Metabolism and Pharmacokinetics of the Cyclin-Dependent Kinase Inhibitor R-Roscovitine in the Mouse

Molecular Cancer Therapeutics 125

Metabolism and pharmacokinetics of the cyclin-dependent kinase inhibitor R-roscovitine in the mouse

Bernard P. Nutley,1 Florence I. Raynaud,1 COOH-R-roscovitine could be inhibited by replacement Stuart C. Wilson,1 Peter M. Fischer,2 of metabolically labile protons with deuterium. After 1 1 Angela Hayes, Phyllis M. Goddard, 60 minutes of incubation of R -roscovitine-d9 or Steven J. McClue,2 Michael Jarman,1 R-roscovitine with mouse liver microsomes, formation of 2 1 f David P. Lane, and Paul Workman COOH-R-roscovitine-d9 was decreased by 24% compared with the production of COOH-R-roscovitine. In 1 Cancer Research UK Centre for Cancer Therapeutics, Institute of addition, the levels of R-roscovitine-d remaining were Cancer Research, Surrey, United Kingdom and 2Cyclacel Ltd., 9 Dundee, United Kingdom 33% higher than those of R-roscovitine. However, formation of several minor R-roscovitine metabolites was enhanced with R-roscovitine-d9, suggesting that metabolic Abstract switching from the major carbinol oxidation pathway had occurred. Synthetic COOH-R-roscovitine and C8-oxo- R-roscovitine (seliciclib, CYC202) is a cyclin-dependent R-roscovitine were tested in functional cyclin-dependent kinase inhibitor currently in phase II clinical trials in patients kinase assays and shown to be less active than with cancer. Here, we describe its mouse metabolism and R-roscovitine, confirming that these metabolic reactions pharmacokinetics as well as the identification of the are deactivation pathways. [Mol Cancer Ther 2005; principal metabolites in hepatic microsomes, plasma, and 4(1):125–39] urine. Following microsomal incubation of R-roscovitine at 10 Ag/mL (28 Amol/L) for 60 minutes, 86.7% of the parent drug was metabolized and 60% of this loss was due to Introduction formation of one particular metabolite. This was identified Mammalian cell division is tightly regulated by the as the carboxylic acid resulting from oxidation of the activation of the cyclin-dependent kinase (CDK) family of hydroxymethyl group of the amino alcohol substituent at cell cycle regulatory proteins (1). This regulation is required C2 of the purine core present in R-roscovitine. Identifica- for the processes that govern cell proliferation and to allow tion was confirmed by chemical synthesis and comparison DNA replication and mitosis to occur in proper sequence of an authentic sample of the R-roscovitine-derived and at the correct time. Activation of the CDKs is controlled carboxylate metabolite (COOH-R-roscovitine). Other minor by various signal transduction pathways and requires metabolites were identified as C8-oxo-R-roscovitine and 9 formation of a complex consisting of the catalytic CDK unit N -desisopropyl-R-roscovitine; these accounted for 4.9% and the appropriate regulatory cyclin subunit (2). Once and 2.6% of the parent, respectively. The same metabolic activated, these cyclin-CDK complexes in turn regulate pattern was observed in vivo, with a 4.5-fold lower AUC11 other factors (e.g., phosphorylation of the retinoblastoma for R-roscovitine (38 Amol/L/h) than for COOH-R-roscovi- protein) leading to an orderly progression through the cell tine (174 Amol/L/h). Excretion of R-roscovitine in the urine cycle. Several such cyclin-CDK complexes have been up to 24 hours post-dosing accounted for an average of identified and these were shown to regulate various points only 0.02% of the administered dose of 50 mg/kg, of the cell cycle (1). Control of the cell cycle is commonly whereas COOH-R-roscovitine represented 65% to 68% deregulated in human cancers, for example, by mutation or of the dose irrespective of the route of administration (i.v., altered expression of CDKs, cyclins, and their regulatory i.p., or p.o.). A partially deuterated derivative (R-roscovi- molecules (2). Because of this, synthetic inhibitors of CDKs tine-d9) was synthesized to investigate if formation of may provide important new cancer treatments that are more selective than many of the cytotoxic drugs currently in use (2–4). Previous reports have described the discovery of 2,6,9- trisubstituted purines that were effective at inhibiting the Received 7/29/04; revised 11/3/04; accepted 11/10/04. activity of several protein kinase classes, including the CDK Grant support: Cancer Research UK and Cyclacel Ltd. D. Lane is a Gibb family and especially CDK2 (5–8). Since the first report (5) Fellow and P. Workman is a Life Fellow of Cancer Research UK. of the selective inhibition of CDK2 by the 2,6,9-trisubsti- The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked tuted purine olomoucine (Fig. 1), there has been substantial advertisement in accordance with 18 U.S.C. Section 1734 solely to progress in the development of more potent analogues. indicate this fact. Among these, CYC202 (seliciclib), a pure and chirally Requests for reprints: Florence I. Raynaud, Cancer Research UK Centre for defined form (9) of R-roscovitine (refs. 8, 10; Fig. 1), was Cancer Therapeutics, Institute of Cancer Research, Cotswold Road, Sutton, Surrey SM2 5NG, United Kingdom. Phone: 44-20-8722-4212; chosen for development from a large set of 2,6,9-trisubsti- Fax: 44-20-8770-7899. E-mail: [email protected] tuted purine analogues (11) and is currently undergoing Copyright C 2005 American Association for Cancer Research. phase II clinical trials (12, 13).

Mol Cancer Ther 2005;4(1). January 2005

atmosphere cooled to 0jC was added diisopropylethylamine (4.0 mL, 2.64 eq, 22.96 mmol) followed by benzyl- amine (1.15 mL, 1.21 eq, 10.53 mmol). The reaction mixture was stirred at 0jC for 3 hours, allowed to return to room temperature over 30 minutes, and stirred at this temperature for 16 hours, when thin-layer chromatography with CH2Cl2/Et2O/MeOH (55:43:2) indicated that the reaction had gone to completion. The solvent was evaporated in vacuo, and the residue was purified by gradient column chromatography on silica gel eluted with CH2Cl2/Et2O/ MeOH (55:45:0–55:43:2) to afford 2 as a white solid (1.36 g, 64%). Melting point 240-241jC. 1Hnuclearmagnetic y resonance (NMR) (d6-DMSO, 250 MHz): 4.62 (d, 2H, J = 5.60 Hz, -HNCH2-Bz), 7.25-7.33 (m, 5H, Bz), 8.10 (s, 1H, -N = CH-NH-), 8.81 (brs, 1H, -HNCH2-Bz), 13.06 (brs, 1H, -N = CH-NH-). Fast atom bombardment mass spectroscopy (FABMS) m/z: 244 ([M + H]+, 100), 180 (15), 166 (9), 136 (5), 91 (10). Accurate mass (M + H): actual: 244.0998, measured: 244.1002. Microanalysis (expected/measured): C12H10N5F: C: 59.25:59.12, H: 4.14:4.06, N: 28.79:28.47. Benzyl-(2-fluoro-9-isopropyl-9H-purin-6-yl)amine (3a). To a stirred solution of 2 (0.83 g, 1 eq, 3.41 mmol) in Figure 1. Chemical structures of olomoucine (OLO), R-roscovitine dimethylacetamide (10 mL) at room temperature under an (ROS), COOH-R-roscovitine (COOH-ROS), and bohemine (BOH). argon atmosphere was added powdered, anhydrous K2CO3 (2.35 g, 5 eq, 17.00 mmol) followed by 2-bromopropane (3.2 mL, 10 eq, 34.08 mmol). The reaction mixture was stirred at room temperature for 48 hours, when thin- Identification of the biotransformation pathways is an essential component of the rational development and use of molecular therapeutics. In this article, we present the results of our in vitro and in vivo studies of the metabolism of R-roscovitine in the mouse. Our findings are supported by the synthesis of putative metabolites and of selectively deuterated derivatives. We identified the major metabolite of R-roscovitine as a carboxylic acid formed by oxidation of the hydroxymethyl moiety of the purine C2 substituent and show that the formation of this metabolite is a deactivation reaction. The results presented here contributed to the initiation of clinical trials with R- roscovitine and the basic murine pharmacokinetics and metabolism variables have been observed recently to be similar in humans (14).

Materials and Methods General SKF-525A, NADPH, and formic acid (96% ACS grade) were purchased from Sigma-Aldrich (Gillingham, Dorset, United Kingdom). High-performance liquid chromatography–grade methanol was purchased from Laserchrom (Rochester, Kent, United Kingdom). Olomoucine and R-roscovitine were prepared as described previously (8, 9). Schemes for the synthesis of other compounds used in this study are given in Fig. 2. Protein kinase assays were Figure 2. Scheme for synthesis of compounds used in this study. a, j done as described previously (12). BnNH2,iPr2NEt, n-butanol, 95 C, 4 h. b, iPr-Br (for 3a)oriso-C3D7Br (for 3b), K2CO3, DMA, room temperature, 24 h. c, 8 (for 4a)or Syntheses NH2CH(CH2CH3)COOH (for 4b), DBU, NMP, 160jC, 1 h (racemiza- Benzyl-(2-fluoro-9H-purin-6-yl)amine (2). To a stirred tion during reaction). d, DIEA, n-butanol, DMSO, CH3CH2C(NH2)CH2OH, 140jC, 48 h. DMA, -bromosuccinimide, room temperature, 16 h. solution of 6-chloro-2-fluoro-9H-purine 1 (1.5 g, 1 eq, 8.69 e, N f, DMSO, NaOH aqueous, 140jC, 32 h. g, LiAlD4, monoglyme, 95jC, 16 mmol; ref. 15) in n-butanol (100 mL) under an argon h. h, 9, DIEA, n-butanol, DMSO, 140jC, 48 h.

Mol Cancer Ther 2005;4(1). January 2005

layer chromatography with CH2Cl2/Et2O/MeOH (55:40:5) (R)-2-(6-Benzylamino-9-isopropyl-9H-purin-2-ylami- indicated that the reaction had gone to completion. The no)butan-1-ol (5a; R-Roscovitine). To a stirred solution of solvent was evaporated in vacuo, and the residue was 3a (0.50 g, 1 eq, 1.75 mmol) in n-butanol/DMSO (10 mL, partitioned between EtOAc (100 mL) and water (200 mL). 4:1) at room temperature under an argon atmosphere was The aqueous phase was extracted with more EtOAc (2 50 added diisopropylethylamine (3.0 mL, 9.82 eq, 17.22 mL), and the combined organic phases were washed with mmol) followed by (R)-()-2-aminobutan-1-ol (1.6 mL, brine (50 mL), dried (MgSO4), and evaporated in vacuo. The 10 eq, 17.0 mmol). The reaction mixture was placed in a residue was purified by silica gel column chromatography preheated oil bath at 140jC and was stirred at this eluted with CHCl3 to afford 3a as a white solid (0.78 g, temperature for 48 hours, when thin-layer chromatogra- j 1 80%). Melting point 158-160 C. H NMR (d6-DMSO, 250 phy with CH2Cl2/Et2O/MeOH (55:40:5) indicated that the y MHz): 1.49 (2 s, 6H, CH(CH3)2), 4.64 (m, 3H, - reaction had gone to completion. The reaction mixture CH(CH3)2 + -HNCH2-Bz), 7.26 (m, 5H, Bz), 8.25 (s, 1H, -N = was allowed to cool to room temperature, and the solvent CH-N-), 8.90 (brs, 1H, -HNCH2-Bz). FABMS m/z: 286 ([M + was evaporated in vacuo. The residue was partitioned + H] , 100), 242 (17), 176 (22), 154 (65), 136 (60). Accurate between CH2Cl2 (100 mL) and water (150 mL), the mass (M + H): actual: 286.1468, measured: 286.1462. aqueous phase was extracted with more CH2Cl2 (2 50 Microanalysis (expected/measured): C15H16N5F: C: mL), and the combined organic phase was washed with 63.14:63.08, H: 5.65:5.58, N: 24.54:24.46. brine (50 mL), dried (MgSO4), and evaporated in vacuo. 2 Benzyl-(2-fluoro-9-[ H7]isopropyl-9H-purin-6-yl)amine The residue was purified by gradient column chroma- (3b). Methodology as for 3a but using 2-bromopropane-d7 tography on silica gel eluted with CH2Cl2/Et2O/MeOH (1.00 g, 1.95 eq, 7.69 mmol). White solid (0.63 g, 55%). (50:50:0–50:50:1) to afford 5a as a white solid (0.58 g, j 1 y j 1 Melting point 154-155 C. HNMR(d6-DMSO, 250 MHz): 93%). Melting point 104-105 C. HNMR(d6-DMSO, 4.63 (d, 2H, J = 5.48 Hz), 7.24-7.33 (m, 5H), 8.22 (s, 1H), 8.82 250 MHz): y 0.82 (t, 3H, J =7.33Hz,-NHCH + (brs, 1H). FABMS m/z: 293 ([M + H] , 100), 286 (25), 154 (CH2CH3)CH2OH), 1.45 (m, 8H, -NHCH(CH2CH3)CH2OH (40), 138 (14), 133 (75), 107 (12). Accurate mass (M + H): +-CH(CH3)2), 3.30-3.42 (m, 2H, -NHCH(CH2CH3)CH2OH), actual: 293.1907, measured: 293.1894. Microanalysis 3.70 (m, 1H, -NHCH(CH2CH3)CH2OH), 4.51 (m, 4H, - (expected/measured): C: 61.62:61.49, H: 5.67:5.51, N: HNCH2-Bz +, OH + -CH (CH3)2), 5.80 (d, 2H, J = 8.37 Hz, 23.95:23.82. -NHCH(CH2CH3)CH2OH), 7.17-7.35 (m, 5H, Bz), 7.76 2-(6-Benzylamino-9-isopropyl-9H-purin-2-ylamino) (brs, 2H, -N = CH-N- + -HNCH2-Bz). FABMS m/z: 355 butyric acid (4a, COOH-R-Roscovitine). To a solution of ([M + H]+, 100), 323 (52), 154 (10), 134 (14). Accurate 3a (151 mg, 0.05 mmol) in 1-methyl-2-pyrrolidinone (5 mL) mass (M + H): actual: 355.2246, measured: 355.2260. and 1,8-diazabicyclo[5.4.0]undec-7-ene (1.5 mL) was added Microanalysis (expected/measured): C19H26N6O: (R)-()-2-aminobutyric acid (8, 99% ee/GLC: 1.03 g, 10 C: 64.38:64.21, H: 7.39:7.44, N: 23.71:23.37. mmol), and the mixture was stirred under nitrogen at (R)-2-(6-Benzylamino-8-bromo-9-isopropyl-9H-purin-2- 160jC for 1 hour. After cooling, the mixture was diluted ylamino)butan-1-ol (6). To a stirred solution of 5a (0.10 g, 1 with 10% aqueous citric acid and CH2Cl2 (25 mL each). The eq, 0.28 mmol) in dimethylacetamide (2.5 mL) at room phases were separated, and the organic fraction was temperature under an argon atmosphere was added N- extracted with brine (2 10 mL), dried over MgSO4, bromosuccinimide (0.06 g, 1.15 eq, 0.33 mmol). The reaction filtered, and evaporated. The residue was fractionated by mixture was stirred in the dark at room temperature for 16 preparative reverse-phase high-performance liquid chro- hours, when thin-layer chromatography with CH2Cl2/ matography. Appropriate fractions were pooled and Et2O/MeOH (55:40:5) indicated that the reaction had gone lyophilized to afford 4a (137 mg, 74%) as an amorphous to completion. The reaction mixture was cooled to 0jC, and 1 y off-white solid. H NMR (d6-DMSO, 300 MHz) 0.95 (t, J = sodium hydrosulfite solution (10% in water, 5 mL) was 7.3 Hz, 3H, CH2CH3); 1.51 (d, J = 6.7 Hz, 2H, CH2CH3); 1.78 added. The solvent was evaporated in vacuo, and the (m, J = 7.3 Hz, 2H, CH2CH3); 4.27 (m, 1H, CHCH2); 4.64 residue was partitioned between EtOAc (50 mL) and (septet, J = 6.7 Hz, 1H, CH(CH3)2); 4.69 (m, 2H, CH2Ph); saturated aqueous NaHCO3 solution (50 mL), the aqueous 7.26-7.41 (m, 5H, ArH). FABMS accurate mass [M + H]+: phase was extracted with more EtOAc (2 20 mL), and the found 369.2033, calculated for C19H25N6O2 369.2039. combined organic phase was washed with brine (20 mL), 2 2-(6-Benzylamino-9-[ H7]isopropyl-9H -purin-2- dried (MgSO4), and evaporated in vacuo. The residue ylamino)butyric Acid (4b, COOH-R-Roscovitine-d7). was purified by column chromatography on silica gel Methodology as for 4a but using 3b (0.12 g, 1 eq, 0.41 eluted with CHCl3 to afford 6 as a white solid (0.12 g, j j 1 mmol); white solid (0.081 g, 53%). Melting point 86-88 C. 98%). Melting point 54-56 C. HNMR(d6-DMSO, 250 1 y y H NMR (d6-DMSO, 250 MHz): 0.91 (t, 3H, J = 7.35 Hz), MHz): 0.82 (t, 3H, J = 7.28 Hz, -NHCH(CH2CH3)CH2OH), 1.38-1.74 (m, 2H), 4.09-4.24 (m, 1H), 4.32-4.86 (m, 3H), 6.39 1.42-1.57 (m, 8H, -NHCH(CH2CH3)CH2OH + -CH(CH3)2), (m, 1H), 7.15-7.40 (m, 5H), 7.79 (brs, 2H), 12.21 (brs, 1H). 3.30-3.43 (m, 2H, -NHCH(CH2CH3)CH2OH), 3.72 (m, + + FABMS m/z: 398 ([M + Na] , 100), 376 ([M + H] , 27), 330 1H, -NHCH(CH2CH3)CH2OH), 4.61 (m, 4H, -HNCH2-Bz (10), 176 (10), 136 (7). Accurate mass (M + Na): actual: +, OH + -CH(CH3)2), 5.92 (d, 2H, J = 7.95 Hz, - 398.2298, measured: 398.2312. Microanalysis (expected/ NHCH(CH2CH3)CH2OH), 7.18-7.33 (m, 5H, Bz), 7.95 + measured): C: 60.78:60.16, H: 6.57:6.51, N: 22.38:21.50. (brs, 1H, -HNCH2-Bz). FABMS m/z: 433 ([M + H] ,

Mol Cancer Ther 2005;4(1). January 2005

100), 403 (45), 176 (25), 154 (74), 136 (73), 107 (40). Accurate In vitro Kinase Assays mass (M + H): actual: 433.1351, measured: 433.1336. Protein kinase assays were carried out by measurement Microanalysis (expected/measured): C19H25N6OBr: C: of incorporation of radioactive phosphate from ATP into 52.66:52.77, H: 5.81:5.59, N: 19.39:18.15. appropriate polypeptide substrates by purified recombi- 2-(R)-6-Benzylamino-2-(1-hydroxymethyl-propyla- nant human protein kinases and kinase complexes as mino)-9-isopropyl-7,9-dihydro-purin-8-one (7, C8-oxo-R- described previously (12). Assays were done using 96-well Roscovitine). To a stirred suspension of 6 (0.03 g, 1 eq, 0.07 plates and appropriate assay buffers [typically 25 mmol/ mmol) in DMSO (1 mL) at room temperature in a pressure L a-glycerophosphate, 20 mmol/L MOPS, 5 mmol/L vessel was added NaOH solution (50% w/v in water; 5 EGTA, 1 mmol/L DTT, 1 mmol/L Na3VO3 (pH 7.4)] into mL). The vessel was sealed and placed in a preheated oil which were added 2 to 4 Ag of active enzyme with bath at 140jC, and the vessel contents were stirred at this appropriate substrates. The reactions were initiated by temperature for 32 hours. The reaction mixture was addition of Mg/ATP mix (15 mmol/L MgCl2 and 100 allowed to cool to room temperature, and the pH was mol/L ATP with 30–50 kBq per well of [g-32P]ATP), and adjusted to 7.0 with citric acid solution (10% in water). This mixtures were incubated as required at 30jC. Reactions was extracted with EtOAc (3 50 mL), and the combined were stopped on ice followed by filtration through p81 organic phase was washed with brine (20 mL), dried filter plates or GF/C filter plates (Whatman Polyfil- (MgSO ), and evaporated in vacuo.Theresiduewas 4 tronics, Kent, United Kingdom). After washing thrice purified by gradient column chromatography on silica gel with 75 mmol/L aqueous orthophosphoric acid, plates eluted with CH2Cl2/Et2O/MeOH (50:50:0–50:50:15) to afford 7 as a white solid (3.4 mg, 13%). Melting point were dried, scintillant was added and incorporated j 1 y radioactivity was measured in a scintillation counter 68-70 C. HNMR(d6-DMSO, 250 MHz): 0.83 (TopCount, Packard Instruments, Pangbourne, Berks, (t, 3H, J = 7.63 Hz, -NHCH(CH2CH3)CH2OH), 1.42-1.57 United Kingdom). Compounds for kinase assay were (m, 8H, -NHCH(CH2CH3)CH2OH + -CH(CH3)2), 3.31-3.40 (m, 2H, -NHCH(CH2 CH3 ) CH2 OH), made up as 10 mmol/L stocks in DMSO and diluted into 3.72 (m, 1H, -NHCH(CH2CH3)CH2OH), 4.50-4.60 10% DMSO in assay buffer. Data were analyzed using (m, 4H, -HNCH2-Bz +, OH + -CH(CH3)2), 5.60 (d, 2H, J = curve-fitting software (GraphPad Prism version 3.00 for 8.84 Hz, -NHCH(CH2CH3)CH2OH), 7.18-7.36 (m, 5H, Bz). Windows, GraphPad Software, San Diego, CA) to + FABMS m/z: 371 ([M + H] , 100), 339 (40), 323 (20), 286 determine IC50 values (concentration of test compound (17), 133 (55), 113 (10). Accurate mas s (M + H):actual: that inhibits kinase activity by 50%). 371.2195, measured: 371.2206. In vitro Metabolism Studies 2 [1- H2]-(R)-(-)-2-Amino-butan-1-ol (9). Asolutionof Mouse hepatic S9 fraction and microsomes were LiAlD4 (1.22 g, 1.5 eq, 29.05 mmol) in anhydrous mono- purchased from Tebu-bio Ltd. (Peterborough, United glyme (40 mL) cooled to 0jC was added to a stirred Kingdom). S9 fraction or microsomes (100 ALofa suspension of (R)-(=)-2-aminobutyric acid (8, 2.0 g, 1 eq, solution of 1 mg protein/mL in PBS) were mixed with 19.39 mmol) in anhydrous monoglyme (20 mL) at 0jC either R-roscovitine (100 AL of 100 Ag/mL solution) or a underanatmosphereofargon.Whenadditionwas mixture of R-roscovitine and R-roscovitine-d9 (100 AL complete, the reaction mixture was allowed to reach room solution containing 50 Ag/mL of each) and 200 AL PBS. temperature over 20 minutes, placed in a preheated oil bath For inhibitor studies, SKF-525A was added to either S9 at 95jC, and refluxed at this temperature under argon for 16 fraction or microsomes at a final concentration of hours. The solvent was evaporated in vacuo, and the residue 1 mmol/L. Reaction was started by the addition of was refluxed in acetonitrile (60 mL) for 3 hours. The solution 100 AL NADPH (5 mmol/L). Following incubation at was allowed to cool to room temperature, the precipitate 37jC for 0, 15, and 60 minutes, the reaction was was removed by filtration and washed with more acetoni- terminated by the addition of methanol (1.5 mL) and trile (3 30 mL), and the combined filtrate was evaporated thoroughly mixed. Following centrifugation, the superna- in vacuo to afford 9 as a pale yellow oil (1.34 g, 76%). 1H tant was transferred to clean tubes. Control incubations, y NMR (d6-DMSO, 250 MHz): 0.84 (t, 3H, J = 7.35 Hz), 1.03- prepared by replacing either microsomes or drug with an 1.53 (m, 2H), 2.42-2.53 (m, 1H). equivalent volume of PBS, were included and treated as 2 2 [1- H2]-2-(6-Benzylamino-9-[ H7]isopropyl-9H-purin-2- described above. ylamino)-butan-1-ol (5b, R-Roscovitine-d9). Methodology In vivo Studies as for 5a but using 3b (0.58 g, 1 eq, 1.98 mmol) and Before use, female BALB/c mice (Charles River, 2 [1- H2]-(R)-()-2-amino-butan-1-ol (9, 1.1 g, 6 eq, 12.06 Margate, United Kingdom) were housed in a maximizer mmol). White solid (0.36 g, 51%). Melting point 102- positive, individually vented caging system (Thoren, j 1 y 103 C. H NMR (d6-DMSO, 250 MHz): 0.82 (t, 3H, J = Hazleton, PA) and allowed food (Lillico, Betchwork, 7.38 Hz), 1.29-1.63 (m, 2H), 3.72-3.80 (m, 1H), 4.40-4.85 Surrey, United Kingdom) and tap water ad libitum.All (m, 3H), 5.83 (d, 1H, J = 8.50 Hz), 7.12-7.36 (m, 5H), 7.77 animal procedures complied with local and national (brs, 2H). FABMS m/z: 364 ([M + H]+, 100), 330 (28), 240 (United Kingdom Coordinating Committee for Cancer (8), 154 (25), 136 (20). Accurate mass (M + H): actual: Research) guidelines for animal welfare (16). 364.2811, measured: 364.2805. Microanalysis (expected/ Mice were given 100 mg/kg i.v. in the tail vein (0.1 measured): C: 62.78:62.87, H, 7.38:7.34, N: 23.12:22.82. mL/kg) of either R-roscovitine, R-roscovitine-d9,ora1:1

Mol Cancer Ther 2005;4(1). January 2005

mixture of R-roscovitine and R-roscovitine-d9 (i.e., 50 methanol (400 AL) to precipitate proteins and salts and mg/kg of each). For plasma pharmacokinetic studies, centrifuged, and the supernatant was transferred to clean blood was sampled by cardiac puncture under transient tubes for subsequent analysis. Details of LC/MS equip- anesthesia with halothane at 0, 0.25, 0.5, 1, 2, 4, 6, and 24 ment and conditions are given below. hours with three animals for each time point. All samples Liquid Chromatography/Mass Spectrometry and Liq- were then stored frozen at 20jC until analyzed. Levels uid Chromatography/Mass Spectrometry/Tandem of parent compound(s) and metabolites were determined Mass Spectrometry Analysis in samples prepared for pharmacokinetic analysis. Qualitative Analysis. Metabolite detection and LC/MS/ Metabolites were identified in 0.25-hour plasma samples, MS product ion spectra studies were done using a whereas plasma samples from vehicle-treated mice were Thermoseparations (Manchester, United Kingdom) used as controls. AS3000 autosampler, P4000 quaternary pump, and In separate experiments to quantify urinary excretion of UV1000 detector set to 254 nm attached to a Thermofinnigan R-roscovitine and the major metabolite COOH-R-roscovi- (Hemel Hempstead, United Kingdom) LCQ ion trap mass tine, mice were given 50 mg/kg i.v. (n = 4), i.p. (n =2), spectrometer. Samples were injected into a Supelco LC- or p.o. (n =2)R-roscovitine. Animals were kept ABZ, 50 4.6 mm, 5 Am column (Supelco, Inc., Supelco individually in metabowls for 24 hours, and food and Park, Bellefonte, PA). The mobile phase consisted of 0.1% water were provided ad libitum. To ensure urination, mice formic acid (A) and methanol (B). The gradient started were given 0.1 mL saline i.p., and urine was collected with 90:10 (A/B, v/v), which was held isocratically for into labeled tubes. Cages were then washed with double- 0.5 minute, followed by a linear increase to 10:90 (A/B, v/v) distilled water (1 mL), which was collected into the same over 6 minutes and maintained at these conditions for a tube as the urine, and stored frozen at 20jC until further 3.5 minutes; a flow rate of 1 mL/min was used analyzed. throughout the analysis. Sample Preparation Eluant from the UV detector passed, without splitting, Identification of Metabolites of R-Roscovitine and R- into a standard LCQ electrospray source operated in positive Roscovitine-d9 in Mouse Plasma, Urine, and Microsomal mode. Mass spectrometer conditions were sheath gas 80, Incubations. Pooled 0.25-hour mouse plasma samples from auxiliary gas 20 (both arbitrary units), capillary voltage 4.5 treated animals were used for liquid chromatography/ kV, and heated capillary temperature 250jC. The mass range mass spectrometry/mass spectrometry (LC/MS/MS) anal- was 50 to 650. Scan time was controlled by the ion trap and ysis to obtain collision-induced dissociation spectra of was set to a maximum injection time of 200 ms or the time metabolites to aid identification. For these studies, aliquots required to inject 2 108 ions; for each scan, the system (33 AL) of plasma from three animals were combined and automatically used whichever time was reached first. precipitated with methanol (300 AL), mixed, and centri- For MS/MS analysis, spectra were obtained by setting fuged, and the supernatant was transferred to clean tubes. the ion to be retained in the ion trap and the ion energy Further details of the analyses are given below. Aliquots (usually 30–50 V) used for fragmentation. Helium gas, (400 AL) of the microsome incubations treated as described continually present as damping gas in the trap, served as above were mixed with olomoucine (20 ALofa10Amol/L collision gas at the higher fragmentation voltages. solution) as an internal standard and then analyzed by full- Quantitative Analysis. Multiple reaction monitoring scan LC/MS (details below). LC/MS/MS analysis used a Waters (Watford, United Quantification of R-Roscovitine, COOH-R-Roscovi- Kingdom) high-performance liquid chromatography sys- tine, R-Roscovitine-d9, and COOH-R-Roscovitine-d7 in tem consisting of a WISP 717 autosampler and 600 MS Mouse Plasma and Urine Samples. Calibration curves of quaternary pump and controller. Samples were injected the four analytes were prepared by dilution of stock into the column as described in Qualitative Analysis, and solutions (2 mmol/L in DMSO) in methanol to provide a the same mobile phase was used. However, a different series of calibrant stocks. These were spiked (10 ALof gradient was employed with a lower mobile-phase flow each calibrant stock into either 100 AL control mouse rate to enhance sensitivity. The gradient (flow rate 0.6 mL/ plasma or urine) to give a calibration range of 50 to min) started with 80:20 (A/B, v/v), which was held 25,000 nmol/L in plasma. Additional stocks were isocratically for 0.5 minute followed by a linear increase prepared and used to spike control plasma (100 AL) to to 10:90 (A/B, v/v) over 6 minutes and maintained at these provide quality-control samples at nominal concentra- conditions for a further 3.5 minutes. tions of 150, 1,500, and 15,000 nmol/L. In circumstances Column eluant entered a Thermofinnigan TSQ700 triple- where urinary analyte levels were outside the calibration stage mass spectrometer without splitting via a standard range, urine samples were diluted (1:10, 1:100, or 1:1,000) Thermofinnigan electrospray source operated in positive and analyzed against calibration curves, and quality- ion mode. MS conditions were sheath gas 70 p.s.i., control samples were prepared in suitably diluted control auxiliary gas 15 (arbitrary units), capillary voltage 4.5 kV, urine. Calibrants, quality-control, and test plasma or and heated capillary 250jC. Multiple reaction monitoring urine samples (100 AL) from treated mice were all added fragmentation was achieved using an argon pressure of 1.5 to olomoucine (20 ALof10Amol/L solution) as internal mTorr and collision offset energy of 30 eV; electron standard and mixed. All samples were then treated with multiplier gain was set to 1,500 to 1,800 V.

Mol Cancer Ther 2005;4(1). January 2005

Figure 3. Extracted ion traces for plasma metabolites of R-roscovitine and R-roscovitine-d9.

The following transitions were used for all analyses: lites. These traces show that several metabolites were R-roscovitine m/z 355 to 241, R-roscovitine-d9 m/z 364 to 250, identified for which the number of deuterium atoms COOH-R-roscovitine m/z 369 to 255, COOH-R-roscovitine- present in each metabolite could be determined d7 m/z 376 to 232, and olomoucine m/z 299 to 91. (see Table 1). Identified metabolites derived from R-roscovitine were then subjected to LC/MS/MS product ion analysis as described in Materials and Methods. Table 1 lists the Results molecular ions detected and retention times of the parent Characterization of Plasma Metabolites of R-Roscovitine drug and metabolites together with major fragmentation Detection of Metabolites in Plasma Samples from Mice ions produced from these parent ions. Metabolites were Given R-Roscovitine and R-Roscovitine-d9. Plasma labeled from M1 onwards. metabolites of R-roscovitine were located by comparison Interpretation of MS/MS Spectra of R-Roscovitine and of data from control and treated samples subjected to LC/ Its Metabolites. R-roscovitine and its metabolites show a MS analysis. MS ions that could not be detected in control characteristic neutral loss of 42 Da derived from the samples but were present in plasma from treated mice were isopropyl group, which is lost as propene. The hydrox- identified. Metabolites were identified as being derived yalkyl group of R-roscovitine and some hydroxylated from R-roscovitine if they could not be detected in plasma metabolites can undergo several different fragmentation samples treated with R-roscovitine-d9 and vice versa. reactions and may lose water (neutral loss of 18 Da), Metabolites were identified in plasma extracts from mice methanol (loss of 32 Da), or the entire hydroxyalkyl moiety. given 1:1 mixture of R-roscovitine and R-roscovitine-d9. Neutral loss of these hydroxyalkyl moieties gives charac- Extracted ion traces of these metabolite ion pairs are shown teristic mass differences of 72 Da (loss of hydroxybutene) in Fig. 3. The mass difference between the lower mass ion and 88 Da (loss of dihydroxybutene). Sequential fragmen- (derived from R-roscovitine) and the higher mass ion tation reactions are indicated by the neutral loss of 60 Da (derived from R-roscovitine-d9) of each pair indicates the seen in several spectra attributable to the initial loss of number of deuterium atoms remaining in these metabo- water (18 Da) and then a propene group (42 Da) or vice

Mol Cancer Ther 2005;4(1). January 2005

Table 1. Identification of metabolites in plasma samples following administration of R-roscovitine or R-roscovitine-d9 to BALB/c mice

Compound Parent ion m/z No. deuterated Retention Product ions atoms time (min) (neutral loss from parent ion)

R-roscovitine 355 9 4.74 337 (18), 313 (42), 295 (18 + 42), 283 (72), 241 (42 + 72), 233 (122), 192 (163) M1 313 2 3.85 295 (18), 281 (32), 241 (72), 191 (122), 163 (150) M2 353 8 3.66 335 (18), 311 (42), 283 (70), 281 (72), 262 (91), 247 (110), 231 (122) M3 369 7 5.53 327 (42), 323 (46), 281 (42 + 46), 217 (152), 205 (164), 151 (218), 131 (238) M4 371 9 3.71 353 (18), 329 (42), 311 (18 + 42), 283 (88), 249 (122), 230 (141), 193 (178) M5 371 9 3.99 353 (18), 335 (36), 329 (42), 323 (48), 311 (18 + 42), 309 (62), 295 (36 + 42), 293 (78), 283 (88) M6 371 9 5.96 353 (18), 329 (42), 311 (58), 299 (72), 280 (91), 249 (122), 208 (163)

NOTE: Number of deuterium atoms in R-roscovitine-d9 derived metabolites, metabolite retention times, and MS/MS fragmentation ions. versa. These reactions are indicated in the product ion M2: m/z 353. This metabolite has a coeluting ana- spectra list of Table 1 as the sum of several smaller neutral logue derived from R-roscovitine-d9 with a m/z of 361, losses when these can be assigned. Another important loss suggesting loss of one deuterium atom. The MS/MS seen for some metabolites is a fragment of 46 Da. This spectrum (Table 1; Fig. 4A) shows neutral losses of 42 Da derives from neutral loss of formic acid (HCOOH) from a (forming isopropenyl [CH2 = CHCH3] group) producing an carboxylic acid moiety of the modified purine C2 substit- ion at m/z 311, 70, and 72 (different rearrangements of uent of R-roscovitine. hydroxybutyl group) producing ions at m/z 283 and 281 Although odd electron reactions are uncommon in together with loss of 91 Da (benzyl radical) producing an collision-induced dissociation MS/MS spectra, they can ion at m/z 262. The spectrum of deuterated analogue with a still occur (17). R-roscovitine and its metabolites show an m/z of 361 forms similar ions (see Fig. 4B), but the masses apparent neutral loss of 91 Da. This is difficult to explain if are different due to the presence of deuterium. For it is a true neutral loss and is probably due to formation of example, the ion m/z 313 is derived from loss of a an energetically stable benzyl radical (18) and a positively deuterated isopropenyl moiety (CD2 = CDCD3) of mass charged ion. 48 Da, leaving one of the deuteriums of the d7-isopropyl Identification of R-Roscovitine Metabolites from MS/ group attached to the purine nucleus of the m/z 313 ion. MS Spectra and Deuterated Analogue Experiments. This and the presence of the other deuterium in the R-Roscovitine. The compound eluting at 4.74 minutes hydroxybutyl group account for the 2 Da mass difference of has the same retention time and same spectrum as R- this ion from the m/z 311 ion seen for the m/z 353 metabolite roscovitine. R-roscovitine shows major fragmentation ions derived from R-roscovitine. The ions in the R-roscovitine- at m/z 337 (loss of water, 18 Da), 313 (loss propene from d9 spectrum at m/z 288 and 290 (neutral losses of 72 and 70 isopropyl group, 42 Da,), 295 (loss of 18 + 42 Da), 283 (loss Da, respectively) may be due to equivalent rearrangements hydroxybutene moiety from C2 substituent, 72 Da), and of the hydroxybutyl moiety but with migration of the 241 (loss of 42 + 72 Da). Other fragment ions seen included remaining deuterium atom to the C2-amino nitrogen atom. m/z 233 (122 Da), 192 (163 Da), and 91 (benzylic ion). The spectrum also shows a loss of 91 Da to give an ion of M1: m/z 313. This ion is 42 Da less than R-roscovitine, m/z 270. This is probably due to loss of a benzyl radical, which suggests loss of the isopropyl group attached to N9 which suggests any metabolic change has not occurred of the purine nucleus. The selected ion trace for this in the benzyl moiety itself. Comparison of the ions and metabolite derived from R-roscovitine-d9 (see Fig 3, m/z neutral losses for the corresponding metabolites derived 315 trace) is only 2 a.m.u. higher than the ion derived from from R-roscovitine and R-roscovitine-d9 suggests that this R-roscovitine; this can be explained by loss of the d7- metabolite is an aldehyde produced by oxidation of the isopropyl group. The MS/MS spectrum (Table 1) shows hydroxybutyl carbinol. neutral losses of 18 Da (m/z 295), 32 Da (m/z 281), and 72 Da M3: m/z 369. This major metabolite was identified as (m/z 241) from the molecular ion, which could arise from COOH-R-roscovitine. The authentic compound was syn- loss of water, methanol, and 1-hydoxybutene, respectively. thesized and characterized as described in Materials and Formation of these neutral moieties from the C2 substitu- Methods and analyzed by LC/MS and LC/MS/MS, and the ent, and the lack of any ions or neutrals from the isopropyl results were compared with those for the corresponding group, support identification of this metabolite as N9- metabolite obtained from plasma and urine samples. desisopropyl-R-roscovitine. Retention time and molecular ion for the synthetic

Mol Cancer Ther 2005;4(1). January 2005

Figure 4. MS/MS spectra of metabolite M2 identified in plasma 0.25 following i.v. administration of 50 mg/kg (A) R - roscovitine and (B) R-roscovitine-d9.

compound were identical to those for the plasma metabo- the incorporation of oxygen into these metabolites. They lite. The MS/MS spectrum of the synthetic compound (Fig. also have peaks in the m/z 380 selected ion trace derived 5A) corresponds with that of the metabolite seen in plasma from R-roscovitine-d9, which coelute with those of the m/z (see Fig. 5B) and urine (data not shown). In addition, the 371 trace and which therefore contain nine deuterium analogue of this metabolite (COOH-R-roscovitine-d7) de- atoms. Interpretation of the spectra obtained from each of rived from administration of R-roscovitine-d9 had the same these metabolites is discussed below. retention time and MS/MS spectrum (data not shown) as M4: m/z 371. The MS/MS spectrum for the first of these synthetic COOH-R-roscovitine-d7 described in Materials metabolites contains the following ions: m/z 353 ( 18 Da, andMethodsandusedasananalyticstandardfor loss of water), 329 (42 Da, loss of the propenyl group from 9 quantitative pharmacokinetic analysis. the N group), 311 (60 Da, from sequential loss of 18 and Generally, only one metabolite was detected for each of 42 Da), 283 (88 Da, loss of dihydroxybutene from C2 the pairs of ions, except for the R-roscovitine derived ion substituent, which contains additional oxygen; i.e., 72 + 16 m/z 371. Three high-performance liquid chromatography Da), and 91. These data suggest that the metabolite in peaks can be identified in this trace, suggesting the question is derived from oxidation of R-roscovitine in the presence of three isomeric metabolites. The 16 Da mass hydroxybutyl group. Although the precise position of this difference of this ion from R-roscovitine (m/z 355) suggests hydroxylation cannot be ascertained from these results, the

Mol Cancer Ther 2005;4(1). January 2005

presence of nine deuterium atoms in the metabolite derived Da), and 91. Many of the neutral losses seen in this from R-roscovitine-d9 suggests that hydroxylation has spectrum are the same as those of R-roscovitine, and the occurred on part of the hydroxybutyl side chain in which product ions are 16 Da heavier. Oxidation at C8 of the deuterium was not incorporated. purine core of R-roscovitine would account for these M5: m/z 371. This is another hydroxylated metabolite. findings. The identity of this metabolite was confirmed Although the MS/MS spectrum of this metabolite is more by synthesis of authentic C8-oxo-R-roscovitine as de- complex, it has many of the same ions as M4. It is probably scribed in Materials and Methods. The synthetic com- derived from oxidation at a different carbon atom of the pound had the same retention time and mass spectrum hydroxybutyl group of R-roscovitine to that of M4. Once (see Fig. 6A) as M6 (Fig. 6B). again, the presence of nine deuterium atoms in the R- Although other minor metabolites could be detected in roscovitine-d9-derived analogue indicates that hydroxyl- both plasma or urine, including several possible glucur- ation has not occurred at a deuterium-substituted carbon onides, interpretable MS/MS spectra could not be obtained atom. for these ions and so they are not discussed further. M6: m/z 371. This is the third hydroxylated metabolite. Microsomal Metabolism of R-Roscovitine and R- The MS/MS spectrum shows m/z 353 (18 Da, loss of Roscovitine-d9 water), 329 (42 Da, loss of propene), 299 (72 Da, loss After LC/MS analysis of microsomal incubations, peak of hydroxybutene), 280 (91 Da, benzyl radical), 257 (loss areas of internal standard, parent compound, and metab- of both propene and hydroxybutene moieties, 42 + 72 olites were obtained from the appropriate extracted ion

Figure 5. MS/MS spectra of (A) synthetic COOH-R-roscovitine and (B) corresponding metabolite detected in plasma.

Mol Cancer Ther 2005;4(1). January 2005

traces. Metabolite levels were then expressed as peak area somes in either the absence or the presence of the ratios of analyte/internal standard, and the results were cytochrome P450 (CYP) inhibitor SKF-525A (19) are shown plotted against time. Microsomal metabolism of both R- in Table 3. Control incubations of either S9 fraction or roscovitine and R-roscovitine-d9 was rapid and extensive. microsomes with R-roscovitine in which NADPH was not Results for the parent compound and six phase I added did not show formation of any appreciable levels of metabolites described earlier are shown in Table 2. Of metabolites (data not shown). The results show that these metabolites, COOH-R-roscovitine and C8-oxo-R- formation of COOH-R-roscovitine is higher in microsomes roscovitine had the highest peak area ratios. Although the than in S9 fraction. This difference is not due to increased corresponding ratio for the aldehyde intermediate was metabolism by other pathways, because the relative levels comparatively high in the 15-minute incubation, the levels of other metabolites are very similar and the levels of were 4- to 5-fold lower in the 1-hour incubations. This R-roscovitine after 60 minutes are higher in the S9 fraction is presumably due to rapid further metabolism of the than in the microsomes. In addition, incubation of S9 aldehyde to COOH-R-roscovitine. fraction or microsomes with SKF-525A led to decreased Comparative In vitro Metabolism of R-Roscovitine by formation of COOH-R-roscovitine for both cellular frac- Mouse S9 Fraction and Microsomes. Metabolite levels tions. Reduction in the formation of this metabolite is detected in incubations of mouse S9 fraction and micro- accompanied by a concomitant increase in R-roscovitine

Figure 6. MS/MS spectra of (A) synthetic C8-oxo-R- roscovitine and (B)corresponding metabolite detected in plasma.

Mol Cancer Ther 2005;4(1). January 2005

Table 2. Relative levels of metabolites of R-roscovitine and R-roscovitine-d9 in mouse liver microsomal incubations

Analogue Peak area of metabolites as % of R-roscovitine peak area at t =0

Parent M1 M2 M3 M4 M5 M6

0min R-roscovitine 100.0 0.0 0.0 0.0 0.0 0.0 0.0 R-roscovitine-d9 100.0 0.0 0.0 0.0 0.0 0.0 0.0 15 min R-roscovitine 17.6 1.9 2.6 66.1 0.3 1.0 9.6 R-roscovitine-d9 25.2 0.6 3.2 48.4 0.5 2.4 19.0 60 min R-roscovitine 15.7 1.8 0.6 70.9 0.3 0.9 8.6 R-roscovitine-d9 20.9 0.6 0.6 57.0 0.6 2.2 16.7

levels after incubation for 60 minutes. These data suggest tine of selected kinases. The results show that COOH-R- that formation of COOH-R-roscovitine is primarily micro- roscovitine is a considerably less potent inhibitor somal and NADPH dependent and that CYP is important of both CDK2 (26-fold) and CDK4 (11-fold) compared in this metabolic conversion. with R-roscovitine. Thus, metabolism to the carboxylic Interestingly, formation of other, more minor, metabo- acid represents a deactivation pathway. Conversion of lites is not so affected by SKF-525A. Indeed, C8-oxo- R-roscovitine to the C8-oxo-R-roscovitine metabolite is R-roscovitine levels are higher in mouse microsomal also a deactivation pathway for CDK2 (35 times less incubations with the inhibitor than without. In addition, active) but not for CDK4, where IC50 values are 14.6 and levels of this metabolite are very similar between S9 17 Amol/L for R-roscovitine and C8-oxo-R-roscovitine, fraction and microsomal incubations (Table 3, column respectively. M6). However, this metabolite was not formed in control Pharmacokinetics of R-Roscovitine, R-Roscovitine- incubations with no NADPH. These results suggest that d9,andtheMetabolitesCOOH-R -Roscovitine and C8-oxo-R-roscovitine is formed by a microsomal NADPH- COOH-R-Roscovitine-d7 in Mice dependent enzyme system. Figure 7 shows the plasma levels of R-roscovitine and R- Comparative Kinase Inhibitory Activities of R- roscovitine-d9 as well as the respective carboxylic acid Roscovitine, COOH-R -Roscovitine, and C8-oxo- metabolites. Plasma pharmacokinetic variables derived R-Roscovitine from quantitative analysis by multiple reaction monitoring

Table 4 shows the IC50 values for the inhibition by R- LC/MS are summarized in Table 5. There was no roscovitine, COOH-R-roscovitine, and C8-oxo-R-roscovi- difference in R-roscovitine-d9 pharmacokinetic variables,

Table 3. Quantitative data for the metabolism of R-roscovitine by incubation data for mouse microsomes and S9 fraction in the absence and presence of the CYP inhibitor SKF525A (I)

Fraction Time (min) Peak area of metabolites as % of R-roscovitine peak area at t =0

Parent M1 M2 M3 M4 M5 M6

I() Microsomes 0 100.0 0.0 0.0 0.0 0.0 0.0 0.0 15 17.6 3.6 0.5 83.5 0.6 1.6 4.4 60 13.3 2.6 0.4 60.0 0.4 1.1 4.9 S9 0 100.0 0.0 0.0 0.0 0.0 0.0 0.0 15 45.7 3.8 1.6 49.4 1.1 3.0 4.4 60 30.0 4.1 0.9 52.0 0.7 2.1 4.4 I (+) Microsomes 0 100.0 0.0 0.0 0.0 0.0 0.0 0.0 15 73.3 4.3 4.7 17.9 1.6 3.6 8.3 60 67.7 4.7 5.8 26.6 2.1 3.6 8.8 S9 0 100.0 0.0 0.0 0.0 0.0 0.0 0.0 15 74.5 3.3 3.3 17.6 0.8 1.3 5.5 60 54.7 3.8 1.6 18.5 1.1 2.3 4.9

Mol Cancer Ther 2005;4(1). January 2005

Table 4. Inhibitory activity on selected kinases for R-roscovitine, COOH-R-roscovitine, and C8-oxo-R-roscovitine

Enzyme IC50 (Amol/L)

R-roscovitine COOH-R-roscovitine C8-oxo-R-roscovitine

CDK2/cyclin E 0.049 F 0.037 1.27 F 0.17 1.7 CDK4/cyclin D1 14.6 F 6.0 160 F 20 17.0 CDK1/cyclin B 1.9 F 0.1 >50 ND Protein kinase A >20 >50 ND ERK-2 3.7 F 1.9 >50 ND p70 S6 kinase >20 >50 ND SAPK2a >20 >50 ND Casein kinase >20 >50 ND Protein kinase C >20 >50 ND

Abbreviation: ND, not determined.

including AUC0-1 or Cmax, compared with R-roscovitine. of the administered dose, are shown in Table 6. These data However, the AUC1 for COOH-R-roscovitine was 35% show that, irrespective of administration route, only trace amounts of excreted R-roscovitine were present as the higher and the Cmax 40% higher than for COOH-R- parent (0.01–0.05%), whereas the majority of the dose roscovitine-d7. These data suggest that the presence of (65.2–68.5%) was excreted as the carboxylic acid. Thus, this deuterium decreases in vivo metabolism of R-roscovitine- metabolic conversion is an important factor affecting d to COOH-R-roscovitine-d compared with the non- 9 7 clearance of R-roscovitine in mice. deuterated analogues, but this does not result in

increased levels of circulating R-roscovitine-d9 levels compared with R-roscovitine. One possible explanation of this was provided by the semiquantitative analysis of Discussion minor metabolites, which indicated that levels of all three We have shown in detail the metabolism of R-roscovitine in the mouse. A putative metabolic pathway for this com- hydroxylated metabolites were f2-fold higher for R- pound is shown in Fig. 8. Identification of two major roscovitine-d when compared with R-roscovitine (data 9 metabolites was confirmed by synthesis of the authentic not shown). This suggests metabolic switching to compounds. LC/MS and LC/MS/MS were used to show alternative pathways. identical chromatographic and MS/MS fragmentation Urinary Excretion of R-Roscovitine and COOH-R- behavior of the analytic standards with the plasma and Roscovitine urinary metabolite. The formation of the most abundant Results for the urinary excretion of R-roscovitine and the metabolite COOH-R-roscovitine represents a deactivation metabolite COOH-R-roscovitine, expressed as a percentage reaction, as the metabolite is less biologically active than the parent compound at inhibiting CDK2/cyclin E and CDK4/cyclin D1 (26- and 11-fold, respectively). The significance of this deactivation pathway is emphasized by the observation that the carboxylate is the major excretion product with 68% of the administered dose (50 mg/kg i.v.) being recovered in the 0- to 24-hour urine samples. The importance of this same metabolic pathway has also been noted in healthy volunteers in whom nonsaturable first-pass metabolism was found to produce the carboxylate which again was renally cleared as we have shown in the mouse (14). Metabolic conversion of xenobiotics that contain a primary aliphatic alcohol group to carboxylic acids has been observed on many occasions (20). Either initial conversion of R-roscovitine to an aldehyde or direct conversion to the carboxylic acid is mediated by a Figure 7. Plasma pharmacokinetic profiles for R-roscovitine, R-roscovi- NADPH-dependent system. In either case, the low levels tine-d9, COOH-R-roscovitine, and COOH-R-roscovitine-d7 in mice after i.v. administration of R-roscovitine and R-roscovitine-d9 (1:1) at a total dose of of the R-roscovitine-derived aldehyde metabolite detected 100 mg/kg. and its apparent rapid further oxidation to the carboxylate

Mol Cancer Ther 2005;4(1). January 2005

Table 5. Pharmacokinetic variables for R-roscovitine, R-roscovitine-d9, COOH-R-roscovitine, and COOH-R-roscovitine-d7 in mice given R-roscovitine and R-roscovitine-d9 (1:1) at a total dose of 100 mg/kg

Compound Cmax (observed) tmax AUClast t1/2 Ez AUC1 (observed) Cl(observed) Vz (observed) (nmol/L) (h) (nmol/L h) (h) (nmol/L h) (L/h) (L)

R-roscovitine 21,713 0 38,123 3.54 38,205 0.07 0.38 R-roscovitine-d9 20,321 0 40,256 3.47 40,351 0.07 0.35 COOH-R-roscovitine 97,298 1 173,901 2.08 173,955 0.02 0.05 COOH-R-roscovitine-d7 57,514 1 113,540 1.99 113,563 0.03 0.07

(Tables 2 and 3) suggest that pharmacologic complications oxidase. Studies have shown that these enzymes have arising from this potentially reactive metabolite are different substrate specificities and subcellular localiza- unlikely to be significant. Chmela et al. (21) suggested that tions. Xanthine and aldehyde oxidase are present in the a cytosolic class 1 alcohol dehydrogenase was responsible soluble cytosol, unlike the microsomal mixed function for conversion of bohemine, a structural analogue of R- oxidase. As C8 hydroxylation of R-roscovitine occurs in roscovitine (see Fig. 1), to an analogous carboxylic acid both microsomes and S9 fraction (which contains micro- metabolite. We have found that metabolism of R-roscovi- somal enzymes) is NADPH dependent but does not seem tine is primarily microsomal, is inhibited in the presence of to be inhibited by SKF-525A (see Table 3), it is possible that the CYP inhibitor SKF-525A (19), and is NADPH depen- more than one enzyme system may be responsible for this dent. This suggests that metabolism of R-roscovitine, at metabolic conversion. least in mouse microsomes, is principally CYP mediated. A Literature reports of the metabolism of O6-benzylgua- subsequent study (22) showed that bohemine carboxylation nine, which shares some structural features with the 2,6,9- was sensitive to certain CYP enzyme inhibitors and trisubstituted purines, have shown that C8 hydroxylation is concluded that CYP2A and CYP3A contributed substan- an important metabolic pathway in mice, rats, and man tially to this biotransformation. In the case of R-roscovitine, (24–28). In vitro experiments have shown hepatic forma- it remains to be determined which CYP enzymes are tion of O6-benzyl-8-oxoguanine is primarily microsomal responsible for the formation of the major metabolite and NADPH dependent in both rats and man (24 25). COOH-R-roscovitine. Further studies are required to elucidate, in more detail, the Oxidative N-dealkylation is a metabolic pathway seen role of different enzyme systems in the metabolism of with many xenobiotics and is usually due to activity of R-roscovitine. the microsomal mixed function oxidase system (20). Carbon-deuterium bonds are well known to be harder Our results show that the isopropyl group of R-roscovitine to break than equivalent carbon-hydrogen bonds. The is susceptible to N-dealkylation. However, this is a difference in energy requirements has the effect of slowing relatively minor metabolic pathway for these substituted some reaction processes; this is known as the deuterium purines as shown by the low levels of the dealkylated isotope effect (18). For enzymatic reactions, this slowing metabolite produced in microsomal incubations of R- effect can be detected by changes in the rate and extent of roscovitine. product formation. A 2- to 5-fold change in the rate of some Another route of metabolism observed with R-roscovi- metabolic pathways has been shown previously where tine was hydroxylation, especially at the C8 position of the carbon-deuterium bonds are involved in the metabolic purine nucleus. Metabolism of purine analogues at this process (29). For this reason, we synthesized the deuterated position has been seen for several compounds, including analogue R-roscovitine-d9 and compared its quantitative caffeine and theophylline (23). Several enzyme systems and qualitative metabolism with that of R-roscovitine. may be involved in this reaction, including xanthine The difference in formation of metabolites seen between oxidase, aldehyde oxidase, and CYP mixed function R-roscovitine and R-roscovitine-d9 can mostly be explained

Table 6. Urinary excretion for R-roscovitine and the major metabolite COOH-R-roscovitine following i.v., i.p., or p.o. administration of 50 or 100 mg/kg R-roscovitine (expressed as a percentage of administered dose)

Administration route No. animals Percentage of dose (mean F SD) excreted as

R-roscovitine COOH-R-roscovitine i.v. 4 0.05 F 0.06 67.98 F 11.04 i.p. 2 0.01 F 0.01 68.53 F 5.63 p.o. 2 0.01 F 0.00 65.22 F 7.39

Mol Cancer Ther 2005;4(1). January 2005

less active at inhibiting CDK2 than R-roscovitine and so would not be expected to result in an overall improvement in activity in vivo of R-roscovitine-d9 compared with R-roscovitine. We have reported previously that plasma clearance of R-roscovitine is lower than that of the analogue bohemine (0.14 and 1.27 L/h, respectively; ref. 11). Studies on the metabolism of bohemine identified glycosidation as an important metabolic pathway (21, 22). Furthermore, a recent report (30) suggests that the aliphatic hydroxyl groups present in olomoucine, bohemine, and R-roscovitine can be conjugated with glycoside donors, especially UDP-glucose and UDP-glucuronic acid. Glucuronidation of bohemine was detected in mouse, rat, monkey, and human microsomes in the presence of UDP-glycosyl donors, Figure 8. Putative scheme for the metabolism of R-roscovitine. whereas extensive glucosidation seemed to be confined to mouse microsomes. Furthermore, the order of susceptibil- ity to both glucosidation and glucuronidation in mouse microsomes was bohemine >> R-roscovitine > olomoucine by the presence of deuterium leading to metabolic switch- (30). Although detailed phase II microsomal incubations ing between alternative pathways. Interestingly, formation have not been carried out, glycosilated and glucuronidation of R-roscovitine aldehyde did not seem to be inhibited by products were only detected as traces in mice urine and the presence of deuterium, although breaking of a carbon- represented <2% of the administered dose. deuterium bond was involved in this reaction (Table 2, The high plasma clearance observed together with the column M2). However, formation of COOH-R-roscovitine high levels of carboxylic acid formed that we observed in from the aldehyde, which involves the breaking of another the mouse were confirmed in subsequent studies in human carbon-deuterium bond, was inhibited (Table 2, column healthy volunteers and were predictive of both parent and M3). Comparison of the peak area ratios for the analogues metabolite levels seen in humans (14). The presence of of this metabolite shows that f14% less COOH-R- minor metabolites, however, was not examined in the roscovitine-d7 was formed from R-roscovitine-d9 compared human studies. with COOH-R-roscovitine derived from R-roscovitine after Metabolism studies provide important information for incubation for 1 hour. a modern drug development program. Identification of That inhibition of this major metabolic pathway by the major metabolic pathways of candidate drugs can help presence of deuterium led to metabolic switching was drive drug design, especially when metabolism is an shown by increased formation of several minor metabo- important factor affecting bioavailability or clearance. lites, most notably C8-oxo-R-roscovitine-d9, for which peak Traditionally, identification of metabolites has relied on area ratios were double those seen for the corresponding (a) extraction of potential metabolites from biological R-roscovitine metabolite (Table 2, column M6). One fluids, such as plasma and urine; (b) identification by additional metabolite showed reduced formation from R- mass spectrometry; (c) chemical synthesis and character- roscovitine-d9 and is derived from breakage of a carbon- ization of all potential metabolite standards by spectro- deuterium bond; this was N-d7-desisopropyl-R-roscovitine scopic means; and (d) coelution of metabolites and (Table 2, column M1). About 33% of this metabolite standards following chromatographic separation. Howev- were formed from R-roscovitine-d9 compared with R- er, with the need for speedy provision of metabolism roscovitine. data to facilitate drug design, insufficient time or The pharmacokinetic profile of R-roscovitine following resources usually limit the amount of interpretable data i.v. administration confirms that clearance is mainly a result available. We have attempted to overcome some of these of metabolism to COOH-R-roscovitine. Although the levels problems by using deuterium-labeled analogues of of the latter were slightly decreased when R-roscovitine-d9 R-roscovitine. Mass differences of metabolites derived was given, this did not result in an appreciable increase in from R-roscovitine and R-roscovitine-d9 were used to either Cmax or AUC for the parent compound. A major determine the number of deuterium ions present in the change in pharmacokinetics was not expected for metabolites and to help elucidate the major metabolic R-roscovitine-d9 compared with R-roscovitine given that pathways. the rate of conversion to the carboxylate was not greatly In conclusion, we have characterized the major mouse decreased. In addition, the fact that levels of the hydroxyl- urinary metabolite of R-roscovitine as the carboxylic acid ated metabolites were increased (data not shown) suggests analogue and identified several other metabolites of this that metabolic switching, as described above for microsom- compound. Metabolism is both rapid and extensive, with al incubations, did occur in the mouse. Unfortunately, the COOH-R-roscovitine accounting for the majority of urinary major hydroxylated metabolite (C8-oxo-R-roscovitine) is metabolites. These data suggest that metabolism is an

Mol Cancer Ther 2005;4(1). January 2005

important factor affecting the plasma clearance of R- 14. De la Motte S, Gianella-Borradori A. Pharmacokinetic model of R-roscovitine and its metabolite in healthy male subjects. Intl J Clin roscovitine. Metabolism to the carboxylic acid is a Pharmacol Ther 2004;42:232 – 9. deactivation pathway, because this metabolite is a less 15. Gray NS, Kwon S, Schultz PG. Combinatorial synthesis of 2,9- potent CDK inhibitor than the parent compound. disubstituted purines. Tetrahedron Lett 1997;38:1161 – 4. 16. Workman P, Twentyman P, Balkwill F, et al. United Kingdom Co- ordinating Committee on Cancer Research (UKCCCR) Guidelines for the Acknowledgments Welfare of Animals in Experimental Neoplasia. 2nd ed. Br J Cancer 1998;77:1 – 10. We thank Dr. Ted Mc Donald for valuable discussions. 17. Busch KL, Glish GL, McLuckey SA. Mass spectrometry/mass spectrometry: techniques and applications of tandem mass spectrometry. References New York (NY): VCH Publishers; 1988. 18. Morrison RT, Boyd RN. Organic chemistry. 5th ed. Boston (MA): Allyn 1. Grana X, Reddy EP. Cell cycle control in mammalian cells: role of and Bacon, Inc.: 1987; p. 1370. cyclins, cyclin dependent kinases (CDKs), growth suppressor genes and cyclin-dependent kinase inhibitors (CKIs). Oncogene 1995;11:211 – 9. 19. Netter KJ. Inhibition of oxidative drug metabolism in microsomes. Pharmacol Ther 1980;10:515 – 35. 2. Garrett MD, Fattaey A. CDK inhibition and cancer therapy. Curr Opin Genet Dev 1999;9:104 – 11. 20. Jakoby W, Bend JR, Caldwell J. Metabolic basis of detoxication: metabolism of functional groups. New York (NY): Academic Press; 1982. 3. Brooks G, La Thangue N. The cell cycle and drug discovery: the promise and the hope. Drug Design Today 1999;4:455 – 64. 21. Chmela Z, Vesely J, Lemr K, et al. In vivo metabolism of 2,6,9- trisubstituted purine-derived cyclin-dependent kinase inhibitor bohemine in 4. Sielecki TM, Boylan JF, Benfield PA, Trainor GL. Cyclin-dependent kinase inhibitors: useful targets in cell cycle regulation. J Med Chem mice: glucosidation as the principal metabolic route. Drug Metab Dispos 2001;29:326 – 34. 2000;43:1 – 18. 5. Vesely J, Havlicek L, Strnad M, et al. Inhibition of cyclin-dependent 22. Rypka M, Vesely J, Chmela Z, et al. In vitro biotransformation of kinases by purine analogues. Eur J Biochem 1994;224:771 – 86. 2,6,9-trisubstituted purine-derived cyclin-dependent kinase inhibitor bohemine by mouse liver microsomes. Xenobiotica 2002;32:1017 – 103. 6. Legraverend M, Ludwig O, Bisagni E, Leclerc S, Meijer L. Synthesis of C2 alkynylated purines, a new family of potent inhibitors of cyclin- 23. Testa B, Jenner P. Drug metabolism: chemical and biochemical dependent kinases. Bioorg Med Chem Lett 1998;8:793 – 8. aspects. New York (NY): Marcel Dekker; 1976. 6 7. Legraverend M, Ludwig O, Bisagni E, et al. Synthesis and in vitro 24. Roy SK, Gupta E, Dolan ME. Pharmacokinetics of O -benzylguanine in evaluation of novel 2,6,9-trisubstituted purines acting as cyclin rats and its metabolism by rat liver microsomes. Drug Metab Dispos D-dependent kinase inhibitors. Bioorg Med Chem Lett 1999;7:1281 – 93. 1995;23:1394 – 9. 8. Havlicek L, Hanus J, Vesely J, et al. Cytokinin-derived cyclin- 25. Roy SK, Korzekwa KR, Gonzalez FJ, Moschel RC, Dolan ME. Human 6 dependent kinase inhibitors: synthesis and cdc2 inhibitory activity of liver oxidative metabolism of O -benzylguanine. Biochem Pharmacol olomoucine and related compounds. J Med Chem 1997;40:408 – 12. 1995;50:1385 – 9. 9. Wang S, McClue SJ, Ferguson JR, et al. Synthesis and configuration of 26. Dolan ME, Chae MY, Pegg AE, Mullen JH, Friedman HS, Moschel RC. 6 6 the cyclin-dependent kinase inhibitor roscovitine and its enantiomer. Metabolism of O -benzylguanine, an inactivator of O -alkylguanine-DNA Tetrahedron: Asymmetry 2001;12:2891 – 4. alkyltransferase. Cancer Res 1994;54:5123 – 30. 10. Whittaker SR, Walton MI, Garrett MD, Workman P. The cyclin- 27. Berg SL, Murry DJ, McCully CL, Godwin K, Balis FM. Pharmacoki- dependent kinase inhibitor CYC202 (R-roscovitine) inhibits retinoblastoma netics of O6-benzylguanine and its active metabolite 8-oxo-O 6-benzylgua- protein phosphorylation, causes loss of cyclin D1, and activates the nine in plasma and cerebrospinal fluid after intrathecal administration mitogen-activated protein kinase pathway. Cancer Res 2004;64:262 – 72. of O 6-benzylguanine in the nonhuman primate. Clin Cancer Res 1998;4:2891 – 4. 11. Raynaud FI, Fischer PM, Nutley BP, Goddard PM, Lane DP, Workman P. Cassette dosing pharmacokinetics of a library of 2,6,9-trisubstituted 28. Dolan ME, Roy SK, Fasanmade AA, Paras PR, Schilsky RL, Ratain MJ. purine CDK2 inhibitors prepared by parallel synthesis. Mol Cancer Ther O 6-benzylguanine in humans: metabolic, pharmacokinetic, and pharma- 2004;3:353 – 62. codynamic findings. J Clin Oncol 1998;16:1803 – 10. 12. McClue SJ, Blake D, Clarke R, et al. In vitro and in vivo antitumor 29. Farmer PB, Foster AB, Jarman M, Oddy MR, Reed DJ. Synthesis, properties of the cyclin dependent kinase inhibitor CYC202 (R-roscovi- metabolism, and antitumor activity of deuterated analogues of 1-(2- tine). Int J Cancer 2002;102:463 – 8. chloroethyl)-3-cyclohexyl-1-nitrosourea. J Med Chem 1978;21:514 – 20. 13. Fischer PM, Gianella-Borradori A. CDK inhibitors in clinical develop- 30. Cervenkova K, Belejova M, Chmela Z, et al. In vitro glycosidation ment for the treatment of cancer. Expert Opin Investig Drugs potential towards olomoucine-type cyclin-dependent kinase inhibitors in 2003;12:955 – 70. rodent and primate microsomes. Physiol Res 2003;52:467 – 74.

Mol Cancer Ther 2005;4(1). January 2005

Downloaded from mct.aacrjournals.org on September 28, 2021. © 2005 American Association for Cancer Research. Metabolism and pharmacokinetics of the cyclin-dependent kinase inhibitor R-roscovitine in the mouse

Bernard P. Nutley, Florence I. Raynaud, Stuart C. Wilson, et al.

Mol Cancer Ther 2005;4:125-139.

Updated version Access the most recent version of this article at: http://mct.aacrjournals.org/content/4/1/125

Cited articles This article cites 24 articles, 7 of which you can access for free at: http://mct.aacrjournals.org/content/4/1/125.full#ref-list-1

Citing articles This article has been cited by 9 HighWire-hosted articles. Access the articles at: http://mct.aacrjournals.org/content/4/1/125.full#related-urls

E-mail alerts Sign up to receive free email-alerts related to this article or journal.

Reprints and To order reprints of this article or to subscribe to the journal, contact the AACR Publications Subscriptions Department at [email protected].

Permissions To request permission to re-use all or part of this article, use this link http://mct.aacrjournals.org/content/4/1/125. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.