Abstract
Synthetic cyclin-dependent kinase inhibitors have recently been referred to as effective antiproliferative agents. This study was conducted to characterize clearance of a 3H-labeled, trisubstituted purine-type inhibitor, 8-[3H]bohemine [6-benzylamino-2-(3-hydroxypropylamino)-9-isopropylpurine], in mice. Radioactivity profiles were analyzed by liquid scintillation counting and by thin layer chromatography followed by autoradiography. Metabolite structures were elucidated by mass spectrometry, NMR, and enzymatic analyses. Bohemine was rapidly and completely metabolized in vivo and disappeared from circulation during the first 60 min following intravenous administration. The metabolites were partly eliminated by the hepatobiliary tract and partly by renal excretion. The terminal hydroxyl group located at the C2 side chain of bohemine made the compound susceptible to main metabolic attacks, i.e., distinct types of conjugation reactions with glycosyl donors as well as an oxidative reaction. Other pathways were of relatively minor significance. Bohemine O-β-d-glucoside was the most abundant metabolite to be excreted. The enzymatic mechanism responsible for bohemine glucosidation in vitro required the presence of a UDP-glucoside donor. Additional glycosidation products were observed after inclusion of UDP-glucuronide, UDP-xylose, UDP-galactose, or UDP-N-acetylglucosamine into microsomal incubates. Glycosidations occurred faster in the kidney incubates than in hepatic ones. The second principal bohemine metabolite was a carboxylic acid, 6-benzylamino-2-(2-carboxyethylamino)-9-isopropylpurine. A cytosolic, 4-methylpyrazole-sensitive alcohol dehydrogenase class I was shown to mediate oxidation of the terminal hydroxyl group of bohemine into this acid, which was the only metabolite found in the blood in significant amounts. However, it displayed only weak cyclin-dependent kinase-1-inhibitory activity (IC50 > 100 μM) when compared with that of bohemine (IC50 ∼ 1 μM).
During evolution, a highly ordered and conserved array of mechanisms regulating cyclin-dependent kinases (CDKs1), which govern the timing of cell cycle progression and cell division, has developed in eukaryotes. Recently, natural peptide CDK inhibitors have been uncovered and shown to play an important regulatory role in cell differentiation, proliferation, senescence, and programmed death (Chellappan et al., 1998). It has also been demonstrated that the effects of these endogenous inhibitors can be partly mimicked by several different types of synthetic, small-molecule inhibitors including butyrolactone I, flavopiridol, 2,6,9-trisubstituted purines such as olomoucine, roscovitine, and purvalanol, paullones, indirubins, and others (Gray et al., 1999; Meijer et al., 1999; Sielecki et al., 2000).
Recent studies have shown the significant antiproliferative activity of synthetic CDK inhibitors in vivo, which has encouraged investigators to evaluate their potential use as new anticancer agents. During the short period that has elapsed since the unusual binding mode of olomoucine to CDK2 was elucidated (Veselý et al., 1994; Schulze-Gahmen et al., 1995), more than 3000 novel 2,6,9-trisubstituted purine compounds have been synthesized and screened for their effects. Their selectivity for molecular targets has made them invaluable reagents for molecular biological, biochemical, and morphological studies of the cell cycle. Importantly, the inhibitors suppress a subset of CDKs related to CDK1. These include cyclin/CDK1, cyclin/CDK2, and cyclin/CDK7 complexes. They appear to play a more critical role in the cell cycle than a subset of CDKs involving cyclin/CDK4 and cyclin/CDK6. Clinical applications of the inhibitors are also being extensively investigated in the treatment of atherosclerosis and vascular diseases, dermatological, nephrological, and neurological diseases, and parasitic and viral infections (Hung et al., 1996; Schow et al., 1997; Gray et al., 1998,1999; Walker, 1998; Chang et al., 1999; Imbach et al., 1999; Meijer et al., 1999; Sielecki et al., 2000).
However, ongoing research has revealed a dearth of knowledge concerning the fate of CDK inhibitors at the level of cells, in animals and in humans. Since the olomoucine-type (i.e., a primary hydroxyl group possessing) CDK inhibitors roscovitine and 6-benzylamino-2-(3-hydroxypropylamino)-9-isopropylpurine, designated as 8-[3H]bohemine (Strnad et al., 1997; Sielecki et al., 2000), have been selected by some investigators as potential candidates for therapeutic uses (M. Strnad, personal communication), it is important to understand more about the fate of these inhibitors. We chose bohemine to initiate an investigation into the metabolism of CDK inhibitors using an in vivo mouse model.
This article describes the relatively rapid disappearance of bohemine from animal circulation due to its fast metabolism in liver and kidney. The CDK-inhibitory potential of the major metabolite found in blood was examined in vitro and demonstrated to be insufficient to substitute for the CDK-inhibitory effect of the metabolically unstable parent compound. Hopefully, our article will facilitate the design of new potent purine CDK inhibitors with higher metabolic stability.
Experimental Procedures
Chemicals.
Alcohol dehydrogenase (ADH), EC 1.1.1.1, from equine liver (A9589); alcohol dehydrogenase, EC 1.1.1.1, from baker's yeast (A7011); β-NAD (N6522); β-glucuronidase, EC 3.2.1.31, from bovine liver (G0251); sulfatase, EC 3.1.6.1, from Aerobacter aerogenes (S1629); sulfatase, EC 3.1.6.1, from Limpets (S8629); β-glucosidase, EC3.2.1.21 (G4511); α-glucosidase, EC 3.2.1.20, recombinant form, fromSaccharomyces cerevisiae (G0660);d-gluconic acid 1,5-lactone, UDP-glucuronic acid, UDP-glucose, UDP-galactose, UDP-mannose, UDP-xylose, UDP-glucosamine, UDP-galactosamine, UDP-N-acetylglucosamine, UDP-N-acetylgalactosamine,n-octyl-β-d-glucoside,n-octyl-α-d-glucoside, ATP, histone (type III-S), 4-methylpyrazole hydrochloride (4-MP), disulfiram, and the protease inhibitors phenylmethylsulfonyl fluoride, leupeptin, and aprotinin were obtained from Sigma-Aldrich (St. Louis, MO). [γ-32P]ATP was purchased from Amersham Pharmacia Biotech (Little Chalfont, England, Uppsala, Sweden, and Piscataway, NJ). Bovine serum albumin (fraction V) was produced by Bioveta (Ivanovice n.H., Czech Republic). Silica gel for column chromatography (Kieselgel 60, particle size of 40–63 μm) and silica gel 60 F254 TLC sheets were purchased from Merck (Darmstadt, Germany). All other chemicals and reagents used were of analytical or high-performance liquid chromatography grade.
Radiolabeled Compounds and Chemical Syntheses.
Bohemine was synthesized as described by Havlı́cek et al. (1997). 8-[3H]Bohemine was prepared by an isotope exchange reaction. Briefly, a reaction mixture containing bohemine, palladium oxide/barium sulfate as a catalyst, and dioxane-0.1 M aqueous disodium carbonate (9:1) as a solvent was stirred with a tritium gas (3H2 with about 60% carrier-free radioactivity) at 23°C for 2 h. The solvent was then removed by lyophilization, the residue was dissolved in a mixture of tert-butanol-H2O (9:1), and the catalyst was removed by centrifugation. The product was purified by TLC on aluminum oxide 150 F254 sheets (Merck) with toluene-methanol (97:3) as a mobile phase. The specific activity of the [3H]bohemine product was 300 TBq/mol, and its radiochemical purity was at least 98% when checked by TLC on Silicagel 60 F254 sheets (Merck) with chloroform-methanol (95:5).
Bohemine metabolite M-3, extracted from a TLC spot (see below), was esterified with diazomethane into its corresponding methylester. The structural identification of the esterified product of this reaction was performed by MS, 1H NMR, and Fourier-transform infrared spectroscopy, and the product was ascertained as 6-benzylamino-9-isopropyl-2-(2-methoxycarbonylethylamino)purine (seeResults).
In Vivo Study Using Anesthetized Mice.
All the experiments using animals were authorized by the Institutional Ethical Committee and carried out according to institutional guidelines in compliance with Czech national laws, and in agreement with the Declaration of Helsinki. The in vivo experiments were carried out in monitored male NMRI outbred mice of mixed genetic background and of approximately 20 to 25 g of body weight between 5 and 7 weeks of age (VUFB, Konarovice, Czech Republic).
To determine bohemine distribution and excretion, the animals were anesthetized using 0.05 to 0.08 mg of pentobarbital/g of body weight administered intraperitoneally. A stock solution of unlabeled bohemine (11.3 mg/ml) was prepared in 0.07 M HCl, and shortly before use it was further diluted in 0.9% NaCl to a final concentration of 1.13 mg/ml unlabeled drug. [3H]Bohemine (74 kBq/g of body weight) was added to 300 μl of this solution. To achieve a dosage of about 13 μg (≈40 nmol) of bohemine/g of animal body weight, an appropriate volume of the diluted bohemine solution, proportional to the weight of the animal, was injected into either the tail vein or subcutaneously. Animals were killed 2, 5, 10, 30, 60, or 120 min after drug administration (n = 3). Control short-term i.v. experiments (5 and 10 min) were also performed using transient (1- to 2-min) ether narcosis instead of pentobarbital. Liver, stomach, kidney, gall bladder, two parts of intestine (a proximal 6- to 8-cm segment and the remaining part of the bowel), spleen, and urinary bladder were removed and weighed. Samples of blood, lungs, brain, bone marrow, retroperitoneal fat, cardiac muscle, and posterior femoral muscle were also removed for further analysis.
Preparation of Mouse Liver and Kidney Cytosolic and Microsomal Fractions.
Liver and kidney subcellular fractions were prepared by differential centrifugation (10,000g for 15 min followed by 100,000g for 60 min using an Avanti 30 Centrifuge and Optima LE-80K Ultracentrifuge, Beckman, Fullerton, CA) in 0.25 M sucrose solution. The fraction aliquots were stored at −80°C until used. Proteins were determined according to Bradford (1976) with bovine serum albumin as a standard.
In Vitro Transformation Studies: Incubations with Cytosolic Fractions.
Final volumes of incubation mixtures amounted to 0.2 ml. Incubations with cytosolic fractions were run in a phosphate buffer (0.1 M, pH 7.4) with MgCl2 (5 mM), unlabeled bohemine (final concentration, 3.6 μM), and 240 kBq (≈720 pmol) of [3H]bohemine (final concentration of [3H]bohemine ≈ 3.6 μM), and 7.5 mM NAD. The final mixture contained 25 μl of cytosol (0.3–0.5 mg of cytosolic protein).
Incubations with Microsomal Fractions.
The microsomes were transferred by repeated centrifugation (100,000g, 60 min) into the phosphate buffer (0.1 M, pH 7.4) shortly before use. The incubations were conducted in the presence of glycosyl donors (final volume, 0.2 ml) that used 25 μl of microsomes (0.6 mg of protein) and either UDP-glucuronic acid (3.9 mM), UDP-glucose (4.1 mM), UDP-galactose, UDP-mannose, UDP-xylose, UDP-glucosamine, UDP-galactosamine, UDP-N-acetylglucosamine, UDP-N-acetylgalactosamine, n-octyl-β-glucoside, or n-octyl-α-glucoside (2.0 mM). These were further supplemented with 240 kBq (≈720 pmol) of [3H]bohemine (final concentration of [3H]bohemine ≈ 3.6 μM). The reactions were allowed to proceed in open tubes at 37°C under continuous mixing for 2 h, and aliquots were removed at 5, 10, 30, and 120 min. The volume of the incubation mixtures was increased 10 to 20 times and supplemented with up to 7.5 to 15 μmol of unlabeled bohemine for preparatory purposes.
Incubations with Enzymes.
An ADH incubation mixture was of the same composition as the cytosol mixture except that it was supplemented with either 0.43 Sigma units of ADH from equine liver or 135 Sigma units of ADH from baker's yeast instead of cytosol. Incubates of metabolites extracted from TLC spots (see below) contained about 10 kBq of radioactivity and were supplemented either with 0.67 Sigma units of β-glucuronidase, 7.6 Sigma units of sulfatase, or 2.5 Sigma units of β-glucosidase in a citrate buffer (0.05 M, pH 5.0). The α-glucosidase incubation mixture contained about 10 kBq of spot extract and the recombinant enzyme (2.5 Sigma units) in the phosphate buffer. Enzyme inhibitors were added to the mixtures at concentrations as indicated in the text (see Results). The mixtures were then incubated under continuous gentle mixing at 37°C, and aliquots were removed at different time intervals (see Results). The volumes of incubation mixtures were increased 10 to 20 times for preparatory purposes.
Cyclin-Dependent Kinase Inhibition Assay.
CDK1-cyclin B complex was produced in Sf9 insect cells coinfected with baculoviral constructs. Seventy hours after infection (50 mM Tris/Cl buffer, pH 7.4, 150 mM NaCl, 5 mM EDTA, 20 mM NaF, 1% Tween 20, and protease inhibitors), the cells were harvested on ice for 30 min. A soluble fraction was centrifuged at 14,000g for 10 min, and the supernatant was stored at −80°C until used. An assay mixture contained 1 mg/ml histone, 15 μM ATP, 7.4 kBq of [γ-32P]ATP, 10 mM MgCl2, 5 mM EGTA, 10 mM 2-glycerolphosphate, 1 mM NaF, 1 mM dithiothreitol, protease inhibitors, and tested compound in a final volume of 10 μl in a HEPES buffer (50 mM, pH 7.4). After 10 min, the incubation was stopped by adding SDS sample buffer and the product resolved using 12.5% SDS-polyacrylamide gel electrophoresis (Mini-Protean II System, Bio-Rad Laboratories, Hercules, CA). The gels were analyzed by densitometry as described previously (Veselý et al., 1994) with the only exception that an Imaging Analyzer BAS 1800 (Fujifilm, Tokyo, Japan) was used to quantify the phosphorylated histone. Experiments were performed in duplicate.
Extraction of Metabolites.
For 3H-metabolite analysis and identification, blood, tissues, and organs were collected, weighed, and homogenized in a 9-fold volume of ethanol. The stomach, intestine, urinary bladder, and gall bladder were processed with their contents included. The in vitro incubates were stopped and extracted using ethanol in the same manner. The ethanolic supernatants (4000g, 5 min) were then decanted, concentrated, and used for the isolation and identification of metabolites. The sediments were found to contain less than 12% of the total sample dose when measured for radioactivity. Moreover, crude blood and liver samples were subject to distillation, in parallel with the extraction. The evaporated water phase was collected in a condensate receiver. Less than 10% of the 3H radioactivity was found to pass from samples to the condensate, indicating that the 3H label, confined to the C8 of the bohemine purine ring, was a marker complying with the requirements of this study.
TLC.
The ethanolic extracts (supernatants) were concentrated in vacuo under a nitrogen atmosphere at 40°C and chromatographed on Silicagel 60 F254 sheets (Merck). TLC was routinely run in acidic mobile phase chloroform-methanol-acetic acid (90:10:1) (v/v; phase A). Acetic acid was replaced by 1 volume of concentrated aqueous ammonia (phase B) to diagnose (by changing TLC mobility of the spots) the acidic/basic character of substances. The spots were visualized by autoradiography. For structure identification, the spots were scraped out into tubes, extracted with ethanol, and dried, and chemical structures of the isolated metabolites were elucidated by means of enzymatic analyses, MS, and NMR spectroscopy (see Results).
Liquid Scintillation Analysis.
Radioactivity was measured in a Canberra Packard TriCarb 2700 TR Liquid Scintillation Analyzer (Canberra Packard, Groningen, The Netherlands, and Meriden, CT) with an automatic quench correction. Canberra Packard Ultima Gold scintillation fluid, 2 ml per 100-μl aliquots, was used as a scintillation cocktail.
Autoradiography.
Autoradiography of the TLC spots was performed using either3H-Hyperfilm or Kodak BioMax MS film with Kodak BioMax TranScreen LE at −70°C (Amersham Pharmacia Biotech). The detection limit of both techniques was about 100 Bq/spot (equal to about 0.3 pmol of a labeled substance per spot), provided the duration of the exposure was about 72 h in the first case, and about twice as short in the second. The spot intensities were computerized and data processed using GeneGenius (Syngene, Cambridge, UK) and/or Imaging Analyzer BAS 1800 (Fujifilm);32P-SDS-polyacrylamide gel electrophoresis gels were evaluated using the Imaging Analyzer BAS 1800 only.
Spectroscopic Methods.
Electrospray ionization MS and MSn were carried out using an ion trap mass spectrometer LCQ (Finnigan-MAT, San Jose, CA). The analyzed samples were either standards, diluted by 1% acetic acid in methanol, or ethanolic extracts of TLC spots acidified with 1 μl of acetic acid. The samples were introduced with a flow rate of 2 μl/min to the electrospray interface with a source voltage of 5.6 kV, sheath gas flow of 28, and a capillary temperature of 200°C. The selected ions were fragmented in the ion trap (isolation width, 6.0; collision energy, 25%). Mass spectra were scanned in a full scan regime as well as in zoom scan and MSn.1H NMR and 13C NMR spectra were measured on a Varian VXR-400 instrument (Palo Alto, CA) at 400 and 100 MHz, respectively. J values are given in Hz.
Calculations.
To evaluate the total radioactivity content in the circulatory system of mice, the concentration of radioactivity in blood was multiplied by a factor amounting to 8% of mouse body weight. The relative accumulation of 3H radioactivity in different mouse body tissues was represented as multiples of its blood concentration [y = (Bq/g of wet tissue)/(Bq/ml of blood)]. The relative distribution of metabolites present in the respective tissues was calculated from radiochromatograms as the area of the metabolite/(areas of all metabolites + start area + area of parent compound residue). The results were then expressed graphically after being converted to a corresponding fraction of the total administered dose, provided the relative distribution of radioactivity among organs was known from scintillation counting measurements (seeResults).
Results
3H-labeled trisubstituted purine bohemine was used to study the metabolism of this selective CDK inhibitor in mice in vivo. First, we focused on the distribution of radioactivity among different mouse body tissues as well as on the identification of pathways of radioactivity elimination from the animal body. Next, we concentrated on resolution of tissue extracts and identification of major products of bohemine metabolism by TLC, enzymatic analyses, and spectroscopic methods.
In Vivo Experiments. Distribution of 3H Radioactivity in Tissues.
Two minutes after [3H]bohemine i.v. injection, the radioactivity concentration in blood amounted to about 2.9% of the total dose/ml of blood while being stored in several organs above concentrations present in the blood. It was elevated in the liver, kidney, in the first 8 cm of the proximal intestine, and in gall bladder. Table 1 demonstrates that mean concentrations of 3H radioactivity in the liver and kidney exceeded those in blood about 3- to 5-fold. Later, the radioactivity appeared in more distal parts of the intestine as well as in the urinary bladder. There was no 3H radioactivity accumulation in any of the other tissues examined (seeExperimental Procedures), as indicated by comparing radioactivity concentrations in the above tissues and blood. Table2 shows the time course of the total3H radioactivity dose distribution among different mouse body tissues. The total radioactivity content in blood, kidney, and liver peaked within the first 2 min after i.v. injection and then declined progressively. The radioactivity dropped very steeply in blood, amounting to less than 1/10 of the total dosage within 2 min after its i.v. administration. It is apparent (Table 2) that during the first 30 min, a fraction of the radioactivity sequestered in the liver, intestine, kidney, and urinary bladder achieved more than 50% of the total dose. The portion of the total radioactivity dosage excreted into the intestine was approximately the same as that sequestered into the urinary bladder during the first 60 min (Table 2). Both peaking and elimination of the radioactivity were somewhat more protracted after [3H]bohemine s.c. injection than after the i.v. administration. About twice as much time (60 min) was necessary to sequester 50% of the total s.c. radioactivity dose in the same target organs. Peak concentrations of 3H radioactivity (1.0–1.5% of the total dose/ml) were contained in blood between 10 and 120 min following s.c. administration (data not shown).
Tissue Metabolite Profiles.
In the next part of our experiments, we were interested in whether the radioactivity contained in blood and concentrated in mouse organs was the parent compound [3H]bohemine or a product of its metabolism.
A blood 3H radioactivity profile is given in Fig.1. It is obvious that after i.v. administration, [3H]bohemine completely disappeared from the circulation in about 60 min. As illustrated, by far the most prevalent portion of 3H radioactivity in the profile was accounted for by the presence of a spot designated as M-3 (Fig. 1; Table 3). The intensity of this spot declined very slowly over time and persisted in the circulation beyond our observation period. The transient presence of two other spots designated as M-1 and M-2 was also registered in the blood radioactivity profile (Fig. 1; Table 3).
The time course of 3H radioactivity distribution among the principal 3H radioactivity spots observed in mouse liver, proximal intestine, kidney, and urinary bladder ex vivo extracts can be seen in Fig.2. It is evident that there were some quantitative differences in radioactivity profiles within these organs. Ten minutes after bohemine i.v. administration, the principal spot accounting for 54% of the radioactivity deposits in liver was the same as that found in blood and designated as M-3 (Fig. 2). On the other hand, in kidney, intestine, and urinary bladder the predominant spot was M-2. This made up for 39, 55, and 75% of the radioactivity pools present in the respective organ extracts 10 min after the administration, but only about 5% of the radioactivity found in the liver extract. A small amount of bohemine was detected in the proximal intestine, whereas there was no bohemine in the urinary bladder (Fig.2). Furthermore, in the liver and kidney, the rather steep curves of parent bohemine and M-2 spots contrast with a relatively slow decline in M-3 intensity. Moreover, between 10 and 30 min following i.v. administration, the intensity of the M-3 spot outweighed that of M-2 in the kidney because of much faster disappearance of M-2.
To obviate potential interference between pentobarbital and bohemine metabolism, we used animals transiently narcotized with ether, instead of pentobarbital (see Experimental Procedures). However, in these short-term experiments, no apparent differences could be demonstrated between metabolite profiles in ex vivo organ extracts obtained from animals anesthetized by pentobarbital or by ether (data not shown).
Structure Elucidation of the Principal Metabolites Obtained in the in Vivo Experiments.
When the polar spot M-1 remaining in the close vicinity of the start (see Table 3 for RF values) was extracted and incubated with the enzyme β-glucuronidase, it yielded a single3H-labeled product with chromatographic mobility identical to that of bohemine. No products were detected after incubation of the M-1 spot extract with sulfatases or glucosidases as described under Experimental Procedures. Soft ionization MS of the M-1 spot extract indicated the presence of a compound with anMr of 516 (m/z = 517 [M + 1]+), which corroborated the structure of bohemine glucuronide. Consequent fragmentation in an ion trap led to a neutral loss of 176 that could be explained by elimination of a glucuronide moiety given that the fragmentation of the residual ionic form (m/z = 341 [M + 1]+) was identical to that of authentic bohemine (Table 3). Thus, it can be concluded that bohemine metabolite M-1 was bohemine β-glucuronide. However, it is apparent that this conclusion is still open to verification by a direct structural analysis of the supposed glucuronide moiety.
When analyzing an M-2 spot extract by soft ionization MS, a peak of aMr of 502 (m/z = 503 [M + 1]+) was obtained, suggesting an unexpected bohemine conjugate with a hexose. Indeed, consequent fragmentation in an ion trap led to a neutral loss of 162 that could represent elimination of a hexose moiety since, again, fragmentation of the rest of the molecule (an ion of m/z = 341 [M + 1]+) was indistinguishable from that of bohemine (Table 3). The treatment of the M-2 spot extract with the enzyme β-glucosidase produced a single radioactive spot with the same mobility as authentic reference [3H]bohemine. To gain deeper insight into what the configuration of a putative glycosidic bond might be, the M-2 metabolite was treated with either recombinant α-glucosidase or β-glucosidase in both the presence and absence of a selective β-glucosidase inhibitor, glucono-1,5-lactone (10 mM). While the sample was decomposed by β-glucosidase in the absence of glucono-1,5-lactone, no cleavage was observed in the presence of the inhibitor. Furthermore, α-glucosidase did not decompose M-2. Finally, NMR data unequivocally demonstrated the presence of the β-glucoside anomer in the M-2 metabolite structure (Fig. 3; Table 3). On the basis of all these findings, we concluded that metabolite M-2 was bohemineO-β-d-glucoside.
A metabolite contained in an M-3 spot extract was tentatively identified as an acid based on its dissimilar TLC mobilities in chromatographic phases (Table 3). In accord with that conjecture, a soft ionization MS analysis suggested a structure generated by oxidation of the 2-(3-hydroxypropylamino) group of bohemine to a 2-(2-carboxyethylamino) group (m/z = 355 [M + 1]+). Consequent methylesterification of the M-3 spot extract led to a single product with a chromatographic mobility characterized by RF values of 0.85 and 0.91 in phases A and B, respectively. TheseRF values were entirely distinct from those of M-3 (Table 3). Further analysis of the methylesterified product by MSn, 1H NMR, and IR ascertained it as 6-benzylamino-9-isopropyl-2-(2-methoxycarbonylethylamino)purine (m/z = 369 [M + 1]+; Table 3). Thus, the M-3 spot metabolite, representing the principal portion of the radioactivity contained in blood and liver and persisting in the mouse body beyond our observation period (Figs. 1 and2) was a carboxylic acid, 6-benzylamino-2-(2-carboxyethylamino)-9-isopropylpurine. This conclusion was further corroborated by a subsequent analysis of a product of the bohemine reaction with enzyme ADH (see below).
Incubation of Bohemine in Microsomal Incubates Supplemented with Activated Glycosides.
Having obtained data indicating that bohemine β-glucuronide and bohemine β-glucoside were the main phase II products of bohemine metabolism in mice in vivo, we incubated [3H]bohemine with mouse liver or kidney microsomes in the presence of either UDP-glucuronic acid, UDP-glucose, or a lipophilic glucose donor, n-octyl-glucoside. In both types of microsomes, the activity of a UDP-glucose-consuming system governed the kinetics (Fig. 4). Only small amounts of the corresponding product were formed in the presence of UDP-glucuronic acid donor (Fig. 4), whereas the artificial donorn-octyl-glucoside was ineffective in our assays (not shown). Subsequent TLC, MS, and enzymatic analyses showed that the UDP-glucuronide reaction product was identical to spot M-1 bohemine β-glucuronide, and the UDP-glucoside reaction product was identical to spot M-2 bohemine β-glucoside. Bohemine was then incubated in microsomes supplemented with various activated glycosides, such as UDP-galactose, UDP-mannose, UDP-xylose, UDP-glucosamine, UDP-galactosamine, UDP-N-acetylglucosamine, UDP-N-acetylgalactosamine, orn-octyl-α-glucoside, with the aim to further explore the glycosidation potential of the microsomes. As is apparent from Fig. 4, UDP-xylose was as effective a sugar donor as UDP-glucose, both in the kidney and in liver microsomes. UDP-Galactose and UDP-N-acetylaminoglucose were also remarkably good donors (Fig. 4). With each of these active glycosides, the kidney microsomes exhibited significantly higher specific glycosyltransferase activity toward bohemine than did liver microsomes (Fig. 4). The remaining compounds were inactive as donors.
The in vitro UDP-xylose reaction product was tentatively identified as bohemine xyloside based on soft ionization MS analysis (m/z = 473 [M + 1]+); RF values of this product were different from those of the product of UDP-glucose reaction (Table 3). Similarly, the UDP-N-acetylglusosamine reaction product was tentatively identified as bohemineN-acetylaminoglucoside (m/z = 544 [M + 1]; Table 3). On the contrary, neither MS nor TLC mobility characteristics of the UDP-galactoside reaction product were sufficient to differentiate it from those of bohemine glucoside (Table 3).
The M-3 Spot Metabolite Is Produced in Cytosolic Incubates—Bohemine Is a Substrate for ADH.
When monitored by TLC, [3H]bohemine incubation with either a liver or kidney cytosolic fraction in the presence of NAD+ coenzyme resulted in detection of a single metabolite. The TLC mobility, diazomethylation, and consequent MS/1H NMR analyses identified this product as metabolite M-3 (see above). This in vitro formation of the metabolite was inhibited by a 5 μM classical ADH inhibitor, 4-MP, while a 100 μM aldehyde dehydrogenase inhibitor, disulfiram, exerted no significant effects (data not shown). Bohemine was then incubated with commercial equine liver ADH. It was found to change bohemine into M-3. This reaction was also sensitive to 4-MP and resistant to disulfiram. Consequently, it can be concluded that bohemine is a substrate for a cytosolic, 4-MP-sensitive ADH class I that is able to catalyze its oxidation into the corresponding carboxylic acid, 6-benzylamino-2-(2-carboxyethylamino)-9-isopropylpurine. By contrast, no products were detected in incubates of bohemine with yeast ADH (data not shown).
CDK Inhibition by Bohemine Metabolite M-3.
The only metabolite that was found to circulate in blood in significant amounts was bohemine carboxylic acid (the M-3 spot metabolite). This was characterized by rather slow elimination and prolonged persistence in the mouse body (Figs. 1 and 2). For practical reasons, it was important to ascertain whether this metabolite could mimic CDK-inhibitory, antiproliferative effects of the metabolically unstable parent compound. However, the carboxylic acid derivative failed to inhibit CDK1-cyclin B substantially in vitro, even when tested in a concentration range up to 100 μM, while bohemine displayed an IC50 value for CDK1-cyclin B complex of 1 μM (see Experimental Procedures; data not shown).
Discussion
Currently, investigations on novel purine-derived CDK inhibitors are being conducted to evaluate their potential therapeutic effects. As this trend is likely to continue, there is a need to establish the relationship between the chemical structure of the inhibitors and their pharmacokinetic properties in vivo.
Using [3H]bohemine-injected male mice, we identified the liver and kidney as the organs responsible for clearance of this selective CDK inhibitor. Principal bohemine metabolites were isolated from mouse body fluids and tissues and their chemical structures were elucidated. Our data indicate that the predominant bohemine metabolite to be eliminated was a bohemine glucoside conjugate (an M-2 spot metabolite).
Generally, the most common xenobiotic conjugates to be encountered in mammals are glucuronic acid derivatives. If glucuronidation is possible, the remaining types of glycosidations are considered to be of minor importance in animals, although they are common in plants and invertebrates (Tang, 1990). However, novel pathways forO-glycoside conjugate formation that play a role in the metabolism of foreign and/or endogenous compounds in mammals have been identified recently, namelyO-β-d-glucosidation (Gessner et al., 1973), O-α-d-glucosidation (Kamimura et al., 1988), andO-β-d-N-acetylglucosaminidation (Marschall et al., 1989). Our study demonstrated that the terminal hydroxyl group of bohemine was extensively conjugated with a hexoside moiety different from glucuronide. As a matter of fact, the TLC and MS characteristics obtained in our experiments did not discriminate between hexoside epimers. Nevertheless, the enzymatic analysis performed according to Matern et al. (1984) and Tang (1990) suggested the bohemine O-β-d-glucoside as the most probable candidate structure for metabolite M-2. This was confirmed by 1H NMR and 13C NMR measurements that established the presence ofO-β-glucoside anomer in the metabolite. Nakano et al. (1986a,b) were likely the first to describeO-β-glucosidation of nonacidic hydroxyl groups in mammals (in dogs). Boberg et al. (1998) uncovered similar preferential formation of O-β-d-glucoside conjugate, as compared with production of glucuronide, using cerivastatin as a foreign aglycone substrate in dogs. Interestingly, the same detoxification reaction was absent in mice and rats (Boberg et al., 1998). O-β-Glucosidation of the nonacidic primary hydroxyl group of bohemine as observed in our experiments might, therefore, be a reaction of similar significance in mice.
The glucoside conjugate may be synthesized by either a UDP-sugar-dependent microsomal glucosidation system or a nucleotide-independent, i.e., lipid-soluble sugar-utilizing system (Matern et al., 1984; Matern and Matern, 1990). Our microsomal incubates synthesized bohemine β-glucoside when supplemented with UDP-glucose, whereas those withn-octyl-β-d-glucoside remained inactive. Thus, the favored enzymatic mechanism responsible for the bohemine glucoside formation in mice was sugar nucleotide-dependent. Formation of additional products was observed after inclusion of UDP-xylose, UDP-galactose, and UDP-N-acetylglucosamine into incubates. MS data strongly suggest that the products of UDP-xylose reaction and UDP-N-acetylglucosamine reaction were the corresponding glycosides bohemine xyloside and bohemineN-acetylaminoglucoside, respectively. The glycosidation reactions were significantly faster in mouse kidney microsomes than in the liver fraction. No conclusions can be drawn at this time concerning the nature of the enzyme systems catalyzing these reactions and their physiological significance. It is noteworthy that the presence of UDP-xylose- and UDP-galactose-dependent transglycosylation reactions has been demonstrated in human liver microsomes (Radominska et al., 1993). Moreover, N-acetylglucosaminidation has been established as an important conjugation pathway in humans (Marschall et al., 1992).
The principal product of phase I bohemine metabolism found in mice ex vivo extracts was a metabolite formed by oxidation of the bohemine terminal hydroxyl group into a carboxylic one (an M-3 spot metabolite). This metabolite predominated in blood and liver but, surprisingly, represented only a minor fraction of the radioactivity found in intestine, kidney, and in urinary bladder during the first 60 min after the i.v. injection. M-3 is likely secreted by hepatocytes into both the intestinal lumen (via the hepatobiliary system) and the blood stream (via sinusoidal poles). Alternatively, the metabolite in question might undergo extensive reabsorption in the proximal intestine. Both assumptions are in accord with the presence and persistence of significant amounts of M-3 in blood, liver, and kidney.
Oxidation of the primary hydroxyl group of bohemine into the carboxylic acid might have been catalyzed by a variety of enzyme systems. Here we showed that a 4-MP-sensitive member of the mammalian ADH family of oxidoreductases was responsible for the formation of at least a portion of metabolite M-3. It is worth recalling that medium-chain, dimeric, mostly cytosolic ADHs that are widely distributed in vertebrates as well as in various tissues of individual mammalian species divide into classes and isoforms according to their different subunit composition (Riveros-Rosas et al., 1997). The horse liver ADH, used in our in vitro experiments, consists of two subunits, E and/or S, determining the horse liver enzyme as a high-affinity/low-capacity mammalian ADH class I (Jörnval and Höög, 1995). Other authors have shown that horse liver ADH catalyzes NAD+-dependent alcohol oxidation into aldehydes as well as dismutation of aldehydes into their corresponding alcohols and acids (Abeles and Lee, 1960;Henehan and Oppenheimer, 1993; Svensson et al., 1996). In accord with those findings, we observed that horse liver ADH also oxidized bohemine into its corresponding carboxylic acid, at least in vitro. On the other hand, we were unable to identify any products of the bohemine reaction with yeast ADH. This is in agreement with data reporting yeast NAD+-dependent ADH to lack aldehyde dehydrogenase properties (Dickinson and Monger, 1973).
The participation of ADH class I in bohemine transformations in mice is strongly suggested by an inhibitory effect of low (5 μM) concentration of a classic ADH inhibitor, 4-MP (Eklund et al., 1990;Kemper and Elfarra, 1996), on the bohemine acid production observed in cytosolic incubates. The mouse cytosolic form of ADH class I has been designated as A2 (Algar et al., 1983). As expected, the mouse liver cytosol displayed much higher NAD+-dependent bohemine dehydrogenase activity than did the kidney cytosol. Furthermore, disulfiram, a potent inhibitor of a cytosolic aldehyde dehydrogenase (Kitson, 1975; Eckfeldt and Yonetani, 1976; Hempel et al., 1984), did not exert significant effects on bohemine oxidation in the incubates, which argues against the importance of this enzyme in cytosolic, NAD+-dependent bohemine oxidation.
Naturally, the present data do not exclude a role for other ADH classes and/or other oxidoreductases in oxidation of bohemine in mice in vivo. The most striking result of this study was a lack of metabolites detected in the organ extracts that would unequivocally imply a role of the microsomal, NADPH-dependent cytochrome P450 enzyme system in bohemine metabolism in mice. It could be speculated that microsome activity was suppressed, at least partly, through the use of pentobarbital anesthetic. However, having examined this assumption using control animals under brief ether narcosis, we were unable to show any differences between profiles of the principal metabolites in these animals and those anesthetized by pentobarbital. These facts substantiated the interpretation of our data as correctly reflecting the principal bohemine metabolites in male mice in vivo, with the glucoside conjugation being the main bohemine metabolic route. In the context of our study, it is very significant that even if mutual competition between bohemine and anesthetic agents was present in vivo, suppressing cytochrome P450s, this would only mean that we underestimated the rapidity of bohemine elimination from the mouse body, consequently underscoring our recognition of the powerful potential of mouse organs to detoxify and excrete bohemine.
This aside, we do not exclude the possibility that several metabolites were produced that remained unanalyzed by us. Highly polar minor products might be formed that did not move on TLC plates. Analysis of such spots was hampered by their immobility and by small quantities of single metabolites present in them that could not be identified and/or quantified individually. More recently, we have explored this puzzle using in vitro models such as subcellular fractions, liver cell suspensions, and precision-cut organ tissue slices, enabling us to acquire minor metabolites in sufficient amounts. Our results will be communicated in this regard in a forthcoming paper (M. Rypka, J. Veselý, Z. Chmela, L. Havlíček, K. Lemr, K. Cervenkova, B. Cerny, K. Fuksová, J. Hanuš, M. Belejova, Z. Cervinkova, H. Lotkova, J. Lukeš, and K. Michalikova, manuscript in preparation).
It can be concluded from this study that the tissue availability of bohemine in mice was grossly reduced shortly after its application, owing to fast clearance of bohemine. Significantly, bohemine carboxylic acid, the only bohemine metabolite circulating in the blood in significant amounts, can hardly substitute for the beneficial potential of metabolically unstable bohemine since it displayed only weak CDK1-inhibitory activity in our assay, in contrast to the parent compound. Thus, the remarkable CDK-inhibitory capacity of bohemine exerted in vitro may not be expressed in vivo. Hopefully, this knowledge will facilitate the design of new potent purine CDK inhibitors with higher metabolic stability.
Acknowledgments
We thank Helena Rozsypalová and Sylva Snášelováfor their excellent technical assistance. We also acknowledge P. Sedmera, M. Kuzma, and V. Havlı́cek from the Institute of Microbiology, Academy of Sciences, Czech Republic, for NMR measurements and for skillful interpretation of MS spectra. We are grateful to Jirı́ Bartek (Copenhagen, Denmark) and David P. Lane (Dundee, UK), who kindly provided us with baculoviral constructs. Last but not least, we are indebted to Jarmila Potomková and Alexander Oulton for their editorial assistance.
Footnotes
-
Send reprint requests to: Jaroslav Veselý, Dept. of Pathological Physiology, Medical Faculty, Palacký University, Hnevotı́nská 3, CZ-775 15 Olomouc, Czech Republic. E-mail:vesely{at}risc.upol.cz
-
This work was supported in part by the Ministry of Education, Youth and Sports of the Czech Republic (MSM 151100001, MSM 151100005, MSM 153100013, VS 96021, and EU COST B5/B17 Actions), the Grant Agency of the Czech Republic (204/96/K235), and Palacký University (UP 12101101).
- Abbreviations used are::
- CDK
- cyclin-dependent kinase
- ADH
- alcohol dehydrogenase
- Bq
- becquerel
- 4-MP
- 4-methylpyrazole
- TLC
- thin layer chromatography
- MS
- mass spectrometry
- Received April 26, 2000.
- Accepted November 11, 2000.
- The American Society for Pharmacology and Experimental Therapeutics