Abstract
Retigabine (D-23129,N-(2-amino-4-(4-fluorobenzylamino)-phenyl) carbamic acid ethyl ester) is a potent anticonvulsant in a variety of animal models. Rats metabolized [14C]retigabine mainly through glucuronidation and acetylation reactions. Glucuronides were detected in incubates with liver microsomes or slices, in plasma, and in bile and feces but were absent in urine (0–24 h) that contained about 2% of the dose as retigabine and approximately 29% of the dose in > 20 metabolites, which are derived mainly from acetylation reactions. About 67% of the radioactivity was excreted into feces, approximately 10% of the dose as glucuronide. The metabolite pattern in the urine (0–24 h) of dogs was comparatively simple in that retigabine (13%), retigabine-N-glucuronide (5%), and retigabine-N-glucoside (1%) were present. In the same 24-h interval, about 39% of unchanged retigabine was excreted into feces. Plasma profiling and spectroscopic analysis (liquid chromatography with tandem mass spectrometry NMR) of two isolated urinary metabolites obtained after single oral dosing of 600 mg retigabine in healthy volunteers indicated that both acetylation and glucuronidation are major metabolic pathways of retigabine in humans. We found that in vitro assays with liver slices from rat and humans reveal the major circulating metabolites in vivo.
Retigabine (D-23129, N-(2-amino-4-(4-fluorobenzylamino)-phenyl) carbamic acid ethyl ester, CAS: 150812–12-7) was discovered in a series of desaza analogs of Flupirtine1, a centrally acting analgesic. Retigabine has shown pronounced anticonvulsant activities in a variety of animal models after oral administration (Tober et al., 1996; Rostock et al., 1996). The obtained results revealed a unique pharmacological profile associated with an opening effect on potassium channels and an increase of GABA synthesis (Kapetanovic and Rundfeldt, 1996). Its potential for the treatment of epileptic seizures in humans is currently under investigation in clinical studies.
In a recent study, McNeilly et al. (1997) suggested on the basis of liquid chromatography with tandem mass spectrometry results that twoN-glucuronides of [14C]retigabine (Fig. 1) are formed in incubations with human liver slices or microsomes in the presence of uridine 5′-diphosphoglucuronic acid (UDPGA)2. However, the sites ofN-glucuronidation remained unclear.
The aim of the present study was to characterize the metabolism of retigabine in vitro and in vivo by radiometric high-performance liquid chromatography (HPLC) and to identify the major urinary metabolites after oral administration of 8.25 mg/kg [14C]retigabine to rats and dogs or 600 mg of unlabeled retigabine to humans to establish species differences in the metabolism. Furthermore, the role of glucuronidation in the metabolism of retigabine in vivo as predicted from in vitro results (McNeilly et al., 1997) was studied. Metabolite fractions from urine were purified by a multidimensional semipreparative HPLC method and analyzed partly by mass spectroscopic and partly by NMR spectroscopic techniques.
Materials and Methods
Chemicals.
Retigabine was synthesized by ASTA Medica AG (Dresden, Germany). The potential metabolite 1-acetamido-2-amino-4-(4-fluorobenzylamino)-benzene (reference compound for R6 and M2) and 5-acetamido-2-oxo-2,3-dihydro-1H-benzimidazole (reference compound for R3) were provided by P. Meisel and Dr. J. Schäfer, Arzneimittelwerk Dresden GmbH (Dresden, Germany). [14C]Retigabine was synthesized by Amersham International plc (Buckinghamshire, UK) on behalf of Arzneimittelwerk Dresden GmbH with uniformly 14C-labeled 1,2,4-triaminobenzene moiety as shown in Fig. 1. The radiochemical purity was 96.8% and specific activity was 9 mCi/mmol. Quicksafe flow 2 (Zinsser Analytic, Maidenhaid, UK) was used for scintillation counting. All other chemicals came from commercial suppliers in analytical grade. Demineralized water (Milli-Q water purification system, Millipore Waters, Eschborn, Germany) was used for the preparation of aqueous solutions.
Microsomal Preparations.
Frozen human liver samples were obtained from the University of Greifswald, Germany. The microsomal fraction was prepared according to the method of Meier et al. (1983) with modifications. The protein concentration was estimated using the micro-BCA assay (Pierce, Rockford, IL) according to the manufacturer’s instructions. Total cytochrome P-450 content of microsomes was determined by difference spectrophotometric measurement in the presence and absence of carbon monoxide (Omura and Sato, 1964).
Glucuronidation of [14C]Retigabine in Human Liver Microsomes.
Microsomal glucuronidation of retigabine was performed by incubating 500 μg of microsomal protein in a final volume of 200 μl with [14C]retigabine (100 μM) and unlabeled UDPGA (1 mM) or unlabeled retigabine (2.5 mM) and [14C]UDPGA (78 μM) in the presence of the following cofactor solution: 40 mM Tris pH 8.0, 7 mM MgCl2, 0.02% Triton X-100, and 1 mM ascorbic acid (final concentrations). Samples were incubated for 2 h at 37°C. Controls were performed in the absence of [14C]retigabine or [14C]UDPGA and microsomes, respectively. Protein was precipitated by addition of 1 volume ice-cold acetonitrile followed by incubation at −20°C for 20 min and centrifugation for 20 min at 2000g. Volumes of 50 μl from the supernatant were analyzed by HPLC with radioactivity monitoring.
Incubation of [14C]Retigabine with Rat and Human Liver Slices.
Rat liver slices (Wistar, Harlan Sprague-Dawley, Inc. Indianapolis, IN) were freshly prepared. Fresh human liver slices (300 μm × 0.1 cm) were acquired through In Vitro Technologies (Baltimore, MD) from liver samples obtained by the International Institute for the Advancement of Medicine (Exton, PA). Slices were allowed to preincubate in Krebs-Henseleit + 2.25% (w/v) bovine serum albumin buffer for 1 h. At this point, slices were transferred to wells (two slices per well) containing fresh buffer in a total volume of 1 ml. Reactions were started by the addition of retigabine to a concentration of 25 μM for rat and human slices each corresponding to 0.5 μCi [14C]retigabine, respectively, and allowed to proceed for 4 h. Control samples were run using the buffer without liver slices and also slices in the absence of substrate. Samples from supernatants were used for radiometric HPLC analysis.
Studies in Rats.
Urine and plasma samples of rats were taken from a mass balance study as well as from a pharmacokinetic study in Wistar rats after a single oral dose of 8.25 mg/kg of [14C]retigabine, representing 8.73 MBq/kg, which were performed by Hazleton (Madison, WI) or Pharmakon (Waverly, PA), respectively. [14C]Retigabine (200 mg) and unlabeled retigabine (6.25 mg) were given in 13.6 ml of 0.1 N HCl and mixed. Sterile water (5 ml) was added followed by drop-by-drop addition of 0.1 N HCl until a clear solution was achieved. This solution was brought to a volume of 100 ml with water to obtain the final dose solution, from which 4 ml/kg were given to rats via oral gavage. Urine was collected into containers cooled with dry ice at the following time intervals: 0 to 24, 24 to 48, and 48 to 72 h (Hazleton) or 0 to 4, 4 to 8, 8 to 24, and 24 to 48 h postdose (Pharmakon). Plasma at sampling time points of 0.5, 1, 1.5, and 2 h were analyzed. Bile samples (0–1, 1–2, 2–3, 3–4, 4–5, 5–6, 6–24, and 24–48 h postdose) were from a separate study performed at ASTA Medica AG (Frankfurt, Germany) with bile duct-cannulated Wistar rats (n = 2), which orally obtained the same dose and formulation.
Studies in Dogs.
Urine (0–8, 8–24, and 24–48 h postdose) and plasma samples of dogs (sampling time points of 0.5, 1, 2, and 3 h were analyzed) were taken from a pharmacokinetic study in dogs after a single oral dose of 8.25 mg/kg of [14C]retigabine, representing 0.76 MBq/kg. This study was performed by Pharmakon. For oral administration 0.7 mg of [14C]retigabine and 7.55 mg of unlabeled retigabine per kilogram body weight were filled in powder form in a gelatin capsule to produce the test article. Dogs were fasted 12 h before drug administration but had free access to drinking water. All samples were stored frozen from time of collection until transfer to Arzneimittelwerk Dresden GmbH, where they arrived frozen on dry ice and were kept frozen (−30°C) until analysis.
Human Study.
The approval of the study protocol was obtained from an independent ethics committee. Male volunteers were included in the study to assess the tolerability, safety, and pharmacokinetics of orally dosed retigabine after clinical examination and with their written consent. After single dose oral administration of 600 mg retigabine (6 × 100 mg in capsules), urine was collected in containers cooled with dry ice at the following time intervals: 0 to 2, 2 to 4, 4 to 8, 8 to 12, and 12 to 24 h postdose. Plasma at sampling time points of 1.5, 2.5, and 4 h was analyzed. Samples were stored frozen at ASTA Medica AG (Frankfurt/Main, Germany) from time of collection until transfer to Arzneimittelwerk Dresden GmbH, where they arrived frozen on dry ice and were kept frozen (−30°C) until analysis.
Measurement of Radioactivity.
Radioactivity was measured by liquid scintillation counting using a Wallac 1409 scintillation counter (Pharmacia Wallac Oy, Turku, Finland) for 5 min in the range to 105 cpm. Aliquots of urine or chromatographic fractions were added to Quicksafe scintillation cocktail and were shaken 5 min before measurement.
Analytical HPLC.
The Kontron system 450-MT2 (Kontron Instruments, Neufahrn, Germany) consisted of two model 420 pumps, gradient mixer M 800, autosampler 465, and UV detector 432 (set at 220 nm, range 0.5, response time 2 s) connected in series with a radiomonitor LB 507 B (set at peak half width 30 s, response time 1 s, range 10,000 cpm) equipped with a solid scintillator flow cell YG-HP 100 (Berthold, Bad Wildbad, Germany). Analyses were conducted on a Kromasil C18 column (250- × 4.6-mm, 5 μm) protected with a guard (30- × 4.6-mm i.d.) filled with the same material (Resteck, Sulzbach, Germany). The flow rate was 0.5 ml/min. Separations were conducted using a 3-h gradient: 100% A for 20 min, followed by a linear increase to 10% B over 50 min, then followed by a linear increase to 89% B over 126 min. Eluents used in the gradient were solvent A, 20 mM KH2PO4 adjusted with 1 M NaOH solution to pH 7.2; and solvent B, acetonitrile/water (90:10, v/v). The chromatography was performed at 21–23°C. Urine samples and supernatants from liver slice incubations or microsomal incubations were analyzed directly by radiometric HPLC. Bile samples were diluted 1:1 with mobile phase A before HPLC analysis. Plasma samples were diluted (1:1, v/v) with 20 mM KH2PO4 buffer pH 7.2 and enriched by solid phase extraction (SPE) as described below. The14C recovery was between 75 and 100%.
Semipreparative Chromatography (SPHPLC).
The chromatographic system consisted of a Waters 600 E gradient pump (Millipore, Watford, UK), manual loop injector 7125, 1 ml loop volume (Rheodyne Inc., Cotati, CA), a stainless steel column (250- × 10-mm i.d.) packed with Kromasil C18 reversed phase 5 μm material, flow rate 2 ml/min, Kontron UV detector 432 (set at 254 nm, range 2, response time 2 s), radiomonitor LB 507 B (set at peak half width 30 s, response time 1 s, range 10,000 cpm), equipped with a solid scintillator flow cell YG-HP 100 and a 2-channel recorder, type 2210, range 1 V (Kipp & Zonen, Delft, the Netherlands).
Adsorption/Desorption (A/D) Chromatography.
The equipment consisted of a Kontron HPLC pump 420, stainless steel column (250- × 10-mm i.d.) packed with LiChrospher 100 RP-18, 10 μm, endcapped (Merck, Darmstadt, Germany), flow rate 4 ml/min, a Kontron UV detector 432 (set at 254 nm, range 0.5, response time 2 s), radiomonitor LB 507 B (set at peak half width 30 s, response time 1 s, range 100,000 cpm) containing a solid scintillator flow cell YG-HP 100 and a 2-channel recorder, type 2210, range 1 V.
SPE.
Reversed phase extraction columns C18, Bakerbond 7020–08 (Baker Instruments, Inc., Phillipsburg, NJ) were equilibrated with 10 × 4.5 ml methanol after conditioning with 10 × 4.5 ml of water. Fractions from SPHPLC or plasma samples were drawn through the columns with the help of a vacuum, and after washing with 3 × 4.5 ml of water, air was drawn through the column for 1 min. Then the sample was eluted with 4 × 4.5 ml methanol and the eluate obtained was concentrated by a rotary evaporator (VV 2000; Heidolph, Kehlheim, Germany) at a bath temperature of 40°C. By using this procedure, the 14C recovery of a test solution of [14C]retigabine in dog blank plasma (60,000 dpm/ml) was about 100%.
Metabolite Isolation.
Rat urine
Urine from 22 male rats (collecting interval 0–24 h) were pooled. From that, 50 ml were subjected to centrifugation at 5000g for 15 min and filtered through a cellulose filter (pore width 0.8 μm, no. 13304-47-N; Sartorius, Göttingen, Germany) with the help of a Sartorius filter device SM 16510. The filtered urine was pumped at a rate of 4 ml/min onto the A/D column (1.A/D), then washed with 54 ml of water followed by 275 ml of 20 mM KH2PO4 adjusted with NaOH solution (1 M) to pH 7.2. Sequential desorption steps were then carried out at a rate of 2.5 ml/min using acetonitrile/water (90:10, v/v). A flow chart of the isolation procedure used is given in SchemeFS1.
The urine enrichment was repeated, resulting in third and fourth A/D. Fractions were collected when a level of radioactivity of 40,000 dpm was exceeded. In this way, 35 ml of eluate (fraction 8) were collected from the first A/D and 23 ml (fraction 6/2) in the case of the third A/D. These fractions were evaporated to dryness under a vacuum at 40°C. The residue of fraction 8 was reconstituted in 12 ml of water, filtered (minisart NML 0.8 μm, Sartorius) and pumped again onto the A/D column (2.A/D). After washing with 175 ml of buffer solution, 62 ml of acetonitrile/water (5:95, v/v) was pumped onto the column. Nine milliliters of fraction 8.7 were obtained, containing 3.9 × 106 dpm/ml. The elution was continued with acetonitrile/water (90:10, v/v). After 7 ml (fraction 8.8), 25 ml of the fraction 8.9 were obtained, containing 1.1 × 106 dpm/ml. Fraction 6/2.6 was obtained in the same way as fraction 8.9 using third and fourth A/D. The combined fractions 8.9 and 6/2.6 were subjected to SPHPLC in five runs using isocratic separation (20 mM KH2PO4 adjusted to pH 7.2, acetonitrile/water (90:10), 96:4, v/v). The column was loaded with 18 × 106 dpm. There was a severe mass-overloading on the column. Therefore, an unseparated fraction near the void volume was collected from 5 to 8 min. Subsequently the separation was sufficient and metabolite R5 was obtained at 92 min. The early fractions (5–8 min) were combined and separated (6th-11th SPHPLC) using gradient as described above. At a retention time of 69 min, metabolite R4 was obtained. The residue from fraction 8.7 was reconstituted with 6 ml of 20 mM KH2PO4, adjusted with 1 M NaOH solution to pH 7.2, and isocratically chromatographed in five runs (12th-6th SPHPLC, 4 × 106 dpm/run) as mentioned above. Two peaks, R2 at retention time of 50 min and R3 at 55 min, could be baseline-separated (Fig.2).
Human urine.
Urine from male volunteer 11, collecting interval 0 to 12 h was used for 5th and 6th A/D (50 ml each) and 7th A/D (100 ml). The filtered urine was pumped at a rate of 4 ml/min onto the A/D column. It was washed with 230 ml of water and then the column was eluted with 200 ml of methanol. Then both metabolites M1 and M2were isolated from human urine in an analogous way as described above for rat urine (Scheme FS1).
Dog urine.
Ten milliliters of urine from female dog 12, collecting interval 0 to 8 h, were centrifuged at 7500g for 10 min and chromatographed in eight runs (17th-24th SPHPLC) using gradient as described above. The feed was 1 ml, containing about 0.8 × 106 dpm. Two fractions at retention times of 110 (D1) and 119 min (D2) were collected and concentrated by SPE.
Mass Spectroscopy (MS).
Electrospray MS (ESI-MS)
Samples were analyzed using an HPLC system linked to a Quattro tandem quadrupole mass spectrometer equipped with a megaflow electrospray interface (Micromass, Manchester, UK) operating in positive and negative ion mode. Purified metabolite fractions were analyzed by flow injection analysis using a solvent flow of acetonitrile/water (50:50, v/v) at a flow rate of 20 μl/min. The on-line HPLC/ESI-MS analysis of crude samples was performed using the analytical gradient HPLC system. The phosphate buffer was replaced by 1 mM ammonium acetate at pH 7.4. After UV and radiometric detection, the HPLC effluent was split postcolumn to 30 μl/min and introduced into the electrospray source [temperature 80°C, capillary voltage 3.6/2.9 kV (ESI ±), HV lens voltage 0.3/0.2 kV (ESI ±)]. A cone offset from 21 to 55 V and a constant skimmer offset of 5 V was used in positive and negative ion mode. The nebulizer and bath gases were nitrogen delivered at flow rates of 15 and 300 l/h. Mass spectra were acquired with a scan range from 105 to 750 m/z (full scan analysis) in 3 s in both ion modes. Collision-induced dissociation (CID) spectra were acquired with a scan rate of 200 U/s using argon as collision gas at 2.5 × 10−5 mb at a collision energy of 50 eV. All data were processed by MassLynx (Micromass, Manchester, UK) software.
Thermospray MS (TSP-MS).
The same mass spectrometer with a thermospray ion source (VG Biotech, Manchester, UK) were used operating under Lab-base software control in positive ion mode. The same gradient HPLC system as described for ESI-MS was fitted on-line to the MS. To optimize thermospray performance and to dilute the effects of the gradient, 200 mM ammonium acetate solution in 20% (v/v) acetonitrile in water at 0.5 ml/min was added postcolumn. The MS operating conditions were: total solvent flow 1 ml/min, TSP vaporizer temperature 280°C (gradient time 40–70 min) and 275°C (gradient time 70–210 min), source temperature 260°C, and repeller voltage 220 V. The MS run was started at 40 min gradient time. The MS scan parameter and CID conditions were identical with the ESI-MS parameters with the exception of a collision energy of 70 eV.
NMR spectroscopy.
The 1H-NMR experiments for structural elucidation of metabolites M1 and M2 were carried out on a Bruker AMX 600 spectrometer (Bruker Analytik GmbH, Rheinstetten, Germany) in 5-mm tubes. The metabolites were dissolved in 0.5 ml dimethyl sulfoxide (DMSO)-d6 and the1H-NMR spectra were recorded at 27°C. The reference for chemical shift was DMSO-d6 at 2.50 ppm. 1H-NMR studies with metabolite fractions isolated from urinary samples of rats and dogs dissolved in CD3OD were performed at 32°C on a 300 MHz Bruker spectrometer ARX 300 interfaced with an Aspect X-32 computer. Chemical shifts for the 1H-NMR spectra are reported in ppm relative to trimethylsilyl using the residual solvent (CD3OD) signal at 3.50 ppm.
Results
In Vitro Incubations.
The metabolite patterns from incubations of [14C]retigabine with liver slices from rat and humans were similar as shown in Figs. 3and 4. Three metabolites with retention times of 105, 112, and 123 min were formed.
The obvious difference was the dominant appearance of a metabolite at a retention time of 92 min that was only formed in rat liver slices. To identify possible N-glucuronides that were suggested in an earlier in vitro study (McNeilly et al., 1997), we incubated either [14C]retigabine and UDPGA or unlabeled retigabine and [14C]UDPGA with human liver microsomes. Both experiments showed the formation of the same two peaks at 105 and 112 min (data not shown) as formed by the experiment with liver slices from rats and humans (Figs. 3 and 4). In control experiments without UDPGA no metabolite peaks were seen. These results indicated that both metabolites were glucuronides of unchanged retigabine. This could also be demonstrated by the increase of the retigabine peak and the decrease of the peak at 112 min after hydrolysis of glucuronides at pH 5.5 (data not shown).
Excretion Data and Metabolite Profiles.
After oral administration, the total radioactivity excreted in urine and feces of rats over 24 h was 31.0 ± 5.7 and 67.1 ± 10.6% of the dose, respectively (Table1).
Unchanged drug in rat urine represented 2% of the dose in this time interval. In fecal samples of rats the unchanged parent drug was missing (profiles not shown). These results demonstrate extensive metabolism of retigabine in rats. When the metabolite portion of rat plasma is compared with that of liver slices (Fig. 3, liver slices, plasma) it is evident that the most abundant metabolite in plasma is the presumed glucuronide eluting at 112 min. In rats, both glucuronides at retention times of 105 and 112 min were predominantly eliminated via bile (Fig. 3, bile). Both peaks represented about 25% of the cleared dose within 0 to 24 h in rat bile. In contrast, only 4% of the dose was excreted in this time as unchanged retigabine. In accordance with the biliary excretion is the lack of typical mass spectroscopic or chromatographic information for glucuronides of retigabine in rat urine. In contrast, the other metabolite (92 min) that was seen in incubates with rat liver slices was mainly renally cleared in rats (Fig. 3). It is evident that the metabolite profiles in rat urine reveal another level of complexity showing the formation of additional polar metabolites that are neither detected in plasma, bile, or liver slices. The six major metabolites (R1-R6) in rat urine, which we characterize below, represented altogether about 20% of the dose, corresponding to approximately 58% of the radioactivity cleared during 0 to 24 h in urine.
A metabolite pattern similar to that detected in rat plasma was found in plasma from humans when detected by UV after dosing with unlabeled retigabine (Fig. 4). The appearance of the intense peak at 112 min in human urine suggested that in contrast to rats, the main glucuronide was strongly renally excreted in humans. A quantification of the metabolism in humans was not possible because a14C experiment had not been performed. Based on the chromatographic comparison with 14C-labeled fractions isolated from dog or rat urine, both metabolitesM1 and M2 could be identified in human urine (Fig. 4) besides retigabine. In dogs, metabolism of retigabine is less complex than in rats (data not shown). In addition to the parent drug, in plasma as well as in urine only the same two metabolites at 112 (D1) and 121 min (D2) were found. Both metabolites together represented 6% of the dose, corresponding to about 32% of the radioactivity cleared during 0 to 24 h into dog urine. The major metabolite detected at 112 min in urine and plasma was present in most fractions, suggesting the important role of glucuronidation in dogs as well.
Purification of Metabolites from Urine.
Initial concentration and partial purification of metabolites from both rat and human urine was accomplished with a high volumetric throughput A/D chromatographic procedure followed by SPHPLC. This work-up procedure is based on a method published by Vollmer et al. (1986) and is exemplified in Scheme FS1. It avoids unfavorable dividing of radioactivity in the course of metabolite isolation, with the aim of separating metabolite fractions by means of one or two chromatographic runs only at the end. The baseline separation of the both major metabolites R2 and R3 starting from pooled rat urine is illustrated in Fig. 2 and exemplifies the power of this isolation strategy. Metabolites from human urine were isolated in a similar way. To isolate both metabolites in dog urine only a single step chromatographic separation was necessary. Isolated fractions were finally concentrated by SPE to achieve desalted samples for optimal responses in ESI-MS. Before starting to elucidate the structure of the isolated metabolites, their identity and purity were analyzed by analytical HPLC. The retention times detected were compared with those in original samples before isolation to detect any decompositions. The sample preparation technique is described in detail in Materials and Methods.
Identification of Metabolites.
Metabolite R1 .
This metabolite represented in rat urine about 3% of the administered dose in the collecting interval 0 to 24 h. R1 (Fig. 3) could not be enriched from eluate fractions of SPHPLC by SPE in sufficient quantity owing to its high polarity. Thus, R1 was identified in the urine pool of rats by on-line HPLC/TSP-MS. Based on the protonated molecular ion [M+H]+ obtained atm/z 208 (MS data not shown) a molecular mass of 207 was assigned. This odd mass number indicated that it had retained the three nitrogen atoms of retigabine. The characteristic daughter ions atm/z 166 and m/z 43 in the CID product ion spectrum of m/z 208 indicated a loss of ketene and an acetyl group, respectively, which revealed the presence of an acetyl moiety. The detected radioactivity evidenced the presence of the labeled benzene moiety from [14C]retigabine. The structure proposed for R1 (SchemeFS2) is 1,4-bis-acetamido-2-amino-benzene. In comparison with the structures of both metabolites R4 andR6, which were unambiquously determined (see below), we assume that positions 1 and 4 in the triaminobenzene moiety are sites of acetylation. However, the exact positions of the acetyl groups could not be determined for this metabolite.
Metabolite R2.
During the collecting interval of 0 to 24 h, the radioactivity ofR2 (Fig. 3) represented in rat urine about 4% of the administered dose. In the ESI-MS spectrum of the isolated fraction, an intense quasimolecular ion [M+H]+ atm/z 206 was observed, accompanied by cluster ions atm/z 228 [M+Na]+ and atm/z 244 [M+K]+ (MS data not shown). Product ions of m/z 206 were observed at m/z 188,m/z 146, and m/z 43, which represent losses of water and acetyl moieties (data not shown). In the ESI- mode the [M-H]- ion at m/z 204 was obtained, supporting a molecular mass of 205 for R2. The 300 MHz 1H-NMR spectrum of the isolated fraction in methanol-d4 showed the characteristic signals of a benzene ring which is 3-fold substituted (Table 2).
The 2D H,H-cosy experiments provided evidence for two directly coupled protons at 7.13 and 7.36 ppm (doublet) and a further one at 7.84 ppm. Further signals appeared at 1.20 and 2.00 ppm were interpreted as belonging to CH3 and CH3CO groups, respectively. That was confirmed by the inverse H,C-correlation, which revealed the corresponding signals at 23 ppm (CH3CO) and 30 ppm (CH3). The findings from NMR data along with the results obtained by ESI-MS suggested an acetylated benzimidazole structure. The structure proposed for R2 (Scheme FS2) is 5-acetamido-2-methyl-benzimidazol-1-oxide.
Metabolite R3.
Metabolite R3 (Fig. 3) represented most of the radioactivity (about 5% of the dose) among the metabolites detected in rat urine within the collecting interval 0 to 24 h. The molecular mass ofR3 was 191. This was deduced from the ESI+ spectrum which showed a very abundant [M+Na]+ ion at m/z 214, appearing with the [M+H]+ ion at m/z 192 (MS data not shown). In the ESI- spectrum a very intense [M-H]- ion at m/z 190 was obtained. Product ion scanning of m/z 190 in the negative ion mode produced a fragment ion at m/z 147, representing loss of the acetyl moiety. The 1H-NMR spectrum of the isolated fraction R3 presented similarly to that ofR2, three proton signals of the benzene moiety at 7.1, 7.2, and 7.7 ppm, indicating a 1,2,4-substitution (data not shown). At 2.3 ppm the signal of a CH3 group was detected. Based on these MS and NMR results, the proposed structure for R3(Scheme FS2) was 5-acetamido-2-oxo-2,3-dihydro-1H-benzimidazole. To verify this proposal for R3, the corresponding reference compound was synthesized. The MS, NMR, and chromatographic data obtained from synthesized 5-acetamido-2-oxo-2,3-dihydro-1H-benzimidazole (data not shown) confirmed clearly the proposed benzimidazole structure for R3.
Metabolite R4.
The radioactivity of R4 (Fig. 3) represented about 3% of the administered dose in the urine pool of rats. By ESI-MS a quasimolecular ion [M+H]+ at m/z 190 could be observed, which was accompanied by a [M+Na]+ cluster at m/z 212, indicating an odd number of nitrogen atoms (MS data not shown). CID fragmentation of the [M+H]+ ion resulted inm/z 147 and m/z 43 as daughter ions. Negative ion MS gave the [M-H]- ion at m/z 188 and both product ions at m/z 145 and 146, demonstrating the loss of an acetyl group. These results suggested an acetylated benzimidazole structure. The structure proposed for R4(Scheme FS2) is 5-acetamido-2-methyl-benzimidazole.
Metabolite R5.
The radioactivity of the metabolite R5 (Fig. 3) represented about 3% of the administered dose in the urine pool of rats. It cochromatographed with the peak at 92 min intensively formed by rat liver slices (Fig. 3). The TSP+ spectrum ofR5 revealed a protonated molecular ion [M+H]+ at m/z 238 (MS data not shown). Daughter ion scanning of m/z 238 produced a fragment ion at m/z 192, which detected the loss of the ethanol moiety. On the other hand, the presence of the daughter ion atm/z 43 detected the loss of an acetyl group. The appearance of the product ion at m/z 166 indicated the presence of a monoacetylated triaminobenzene moiety. This was supported by the derived molecular mass of 237, which showed that R5 had retained all nitrogen atoms of retigabine. Based on these results, the proposed structure for R5 (Scheme FS2) isN-(2-amino-4-(acetamido)-phenyl) carbamic acid ethyl ester. The exact position of the acetyl group was not determined.
Metabolite R6.
The radioactivity of R6 (Fig. 3) in the pooled rat urine represented only approximately 0.2% of the given dose, although the urinary radioactivity of R6 within the collecting interval 4 to 8 h amounted to about 4%. Therefore, we have investigatedR6 in this urinary fraction by on-line HPLC/TSP-MS. In the full scan mode the quasimolecular ion [M+H]+ atm/z 274 could be clearly detected (MS data not shown). The appearance of the CID product ion at m/z 256 indicated a loss of water. The appearance of m/z 109 in the CID daughter ion spectrum attested to the loss of the fluorobenzylic side chain (fluorotropylium ion). The fragment ions at m/z 231 and 232 as well as m/z 43 as product ions of [M+H]+ indicated the presence of an acetyl group. Furthermore, metabolite R6 could be chromatographically enlarged by spiking with the authentic substance 1-acetamido-2-amino-4-(4-fluorobenzylamino)-benzene in rat urine. In addition, the MS data of R6 were identical with those of 1-acetamido-2-amino-4-(4-fluorobenzylamino)-benzene (data not shown). This strongly suggested that R6 (Scheme FS2) represented the same structure as the reference compound.
Metabolite D1.
D1 was the major metabolite in dog urine, representing about 5% of the administered dose in the collecting interval 0 to 24 h. Metabolite D1 was extremely unstable to acid hydrolysis at room temperature. By using an acidic HPLC eluent, for example, adjusted to pH 5.5, a decrease of the peak at 112 min and a concomitant increase of the peak of the parent drug during the HPLC run could be detected (data not shown). No other radioactive peaks were noted. In the ESI-MS spectrum of fraction D1 isolated from dog urine, cluster ions at m/z 502 [M+Na]+ andm/z 518 [M+K]+ appeared in addition to the protonated ion [M+H]+ at m/z480 (MS data not shown). Moreover, the protonated molecular ion atm/z 304 of the unchanged parent drug could be detected as in-source fragment. The CID product ion spectrum of m/z 480 revealed the typical daughter ions of retigabine as found in a analogous experiment with the parent drug (data not shown). In the negative ion mode, the fragment ion at m/z 175 appeared, representing the loss of glucuronic acid. In addition, 300 MHz1H-NMR data obtained from D1 were compared with those of retigabine. In the NMR spectrum of D1in CDCl3 (data not shown), the signal of the secondary amino group at 6.1 ppm coupled with the adjacent protons of the methylene group. The corresponding signal in the spectrum of retigabine appeared at the same position and was found to be identical. Consequently, this suggested a linkage of the glucuronic acid to the primary amino group. Based on these findings, we suggested forD1 (Scheme FS2) a N-glucuronide conjugate in the 2 position of the triaminobenzene moiety of retigabine.
Metabolite D2.
The radioactivity of the metabolite D2 excreted in dog urine during the collecting interval 0 to 24 represented about 1% of the administered dose. The ESI-MS spectra obtained from fractionD2 were very similar to those of metabolite D1, but a protonated molecular ion [M+H]+ atm/z 466 was monitored, supported by both cluster ions atm/z 488 [M+Na]+ and m/z504 [M+K]+ (MS data not shown). The ESI+ spectrum of D2 was analogous to that from D1 in the mass range < m/z 350. In addition, all characteristic fragment ions of retigabine were present, including the protonated molecular ion at m/z 304. This was also evidenced by the CID product ion spectrum ofm/z 466. Here, daughter ions were detectable atm/z 448 and m/z 430, respectively, yielded by the successive loss of water from protonated molecular ions. The difference of m/z 14 found between the protonated molecular ions ofD1 and D2 could indicate the loss of a CH2 group. But, because this difference is exclusively detected in the mass range > m/z 304 when it is compared to D1, we suggested for D2 (SchemeFS2) the presence of an N-glucoside of the unchanged parent drug.
Metabolite M1.
The ESI-MS data obtained from isolated fraction M1 from human urine (Fig. 4) were in perfect accordance with those detected forD1 found in dog urine, revealing theN-glucuronide structure in position 2 of the triaminobenzene moiety (MS data not shown). Results of a detailed analysis of the1H-NMR spectrum obtained from isolatedM1 at 600 MHz in DMSO-d6 (data not shown) were complementary with these MS data. The NHF signal of the secondary amino group (Fig.5) detected at 5.93 ppm showed strong cross peaks to neighboring methylene protons as well as to a proton in ortho position at the fluorobenzylic ring as detected by nuclear overhauser effect spectroscopy (NOESY) as well as by total correlation spectroscopy (TOCSY).
The position of the urethane side chain adjacent to the aromatic proton HD in position 6 of the triaminobenzene ring at 6.77 ppm results from the interchange broadening of this proton. This is caused by a restricted rotation around the NHA-CO bond in M1. This NHA proton of the urethane moiety was found at 8.11 ppm. Consequently, only the primary amino group is left over as the site of linkage with glucuronic acid. This is evidenced by cross peaks between the NHH function in position 2 of the triaminobenzene moiety at 4.79 ppm and the glucuronide acid residue, as detected by total correlation spectroscopy (TOCSY). The anomeric proton HL at C-1 in the glucuronide moiety could be identified at 5.17 ppm. Nuclear overhauser effects (NOE) were observed between the aromatic proton HE at 6.09 ppm and the glucuronic acid protons. These data indicate that M1 and D1 have identical structures (Scheme FS2).
Metabolite M2.
The interpretation of the ESI-MS spectra of metaboliteM2 (Fig. 4) is made easy by comparison with those of the metabolite R6 in rat urine as well as with those of the reference substance 1-acetamido-2-amino-4-(4-fluorobenzylamino)-benzene. Both full scan (Fig. 6) and CID experiments (data not shown) resulted in the same mass spectroscopic information forM2 as that obtained from reference. Data obtained by1H-NMR (600 MHz) from both compounds dissolved in DMSO-d6 also confirmed an identical structure (Table 3). In both cases, two sets of1H-NMR signals were detected. At increasing temperatures, the major and minor signals approach each other coalescing at the temperature TC ≈ 100°C. The corresponding signals were assigned as major and minor signals (Table 3). For example, the amide NHA major signal at 8.81 ppm corresponded to the minor signal at 8.12 ppm and the major signal of the neighboring aromatic proton HD at 6.69 ppm corresponded to the minor signal at 6.58 ppm. The detailed analysis of 600 MHz1H-NMR data revealed the presence of E/Z conformers (major/minor signals) of the amide structure. Integration of the signals yielded the approximate ratio of the two isomers as 90:10 at 27°C. Repeated heating and cooling of analyzed solutions did not change this ratio.
We could support these NMR results by semiempirical molecular orbital calculations using the Austin model 1 approach (AM1) conducted with the VAMP 5.6 program (Oxford Molecular Ltd., Oxford, UK) on the basis of the model structure 1-acetylamino-2-aminobenzene. The calculated differences between the energies of the transition state (saddle point) and the stable conformations (minima) are in a range from 8.6 to 9.4 kcal/mol. These data suggest that both conformers are stable at room temperature. Based on the energy difference of 0.8 kcal/mol found between the two conformers, a ratio of conformers was calculated to be 80:20 from Boltzmann statistics. The outcome of the calculation shows that the energetically more stable conformer corresponds to the E conformer, as expected. This ratio, 80:20, calculated on the basis of the model structure is in good accordance with the ratio 90:10, which we have estimated from the NMR data forM2.
Discussion
We have optimized a method for the isolation and characterization of metabolites from body fluids by chromatographic and spectroscopic techniques (Vollmer et al., 1986). This method is based on the repeated chromatographic preparation of radioactive fractions, their recombination, and rechromatography. The power of this method is demonstrated by the successful baseline separation of the closely eluting metabolites R2 and R3 (Fig. 2). Earlier studies have identified glucuronides of retigabine as metabolites in vitro in humans (McNeilly et al., 1997). Here we have extended these investigations to further in vitro experiments, profiling of in vivo samples from rats, dogs, and humans, and the structural identification of the major metabolites produced in vivo. Our A/D-method, when applied to samples from clinical studies, produced sufficient material for MS/MS and additionally NMR characterization. As a result, we provide unambiguous evidence on the basis of both detailed 600 MHz1H-NMR analysis and ESI-MS/MS results that the metabolite fraction M1 isolated from human urine at retention time 112 min is an N-glucuronide of unchanged [14C]retigabine substituted at the primary amino group (M1, Scheme FS2). The predominance of this major metabolite in the plasma profiles of rats, dogs, and humans suggests that glucuronidation also plays an important role in in vivo metabolism of retigabine after oral administration (Scheme FS2). AnotherN-glucuronide substituted at the secondary amino group, which was first suggested by McNeilly et al. (1997), was chromatographically identified at 105 min in lower concentrations in liver slices as well as in rat bile and in liver microsomal experiments using [14C]retigabine or in the inverted mode using [14C]UDPGA. But, to date, it could be identified by mass spectroscopy only in in vitro assays. RetigabineN-glucuronide at retention time of 112 min is the major metabolite in the plasma of rat, dog, and humans. Interestingly, no glucuronides could be detected in rat urine. The urinary rat metabolites are an example of extensive multiple acetylation and ring closure reactions. Radiometric profiling of rat urine revealed more than 20 metabolites (Fig. 3) in addition to only a minor portion of unchanged [14C]retigabine (Table 1). From these metabolites we have identified seven structures which represent >64% of the radioactive dose excreted in urine in 24 h (Table 1). Furthermore, 100% of the radioactive dose present in 24-h dog urine can be attributed to metabolites identified in this article. From this we conclude that, in dogs, retigabine glucuronide and glucoside (see below) are the only existing metabolites in urine.
In dogs, low amounts of an N-glucoside (D2) of the unchanged parent drug were detected in urine (1% of the dose). This metabolite was also identified as a minor fraction in plasma as suggested by cochromatography with isolated fraction D2 from dog urine (data not shown). Moreover, we have shown that in our studiesN-glucosidation of retigabine seems to be a pathway unique to the dog. The formation of N-glucuronide, however, is common to other species. N-glucosides are not very common drug metabolites but have been reported for amino salicylic acid and barbiturates (Tjornelund et al., 1989; Soine et al., 1991). Because this metabolite was only detected in dogs, we regard its formation as a species-specific effect with no relevance to the human metabolism. In terms of species specificity, the dog is also unique in the metabolism of retigabine in that it expectedly does not form acetylated metabolites.
The metabolite structures identified in rat urine showed thatN-acetylation and ring closure reactions, which lead to benzimidazole structures are further important pathways in the metabolism of retigabine in this species. The elucidation of the structure of R3 (Scheme FS2), the major metabolite in rat urine, as 5-acetamido-2-oxo-2,3-dihydro-1H-benzimidazole is evidence for the formation by ring closure from retigabine, whereas another mechanism leading to benzimidazoles is the ring closure from acetylated metabolites such as R6, which lead to methyl benzimidazoles like R4. The urinary metabolite R5is an interesting rat-specific metabolite. It appeared in plasma and urine of rats (Fig. 3). On the basis of its intense formation in rat liver slices, we suggest that R5 is the starting product for the formation of R3 by ring closure under concomitant loss of ethanol.
Our data on the metabolism in humans are preliminary because they rely mostly on nonradioactive methods, as a 14C-study in humans has not been performed. The advantage of liver slice experiments is the possibility of the prediction of the human metabolism with radioactive substance in an in vitro system. Using the collective information of the above described experiments together with information from synthesized reference compounds we can now assign structures to the metabolite profiles from human liver slices as well as human plasma as shown in Fig. 4. Retigabine glucuronideM1 (112 min) is the dominant metabolite, presumably followed by the alternative glucuronide (105 min) and the acetylated metaboliteM2 (123 min, Scheme FS2). Our data show the predominant role of phase II reactions in the metabolism of retigabine. Especially the efficient N-glucuronidation, which is apparent from in vivo and in vitro data, warrants a further investigation.
Acknowledgments
We thank A. Scherf for skillful technical assistance and M. Schwahn (ASTA Medica, Frankfurt, Germany) for surgically preparing the bile duct-catheterized rats and performing the bile excretion study.
Footnotes
-
Send reprint requests to: Dr. Roland Hempel, Corporate Research & Development, ASTA Medical Group, Biochemistry Dresden, Arzneimittelwerk Dresden GmbH, Meissner Str. 191, D-01445 Radebeul, Germany. E-mail:Dr-Roland.Hempel{at}astamedica.de
-
This work was supported by the Sächsisches Ministerium für Arbeit und Wirtschaft (project number 1791).
-
1 Introduced on the market by ASTA Medica AG, Frankfurt/Main, Germany; trade name: Katadolon.
- Abbreviations used are::
- CID
- collision induced dissociation
- DMSO
- dimethyl sulfoxide
- HPLC
- high-performance liquid chromatography
- MS
- mass spectroscopy
- LC
- liquid chromatography
- UDPGA
- uridine 5′-diphosphoglucuronic acid
- SPE
- solid phase extraction
- SPHPLC
- semipreparative chromatography
- A/D
- adsorption/desorption
- ESI-MS
- electrospray mass spectroscopy
- TSP-MS
- thermospray mass spectroscopy
- NMR
- nuclear magnetic resonance
- Received July 22, 1998.
- Accepted January 26, 1999.
- The American Society for Pharmacology and Experimental Therapeutics