver Microsomes. Liver microsomes from three individual rat and human livers were incubated with FLAP (3 mM) for 120 min and subsequently analyzed by HPLC. Typical HPLC chromatograms from these experiments are shown in Fig. 1 for rat liver microsomes in the absence (Fig. 1A) and presence (Fig. 1B) of UDPGA. Qualitatively similar metabolic profiles were observed in human liver microsomes in the presence of UDPGA (Fig. 2, A and B).

View larger version (18K): [in this window] [in a new window]   Fig. 1.   Representative HPLC chromatogram of FLAP and its two metabolites M1 and M2 in rat liver microsomes. Two milligrams of microsomal protein was incubated with FLAP (3 mM) for 120 min at 37°C without (A) and in the presence of (B) UDPGA. For details, see Experimental Procedures.

View larger version (18K): [in this window] [in a new window]   Fig. 2.   Representative HPLC chromatogram of FLAP and its two metabolites M1 and M2 in human liver microsomes. Two milligrams of microsomal protein was incubated with FLAP (3 mM) for 120 min at 37°C without (A) and in the presence of (B) UDPGA. For details, see Experimental Procedures.

Effect of Detergents on FLAP Glucuronidation by Human Liver Microsomes. UGT proteins are located in the endoplasmic reticulum and are subject to latency of their enzymatic activities. This has resulted in the use of detergents to activate the enzymes. We therefore examined the effect of low solubilizing concentrations (0-1 mg of detergent/mg of protein) of some common detergents on the catalytic activity of UGTs in human liver microsomes (Fig. 3). Triton X-100 and sodium cholate inhibited M2 formation by 24 to 77 and 10 to 71%, respectively, at a concentration between 0.1 to 1 mg/mg of protein. In contrast, Brij 35, Brij 58, and digitonin highly activated M2 glucuronidation whereby the highest increase in M2 formation (3.4-3.9-fold) could be found when a detergent concentration of 0.5 mg/mg of protein was used. Brij 35, Brij 58, and especially digitonin are therefore the detergents of choice for studying FLAP glucuronidation in vitro. Qualitatively similar detergent-dependent metabolic profiles for M1 formation were found (data not shown).

View larger version (49K): [in this window] [in a new window]   Fig. 3.   Effect of detergents on M2 formation. For details, see Experimental Procedures.

Kinetics of FLAP Glucuronidation by Rat and Human Liver Microsomes. Formation of both metabolites M1 and M2 was linear with time, up to 40 min, and with respect to microsomal protein concentrations, of 0.5 to 2 mg/ml (data not shown). The kinetic constants for these reactions were estimated using FLAP concentrations up to 3 mM. Figure 4 depicts representative Michaelis-Menten plots for M1 and M2 formation by rat (Fig. 4A) and human (Fig. 4B) liver microsomes. The formation rate of both glucuronides substantially differed in rat and human liver preparations. The apparent K_M values for M1 and M2 were significantly lower in rat liver (87 ± 13 and 178 ± 31 µM, respectively; p < 0.001), compared with human liver preparations (312 ± 137 and 312 ± 57 µM, respectively). V_max values for M1 and M2 determined in the same microsomal incubations yielded 16- and 2.8-fold higher values in rat than in human liver microsomes (139 ± 39 and 1286 ± 188 pmol/mg of protein/min V_max/K_M). Formation of M1 and M2 was about 50- and 5-fold higher in rat liver microsomes (1.58 ± 2.23 and 7.22 ± 1.17 µl/min/mg) compared with human liver microsomes (0.032 ± 0.016 and 1.52 ± 0.93 µl/min/mg), indicating species-related differences in FLAP glucuronidation. Table 1 summarizes the kinetic data for the formation of M1 and M2 in rat and human liver.

View larger version (16K): [in this window] [in a new window]   Fig. 4.   Rate of M1 (A) and M2 (B) formation normalized to protein content as a function of FLAP concentration. Rat and human liver microsomes were incubated with FLAP for 120 min at 37°C in the presence of UDPGA. Data are expressed as mean ± S.D. (n = 3 individual liver microsomal preparations).

Glucuronidation of FLAP by Recombinant UGTs. UGTs of family 1 are able to metabolize FLAP to two glucuronides, albeit with different ratios (Fig. 5). Incubation in the presence of recombinant UGT isoenzymes exhibited that 87.3% of M1 is catalyzed by UGT1A1 and 12.7% by UGT1A9. M2 is formed by UGTs 1A9 (83.8%), 1A1 (13.6%), and 1A10 (2.6%). As illustrated in Fig. 5, no glucuronide formation was detected with recombinant UGT1A3, UGT1A6, UGT2B7, and UGT2B15.

View larger version (17K): [in this window] [in a new window]   Fig. 5.   Rate of M1 and M2 formation from recombinant human UGTs. Data represent the mean of two determinations.

Chemical Inhibition of FLAP Metabolism. To further prove the involvement of UGT1A1 and UGT1A9 in the formation of M1 and M2, incubation of human liver microsomes with FLAP either in the presence of the UGT1A1 substrate bilirubin (Green and Tephly, 1998) or the UGT1A9 substrate propofol (McGurk et al., 1998) was performed. Bilirubin (0-100 µM) and propofol (0-800 µM) application resulted in a concentration-dependent decrease in FLAP glucuronidation. As expected from the incubation experiments using recombinant human UGT isoenzymes, bilirubin preferably inhibited M1 (Fig. 6A) over M2 (Fig. 6B), resulting in K_i values of 36 and 258 µM, respectively. The nature of UGT1A9-mediated M2-glucuronidation was demonstrated by simultaneous incubation with propofol. In contrast to bilirubin, propofol showed a more prominent inhibition of the formation of M2 (K_i: 46 µM; Fig. 7B) compared with M1 (K_i: 150 µM; Fig. 7A). Lineweaver-Burk analyses revealed that the nature of bilirubin and propofol inhibition was competitive (data not shown).

View larger version (17K): [in this window] [in a new window]   Fig. 6.   Representative Dixon plots for bilirubin inhibition of M1 (A) and M2 (B) formation by human liver microsomes. For details, see Experimental Procedures.

View larger version (17K): [in this window] [in a new window]   Fig. 7.   Representative Dixon plots for propofol inhibition of M1 (A) and M2 (B) formation by human liver microsomes. For details, see Experimental Procedures.

Identification of FLAP Metabolites. After incubation of microsomal samples with -glucuronidase before HPLC analysis, M1 and M2 were no longer detectable. The concomitant increase of parent FLAP indicates glucuronidated metabolites. Positive ion mass spectra of FLAP and both biotransformation products showed stable protonated molecular ions at m/z 402, 578, and 578 amu, in agreement with the molecular mass of FLAP and FLAP monoglucuronides (data not shown). Mass spectrometry and retention times indicated that the two monoglucuronides formed in liver microsomes were identical to those found in the bile of rats (Jäger et al., 1998).

View larger version (15K): [in this window] [in a new window]   Fig. 8.   Metabolic pathway of FLAP in rat and human liver microsomes.

	Discussion

Top Abstract Introduction Experimental Procedures Results Discussion References

FLAP is a semisynthetic flavonoid that induces cell cycle arrest and apoptosis in a variety of cell lines and is currently in phase II clinical testing. In a phase I infusional trial, postinfusional plasma concentration peaks that are compatible with enterohepatic circulation of FLAP were seen (Senderowicz et al., 1998). Furthermore, a dose-dependent increase in serum bilirubin was observed in these patients, indicating a possible interference with bilirubin transport and/or metabolism. Moreover, a previous study showed that FLAP is extensively metabolized in isolated rat liver, obtaining two distinct monoglucuronides excreted mainly into the bile (Jäger et al., 1998). Analysis of plasma samples in a very recent phase II trial also exhibited the formation of a glucuronide, most likely M2, detected previously in rat bile (Innocenti et al., 2000).

Glucuronidation is one of the most common phase II biotransformation reactions for therapeutic drugs catalyzed by UGTs. To date, at least 10 different UGT isoforms have been identified divided into two families (UGT1 and UGT2). The UGTs are not only involved in the metabolism of many drugs (e.g., morphine, paracetamol) but also capable of the biotransformation of endogenous substrates (e.g., bilirubin and ethynylestradiol) and several xenobiotics (Green and Tephly, 1996; Radominska-Pandya et al., 1999). Glucuronidation leads to an increased polarity of hydrophobic compounds and results, in many cases, in a loss of biological activity. UGTs also play a role in the generation of bioactive compounds. Particularly, morphine, steroids, bile acids, and retinoids are all glucuronidated to more active or, even in some instances, to more toxic compounds (Radominska-Pandya et al., 1999).

Using recombinant UGT isoenzymes, we demonstrated that FLAP is glucuronidated by UGT1A1 and UGT1A9 and to a small extent also by UGT1A10. While M1 is preferentially catalyzed by UGT1A1 (>80%), M2 is predominantly formed by UGT1A9 (>75%), and to a minor extent by UGT1A10 and UGT1A1.

The involvement of UGT1A1 and UGT1A9 in the glucuronidation of M1 and M2 was also proved by inhibition studies with the UGT1A1-specific substrate bilirubin and propofol, a drug used for anesthetic induction and maintenance, which is primarily conjugated by UGT1A9. Using this approach, a competitive inhibition of M1 formation (K_i: 36 µM) by bilirubin was observed. As expected, the main metabolite M2 was only slightly affected by bilirubin (K_i: 258 µM). On the basis of plasma levels, bilirubin will likely have no effect on the metabolism of FLAP in humans, where the serum concentration of bilirubin is approximately 1.7 to 3.4 µM. However, particularly in patients with liver disease, liver bilirubin concentration may be higher, resulting in significantly higher unconjugated FLAP blood levels and concomitant increase in FLAP toxicity. Indeed, FLAP treatment during a phase I infusional trial increased serum bilirubin in a number of patients for up to 48 h. Furthermore, one patient died (out of 18 patients) while receiving FLAP at a dose tolerated by all other patients (78 mg/m²/day). This patient had metastatic rectal carcinoma to the liver and lung with mild exertional dyspnea and had received radiation to the liver. On the second day of FLAP infusion, dyspnea increased, which worsened by day 3, when grade 3 hyperbilirubinemia and grade 2 elevated transaminases were noted. Although the infusion was stopped, the respiratory status continued to deteriorate. Therefore care should be taken in patients with extensive liver disease and increased bilirubin.

Besides competition of FLAP and bilirubin to UGT1A1, FLAP may also compete for the canalicular multispecific organic anion transporter (cMOAT; MRP2) in the liver responsible for the excretion of bilirubin glucuronides as observed with the flavonoid genistein (Jäger et al., 1997).

On the other hand, propofol inhibited M2 formation to a higher extent than M1 (K_i: 47 and 142 µM, respectively). As propofol exhibits its anesthetic activities in steady state blood concentration at approximately 3 µg/ml (16 µM) and as FLAP reaches plasma concentrations of 300 to 600 nM, an interaction of propofol with the metabolism of FLAP and vice versa during therapy seems unlikely.

NMR proved to be a powerful tool to elucidate the structure of FLAP phase II metabolites. Using 1D ¹H, ¹³C, 2D DQF-COSY, TOCSY, HSQC, and HMBC experiments we determined the structure of M1 as 5-O-- and that of M2 as 7-O--glucopyranuronosyl-flavopiridol. This is in line with previous results from our lab where we proposed monoglucuronidation of the hydroxyl groups C-5 and C-7 for M1 and M2 in rat liver based on LC/MS and UV experiments (Jäger et al., 1998).

M2 represents the major pathway for FLAP metabolism in human liver catalyzed by more than one UGT. The formation of metabolite M1 is about 50 times lower than M2. Although Innocenti and coworkers (2000) did not show this minor biotransformation product in cancer patients, we think M1 should be detectable in human plasma based on our in vitro data. Pharmacological effects of FLAP glucuronides may be considered because structure-activity relationship studies revealed that replacing the hydroxyl groups in positions 5 and 7 by a methoxyl group still exhibited inhibitory potency on cyclin-dependent kinases and epidermal growth factor receptor tyrosine kinase (Sedlacek et al., 1996).

Our data further demonstrate a possible extrahepatic biotransformation of FLAP as both UGT1A1 (present mainly in liver) and UGT1A10 (absent in liver) are found in the intestine. As we showed in our previous study, FLAP glucuronides are excreted into bile where they can be cleaved by -glucuronidase. During enterohepatic circulation, FLAP might also be glucuronidated by intestinal UGT1A1 and UGT1A10. These isoenzymes may therefore contribute to the FLAP metabolism.

Taken together, our results elucidate the enzymatic pathways of FLAP in human liver, which must be taken into consideration during cancer therapy of patients.