Abstract
Ruboxistaurin (LY333531), a potent and isoform-selective protein kinase C β inhibitor, is currently undergoing clinical trials as a therapeutic agent for the treatment of diabetic microvascular complications. The present study describes the disposition and metabolism of [14C]ruboxistaurin following administration of an oral dose to dogs, mice, and rats. The study revealed that ruboxistaurin was highly metabolized in all species. Furthermore, the results from the bile duct-cannulated study revealed that ruboxistaurin was well absorbed in rats. The primary route of excretion of ruboxistaurin and its metabolites was through feces in all species. The major metabolite detected consistently in all matrices for all species was the N-desmethyl metabolite 1, with the exception of rat bile, in which hydroxy N-desmethyl metabolite 5 was detected as the major metabolite. Other significant metabolites detected in dog plasma were 2, 3, 5, and 6 and in mouse plasma 2, 5, and 19. The structures of the metabolites were proposed by tandem mass spectrometry with the exception of 1, 2, 3, 5, and 6, which were additionally confirmed either by direct comparison with authentic standards or by nuclear magnetic resonance spectroscopy. To assist identification by nuclear magnetic resonance spectroscopy, metabolites 3 and 5 were produced via biotransformation using recombinant human CYP2D6 and, likewise, metabolite 6 and compound 4 (regioisomer of 3 which did not correlate to metabolites found in vivo) were produced using a microbe, Mortierella zonata. The unambiguous identification of metabolites enabled the proposal of clear metabolic pathways of ruboxistaurin in dogs, mice, and rats.
Protein kinase C (PKC) is a family of phospholipid-dependent serine/threonine-specific intracellular enzymes, which regulate a variety of cellular functions including cell growth, metabolism, and differentiation (Nishizuka, 1986, 1988; Farago and Nishizuka, 1990). Uncontrolled activation of PKC has been implicated in the progression of numerous disease processes (Bradshaw et al., 1993). Mammalian PKC, grouped into three subclasses, consists of approximately 13 isoforms (Mellor and Parker, 1998) that differ in structure and cofactor requirements (Newton, 2003). Studies have shown that β I and β II isoform-specific activation of PKC has been implicated in the development and progression of several diabetic microvascular complications, including diabetic retinopathy (Ishii et al., 1996; Suzuma et al., 2002; Yokota et al., 2003; Curtis and Scholfield, 2004). In in vivo preclinical models, ruboxistaurin (LY333531) has been found to be an orally effective PKCβ-selective inhibitor with nanomolar potency (Jirousek et al., 1996). Ruboxistaurin has demonstrated efficacy at nontoxic doses in preventing the development and progression of diabetic retinopathy in preclinical animal models (Ishii et al., 1996; Aiello et al., 1997; Danis et al., 1998; Nakamura et al., 1999), was well tolerated, and improved diabetes-induced retinal blood flow abnormalities in patients (Aiello et al., 1999). Ruboxistaurin is currently in phase III clinical trials for the treatment of diabetic microvascular complications.
Previously, the disposition of ruboxistaurin (LY333531) and its N-desmethyl metabolite 1 (major circulating and equally active as that of ruboxistaurin) in rats and beagle dogs following administration of a single oral dose was reported (Burkey et al., 2002). The study found, in general, that disposition of [14C]ruboxistaurin was comparable in rats and dogs. The primary route of excretion in both species was found to be fecal with a substantial biliary component. However, that report did not identify the various phase I and II metabolites detected in plasma and excreta. Therefore, the aim of the present study was to explore in detail the in vivo metabolism following oral administration of [14C]ruboxistaurin to dogs, mice, and rats.
Materials and Methods
Materials. Ruboxistaurin mesylate (Jirousek et al., 1996), two forms of [14C]ruboxistaurin mesylate (O'Bannon et al., 2004), compounds 1 and 2 (Jirousek et al., 1996), and compounds 6 and 7 were prepared at Eli Lilly and Company (Indianapolis, IN). [14C]Ruboxistaurin with two radiocarbons on the C-2 position of the two indole ring carbons (form I; radiochemical purity 98.4% and specific activity 37.2 μCi/mg) was used in the mouse and rat studies, and [14C]ruboxistaurin with radiocarbon on one of the carbonyl carbons of the maleimide group (form II; radiochemical purity 98.8% and specific activity 97.3 μCi/mg) was used in the dog study. Figure 1 shows the structure of [14C]ruboxistaurin. The microorganism Mortierella zonata was obtained from the culture collection of Eli Lilly and Company. Recombinant human CYP2D6 and CYP2D6 reaction mixture were purchased from BioCatalytics Inc. (Pasadena, CA). All other reagents and solvents were either analytical grade or HPLC grade. Male Fischer 344 rats, male CD-1 mice, and female beagle dogs were obtained from Taconic Farms (Germantown, NY), Charles River Laboratories, Inc. (Wilmington, MA), and Marshall Farms (North Rose, NY), respectively.
Dosing and Sample Collection. A single dose of [14C]ruboxistaurin mesylate (form I), suspended in 10% acacia in water, was administered to mice (75 mg/kg; n = 24), rats (5 mg/kg; n = 9), and bile duct-cannulated rats (5 mg/kg; n = 6) by oral gavage. A single dose of [14C]ruboxistaurin mesylate (form II), suspended in 10% acacia and 0.05% Dow Corning Antifoam 1510-US in water, was administered to female beagle dogs (10 mg/kg; n = 4) orally in a capsule. From mice, whole blood was collected by cardiac puncture into heparinized tubes from six mice per time point at 4 and 12 h postdose, and plasma samples were pooled by time point. Urine and feces were collected at 0 to 24, 24 to 48, and 48 to 72 h. From rats, whole blood was collected as described above from three rats per time point at 2, 4, and 8 h. Bile was collected from bile duct-cannulated rats at predose and at 0 to 12, 12 to 24, 24 to 36, and 36 to 48 h postdose. Urine and feces were collected from each cannulated animal at 0 to 24 and 24 to 48 h. From dogs, blood was collected via the jugular vein before dosing and at 2, 4, 8, 16, 24, and 48 h after dosing. Urine and feces were collected at approximately 24, 48, 72, 96, and 120 h. All samples were stored frozen at –70°C until analysis.
Analysis of Radioactivity. Radioactivity in plasma, bile, and urine was determined by liquid scintillation counting on a Tri-Carb model 2300 TR liquid scintillation spectrometer (PerkinElmer Life and Analytical Sciences, Boston, MA) by adding Ultima Gold scintillation fluid to a known amount of the sample. Radioactivity in feces was determined by combustion of a known amount of fecal homogenate with a PerkinElmer Sample Oxidizer model 307 (PerkinElmer Life and Analytical Sciences) followed by liquid scintillation counting. Determination of radioequivalents, data acquisition and storage, and statistical analysis were performed using ADME/WINPET and ADME/LIMS systems (Eli Lilly and Company).
Metabolite Profiling.Sample Preparation. Mouse. Pooled urine samples (approximately 0.5 ml of each sample per time point) and pooled plasma samples (1 ml) were extracted on an Oasis HLB SPE column (60 mg; Waters, Milford, MA), dried, and reconstituted with 300 μl of 1:2 acetonitrile/10 mM aqueous ammonium acetate. The extraction recovery of radioactivity was 100 and 80% for urine and plasma, respectively. Pooled fecal homogenate samples (2 g) per time point were extracted with acetonitrile followed by 1:1 acetonitrile/water. The combined extract was dried and reconstituted as described above. The extraction recovery of radioactivity was 90%.
Rat. Urine and bile samples were analyzed directly. Plasma samples (0.5 ml from each plasma sample per time point) were extracted with an Oasis HLB SPE column, dried, and reconstituted with 250 μl of 1:10 acetonitrile/10 mM aqueous ammonium acetate. Fecal homogenates (0.2 g, each) were extracted with acetonitrile, dried, and reconstituted with 300 μl of 1:2 acetonitrile/10 mM aqueous ammonium acetate. The extraction recovery of radioactivity from plasma and feces was 80%.
Dog. Urine samples (50 μl each) were analyzed directly. Plasma samples (400 μl, each) were extracted with acetonitrile, dried, and reconstituted with 1:4 acetonitrile/10 mM aqueous ammonium acetate. Fecal homogenates (0.3–0.8 g) were extracted with acetonitrile, dried, and reconstituted with 1:2 acetonitrile/10 mM aqueous ammonium acetate. The extraction recovery of radioactivity from plasma and feces was 94 and 93%, respectively.
LC/Radioactivity Analysis. Chromatography was performed on a Waters 2690 Separations Module using either a YMC Basic S5 column (4.6 × 150, 5-μm particle size) or a Phenomenex Synergi Polar-RP column (4.6 × 250 mm, 4-μm particle size; Phenomenex, Torrance, CA) with a flow rate of 1 ml/min. The solvent system consisted of 10 mM aqueous ammonium acetate (mobile phase A) and acetonitrile (mobile phase B), and the analytes were eluted using a gradient profile: 0 min/10% B, 50 min/60% B, 50.1 min/80% B, 52 min/80% B, or a similar gradient profile. Radiodetection was performed either using a Berthold LB509 Radioflow detector (Berthold Technologies, Oakridge, TN) by means of a 500-μl liquid flow cell with a flow rate of liquid scintillant (Ultra Flo AP; PerkinElmer Life and Analytical Sciences) to mobile phase ratio of 3:1, or by collecting the column effluent into a Deepwell Luma-96 solid scintillation-coated plate (PerkinElmer Life and Analytical Sciences), which then was dried and sealed with PerkinElmer Topseal A microplate film. The radioactivity from the plates was analyzed by a Topcount NXT microplate scintillation and luminescence counter (PerkinElmer Life and Analytical Sciences). The HPLC column recovery of radioactivity ranged from 90 to 116% for all matrices.
LC/MS Analysis. The HPLC system consisted of a Shimadzu VP series (Shimadzu Scientific Instruments Inc., Columbia, MD) including a SIL-10AXL autosampler and a Surveyor photodiode array detector. Chromatographic separations were carried out either on a Phenomenex Synergi Polar-RP column (3.0 × 150 mm, 4-μm particle size) with a flow rate of 0.4 ml/min or a Waters YMC Basic column (3.0 × 150 mm, 5-μm particle size). The mobile phase consisted of 10 mM aqueous ammonium acetate (mobile phase A) and acetonitrile (mobile phase B), and the analytes were eluted using a gradient profile: 0 min/10% B, 5 min/10% B, 50 min/75% B, 50.1 min/90% B, 52 min/90% B. Mass spectrometric analysis was conducted on a Finnigan TSQ Quantum mass spectrometer (Thermo Electron Corporation, Waltham, MA) equipped with an electrospray ion source. Accurate mass determination was performed using a Waters Micromass Q-TOF II quadrupole/orthogonal time-of-flight mass spectrometer. The protonated ion (m/z 311.0814) of sulfadimethoxine (Sigma-Aldrich, St. Louis, MO) was used as the lock mass in all accurate mass determinations.
Microbial Transformation of Ruboxistaurin. Frozen microbial stock culture of M. zonata (0.3 ml to each flask) was incubated at 25°C and 250 rpm for 4 days with 20 ml of Traders Fungal Veg medium (total three flasks), which consisted of 25 g of dextrose (Corn Products International, Bedford Park, IL) and 25 g of cottonseed flour (Traders Protein Southern Cotton Oil Co., Lubbock, TX) in 1 liter of tap water. At the end of this period, the seed culture (12 ml to each flask) was inoculated at 25°C and 250 rpm into three flasks, each containing 176 ml of biotransformation medium, which consisted of dextrose (20 g), soybean meal (5 g), NaCl (5 g), yeast extract (5 g), and K2HPO4 (5 g) in 1 liter of distilled water. After 2 days of growth, ruboxistaurin (110 mg in 4.4 ml of DMSO) was added to the fermentation. After 10 days, the fermentation was harvested, and the cells were removed by centrifugation and extracted with 500 ml of methanol. The methanol extract was used for the purification of metabolites.
CYP2D6-Catalyzed Biotransformation of Ruboxistaurin. Ruboxistaurin (55 mg) was suspended in 2 ml of isopropanol. CYP2D6 Mix consisting of buffer, stabilizers, and the NADPH recycling system (35 g) was dissolved in 90 ml of water. CYP2D6 (100 nmol) was suspended in 20 ml of water. To the CYP2D6 Mix solution, 8 ml of the CYP2D6 suspension followed by 400 μl of compound suspension were added. The reaction mixture was shaken at 150 rpm at 30°C. Subsequent additions of CYP2D6 (4 ml each) and the compound (200 μl each) were made at 2, 4.5, and 6 h after the start of the reaction. After the final addition, the reaction mixture was incubated at 30°C at 150 rpm overnight. The reaction was terminated by the addition of 90 ml of ethanol, shaken at 150 rpm at 30°C for 1 h, and centrifuged at 3000 rpm for 25 min, and the supernatant was used for the purification of metabolites.
Purification of Metabolites. The HPLC instrumentation consisted of either a Waters 600 model HPLC system equipped with a Waters Micromass ZMD mass spectrometer (Waters) or a Shimadzu VP series (Shimadzu Scientific Instruments Inc.) equipped with a Thermo Electron LCQ spectrometer. Preparative separations were performed on a Supelco Discovery C18 column (10 × 250 mm; Supelco, Bellefonte, PA) with a flow rate of 4 ml/min. Analytical separations were performed on a Supelco Discovery C18 column (4.6 × 250 mm) with a flow rate of 1 ml/min. The mobile phase consisted of 10 mM aqueous ammonium acetate (mobile phase A) and acetonitrile (mobile phase B).
The supernatant from the 2D6 incubation was evaporated to approximately 100 ml and applied on two C18 preconditioned SPE columns. Each column was washed with 50 ml of water followed by 50 ml of methanol. The combined methanol extract was dried, dissolved in 20 ml of 4:1 methanol/water, and divided into six portions, and each portion was chromatographed separately on a Supelco Discovery C18 column (10 × 250 mm) with a flow rate of 4 ml/min using a gradient profile: 0 min/10% B, 5 min/10% B, 50 min/60% B, 50.1 min/90% B, and 55 min/90% B. Column effluents were appropriately combined to yield fractions A (mostly containing 5), B, C (mostly containing 5), and D (mostly containing 3). Fractions A and C were combined and repurified over a Supelco Discovery C18 column (4.6 × 150 mm) using a gradient profile: 0 min/15% B, 40 min/40% B, 40.1 min/90% B, and 45.1 min/90% B to yield 5 suitable for NMR analysis. Similar purification of fraction D afforded 3.
The methanol extract from the M. zonata incubation was dried down, suspended in 20 ml of DMSO, and centrifuged, and the supernatant was chromatographed over a Supelco Discovery C18 column (10 × 250 mm) using a gradient profile: 0 min/10% B, 3 min/10% B, 50 min/60% B, 50.1 min/90% B, and 55 min/90% B. The fractions containing primarily the mixture of 4 and 6 were combined and subjected to LC/NMR.
LC/NMR and NMR Spectroscopy. The HPLC instrumentation consisted of a Varian Star Chromatography system (Varian, Inc., Palo Alto, CA) equipped with a ProStar 230 solvent delivery module and a ProStar 330 photodiode array detector. NMR spectra were acquired on an Inova 500 MHz system using either an IFC 3-mm ID probe or a 5-mm cold triple-resonance probe (Varian, Inc.). LC separations were carried out on a Phenomenex Synergi Polar-RP column (4.6 × 250 mm) with a flow rate of 1 ml/min. The mobile phase consisted of 10 mM ammonium formate in D2O (mobile phase A) and 85% DCD3CN (mobile phase B), and the analytes were separated by a gradient profile: 0 min/20% B, 5 min/20% B, 50 min/60% B, 51 min/90% B, and 53 min/90% B. NMR experiments were conducted by stopping the LC flow after each component entered the NMR probe. To collect quality two-dimensional NMR data in a reasonable time, the effluents containing 4 and 6 were collected, freeze-dried, and examined by conventional NMR. Compounds were dissolved in DMSO-d6 and the spectra were referenced with respect to residual solvent signals at 2.49 ppm (1H) and 39.5 ppm (13C).
Results
Excretion Profiles. The recovery of radioactivity in urine and feces from female beagle dogs and male CD-1 mice, and in bile, urine, and feces from male bile duct-cannulated Fischer 344 rats following oral administration of a single dose of [14C]ruboxistaurin is summarized in Table 1. In dogs, ∼1 and 90% of the dose was excreted in urine and feces, respectively, over 120 h. In mice, ∼3 and 88% of the administered dose was recovered in urine and feces, respectively, over 72 h. In bile duct-cannulated rats, ∼3, 34, and 59% of the dose was recovered in urine, feces, and bile, respectively, over 48 h. Thus, no significant difference was observed in the elimination routes of ruboxistaurin in any species.
Metabolite Profiles of Ruboxistaurin. The plasma and excreta samples from the dog study were analyzed in detail to profile the metabolites in these matrices, and the results from this study were extended to the other species. Representative radiochromatograms of 2-h plasma, 0- to 24-h feces, and 0- to 24-h urine after oral administration of a single dose of [14C]ruboxistaurin in dogs are shown in Fig. 2. The mean concentration-time profiles of ruboxistaurin and its major metabolites with measurable plasma concentrations following oral administration of a single dose of [14C]ruboxistaurin in female beagle dogs are shown in Table 2, and male CD-1 mice and male Fischer 344 rats are shown in Table 3. In all species, the N-desmethyl metabolite 1 was formed rapidly and was the predominant compound detected in plasma.
Table 4 lists the relative amounts of each metabolite and parent detected in urine and feces over 48 h as percentage of dose excreted in dogs. In urine, parent and a total of 13 metabolites were detected, with both parent and metabolite 1 each comprising approximately 0.23% and other metabolites comprising less than 0.2% of the dose. In feces, parent and a total of 16 metabolites were detected, with parent accounting for approximately 39% of the dose. In plasma, parent and five metabolites were detected. In the 2-h plasma sample radiochromatogram, metabolite 6 was not detected but was detected in all later samples.
In mice, parent (11% of the radioactive material in the sample), and metabolites 1 (47%), 2 (<1%), 5 (2%), and 19 (9%) were detected in 4-h pooled plasma samples. In 0- to 24-h urine, the radioactivity excreted was associated predominantly with metabolite 1 (41% of 2.5% of the dose excreted in this matrix), and in feces, the radioactivity excreted was associated with mostly parent (51.8% of 88% of the dose excreted in this matrix) and metabolite 1 (15.2%). In addition, 15 and 13 other metabolites were recovered from urine and feces, respectively (Table 5).
In rats, parent and only metabolite 1 were detected in the 2-h plasma sample. In bile duct-cannulated rats, over the 0- to 24-h collection period, approximately 63% of the radioactivity (data not shown) from 57% of the dose excreted in the bile was found to be hydroxy N-desmethyl metabolite 5. In addition, 12 other metabolites were detected in this matrix. In urine and feces, over the 0- to 24-h collection period, the predominant metabolite excreted was 1. In addition, four and nine other metabolites were excreted in urine and feces, respectively (Table 6).
Biosynthesis and Isolation of Ruboxistaurin Metabolites. Certain metabolites were of particular interest due to the fact that they were circulating in the plasma of dogs. To unambiguously determine the structures of these metabolites by NMR spectroscopy, recombinant human P450s and microbial cultures were screened for the production of sufficient amounts of these metabolites. Ruboxistaurin was incubated with recombinant human CYP2D6, which produced metabolites 3 and 5 in approximately 2 to 3% yields. Subsequent large-scale incubation with 55 mg of ruboxistaurin followed by chromatographic purification resulted in the isolation of approximately 0.2 and 0.1 mg of 3 and 5, respectively, suitable for structure determination by NMR. Similarly, microbial incubation with M. zonata produced metabolites 4 (3% yield) and 6 (1% yield). Metabolite 4 did not correlate with any in vivo dog metabolite but was found to be unique to microbial metabolism (similar direct comparison of 4 with in vivo metabolites of mice and rats was not carried out). After initial chromatographic purification, the mixture of 4 and 6 was subjected to LC/NMR. Although LC/NMR detected the two compounds cleanly, the need to perform extensive two-dimensional NMR experiments, at least with 4, demanded drying down the sample for conventional NMR using a fully deuterated solvent.
Structure Identification of Metabolites of Ruboxistaurin. In general, structures of most of the metabolites detected in dog plasma and excreta were positively identified, and the results were then extended to the other species. MS and MS/MS data were primarily used in the structure proposal of metabolites of ruboxistaurin. Structures of a number of metabolites were further confirmed either by NMR spectroscopy or by comparison with reference standards. The product ions observed in the MS/MS spectrum of ruboxistaurin were diagnostic and were useful in formulating the structures of its metabolites. Thus, as depicted in Fig. 3A, besides the protonated molecular ion at m/z 469, ruboxistaurin showed product ions at m/z 424 (loss of dimethylamine), m/z 289 (loss of dimethylamine plus a neutral moiety of formula C8H9NO), and m/z 98, 84, and 58 (product ions related to the segment bearing the dimethylamine functionality). In all cases, the product ion elemental composition was confirmed by accurate mass measurement.
Plasma Metabolites.Dog. The major plasma metabolite 1 showed a protonated molecular ion at m/z 455, 14 Da less than the parent, and product ions at m/z 424 and 289 (same as parent), 84, 70, and 44. The structure of 1 was confirmed by direct comparison with a reference standard.
Metabolite 2 showed a protonated molecular ion at m/z 441, 28 Da less than the parent, suggesting that it was a di-demethylated product of the parent. Its structure was also confirmed by direct comparison with a reference standard.
Another predominant circulating metabolite in dog plasma was 3, which displayed a protonated molecular ion at m/z 485, 16 Da higher than the parent. The product ions observed at m/z 440 (loss of dimethylamine) and 289 (loss of dimethylamine plus a neutral fragment of formula C8H9NO2) along with other ions at m/z 440, 98, 84, and 58 supported hydroxylation of indole ring A. To define regiochemistry, metabolite 3 produced via biocatalysis of ruboxistaurin was subjected to NMR analysis. Detailed analysis of 1H and two-dimensional NMR (DQCOSY, HSQC, and HMBC) data of 3 led to the unambiguous assignment of all the protons and carbons (Tables 7 and 8) and structure assignment of 3, including the regiochemistry of oxidation. In contrast to ruboxistaurin, which showed two sets of aromatic four-proton spin systems due to two indole units, 3 showed one set of an aromatic four-proton spin system and one set of a three-proton aromatic spin system (Fig. 4), revealing hydroxylation of the phenyl part of one of the indole rings. Although ruboxistaurin possesses a near C2 symmetry, the two indole units can be distinguished due to the asymmetry of the macrocycle. Thus, the methylene groups alpha to the indole nitrogens can easily be distinguished based on their attachment to the neighboring group (CH2-CH versus CH2-O). Long-range HMBC correlations from these groups will then facilitate identification of the proton and carbon resonances of the respective indole rings. However, due to the low concentration of 3, the HMBC experiment optimized for 8 Hz heteronuclear coupling did not provide the expected correlation from the methylene groups to the indole ring protons and carbons.
Fortuitously, the structure of another aromatic hydroxylated metabolite obtained from a microbial (M. zonata) biotransformation of ruboxistaurin, which was not correlated to any dog metabolite found in vivo and whose structure was unambiguously established by MS and NMR data as 4, was very helpful in the structure determination of 3. The NMR results of 4, reported in Tables 7 and 8, were obtained after a detailed analysis of 1H, DQCOSY, HSQC, and HMBC. Unlike 3, the HMBC spectrum of 4 showed the anticipated correlations (Fig. 5) that established the location of the hydroxyl group. Thus, key multiple-bond heteronuclear correlations observed between H-10 and C-2, and H-21 and C-19, along with COSY cross-peaks observed between H-10 and H-11, and H-19 and H-18, first identified the pyrrolyl proton of the individual indole units with respect to the aliphatic portion of the macrocycle. Then, strong three-bond correlations observed in the HMBC experiment, optimized for 8 Hz heteronuclear coupling, from H-2 (δH 7.35, s) to C-4 (δC127.1), which in turn showed correlation to H-8 (δH 7.28 d, J = 8.5 Hz), and from H-2 to C-9 (δC129.9), which in turn showed correlation to H-5 (δH 7.17 d, J = 2 Hz), established the placement of the hydroxyl group at C-6. Furthermore, the up-field chemical shift (∼6 ppm) of C-9 and the near constant chemical shift of C-4 observed in 4 when compared with ruboxistaurin (Table 8) are also consistent with the placement of the hydroxyl group at C-6. Accordingly, in the MS/MS spectrum of 4, a product ion at m/z 289 was observed as in 3, suggestive of hydroxylation of indole ring A. The fact that 3 and 4 showed the fragment ion at m/z 289, consistent with oxidation of indole ring A (see Fig. 3), and identical 4- and 3-proton spin systems (Fig. 4) for the indole rings, as evidenced by NMR, revealed that 3 and 4 are indeed regioisomers, and the possible location of the hydroxyl group in 3 was at C-7 instead of C-6 as in 4. The placement of the hydroxyl group at C-7 in 3 was further corroborated by the significant up-field shift (∼6 ppm) observed for C-4 and a near constant chemical shift observed for C-9 (Table 8), in contrast to 4, when compared with ruboxistaurin.
Metabolite 5 revealed a protonated molecular ion at m/z 471, 14 Da less than 3, suggesting that it was an N-demethylated product of 3. Accordingly, in the MS/MS, significant product ions were observed at m/z 289 (defining the site of hydroxylation as indole ring A) and m/z 84 (defining the loss of N-methyl group compared with the parent), in addition to ions at m/z 453, 440, 428, 414, and 396. The structure of 5 was further confirmed by NMR. The notable feature in the 1H NMR spectra of 5 and 3 was the close similarity, both in chemical shifts and coupling constants, of the aromatic resonances (Fig. 4), indicating that the hydroxyl group was located at C-7 in both compounds. Table 5 lists the entire proton chemical shifts of 5 in comparison with ruboxistaurin and 3. Due to limited sample availability, no carbon data were obtained for 5.
Metabolite 6 revealed a protonated molecular ion at m/z 485, 16 Da higher than the parent. The N-oxide structure was proposed for 6 based on the product ion observed at m/z 424 (loss of 61 Da from the parent), which was consistent with the loss of hydroxy-dimethylamine when compared with the parent, which showed the same product ion with the loss of dimethylamine. The structure of 6 was again confirmed by NMR as well as by direct comparison with a reference standard. The distinct feature in the 1H NMR spectrum of 6 in contrast to the parent was the downfield shift (∼0.25 ppm) exhibited by both the methyl groups attached to the nitrogen due to the oxidation of the amine to amine N-oxide.
Mouse and Rat. In addition to the metabolites 1, 2, and 5 identified in dog plasma, a glucuronide conjugate, 19, was also identified in mouse plasma. In rat plasma, however, only the N-desmethyl metabolite 1 was observed. Metabolite 19, detected only in the mouse plasma, showed a protonated molecular ion at m/z 647 and product ions at m/z 471, 440, 428, 414, and 289, suggesting hydroxylation of the N-desmethyl metabolite 1 followed by conjugation with glucuronic acid. The presence of the product ion at m/z 289 revealed hydroxylation of indole ring A; however, the exact position of hydroxylation and subsequent glucuronidation within the indole ring could not be determined from these data.
Fecal Metabolites.Dog. The metabolites 1, 2, 3, and 5 observed in plasma were also detected in feces. In addition, 10 other metabolites were detected, each less than 2% of the dose, and their structures were tentatively identified by MS with the exception of some metabolites whose structures could be confirmed by direct comparison with reference standards.
Metabolite 7 revealed a protonated molecular ion at m/z 485, 16 Da higher than the parent, and metabolites 8 and 9, isomeric in nature, showed identical protonated molecular ions at m/z 471, 14 Da less than 7. The product ion spectrum of 7 displayed an array of ions quite different from the ions observed in other oxidative metabolites such as 3, 5, and 6 (Table 9). Formulation of these ions at m/z 440, 271, 176, 169, 157, and 145, along with other ions, as shown in Fig. 3B, suggested oxidation of indole ring A at the C-2 position. The structure thus proposed by MS was further confirmed by direct comparison with a reference standard. The reference standard predominantly existed in the keto form as evidenced by NMR spectroscopy (data not shown). Metabolites 8 and 9 showed identical product ions at m/z 440, 271, 176, 169, 157, 145, and ions 14 Da less in the segment bearing the dimethylamine functionality when compared with 7, suggesting oxidation of indole ring A at the C-2 position and N-demethylation, respectively. On this basis, the structures of the diastereoisomers 8 and 9, as a result of the keto-enol tautomerism, were proposed to be the N-desmethyl products of 7. The presence of both diastereoisomers of 7 was not detected in the fecal sample, and the stereochemistry of the additional chiral center in 7 could not be determined.
Metabolites 10 and 11 are a pair of diastereoisomers with protonated molecular ions at m/z 471. Their structures were proposed based on their mass spectral fragments derived from the observed product ions as depicted in Fig. 3C. It is apparent from the scheme that in 10 and 11, indole ring B was oxidized (C-21 position) instead of indole ring A (C-2 position) as in 8 and 9. The diastereoisomers 12 and 13 showed identical protonated molecular ions at m/z 485, 16 Da higher than the parent, and the same product ions (Table 9), consistent with the oxidation of parent at C-21 of indole ring B.
Metabolite 14 (protonated molecular ion at m/z 485) and its corresponding N-desmethyl metabolite 15 (protonated molecular ion at m/z 471) are distinct from the other hydroxy metabolites described above; in metabolites 14 and 15, the oxidation has occurred on the phenyl of indole ring B. This was borne out specifically by the presence of a fragment ion at m/z 305 (loss of trimethylamine plus a neutral fragment with formula C8H9NO) along with the ion at m/z 440 (loss of dimethylamine) in the product ion spectrum of both 14 and 15. Thus, the presence of a fragment ion at m/z 289 was used to characterize an unchanged indole ring B, and the presence of the ion at m/z 305 was used to characterize hydroxylated indole ring B. However, the exact position of hydroxylation on the phenyl ring could not be determined from these data.
Metabolite 16 (protonated molecular ion at m/z 485) and possibly its corresponding N-desmethyl metabolite 17 (protonated molecular ion at m/z 471) are yet two other hydroxy metabolites of the parent, the structures of which were proposed as shown in the formula on the basis of the fragment ions (Table 9), specifically the ion at m/z 289. Again, the exact position of hydroxylation could not be determined. An additional hydroxy N-desmethyl metabolite, 26 (protonated molecular ion at m/z 471), was also observed; however, the site of hydroxylation could not be determined.
Mouse and Rat. The presence of the identified metabolites 1, 2, 3, 5, 6, 7, 10, 12, 14, 15, and 17 was observed in mouse feces. Additional metabolites tentatively identified in mouse were 20 (N,N-dimethylamino-methyl segment oxidized to a carboxylic acid), 22 (oxidation of the A ring with subsequent glucuronidation), and 24 (hydroxylation at the 2 position; diastereomer of 7). In rat, metabolites 1, 2, 3, 7, 8, 10, 12, 14, 15, and 17 were identified.
Urinary Metabolites.Dog. Plasma and fecal metabolites 1, 2, 3, and 5 and fecal metabolites 7, 8, 10, 12, 13, 14, and 15 were also detected in the urine. In addition, phase II glucuronide conjugates 18 and 19 (detected in mouse plasma) were observed in dog urine. The structure of 18 with a protonated molecular ion at m/z 661 was tentatively proposed to be the glucuronide conjugate of a hydroxylated parent. The location of hydroxylation and subsequent glucuronidation could not be determined.
Mouse and Rat. In mouse, previously identified metabolites 1, 2, 3, 5, 7, 8, 10, 12, 14, 15, 17, 19, and 22 were observed. In addition, three other oxidation products with subsequent conjugation with glucuronic acid (metabolites 21, 23, and 25) were observed. Metabolites 21 and 25 had protonated molecular ions at m/z 647, and their structures were proposed to be the glucuronide conjugates of hydroxylated N-desmethyl metabolites. The product ions observed in the MS/MS spectrum (Table 9) for metabolite 23 suggested oxidation of the B ring and subsequent glucuronidation (product ion at m/z 305). In rat, previously identified metabolites 1, 2, 3, 5, and 10 were observed.
Rat Biliary Metabolites. The major component observed in rat bile was identified as the hydroxy N-desmethyl metabolite 5. Other metabolites, each constituting approximately <1 to 5% of the dose, identified in this matrix were 1, 2, 3, 10, 13, 17, 18, 19, 21, 22, 23, and 25.
Discussion
The disposition of ruboxistaurin (LY333531), a potent and isoform-selective PKCβ inhibitor currently in development for the treatment of diabetic microvascular complications, has been studied previously in rats and dogs (Burkey et al., 2002). That study was limited to the detection and disposition of only the parent and its N-desmethyl metabolite 1 in plasma. The purpose of the present study was to investigate the metabolism of ruboxistaurin in dogs, mice, and rats and to identify distinct metabolites of ruboxistaurin in plasma and excreta since dog was chosen as the primary nonrodent species and mice and rats were chosen as the rodent species for the safety assessment of ruboxistaurin during development.
Following oral administration of a single dose of [14C]ruboxistaurin, the mean total recovery of radioactivity was approximately 92, 91, and 100% in dogs, mice, and rats, respectively. The radioactivity was primarily excreted through feces in dogs (90%) and mice (88%), and through bile (59%) and feces (34%) in bile duct-cannulated rats (Table 1). The fact that 59% of the radioactivity was eliminated in rat bile suggests that fecal excretion in all species may contain a substantial biliary component. Urinary excretion was a minor route of elimination, accounting for only <3% in all species. This is in good agreement with previously published results on the disposition of ruboxistaurin (LY333531) by Burkey et al. (2002).
The major circulating components in plasma were the N-desmethyl metabolite 1 and parent for all three species. The most notable exception was metabolite 19, a glucuronide conjugate of a hydroxy N-desmethyl ruboxistaurin, which was observed in significant amounts (9%) only in mice. In feces, parent and the N-desmethyl metabolite 1 were the major ruboxistaurin-related components observed in all species. Thus, overall, the major metabolic pathway of ruboxistaurin in all three species was N-demethylation.
Preliminary identification of metabolites was achieved by ion spray LC-MS/MS and, when necessary, by accurate mass measurements. The structures of significant metabolites were confirmed either by direct comparison with reference standards or by NMR spectroscopy. In an attempt to establish the exact sites of oxidation, a suite of two-dimensional NMR experiments was conducted. To support characterization by NMR, surrogate biocatalytic systems, including recombinant human P450s and microorganisms, were adopted to produce sufficient amounts of the desired metabolites. These efforts led to the unambiguous identification of all notable metabolites detected in dog plasma. Overall, in addition to parent, five metabolites (1, 2, 3, 5, and 6) were observed in dog plasma by radiochemical detection and identified by mass spectrometry. The structures of metabolites 1 and 2 were confirmed by comparison with reference standards, and those of 3, 5, and 6 were confirmed by NMR. Results from the metabolite profiling and identification experiments demonstrated that ruboxistaurin was primarily metabolized via N-demethylation to metabolite 1 (approximately 60% of the radioactive material observed in the 2-h dog plasma sample and comparable amounts in other species was the N-desmethyl metabolite 1). Minor routes of metabolism included hydroxylation of the indole ring (metabolite 3), oxidation of the tertiary amine to the N-oxide (metabolite 6), combined N-demethylation and hydroxylation of the indole ring (metabolite 5), combined N-demethylation and hydroxylation of the indole ring followed by subsequent glucuronidation (metabolite 19; the exact site of hydroxylation and subsequent glucuronidation could not be determined), and N,N-di-demethylation (metabolite 2). Based on the metabolites identified in plasma, the major metabolic pathways of ruboxistaurin are proposed as shown in Fig. 6.
With the exception of the N-oxide metabolite 6, which was observed only in dog plasma, the metabolites detected in plasma of all species were also detected in one or more of the excreta samples. In addition, several minor oxidative metabolites and oxidative metabolites with subsequent conjugation by glucuronic acid were also detected in these matrices. Figure 7 shows the minor metabolites of ruboxistaurin in excreta not shown in Fig. 6.
In rat, the fact that nearly 60% of the radioactivity was recovered from the bile after an oral dose of 5 mg/kg suggests that ruboxistaurin was well absorbed. Also, the fact that no parent was observed in this matrix indicates that ruboxistaurin was highly metabolized. Extensive metabolism of ruboxistaurin was also evident from the observation that the majority of the radioactivity detected in the circulating plasma was accounted for by the N-desmethyl metabolite 1 in all three species.
In summary, the excretion profiles of ruboxistaurin and its metabolites following oral administration of the compound were very similar in dogs, mice, and rats. Ruboxistaurin was well absorbed and highly metabolized, primarily to the N-desmethyl metabolite. The parent and its metabolites were predominantly eliminated via feces. The structures of most metabolites observed in plasma and significant metabolites observed in the excreta were positively identified by MS/MS, NMR, and by comparison with reference standards. Surrogate biocatalytic systems, including recombinant human P450s and microorganisms, were adopted to generate the desired metabolites in sufficient amounts for characterization by NMR. The positive identification of metabolites in plasma and excreta led to the proposal of metabolic pathways of ruboxistaurin in dogs, mice, and rats.
Acknowledgments
We thank Gregory A. Rener for the accurate mass measurements and Patrick J. Jansen for providing the metabolite reference standard 7. We express a deep gratitude to William J. Ehlhardt for valuable discussions and helpful suggestions during the preparation of the manuscript
Footnotes
-
Article, publication date, and citation information can be found at http://dmd.aspetjournals.org.
-
doi:10.1124/dmd.105.007401.
-
ABBREVIATIONS: PKC, protein kinase C; HPLC, high-performance liquid chromatography; DMSO, dimethyl sulfoxide; LC, liquid chromatography; MS, mass spectrometry; MS/MS, tandem mass spectrometry; P450, cytochrome P450; DQCOSY, double quantum-filtered correlation spectroscopy; HSQC, heteronuclear single quantum coherence; HMBC, heteronuclear multiple bond correlation.
- Received September 15, 2005.
- Accepted October 26, 2005.
- The American Society for Pharmacology and Experimental Therapeutics