Characterization of Dark Liver Pigment Observed in Rats after

Minxia M. He,*,† Robert J. Barbuch,† Trent L. Abraham,† Thomas J. Lindsay,†. David A. Jackson,† Bradley L. Ackermann,† August V. Wilke,‡...
0 downloads 0 Views 131KB Size
912

Chem. Res. Toxicol. 2003, 16, 912-919

Characterization of Dark Liver Pigment Observed in Rats after Subchronic Dosing of the β3-Adrenergic Receptor Agonist LY368842 Minxia M. He,*,† Robert J. Barbuch,† Trent L. Abraham,† Thomas J. Lindsay,† David A. Jackson,† Bradley L. Ackermann,† August V. Wilke,‡ and Charles B. Jensen† Drug Disposition, Lilly Research Laboratories, Eli Lilly and Company, Indianapolis, Indiana 46285, and Toxicology, Lilly Research Laboratories, Eli Lilly and Company, Greenfield, Indiana 46140 Received October 28, 2002

Dark liver pigmentation was observed in F344 rats in a subchronic toxicology study after daily dosing of LY368842 glycolate. In addition, green-colored urine was observed in some animals. To identify the source of the pigment and its potential for toxic consequences, the liver pigment was isolated from the liver tissue of rats. The resulting material was a dark brown to black powder that was insoluble in water, organic solvents, or a tissue-solubilizing agent. Several techniques, such as chemical degradation, HPLC, tandem mass spectrometry (LC/MS/MS), 1H NMR, and matrix-assisted laser desorption ionization mass spectrometry (MALDI-MS), were employed to characterize the dark liver pigment. Following oxidative degradation of the isolated pigment, degradation products related to LY368842 were identified or tentatively identified using LC/MS/MS. Two degradation products had the same protonated molecular ion at m/z 505, which is 30 amu higher than that of LY368842. The major m/z 505 product has been identified as the indole-2,3-dione oxidative product based on 1H NMR data and confirmed by an authentic standard. In addition, monohydroxylated product was also identified in the degradation mixture. These degradation products were consistent with the metabolites found in vivo in rats. MALDI-MS analyses of liver and urine pigment both identified a product with a protonated molecular ion at m/z 977, suggesting formation of indirubin-like and indigo-like pigments. The results obtained suggest that the oxidative metabolites of LY368842 played a key role in the formation of the liver and urine pigments.

Introduction LY368842 is a potent agonist of the human β3-adrenergic receptor and contains the indole moiety (Figure 1). During subchronic toxicology studies in rats and monkeys, black-colored livers were observed in animals receiving high daily doses of LY368842 glycolate salt. In addition, green-colored urine was observed in some of the treated animals. This observation did not correlate with toxicity, but indoles, indole-containing compounds, and their metabolites are known to have biological effects. Among the variety of xenobiotics containing an indole or a substituted indole structure, many appear to have desirable features. For example, indole is known to suppress the hepatotoxicity and carcinogenicity of 2-acetylaminofluorene in rats and hamsters (1). Several indole antioxidants were found to inhibit lipid peroxidation and therefore prevent chemical hepatotoxicity through a radical scavenging mechanism (2). However, a number of other indole-containing xenobiotics have shown undesirable effects. 3-Methylindole was found to be a selective pulmonary toxicant (3). The toxicity of 3-methylindole appeared to be associated with the formation of reactive intermediates (2,3-epoxide) (4), followed by covalent * To whom correspondence should be addressed. Tel: (317)277-9477. Fax: (317)433-6432. E-mail: [email protected]. † Drug Disposition. ‡ Toxicology.

Figure 1. Chemical structure of LY368842.

binding to critical pulmonary proteins (5). Dietary administration of indole-3-carbinol was found to down-regulate FMO1 activity and induce cytochrome P450, which resulted in significant alterations in the metabolism, disposition, and toxicity of xenobiotics (6, 7). Indole undergoes oxidation in the five-membered ring by microbial oxygenase to form indoxyl, which can be further oxidized and dimerized to form indigo (8). Recently, the specific P450 isoforms responsible for metabolism of indole have been investigated with expressed enzyme systems. It was reported that incubation of indole with bacterial membranes containing recombinant human P450 2E1 and human NADPH-P450 reductase produced a blue pigment in a time- and cofactor-dependent manner (9). Further work in the same laboratory has definitively identified the pigments indigo and indirubin as the products of human cytochrome P450-catalyzed metabolism of indole (activity in the order of CYP2A6 > 2C19 > 2E1) (10). Natural eumelanins, dark brown to black insoluble pigments, are commonly found in hair, skin, horns, feathers, and squid ink. They are formed by polymerization of 5,6-indolequinone derived from Tyr (11). How-

10.1021/tx0201016 CCC: $25.00 © 2003 American Chemical Society Published on Web 06/26/2003

Characterization of Dark Liver Pigment from Rats

ever, dark liver pigment has not been commonly observed. It was only found in certain areas of Australia in the 1960s that a dark pigment accumulated in sheep livers, and the liver pigment was identified as an indole melanin (12). Pindolol, an indole-containing β-adrenergic receptor antagonist used for the treatment of angina and hypertension, was also reported to cause tissue and urinary pigmentation in animals (13). The goal of the present investigation was to characterize the nature and the structure of the liver pigment from rats and to determine whether the pigment was derived from LY368842 or its metabolites. Experiments were conducted to isolate the dark liver pigment from rat livers and to characterize the pigment by chemical degradation followed by HPLC, LC/MS/MS,1 and 1H NMR analyses of the degradation products. Attempts were also made to characterize the black liver pigment as well as the green urine pigment by MALDI-MS analyses.

Experimental Procedures Chemicals. LY368842 glycolate was synthesized in Lilly Research Laboratories. HPLC grade acetonitrile and methanol and analytical reagent grade potassium carbonate, hydrogen peroxide, sodium metabisulfite, ethyl acetate, and glacial acetic acid were purchased from Mallinckrodt Baker, Inc. (Paris, KY). Hexane and acetone were purchased from Fisher Scientific (Pittsburgh, PA). Triton X-100 was purchased from Sigma (St. Louis, MO). Soluene 350 was obtained from Packard Instrument Co. DHB was purchased from Sigma Aldrich Fine Chemicals. Dermenkephalin and bradykinin were purchased from the Sigma Chemicals. TFA was purchased from Pierce Chemicals (Rockford, IL). FA was purchased from Fluka (Milwaukee, WI). Isolation of Liver Pigment from Rat Livers. A modified method of Oikawa and Nakayasu (14) was used for isolation of pigment from liver tissue of rats that had received daily doses of 500 mg/kg of LY368842 during the subchronic study. The pigment isolation procedures involved tissue homogenization, detergent treatment, and extensive aqueous and organic washing by centrifugation. Briefly, 25% (w/v) liver homogenates were prepared in water using a Polytron homogenizer (Brinkman Instruments, Inc). The homogenate was centrifuged at 3000g for 20 min, the supernatant was discarded, and the pellet was used for pigment isolation. Initial experiments showed that acid digestion of the rat liver homogenate degraded the LY368842related pigment. Therefore, a milder treatment with 20% (v/v) Triton X-100 was developed to solubilize the rat liver homogenate. Two volumes of 20% Triton X-100 were added to the pellet, and the sample was vortexed. The sample was centrifuged at 9 000g for 10 min, and the dark pellet was washed three times with 10 mL volumes of water. Next, the pellet was washed three times with 10 mL volumes of acetone and once with 10 mL of hexane. The black pellet was air-dried, and the dry pigment was stored in the dark at approximately -70 °C. Chemical Degradation of Liver Pigment. The oxidative degradation reaction was performed according to the method of Napolitano (15). A solution of 1 M K2CO3 (5 mL) in a glass tube was purged with purified nitrogen for approximately 40 min. Liver pigment (5 mg) was suspended in 1.0 mL of MilliQ purified water and added to the tube. The mixture was capped with a rubber stopper supplied with a temperature probe, then stirred and heated to 80 °C, and maintained at 80 °C for 30 1 Abbrevations: DHB, dihydroxy benzoic acid; TFA, trifluoroacetic acid; FA, formic acid; LC/MS/MS, liquid chromatography-tandem mass spectrometry; MS/MS/MS, mass spectrometry to the third power; MALDI-MS, matrix-assisted laser desorption ionization mass spectrometry; ESI, electrospray ionization; CID, collisionally induced dissociation; FAB-MS, fast atom bombardment mass spectrometry; FD-MS, field desorption mass spectrometry; FMO, flavin-containing monooxygenase.

Chem. Res. Toxicol., Vol. 16, No. 7, 2003 913 min while stirring. The reaction mixture was allowed to cool for 15 min at ambient temperature. An aliquot of 30% H2O2 (175 mL) was added to the tube, and the rubber cap was replaced. The mixture was stirred at ambient temperature overnight and then heated and maintained at 80 °C for an additional 30 min. After the mixture was cooled to room temperature, 5% NaHSO3 (1.5 mL) was added. The oxidative reaction mixture was adjusted to pH 1 with concentrated HCl. The reaction mixture was then extracted five times with an equal volume of ethyl acetate. The organic layers were combined and evaporated to dryness, and the residue was analyzed by LC/MS. The aqueous portion was mixed 1:1 with methanol and evaporated to dryness at 50 °C under nitrogen. The residue from the aqueous portion was reconstituted with 150 µL of 0.05% aqueous TFA and analyzed by LC/MS/MS. LC/MS/MS. LC/MS/MS analysis was performed using an Inertsil ODS-3 (MetaChem 3.0 mm × 150 mm, 5 µm) at a flow rate of 0.35 mL/min (Waters 600MS system). A mobile phase gradient of 0.05% TFA (A) and methanol (B) was programmed as follows: started with 100% solvent A; changed to solvent A/solvent B at 80:20 over 1 min; then changed to solvent A/solvent B at 50:50 from 1 to 41 min and then held at 50:50 from 41 to 60 min; then changed to 100% solvent A at 60.1 min and held for 15 min. MS analysis was performed with a Finnigan TSQ mass spectrometer using ESI in the positive ion mode, using a spray voltage of 5000 V, a capillary heater temperature of 250 °C, an indicated sheath gas pressure of 80 psi (N2), and an auxiliary gas flow rate of 30 mL/min (N2). For full scan analysis, the mass spectrometer was scanned from m/z 125 to 1000 in 1 s. MS/MS analysis was performed at a collision energy of -25 eV using argon as the collision gas (1.8 mTorr). Structural analysis for the authentic standard of the indole2,3-dione metabolite (m/z 505) was performed using a Finnigan LCQ ion trap mass spectrometer by positive ion ESI using similar interface conditions as described above. In this experiment, MS/MS/MS data were acquired under direct sample infusion by fragmenting the protonated molecule (m/z 505) under wideband excitation and selecting the predominant product ion at m/z 237 for further decomposition. A relative collision energy of 40% was used for both transitions. The resulting product ion mass spectrum was recorded over an m/z range of 70-510. HPLC Isolation of Chemical Degradation Product. After LC/MS/MS analysis, the remaining degradation sample was used for HPLC isolation of the peaks of interest. HPLC analysis of the chemical degradation mixture was performed using an Inertsil ODS-3 (MetaChem 2.1 mm × 150 mm, 5 µm) at a flow rate of 0.35 mL/min (Waters 600MS system). The same mobile phase and gradient system were used as described in the above section. The peaks of interest (peaks A and B) were isolated and collected separately in 16 mm × 100 mm silylated glass tubes, based on simultaneous detection by UV (218 nm) and LC/MS (m/z 505). The fractions were then evaporated to dryness at 50 °C under nitrogen. 1H NMR. 1. Condition A. The 1H NMR spectrum of isolated peak A was obtained on a Varian Unity 500 MHz system equipped with a 5 mm indirect detection probe. The entire isolated sample was dissolved in 200 µL of 0.05% TFA in D2O. 2. Condition B. The 1H NMR spectrum of the indole-2,3dione standard was obtained on a Bruker AVANCE 300 MHz system equipped with a 5 mm QNP probe. The indole-2,3-dione standard (7 mM) was dissolved in 50 µL of DMSO-d6, which was dispersed in 690 µL of D2O. The NMR analyses were performed at 25 °C in both conditions. Chemical shifts were reported in ppm downfield from tetramethylsilane. MALDI-MS and MALDI-MS/MS. All reported data were acquired using DHB as the MALDI matrix. The DHB matrix solution (10 mg/mL) was prepared by dissolving DHB in 30/70 (v/v) 0.1% TFA/acetonitrile. Prior to MALDI sample preparation, fine suspensions of liver and urinary pigment were prepared by sonicating the pigment (1 mg/mL) in either 0.1% TFA or concentrated FA. Each suspension was diluted 1:10 in the

914

Chem. Res. Toxicol., Vol. 16, No. 7, 2003

He et al.

Figure 2. Product ion mass spectrum and fragmentation of LY368842 standard. corresponding acid prior to analysis. Although both acids yielded qualitatively similar results, FA produced a more dispersed suspension resulting in superior MALDI mass spectra. The influence of concentrated FA on the drug substance was investigated by preparing a 1 mg/mL stock solution of LY368842 in concentrated FA. The MALDI mass spectrum prepared from a dilution of this solution using DHB exhibited the expected protonated mass at m/z 475. No evidence of formylation or degradation was observed. MALDI sample preparation consisted of successive application of 0.5 µL aliquots of sample followed by matrix solution onto a stainless steel target. The samples were allowed to air dry at ambient conditions prior to inserting the sample into the ion source of a TofSpec-2E, MALDI time-of-flight (TOF) mass spectrometer (Micromass, Inc., Beverly, MA). The flight tube length of the instrument was 1 m in the linear mode and 1.7 m in the reflectron mode. Ionization was initiated by a nitrogen laser (337 nm) fired at a rate of 5-10 shots/s. Typically, 40 shots were averaged per spectrum. The acceleration potential used was 20 kV. Detection occurred using a microchannel plate detector with a voltage of 1800 V. Exact mass determination for the ion observed at nominal m/z 977 was performed by coanalyzing an aliquot of the liver pigment (FA suspension) with two peptides serving as internal mass references (dermenkephalin, m/z 955.428, and bradykinin, m/z 1060.569). The mass resolution used for this experiment (ca. 2500 full-width half-maximum) was achieved using the reflectron mode. Mass assignment was accomplished using the software provided by the instrument vendor (MassLynx, v. 3.3). Analysis of the rat urine sample was accomplished under low resolution (500 full-width half-maximum) in the linear mode. Mass assignment for this experiment was obtained using an external mass calibration conducted prior to analysis and verified with peptide standards. Structural information for the ion at nominal m/z 977 in the liver pigment was obtained using the postsource decay mode of the TofSpec-2E. A parent ion selection window of 38 amu was used for this experiment with argon as the collision gas. Nominal m/z assignment was made using the software provided by the vendor under an external mass calibration.

Results Characterization of Liver Pigment by Oxidative Degradation Followed by LC/MS/MS, HPLC, and 1H NMR Analyses. The pigment isolated from rat liver was similar in appearance to synthetic dopamelanin. It was insoluble in water, organic solvents, or soluene (a tissuesolubilizing agent). Attempts to characterize the pigment by direct analysis were not successful due to the high degree of insolubility in aqueous or organic solvents. An indirect method of degrading the pigment and analyzing the degradation products was, therefore, employed. Oxidative degradation of isolated liver pigment was conducted using alkaline hydrogen peroxide treatment, and degradation products were characterized by LC/MS/MS, HPLC, and 1H NMR analyses. The product ion mass spectrum of LY368842 standard (m/z 475) was first obtained (Figure 2). The major ions observed (m/z 207 and m/z 269) result from cleavage of the molecule at the nitrogen-carbon bond. Subsequent loss of ammonia and water from m/z 207 produced the ion at m/z 172. The ion at m/z 134 is consistent with protonated hydroxyindole. The product ion at m/z 74 corresponds to C3H8NO. The liver pigment degradation sample was then analyzed in full scan positive ion mode. LY368842 was not detected in the pigment degradation mixture. The positive ion mass chromatograms indicated a large peak (peak A) and a small peak (peak B), both having a protonated molecular ion at m/z 505, which is 30 amu higher than the parent LY368842. Peak A eluted about 2 min after peak B and was at least 10-fold greater than peak B in relative intensity. The product ion mass spectra of peak A and peak B were identical. Figure 3 shows the representative product ion mass spectrum of peak A (the large m/z 505 peak). The appearance of the product ions at m/z 237 and m/z 269 indicates the addition of 30 amu to the propoxyindole moiety (Table

Characterization of Dark Liver Pigment from Rats

Chem. Res. Toxicol., Vol. 16, No. 7, 2003 915

Figure 3. Product ion mass spectrum of peak A (m/z 505) from degradation of liver pigment. Table 1. Tentatively Identified LY368842-Associated Pigments and Major Degradation Product from Liver Pigment

[M + H]+

characteristic product ions (m/z)

proposed metabolite identification

liver green pigment urine +

+

269, 636, 691, indirubin-like or 709 indigo-like pigment

+

+

NA NA

ND ND

+ +

505a (peak A)

219, 237, 269

977b,c 999b 1015b

indole-2,3-dione

Na adduct of m/z 977 K adduct of m/z 977

a Detected in degradation mixture of liver pigment or in the supernatant of green urine by LC/MS/MS. b Detected in liver or urine pigment by MALDI-MS. c The exact mass of the oxidative dimer was m/z 977.405. NA, not available; ND, not detected.

1). The loss of water from m/z 237 produced the ion at m/z 219. Hydroxylation at the aliphatic positions is unlikely, since it would lead to formation of cleavage metabolites, namely, the diol and amine metabolites. It is likely that the 30 amu was added to the indole ring system, because no other reasonable structures with oxygen attached to other positions of the propoxyindole moiety can be proposed. The product ion spectra of peaks A and B were identical to the product ion mass spectrum of an authentic standard of the indole-2,3-dione metabolite (Figure 4). The structural assignment of peaks A and B was further supported by the presence of a weak ion at m/z 164 (MS/MS/MS data, consistent with protonated hydroxyindole plus 30 amu), confirming that the additional 30 amu was added to the indole ring. The retention time of the authentic standard of indole-2,3-

Figure 4. Product ion mass spectrum of authentic standard of indole-2,3-dione (m/z 505).

dione was the same as the peak A. In addition to peaks A and B, a minor product with a protonated molecular ion at m/z 491 was tentatively identified as hydroxy LY368842. The product ions at m/z 223 and m/z 269 indicate hydroxylation on the 3-aminopropoxyindole moiety. Additional product ions further define the site of hydroxylation: m/z 298, loss of C10H11NO3; m/z 192, C10H10NO3; and m/z 92, C3H10NO2 suggest hydroxylation of the indole ring. The product ions at m/z 192 and m/z 298 indicate hydroxylation at either the 3- or the 5-position of the indole ring. LC/MS/MS analysis of pigment degradation samples also identified a few other minor

916

Chem. Res. Toxicol., Vol. 16, No. 7, 2003

Table 2. Proton NMR Assignments for LY368842 Standard, the Isolated m/z 505 Compound (Peak A), and Indole-2,3-Dione Standard

site

type

2 3 5 6 7 9 10 11 13 14 16 17 19 20 23 24 26

dCHdCHdCHdCHdCH-CH2O-CH〈 -CH2〉C(CH3)2 -CH2dCHdCHdCHdCHdCHdCHdCH-

b

m/z 505 indole-2,3-dione LY368842 compound standard proton (peak A) (m/z 505) (ppm) and proton (ppm) (multiplicity)a proton (ppm) 7.39 (s) b 6.74 (d) 7.20 (dd) 7.24 (d) 4.30 (m) 4.40 (m) 3.58 (d) 1.54 (s) 3.18 (dd) 7.45 (d) 7.08 (d) 7.08 (d) 7.45 (d) 7.19 (d) 8.27 (d) 8.50 (s)

6.96 (d) 7.48 (dd) 7.01 (d) 4.27 4.38 3.48 1.26 3.13 7.40 7.10 7.10 7.40 7.12 8.24 8.48

6.53 (d) 7.58 (dd) 6.75 (d) 4.20 4.28 3.30 1.30 3.05 7.35 7.15 7.15 7.35 7.10 8.28 8.62

a s, singlet; d, doublet; m, multiplet; dd, doublet of doublets. H/D exchange at the 3-position (17).

products, but they were not of sufficient abundance to warrant discussion here. Because mass spectral data of peak A and peak B were unable to define the sites of oxidation on the indole ring, only that 30 amu had been added to the indole, 1H NMR data were acquired for peak A after it was isolated from the liver pigment degradation mixture by HPLC. Data for peak B were not acquired due to insufficient quantities isolated. As a first step in determining the structure of peak A, the 1H NMR spectrum of the parent compound, LY368842, was obtained in deuterium oxide with 0.05% TFA (Table 2). The 1H NMR spectrum of peak A and the expanded view of the aromatic region are shown in Figure 5. In contrast to the parent compound, the hydrogen atom attached to carbon 2 was not present in the spectrum of peak A. In addition, the hydrogen atoms attached to carbons 5, 6, and 7 shifted significantly. Assignments of these hydrogen atoms were based on three features: (i) the coupling pattern, i.e., two doublets (C5 and C7) and a partially overlapped doublet of doublets (C6) (Figure 5); (ii) the coupling constants being the same as the parent (8 Hz); and (iii) the integrations indicating one hydrogen atom at each resonance. The resonance shifts indicate a dramatic change in the electronic environment surrounding these hydrogen atoms. The data indicate oxidation at both the 2- and the 3-positions of the indole (that is, indole-2,3-dione). This result is consistent with the data obtained with an authentic standard of the indole-2,3-dione (LY553298 2HCl salt), showing identical coupling patterns for carbons 5, 6, and 7 (i.e., doublets for C5 and C7 and a partially overlapped doublet of doublets for C6; Table 2). On the basis of the LC/MS/MS and 1H NMR results, peak A (m/z 505) isolated from pigment degradation products has the indole-2,3-dione moiety. Characterization of Liver Pigment by MALDIMS. The MALDI mass spectrum of the rat liver pigment was acquired in exact mass mode. The measured mass of 977.405 is consistent with the proposed elemental composition for the ion (C54H57N8O10), which has a theoretical mass of 977.420. The net mass difference of 0.015 amu corresponds to a relative difference of -15.3

He et al.

ppm. Although this result does not rule out other elemental compositions, it is in reasonable agreement with the proposed composition. Although the ion at m/z 977 is consistent with a product of LY368842, the exact mass data for m/z 977 did not unequivocally confirm that this ion was related to LY368842. Thus, tandem MS analysis using CID was performed on m/z 977. As indicated in Table 1, four product ions were observed (m/z 269, 636, 691, and 709), which are consistent with an oxidation product of LY368842. These observed ions are also consistent with the known CID fragmentation pattern of LY368842 observed by LC/MS/MS. These data strongly suggest that the product of m/z 977 is related to LY368842. While the exact point of attachment cannot be confirmed without an authentic standard, it can be deduced from the product ions observed that attachment involves the indole ring of each monomer unit accompanied by oxidative biotransformation. Characterization of Green Urinary Pigment. In addition to the dark liver pigmentation, green-colored urine was also observed in some rats following daily administration of 500 mg/kg of LY368842. After freezer storage, a dark green precipitate was observed in the samples. The urine samples were centrifuged, and the supernatant was analyzed by LC/MS/MS. A product with m/z 505 was detected in the urine supernatant, and the product ion mass spectrum was identical to peak A detected in the liver pigment degradation products as discussed above (Table 1). The characterization of the precipitated urinary pigment was performed by MALDIMS analysis, and results are described in Table 1. The MALDI-MS spectrum, obtained using DHB as the matrix, was acquired in the linear mode using an external mass calibration. A product at m/z 977 was detected. The peak at m/z 977 was joined by a series of cation adduct ions: (M + Na)+ at m/z 999 and (M + K)+ at m/z 1015. The spectrum was dominated by the cation adducts, presumably due to the relatively large amount of salt in urine.

Discussion Formation of the dark liver pigment in rats and monkeys administered LY368842 daily in subchronic studies was dose-dependent. The pigment was observed at doses of 160 and 500 mg/kg/day in rats and 16 and 50 mg/kg/day in monkeys. The lower effective doses in monkeys suggested that the monkey probably has a greater ability to produce this pigment. Although slight elevation of serum markers of liver toxicity was observed in some rats given 500 mg/kg of LY368842, there was no correlation between the degree of liver darkness and the level of serum markers. The liver pigmentation was not associated with adverse histologic changes in rats or monkeys during the subchronic studies. Histologically, the only observation in liver was the presence of brown intracellular granules in hepatocytes and Kupffer cells. The pigment deposition appeared to be at least partially reversible, since the livers of other rats and monkeys were less dark after a period of recovery from dosing. The isolated rat liver pigment had a dark brown to black color. The extremely poor solubility of this pigment prevented the direct analysis by LC/MS/MS. Also, analyses using FAB-MS and FD-MS failed to provide any information about the molecular weight of the pigment. However, oxidative chemical degradation appeared to be

Characterization of Dark Liver Pigment from Rats

Chem. Res. Toxicol., Vol. 16, No. 7, 2003 917

Figure 5. Proton 1H NMR spectrum and the expanded spectrum (top) of peak A (m/z 505).

a successful approach, which allowed identification of the degradation products (peak A, peak B, monohydroxylated products, etc.) that are structurally related to LY368842. In addition, MALDI-MS analyses of the liver and urine pigments identified the presence of LY368842-related oxidation product(s) (m/z 977), providing further information about the pigment composition. The metabolism of LY368842 has been characterized in F344 rats given a single oral dose of 10 mg/kg of 14C-LY368842 (manuscript in preparation). Metabolites were identified in different biological matrices in rats (urine, bile, feces, and plasma). Interestingly, two m/z 505 metabolites were detected in rat bile with the same product ion mass spectra. The two metabolites eluted about 2 min apart, and there was a 10-fold difference in intensity (the one that eluted later had greater intensity). The peak elution order, relative intensity, and mass spectra of these two m/z 505 metabolites are consistent with those obtained from the two pigment degradation products, peak A and peak B. In addition, a monooxygenated metabolite (m/z 491) with oxidation at the indole ring (site not determined) was detected in rats, which corresponds to the monooxygenated degradation product (m/z 491). The observation of in vivo-generated metabolites identical to the pigment degradation products suggests that these metabolites played an important role in the formation of the pigment. Two possible pathways could be proposed for the in vivo formation of the LY368842-associated pigments

(Scheme 1). The first pathway with oxidation at the fivemembered ring of the indole is in agreement with the pigment formation pathway reported by Gillam et al. (10). By analogy, an indirubin-like pigment (m/z 977) might be formed from reaction of the indole-2,3-dione (m/z 505) and the monohydroxylated indole metabolite (m/z 491). A similar indigo-like pigment (m/z 977) could be derived from dimerization of the monohydroxylated indoles after further oxidation. The liver and urinary pigments both exhibited a molecular ion at m/z 977 and cannot be differentiated based on MALDI-MS data. The urine pigment (m/z 977) could be indigo-like based on its green appearance. It is likely that the alkoxy substituent on the indole influenced its electronic absorption spectrum to produce a green color rather than the blue color of the unsubstituted indigo reported by Gillam et al. (10). Another possible pathway could be via oxidation of the six-membered ring of the indole moiety. This hypothesis is supported by the fact that another dioxygenated product at m/z 505 was also identified as a chemical degradation product (peak B), and it corresponds to one of the m/z 505 metabolites detected in rat bile with an identical product ion spectrum. No 1H NMR spectrum could be generated to define the oxidation positions for peak B. However, by default, this must involve an oxidation on the six-membered ring of the indole. Although two possibilities exist, 5,6-indolequinone is known to lead to the formation of melanin (11). Therefore, it is possible that LY368842 also underwent oxidation to give an

918

Chem. Res. Toxicol., Vol. 16, No. 7, 2003

He et al.

Scheme 1. Proposed Pathways for the In Vivo Formation of LY368842-Associated Pigmentsa

a Two routes are indicated to account for the product observed at m/z 977. The two isomeric possibilities shown could not be differentiated from the data obtained.

indolequinone (m/z 505), leading to formation of melaninlike pigment in vivo. Further work remains to be conducted to explore this melanin hypothesis. It is not known whether such a melanin-like pigment contributed significantly to the composition of the liver pigment. A melanin-like pigment could contribute to the black color and the extremely poor solubility of the liver pigment. It is likely that the dimer (m/z 977) was detected in liver pigment because it had sufficient solubility, whereas a melanin-like pigment could be expected to be insoluble and would fail to interact optimally with the MALDI matrix, which is required for this type of analysis (16). A pindolol-associated urinary pigment was identified as a blue dye with an indigoid structure (13). This indigoid dye has not been detected in human urine following therapeutic doses of pindolol (approximately 1 mg/kg). The pindolol-derived pigment was detected in numerous rat tissues and in rhesus monkey liver Kupffer cells. The tissue pigment was not considered likely to be of toxicological significance, since it did not adversely affect the function of various organ systems (liver, adrenal, and medulla, etc.) and had no effect on the general health of various animal species given the drug for 1-5 years (13). The pindolol-derived pigment was not fully characterized, although it was hypothesized that it may be a product of 5- and 6-hydroxylation of the indole followed by oxidation to a 5,6-indolequinone. The current investigation demonstrates that metabolites of LY368842 play a key role in the formation of the pigment. In vitro incubations of LY368842 with rat, monkey, rabbit, and human liver microsomes clearly demonstrated species differences in the formation of the m/z 505 metabolite (monkey > rabbit > rat > human), correlating well with the different susceptibility observed in rats and monkeys; that is, monkey formed dark liver pigment at lower doses than rats. The in vitro data suggest that humans may be less susceptible to the liver pigment formation as compared to animal species. There-

fore, monitoring these metabolites may be useful in predicting the likelihood of pigment formation in humans. In summary, darkly pigmented livers were observed in subchronic toxicology studies of LY368842 in rats and monkeys. The liver pigment was extracted from rat livers and characterized by various techniques, such as chemical degradation, HPLC, LC/MS/MS, 1H NMR, and MALDI-MS. Two degradation products had the same protonated molecular ion at m/z 505, which is 30 amu higher than that of LY368842. The major m/z 505 product has been identified as the indole-2,3-dione oxidative product based on 1H NMR data and confirmed by an authentic standard. In addition, monohydroxylated product was also identified in the degradation mixture. These degradation products were consistent with the metabolites found in vivo in rats. MALDI-MS analyses of liver and urinary pigment both identified a product with a molecular ion at m/z 977, suggesting formation of indirubin-like and indigo-like pigments. The results obtained suggest that the oxidative metabolites of LY368842 played a major role in the formation of the liver and urine pigments.

Acknowledgment. We thank Joe Tanner for his technical assistance in obtaining MALDI-MS data, Dr. Cynthia Jesudason and Craig Stevens for their technical assistance for chemical degradation of pigment, and Dr. Tianwei Ma for synthesis of authentic standard of indole2,3-dione metabolite. We are also grateful to Drs. Gail D. Williams, William J. Ehlhardt, and Andreas Kaerner for valuable discussions.

References (1) Hopp, M. L., Matsumoto, M., Wendell, B., Lee, C., and Oyasu, R. (1976) Suppressive role of indole on 2-acetylaminofluorene hepatotoxicity. Cancer Res. 36, 234-239. (2) Shertzer, H. G., and Sainsbury, M. (1991) Chemoprotective and liver enzyme induction properties of indole and indenoindole antioxidants in rats. Food Chem. Toxicol. 29, 391-400.

Characterization of Dark Liver Pigment from Rats (3) Lanza, D. L., and Yost, G. S. (2001) Selective dehydrogenation/ oxygenation of 3-methylindole by cytochrome P450 enzymes. Drug Metab. Dispos. 29, 950-953. (4) Skordos, K. W., Skiles, G. L., Laycock, J. D., and Yost, G. S. (1998) Evidence supporting the formation of 2,3-epoxy-3-methylindoline: a reactive intermediate of the pneumotoxin 3-methylindole. Chem. Res. Toxicol. 11, 741-749. (5) Thronton-Manning, J., Appleton, M. L., Gonzalez, F. J., and Yost, G. S (1996) Metabolism of 3-methylindole by vaccinia-expressed P450 enzymes: correlation of 3-methyleneindolenine formation and protein binding. J. Pharmacol. Exp. Ther. 276, 21-29. (6) Katchamart, S., Stresser, D. M., Dehal, S. S., Kupfer, D., and Williams, D. E. (2000) Concurrent flavin-containing monooxygenase down-regulation and cytochrome P450 induction by dietary indoles in rat: implications for drug-drug interaction. Drug Metab. Dispos. 28, 930-936. (7) Larsen-Su, S., and Williams, D. E. (1996) Dietary indole-3-carbinol inhibits FMO activity and the expression of flavin-containing monooxygenase form 1 in rat liver and intestine. Drug Metab. Dispos. 24, 927-931. (8) Russell, G. A., and Kaupp, G. (1969) Oxidation of carbanions. IV. Oxidation of indoxyl to indigo in basic solution. J. Am. Chem. Soc. 91, 3851-3859. (9) Gillam, E. M. J., Notley, L. M., Kim, D., Mundkowski, R. G., Aguinaldo, A. M., Volkov, A., Arnold, F. H., Scoucek, P., DeVoss, J., and Guengerich, F. P. (1999) Formation of indigo by recombinant mammalian cytochrome P450. Biochem. Biophys. Res. Commun. 265, 469-472.

Chem. Res. Toxicol., Vol. 16, No. 7, 2003 919 (10) Gillam, E. M. J., Notley, L. M., Cai, Hongliang, DeVoss, J., and Guengerich, F. P. (2000) Oxidation of indole by cytochrome P450 enzymes. Biochemistry 39, 13817-13824. (11) Prota, G., D’ischia, M., and Napolitano, A. (1998) The chemistry of melanin and related metabolites. The Pigmentary System, Chapter 24, pp 307-332, Oxford University Press, New York. (12) Saram, W. G., Gallagher, C. H., and Goodrich, B. S. (1969) Melanosis of sheep liver. I. Chemistry of the pigment. Aust. Vet. J. 45, 105-108. (13) [FDA] Food and Drug Administration (1982) Summary of basis of approval, pindolol (Visken), Letter to Sandoz Pharmaceuticals, NDA 18-285, pp 9-10. (14) Oikawa, A., and Nakayasu. M. (1973) Quantitative measurement of melanin as tyrosine equivalents and as weight of purified melanin. Yale J. Biol. Med. 46, 500-507. (15) Napolitano, A., Pezzella, A., Vincensi, M. R., and Prota, G. (1995) Oxidative degradation of melanins to pyrrole acids: a model study. Tetrahedron 51, 5913-5920. (16) Wu, K. J., and Odom, R. W. (1998) Characterizing synthetic polymers by MALDI MS. Analytical Chemistry News & Features, July, 456-461. (17) Jones, R. A. (1984) Pyrroles and their benzo derivatives: (ii) reactivity. In Comprehensive Heterocyclic Chemistry (Katritzky, A. R., and Rees, C. W., Eds.) Volume 4, pp 201-312, Pergamon Press Inc., New York.

TX0201016