An Integrated Metabonomic Investigation of Acetaminophen Toxicity in

Chem. Res. Toxicol. 2003, 16, 295-303

295

An Integrated Metabonomic Investigation of Acetaminophen Toxicity in the Mouse Using NMR Spectroscopy Muireann Coen,† Eva M. Lenz,‡ Jeremy K. Nicholson,† Ian D. Wilson,‡ Francois Pognan,§ and John C. Lindon*,† Biological Chemistry, Biomedical Sciences Division, Faculty of Medicine, Imperial College of Science, Technology and Medicine, Sir Alexander Fleming Building, South Kensington, London SW7 2AZ, United Kingdom, Drug Metabolism and Pharmacokinetics, AstraZeneca Pharmaceuticals, Mereside, Alderley Park, Macclesfield, Cheshire SK10 4TG, United Kingdom, and Safety Assessment, AstraZeneca Pharmaceuticals, Wilmington, Delaware 19850 Received September 3, 2002

An integrated metabonomics study using high-resolution 1H NMR spectroscopy has been applied to investigate the biochemical composition of intact liver tissue (using magic angle spinning), liver tissue extracts, and blood plasma samples obtained from control and acetaminophen-treated mice. Principal components analysis was used to visualize similarities and differences in biochemical profiles. The time- and dose-dependent biochemical effects of acetaminophen were related to the drug toxicity, as determined using histopathology. Metabolic effects in intact liver tissue and lipid soluble liver tissue extracts from animals treated with the high dose level of acetaminophen included an increase in lipid triglycerides and monounsaturated fatty acids together with a decrease in polyunsaturated fatty acids, indicating mitochondrial malfunction with concomitant compensatory increase of peroxisomal activity. In addition, a depletion of phospholipids was observed in treated liver tissue, which suggested an inhibition of enzymes involved in phospholipid synthesis. There was also a depletion in the levels of liver glucose and glycogen. In addition, the aqueous soluble liver tissue extracts from high dose animals also revealed an increase in lactate, alanine, and other amino acids, together with a decrease in glucose. Plasma spectra showed increases in glucose, acetate, pyruvate, and lactate. These observations all provide evidence for an increased rate of glycolysis. These findings could indicate a mitochondrial inability to use pyruvate in the citric acid cycle and also reveal the impairment of fatty acid β-oxidation in liver mitochondria of such treated mice.

Introduction High-resolution solution state 1H NMR spectra of biofluids (1-3) and 1H MAS1 NMR spectra of tissues (4, 5) can give information on drug-induced biochemical composition perturbations. When coupled with pattern recognition methods, a thorough analysis of the resulting complex multiparametric spectroscopic data sets can be obtained (6-8). This approach to the study of metabolic processes in biological systems has been termed metabonomics, and this has been defined as “the quantitative measurement of the multiparametric metabolic response of living systems to pathophysiological stimuli or genetic modification” (1, 2). The information obtained from * To whom correspondence should be addressed. Tel: +44(0)20 7594-3194. Fax: +44(0)20 7594-3066. E-mail: [email protected]. † Imperial College of Science, Technology and Medicine. ‡ Drug Metabolism and Pharmacokinetics, AstraZeneca Pharmaceuticals. § Safety Assessment, AstraZeneca Pharmaceuticals. 1 Abbreviations: ALT, alanine aminotransferase; AST, aspartate aminotransferase; CPMG, Carr-Purcell-Meiboom-Gill; DHA, docosahexaenoic acid; DPA, docosapentaenoic acid; EPA, eicosapentaenoic acid; GARP, globally optimized alternating phase rectangular pulse; GB, Gaussian broadening; HMQC, heteronuclear multiple quantum coherence; LB, line broadening; MAS, magic angle spinning; MUFA, monounsaturated fatty acids; PCA, principal components analysis; PUFA, polyunsaturated fatty acids; TOCSY, total correlation spectroscopy; TSP, trimethylsilyl propionic acid.

metabonomics is complementary to that from proteomics and genomics and is applicable to a wide range of problems in diverse biomedical research areas. As well as providing a detailed insight into the biochemical composition of a living organism and how it is affected by xenobiotics, the approach can also give information on the xenobiotic metabolites themselves. The biochemical composition of a biofluid or biological tissue will be modified if cell function varies; hence, metabonomics can inform on the state and severity of organ dysfunction. MAS NMR spectroscopy can be used both for analytical investigations and also for the study of dynamic and physicochemical properties of intact biological tissues. It is a nondestructive technique requiring little sample preparation and small amounts of sample (ca. 20 mg). It relies on the removal of line broadening caused by anisotropic interactions, and it is necessary to recognize that if the residual line broadening seen in the tissue is larger than the spin rate, then some metabolites may not be fully visible. It has recently been applied to investigate the toxin- and disease-induced changes in a variety of biological samples such as liver (4), renal cortex and medulla (5), adipocyte tissue (9), intact cells (10), prostatic tumor tissue (11), and brain (12). As a result of the complexity of 1H NMR spectra, subtle changes in metabolite resonances might be overlooked

10.1021/tx0256127 CCC: $25.00 © 2003 American Chemical Society Published on Web 01/23/2003

296

Chem. Res. Toxicol., Vol. 16, No. 3, 2003

within the natural biological variation; hence, pattern recognition techniques can be used to maximize information recovery from such complex spectra. Pattern recognition methods can be used to reduce the dimensionality or complexity of a data set and hence enable visualization of spectral changes. One widely used method is PCA (6) in which the NMR spectra are reduced to a set of peak intensity descriptors and analyzed to identify similarities and differences between cohorts of samples from control animals and those exhibiting toxicity and to identify novel combinations of metabolites changing as a result of the toxicity (biomarkers). Here, an integrated metabonomic approach (13) has been applied to the investigation of acetaminophen toxicity in the mouse using a range of biological samples. Thus, solution and solid state NMR spectroscopic methods were used to analyze intact tissue, tissue extracts, and blood plasma samples obtained from mice treated with a range of doses of acetaminophen. Acetaminophen is a widely used antipyretic, analgesic drug, which acts as a cyclooxygenase inhibitor. In many species, when taken in high doses, acetaminophen can result in centrilobular hepatic necrosis (14-16) with occasional observation of nephrotoxicity (17, 18).

Experimental Section Drug Administration and Samples. All animal studies were conducted under U.K. Home Office License according to appropriate national legislation. Seventy Alderley Park (AP-1) male mice were fasted overnight and remained fasted after an ip injection of acetaminophen in sterile 0.9% w/v sodium chloride solution. Fasting ensured that an accessible dose of acetaminophen caused observable histopathology lesions, noting that only differences between identically fasted control and dosed groups were examined. Ten mice were injected with vehicle (control, 0 mg/kg acetaminophen), 20 mice were injected with 50 mg/kg of acetaminophen (low dose), 20 mice were injected with 150 mg/kg of acetaminophen (mid dose), and 20 mice were injected with 500 mg/kg of acetaminophen (high dose). Cohorts of the animals were euthanized by CO2 inhalation at 15, 30, 60, 120, and 240 min after dosing. The livers were immediately removed, weighed, and cut into four parts irrespective of lobe distribution, and one of these parts was retained for 1H MAS NMR spectroscopy. All tissues were snap-frozen in liquid nitrogen and stored at -80 °C until required for analyses. Blood samples were withdrawn by cardiac puncture and placed in EDTA-coated tubes from which plasma samples were isolated for NMR analysis by centrifugation. ALT and AST were quantified using colorimetric kits (Sigma 505 and 505-P, respectively), and acetaminophen plasma concentrations were determined using a colorimetric kit based on nitrous acid reaction with acetaminophen (Sigma 430-A). All NMR spectra were then measured at 300 K. 1H MAS NMR Spectroscopy of Intact Liver Tissue. Each tissue sample was thawed, and a portion, weighing between 19 and 23 mg, was placed in a 4 mm MAS ZrO2 rotor with an insert to ensure a spherical cavity and spun in air at the magic angle (θ ) 54° 44’) at a spin rate of 5 kHz, using a Bruker DRX-600 spectrometer (Bruker Biospin, Rheinstetten, Germany). A small amount of TSP solution (50 µL,1 mg/mL TSP in D2O) was added as an internal chemical shift standard (δ ) 0) and to provide a field frequency lock. The CPMG pulse sequence with a total spin-spin relaxation delay (2nτ) of 40 ms was used to measure spin-echo 1H MAS NMR spectra at 600.13 MHz on all liver tissue samples. Typically, 128 scans were collected into 32K data points over a spectral width of 12 000 Hz with a relaxation delay of 5 s and an acquisition time of 2.13 s. A LB factor of 0.3 Hz was applied to all spectra prior to Fourier transformation (FT). 1H NMR Spectroscopy of Aqueous Soluble and Lipid Soluble Liver Tissue Extracts. Liver tissue was extracted

Coen et al. using 50% acetonitrile/50% D2O (1 mL). Preweighed tissue (100 mg) was homogenized in this solvent and then centrifuged at 10 000 rpm for 6 min. The supernatant was collected, lyophilized, and reconstituted in D2O (800 µL) prior to NMR analysis. A TSP solution (50 µL, 1 mg/mL TSP in D2O) was added as above. The pellet that remained from the previous aqueous extraction procedure underwent lipid extraction using 75% CHCl3/25% CH3OH (1 mL). The precipitated protein was spun down at 10 000 rpm for approximately 10 min. The solvent was removed by drying under a stream of nitrogen. This lipid soluble extract was reconstituted in 667 µL of deuterated chloroform/ methanol (CDCl3/CD3OD) (75%/25%) prior to NMR analysis. Solutions were pipetted into 5 mm NMR tubes, and spectra were acquired at a proton observation frequency of 600.13 MHz using a Bruker DRX-600 spectrometer. Aqueous extracts were analyzed using a standard pulse sequence (relaxation delay-90°t1-90°-tm-90°-acquisition). To eliminate the large water resonance, a secondary radio frequency irradiation was applied during the relaxation delay of 2 s and the mixing period (tm ) 100 ms). Typically, 64 transients were accumulated over a spectral width of 12 000 Hz into 32K data points with an acquisition time of 0.68 s. A LB factor of 0.3 Hz was applied to all spectra prior to FT. Lipid extract 1H NMR spectra were also obtained using a single 90° pulse experiment with no solvent suppression. Typically, 128 scans were collected into 64K data points over a spectral width of 12 000 Hz with a relaxation delay of 5 s and an acquisition time of 2.7 s. 1H-decoupled 31P NMR spectra were also acquired on the same instrument at 300 K using a standard pulse sequence. Typically, 1000 transients were collected into 1K data points over a spectral range of 2422.48 Hz (10 ppm). A relaxation delay of 4 s together with an acquisition time of 0.2 s was used. 1H decoupling was achieved using a GARP during acquisition (19). In addition, HMQC spectroscopy (20) was employed to aid assignment of phospholipid moieties and was used in conjunction with the TOCSY sequence (21) to yield multiple bond connectivities. For 1H-31P HMQC-TOCSY experiments (22), 64 transients per increment for 256 increments were collected into 1K data points. A spectral width of 6009 Hz in the F2 dimension (1H) and a spectral width of 2429.73 Hz in the F1 dimension (31P) were used with a relaxation delay of 2 s and an acquisition time of 0.85 s. A TOCSY spin lock time of 100 ms was used, the 31P-1H correlations were based on a J PH of 7 Hz, and GARP 31P decoupling was employed. These data were zero-filled by a factor of 2 and a sine-bell squared apodization function was applied to the FID, in both dimensions, prior to FT. 1H NMR Spectroscopy of Plasma Samples. A 100 µL sample of D2O and 50 µL of TSP solution (0.05 g/4 mL TSP in D2O) were added to 100 µL of plasma, and samples were placed in 5 mm NMR tubes. 1H NMR spectra of these samples were measured at 500.13 MHz 1H observation frequency using a Bruker DRX-500 spectrometer. The CPMG spin-echo pulse sequence with a fixed spin-spin relaxation delay (2nτ) of 96 ms was applied to measure 1H NMR spectra of all plasma samples. Typically, 128 scans were collected into 32K data points over a spectral width of 10 000 Hz with a relaxation delay of 3.4 s and an acquisition time of 1.64 s. The spectra were resolution-enhanced by the use of the Lorentzian-Gaussian transformation (with LB ) -1 and GB ) 0.3) prior to FT.

Data Reduction and PCA. Following NMR spectral phase and baseline correction, each NMR spectrum was data-reduced to 256 regions of equal width (0.04 ppm) using the AMIX (Analysis of MIXtures) software package, version 2.0 (Bruker Biospin). The spectral region containing the water resonance (δ 4.66-5.06) was removed from all data sets prior to normalization and multivariate data analysis in order to eliminate variation due to water suppression efficiency. For the plasma samples, the regions of the spectra containing resonances from free nonendogenous EDTA and EDTA metal complexes were removed (δ2.98-3.38, δ3.54-3.70). Because of broader solvent NMR resonances in the lipid soluble liver tissue extract spectra, which varied in chemical shift, a wider region of all of the

Metabonomics Investigation of Acetaminophen Toxicity

Figure 1. Acetaminophen as a function of time and dose in mouse plasma (n ) 2 for controls and n ) 4 for all treated groups). spectra was removed (δ3.02-4.82). Following a preliminary PCA analysis, the aromatic region (δ6-10) of all spectra was also removed due to the presence of acetaminophen metabolite resonances. All remaining frequency regions of the spectra were then scaled to the total integrated area of the spectra in order to reduce any significant concentration differences. Prior to PCA analysis, all NMR data variables were meancentered and pareto-scaled (23). Pareto scaling gives each variable a variance numerically equal to its standard deviation. PCA was used to reduce the dimensionality of the data sets. The first PC is a linear combination of the original input variables and describes the largest variation in the data set. The second PC is orthogonal to the first PC and describes the next highest variation in the data set. All PCs are orthogonal and describe progressively less of the data variation until PCs will describe purely noise in the data. Two-dimensional scores plots proved an efficient means of visualizing classification of the samples and investigating the regions of the spectra that were altered as a result of acetaminophen dosing. The corresponding loadings plots were used to identify which spectral variables contribute to the positioning of the samples on the scores plot and hence the variables that influence any observed separation in the data set.

Results Histopathology and Clinical Chemistry. Plasma acetaminophen concentrations reached maximal observed values at 15 min following administration of the 500 mg/kg of acetaminophen and at 30 min following administration of the 150 and 50 mg/kg doses, as depicted in Figure 1. ALT and AST levels at 500 mg/kg from 120 min onward were significantly different from those in controls (p < 0.01), but at 150 mg/kg after 240 min, ALT values were significantly increased (p < 0.05, data not shown). Histopathological changes were virtually identical in timing and range to those observed in previous studies (24, 25) and so are not reported here. No changes caused by the xenobiotic were identified at 50 mg/kg at a light microscopic level. All control liver samples were within normal limits at all time points. Control Liver Tissue, Tissue Extract, and Blood Plasma NMR Spectroscopy. A typical 1H MAS NMR spectrum of intact liver tissue from a control mouse measured using the CPMG spin-echo method is shown in Figure 2a. The main peaks arise from glucose, glycogen, and many lipidic moieties, typically predominantly triglycerides and phospholipids. The CPMG spin-echo experiment provided a degree of attenuation of the broad resonances from macromolecular components and im-

Chem. Res. Toxicol., Vol. 16, No. 3, 2003 297

proved visualization of the small molecule metabolites such that the spectra also revealed resonances corresponding to Ala, Gln, Glu, trimethylamine-N-oxide (TMAO), and betaine. To further characterize the biochemical composition of mouse liver, 1H NMR spectra were measured for both lipid and aqueous soluble extracts of the tissue and a representative 1H NMR spectrum of a control lipid soluble liver tissue extract is given in Figure 2b. In comparison to the tissue NMR spectra, the lipid soluble liver tissue extract spectra were less severely overlapped and this enabled the assignment of many additional metabolites such as cholesterol, phosphatidylethanolamine, sphingomyelin, and plasmalogen. It allowed a detailed assignment of unsaturated fatty acyl moieties from components such as phospholipids and triglycerides and enabled the quantitative proportions of individual lipid species to be calculated. By this means, it was possible to assign resonances due to both mono- and polyunsaturated fatty acyl chains and to investigate the effects of acetaminophen toxicity on such species. The predominant unsaturated fatty acid components present in liver tissue were determined to be oleic acid (18:1, n-9), linoleic acid (18:2, n-6), arachidonic acid (20: 4, n-6), and 4,7,10,13,16,19-DHA (22:6, n-3). These assignments were made on the basis of both reported literature results (26) and an analysis of both one- and two-dimensional NMR spectra, including the use of 31P NMR spectra. The molar proportions of MUFAs and PUFAs in control lipid soluble liver tissue extracts were found by NMR peak integration to be approximately 25 and 31%, respectively, the remainder being saturated moieties. The PUFA fraction was calculated to be comprised of approximately 6.5% linoleic acid, 6% arachidonic acid, 5% DHA, and 11.5% of a combination of contributions from DHA, EPA (20:5; 5,8,11,14,17-EPA), DPA (4,7,10,13,16-DPA, 22:5, n-6), and linolenic acid (18:3; 9,12,15-octadecatrienoic acid). These proportions of PUFAs in liver tissue were in agreement with those previously reported in the literature (26). 31 P NMR spectra were also acquired on lipid soluble liver tissue extracts. On the basis of literature values of 31P and 1H chemical shifts, these showed resonances from phosphatidylcholine, phosphatidylethanolamine, sphingomyelin, and lysophosphatidylcholine, with assignments based on a 1H-31P HMQC experiment with 1H-1H TOCSY magnetization transfer. The molar proportions of these phospholipids in the extracts were approximately 47% phosphatidylcholine, 31% phosphatidylethanolamine, 3% for the sum of the two types of lysophosphatidylcholine, 3% sphingomyelin, and 3% of an unidentified phosphorus-containing moiety, and these values were also found to be in agreement with those previously reported in the literature (27). A typical 1H NMR spectrum of a control aqueous soluble liver tissue extract acquired using a standard water resonance presaturation pulse sequence is given in Figure 2c, and such spectra showed only low molecular weight metabolite NMR profiles. There were 1H NMR resonances identifiable from many amino acids and related compounds, soluble membrane components such as choline and phosphocholine, and lactate and glucose. Finally, 1H CPMG NMR spectra were measured on blood plasma samples and these showed resonances from low molecular weight metabolites, with contributions from high molecular weight species such as lipoproteins,

298


Coen et al.

Figure 2. (a) 1H MAS NMR CPMG spectrum (600 MHz) of intact control liver tissue; (b) 1H NMR (600 MHz) spectrum of a control lipid soluble liver tissue extract; (c) solvent presaturation 1H NMR spectrum (600 MHz) of a control aqueous soluble liver tissue extract; (d) 1H NMR CPMG spectrum (500 MHz) of control blood plasma. Key: 3HB, 3-D-hydroxybutyrate; Cho, choline; Chol, cholesterol; Glu, glucose; GPC, glycerophosphorylcholine; Gly, glycerol; LDL, low-density lipoprotein; PCho, phosphocholine; TMAO, trimethylamine-N-oxide; VLDL, very low-density lipoprotein.

partially attenuated through the spin-echo spectral editing. A representative control mouse blood plasma spectrum is given in Figure 2d. On the basis of literature values (28), many resonances were assigned in the spectra including those from various lipoproteins, glucose resonances from both the R- and the β-anomers, and further low molecular weight metabolites such as 3-Dhydroxybutyrate, acetate, Ala, Gln, Glu, pyruvate, and lactate. There were also a number of resonances assigned in the aromatic region of the spectra that corresponded to His, Phe, and Tyr. Observation of Acetaminophen Metabolites. The major metabolites of acetaminophen in mouse urine

collected 4 h after oral treatment with 150 mg/kg [14C]acetaminophen have been shown to be the sulfate and glucuronide conjugates and the cysteine and mercapturate adducts representing 14, 48, 19, and 2%, respectively, of total radioactivity (29). It was possible to observe 1H NMR peaks from the major metabolites of acetaminophen, namely, the sulfate, glucuronide, and N-acetyl-L-cysteine adduct in spectra of plasma samples from animals treated with 150 and 500 mg/kg acetaminophen and euthanized at the later time points (120 and 240 min). Peaks from the glucuronide conjugate were also detected in the liver tissue itself from high dose animals sampled at



Figure 3. (a) 1H MAS NMR CPMG spin-echo spectrum (600 MHz) of intact liver tissue; (b) standard 1H NMR spectrum (600 MHz) of a lipid soluble liver tissue extract; (c) solvent-suppressed 1H NMR spectrum (600 MHz) of an aqueous soluble liver tissue extract; (d) 1H NMR CPMG spectrum (500 MHz) of blood plasma. All spectra are from animals treated with the high dose of acetaminophen (500 mg/kg) and euthanized at 240 min. Key: as in Figure 1.

time points of 60, 120, and 240 min. The aromatic ring proton resonances with characteristic multiplet resonances at chemical shifts of δ7.15 and δ7.37 were clearly visible in the 1H MAS NMR spectra. The acetaminophen-glucuronide metabolites were also identified in aqueous tissue extract NMR spectra from the high dose group of animals sampled at 60, 120, and 240 min postdose. Dose-Dependent Biochemical Alterations after Treatment with Acetaminophen. A number of biochemical alterations were identified for the high dose level of acetaminophen, and these were also observed for the mid dose animals at the later time points although

the scale of the changes was less marked. In contrast, few changes were observed in the low dose group at any time point. Following treatment with acetaminophen, a marked increase was seen in the peaks from the fatty acyl groups of lipid moieties and the glycerol backbone resonances of triglycerides in liver tissue samples. There was also a concomitant decrease in the glucose and glycogen levels together with a marked decrease in the concentration of Ala. A representative 1H MAS CPMG NMR spectrum of intact liver tissue from an animal treated with the high dose of acetaminophen (500 mg/kg) and sampled at a time point of 240 min is shown in Figure 3a.

300


The 1H NMR spectrum of a lipid soluble liver tissue extract after treatment with the high dose of acetaminophen is given in Figure 3b. The main biochemical changes apparent were an overall increase in all fatty acyl chain resonances, an increase in the glyceryl resonances assignable to triglycerides, and a decrease in phospholipids, the latter based on the change in the -N+Me3 resonance of choline headgroups. Further evidence for a depletion of phospholipids in samples from treated mice has been obtained from 31P NMR spectroscopy of lipid soluble liver tissue extracts. This revealed a significant depletion of phospholipid levels in all treated samples in comparison to controls. The 1H NMR spectra revealed an approximate 3-fold elevation in the unsaturated fatty acid fraction, which was calculated to be composed of 25% MUFA and 6% PUFA demonstrating that the percentage composition of MUFAs was unchanged in samples from dosed animals, but the proportion of PUFAs was significantly depleted. The PUFAs in the samples from dosed animals were calculated to be comprised of approximately 6% linoleic acid and 3% arachidonic acid. The levels of the fatty acyl chain methyl resonances corresponding to DHA, DPA, EPA, and linolenic acid were below the detection limit, indicating a marked depletion in all species in treated tissue. A representative spectrum of an aqueous soluble tissue extract from an animal treated with the high dose level of acetaminophen and sampled at a time point of 240 min is given in Figure 3c. The dominant biochemical changes visible were a significant decrease in the glucose resonances, an increase in the choline and phosphocholine resonances, and a marked increase in the resonances corresponding to lactate, Ala, Ile, Leu, Lys, and Val. The resonances in the aromatic region of the spectra (not shown) corresponding to Phe and Tyr were also increased following treatment with the mid and high dose level of acetaminophen in comparison to controls. Formate was also decreased. Finally, a representative spectrum of a blood plasma sample from an animal treated with the 500 mg/kg dose level of acetaminophen 240 min postdose is given in Figure 3d. The major biochemical changes identified in plasma were an increase in the levels of 3-D-hydroxybutyrate, lactate, acetate, pyruvate, glucose, and lipid moieties such as the fatty acyl chains of triglycerides. Although inspection of NMR spectra and integration of individual peaks can give valuable information on biochemical changes, it is difficult to visualize general effects as a function of both dose and time in a large cohort of animals with biological variability. Here, simple pattern recognition methods have been used to map the NMR spectra into a low dimensional metabolic space such that any clustering of the samples based on similarities of biochemical profiles can easily be determined and the biochemical basis for this elucidated. The dose dependence of acetaminophen toxicity was investigated by applying PCA to analyze a data set containing NMR spectra of liver tissue from control and all available treated animals (dose levels of 50, 150, and 500 mg/kg) at the 240 min time point (remembering that some animals were euthanized at earlier time points). The scores plot (Figure 4a) revealed separation between control and treated animals in PC1. The control samples bracket those from the low dose group of animals indicating that major biochemical changes had not occurred following administration of this dose level of acetami-

Coen et al.

Figure 4. Dose dependence of acetaminophen toxicity. (a) Scores plot (PC1/PC2) based on 1H MAS NMR CPMG spectra of intact liver tissue from control and all dose level samples (50, 150, and 500 mg/kg) euthanized at 240 min. (b) Scores plot (PC1/ PC2) based on 1H NMR spectra of control and all dose level lipid soluble liver tissue extract samples (50, 150, and 500 mg/kg) euthanized at 240 min. Key: O, control group (0 mg/kg); 2, low dose group (50 mg/kg); b, mid dose group (150 mg/kg); 9, high dose group (500 mg/kg).

nophen. In comparison, the NMR spectra of the mid and high dose samples were more clearly separated from the control samples in PC1 indicating that biochemical alterations had occurred. An outlying sample in the 150 mg/kg dose group illustrated the presence of interanimal variation. PCA was also applied to a lipid soluble extract data set to investigate the dose dependence of acetaminophen toxicity, and a scores plot revealed clear separation between the high dose sample group and all other samples as shown in Figure 4b. There is one outlying sample from the low dose group seen in the scores plot, and on inspection of the relevant spectrum, it was noted to have a particularly high-intensity resonance corresponding to fatty acyl chain methylene groups. PCA analysis was also applied to investigate metabolic profile alterations in the spectral data of aqueous tissue extracts. On investigation of a data set containing all


control and dosed samples acquired at a time point of 240 min, a scores plot (not shown) revealed clear separation between the high dose level samples (500 mg/kg) and all other samples. Application of PCA to analyze a data set of spectra from blood plasma containing all control and treated animals (dose levels of 50, 150, and 500 mg/kg) at 240 min postdose resulted in a scores plot (not shown) that revealed separation between the high dose group and all other groups in PC2. Time Dependence of Acetaminophen Toxicity. It was apparent from examination of the NMR spectra that major biochemical alterations had occurred following exposure to the high dose of acetaminophen at 60, 120, and 240 min post administration. The scores plot based on NMR spectra of liver tissue spectra from animals treated with the high dose of acetaminophen over the entire time course of the study is shown in Figure 5a. There was clear separation between the control and the high dose group animals at the later time points in PC1. The earliest time point group was largely separated from the control group in PC2. The successive time point groups of 30 and 60 min were overlapped and also separated from the control group in PC2. It can be seen that two of the samples from the 30 min time point group lie beyond their expected position, suggesting that these particular animals were fast responders to acetaminophen. The loadings plot (not shown) revealed the spectral regions, which were responsible for the separation seen in the corresponding scores plot and hence the regions that showed time-dependent alterations in the levels of endogenous metabolites. These spectral regions corresponded to resonances from the fatty acid moieties, lipid triglycerides, glucose, glycogen, choline, and Ala. Visual reexamination of the relevant NMR spectra confirmed that there was a significant decrease in the glucose and glycogen resonances, an increase in the lipid triglyceride resonances, and a less pronounced decrease in the choline and Ala resonances. PCA of all control and high dose level lipid soluble extract samples over all time points resulted in a scores plot (Figure 5b) that revealed very clear separation in PC1 between the dosed samples at the later time points and all other samples. The corresponding loadings plot (not shown) revealed that the separation seen in PC1 of the scores plot for samples treated with the highest dose of acetaminophen at the latest time point was a result of large increases in the resonance intensities for the group of triglyceride peaks, namely, those arising from the methylene and methyl fatty acyl chain resonances, the fatty acyl chain methylene groups adjacent to the carbonyl group, and a resonance from the monounsaturated fatty acyl chain of oleic acid (18:1). The loadings plot also indicated the significance of a decrease in the olefinic proton resonances from polyunsaturated fatty acyl chains. In particular, there was a decrease in the PUFA resonances of linoleic acid, arachidonic acid, and DHA. Analysis of a data set of aqueous soluble tissue extracts containing all control samples and samples treated with the high dose of acetaminophen acquired at all time points revealed very clear separation in PC1 between the treated animals euthanized at the later time points of 120 and 240 min and all other control and treated samples (not shown). The metabolites responsible for the separation seen in the time-dependent scores plots were


Figure 5. Time dependence of acetaminophen toxicity. (a) Scores plot (PC1/PC2) based on 1H MAS NMR CPMG spectra of intact liver tissue from all control and high dose group animals (500 mg/kg acetaminophen) at all time points (15, 30, 60, 120, and 240 min). (b) Scores plot (PC1/PC2) based on 1H NMR spectra of lipid soluble liver tissue extracts from all control and high dose group animals (500 mg/kg acetaminophen) at all time points (15, 30, 60, 120, and 240 min). Key: O, control group (0 mg/kg); treated groups, sampled at various times: 2, 15 min; [, 30 min; 9, 60 min; b, 120 min; +, 240 min.

identified from the loadings plots (not shown) as principally those of glucose, choline, phosphocholine, lactate, Ala, Ile, Leu, and Val. PCA analysis was also applied to a data set of all control and high dose (500 mg/kg) plasma samples sampled at all time points, and this revealed little separation between classes with the exception of the samples obtained at the 240 min time point. The loadings plot revealed that the resonances contributing to this separation were those of the lipid moieties, glucose, lactate, acetate, and pyruvate.

Discussion Significant changes in liver pathology were observed in dose- and time-dependent manners as recorded previously (24, 25), consistent with the observed exposure

302


indicated by the plasma acetaminophen concentrations given in Figure 1. The elevated triglyceride levels evident in spectra of intact liver tissue, chloroform/methanol liver tissue extracts, and blood plasma samples relate to a shift in energy metabolism as a consequence of acetaminophen hepatotoxicity. In addition, the chloroform/methanol liver tissue extracts gave detailed biochemical information on the specific lipidic species that were altered. The level of MUFAs in treated tissue was increased consistent with mitochondrial oxidative phosphorylation impairment, whereas the level of PUFAs was decreased consistent with β-oxidation activity of peroxisomes being increased to counteract depleted energy levels. In addition, histopathological data revealed major impairment of mitochondria (megamitochondria), which would result in such a depletion of energy levels (30) and an accumulation of lipid triglycerides in the liver. 31 P NMR spectroscopy of the lipid soluble tissue extracts indicated depletion of all phospholipid species, in all samples treated with the mid and high dose level of acetaminophen. This novel observation suggests either an increase in the activity of hepatic phospholipase or an inhibition of enzymes involved in phospholipid synthesis. In addition, the depletion of phospholipid species suggests that acetaminophen-induced free radical damage or lipid peroxidation had occurred. Oxidative stress followed by (or inducing) lipid peroxidation has previously been reported in relation to acetaminophen toxicity (31). The increase in phospholipid degradation products, choline and phosphocholine, in the aqueous soluble liver tissue extracts lends further support to the depletion of phospholipids identified in the intact liver tissue. The resonances of arachidonic acid (20:4) were significantly depleted in the lipid soluble tissue extract NMR spectra from animals treated with the mid and high dose level of acetaminophen and euthanized at the later time points of 120 and 240 min. Arachidonic acid, stored in cell membranes in the esterified form at the glycerol C2 of phospholipids, is generated by phospholipid hydrolysis and is also synthesized by elongation and desaturation of dietary linoleic acid. Hence, the significant depletion of phospholipids following treatment with acetaminophen may be a result of the inhibition of the phospholipase-A2 pathway. This is consistent with the genomic study previously carried out (24) that showed that uteroglobin, a phospholipid-binding protein that binds phosphatidylcholine and phosphatidylinositol and is a potent inhibitor of phospholipase-A2, was up-regulated. Significant decreases in the levels of hepatic glucose and glycogen were also observed in animals treated with the high dose of acetaminophen and euthanized at the later time points, and this suggested that the rate of glycogenolysis and glycolysis had increased in these animals. There was a correlated increase in lactate concentrations evident in the corresponding aqueous soluble liver tissue extract and plasma spectra, which offers further support for increased rates of glycogenolysis and glycolysis. In addition, the enhanced levels of glucose and lactate in the plasma from acetaminophen-treated animals are consistent with mitochondrial impairment, which can lead to an inability to use pyruvate in the citric acid cycle. This proposed inability may result from a lack of acetylCoA in the mitochondria as a result of a reduction in the

Coen et al.

rate of oxidative phosphorylation and fatty acid β-oxidation. As the oxidative phosphorylation pathway is an ATP-generating process, this observation supports the hypothesis that the cells switched to glycolysis to compensate for the loss of ATP. A previous study with acetaminophen demonstrated severe mitochondrial changes leading to generation of megamitochondria (24) that are supposedly ATP-depleted and nonfunctional (30). Ala can readily undergo transamination in the cytosol to form pyruvate, and as there was a significant increase in Ala levels in the blood plasma and aqueous soluble liver tissue extracts, this suggests that this pathway was suppressed. The aqueous soluble tissue extracts revealed elevated quantities of many amino acids, in particular those of Leu and Val, following the high dose of acetaminophen. This is indicative that hepatic amino acid transamination reactions were disturbed consistent with an increased amino acid production from pyruvate as a result of the impairment of the citric acid cycle.

Conclusion The application of 1H NMR spectroscopy to an array of biological samples comprising liver tissue, plasma, and tissue extracts enabled the assignment of many endogenous and drug-related metabolites. The spectra revealed both time- and dose-dependent changes in endogenous metabolite levels, which could be related to the biochemical pathways perturbed by the acetaminophen administration. In conclusion, this integrated metabonomics approach for the investigation of acetaminophen toxicity demonstrates the combined use of conventional solution state NMR spectroscopy and MAS NMR spectroscopy for the investigation of biofluids, intact tissues, and extracts together with pattern recognition techniques in probing the biochemical response to a xenobiotic.

Acknowledgment. The Royal Society of Chemistry (Analytical Chemistry Trust Fund), Syngenta, and AstraZeneca Pharmaceuticals are acknowledged for financial support to M.C.

References (1) Nicholson, J. K., Connelly, J., Lindon, J. C., and Holmes, E. (2002) Metabonomics: A Platform For Studying Drug Toxicity And Gene Function. Nat Rev. Drug Discovery 1, 153-161. (2) Nicholson, J. K., Lindon, J. C., and Holmes, E. (1999) Metabonomics: understanding the metabolic responses of living systems to pathophysiological stimuli via multivariate statistical analysis of biological NMR spectroscopic data. Xenobiotica 29, 1181-1189. (3) Lindon, J. C., Nicholson, J. K., Holmes, E., and Everett, J. R. (2000) Metabonomics: metabolic processes studied by NMR spectroscopy of biofluids. Concepts Magn. Reson. 12, 289-320. (4) Bollard, M. E., Garrod, S., Holmes, E., Lindon, J. C., Humpfer, E., Spraul, M., and Nicholson, J. K. (2000) High-resolution 1H and 1H-13C magic angle spinning NMR spectroscopy of rat liver. Magn. Reson. Med. 44, 201-207. (5) Garrod, S., Humpfer, E., Spraul, M., Connor, S. C., Polley, S., Connelly, J., Lindon, J. C., Nicholson, J. K., and Holmes, E. (1999) High resolution 1H NMR and magic angle spinning NMR spectroscopic investigation of the biochemical effects of 2-bromoethanamine in intact renal and hepatic tissue. Magn. Reson. Med. 41, 1108-1118. (6) Lindon, J. C., Holmes, E., and Nicholson, J. K. (2001) Pattern recognition methods and applications in biomedical magnetic resonance. Prog. NMR Spectrosc. 39, 1-40. (7) Holmes, E., Nicholls, A. W., Lindon, J. C., Connor, S. C., Connelly, J., Haselden, J. N., Damment, S. J. P., Spraul, M., Neidig, P., and Nicholson, J. K. (2000) Chemometric models for toxicity


(8)

(9) (10)

(11)

(12)

(13)

(14) (15) (16)

(17) (18) (19)

classification based on NMR spectra of biofluids. Chem. Res. Toxicol. 13, 471-478. Beckwith-Hall, B. M., Nicholson, J. K., Nicholls, A. W., Foxall, P. J. D., Lindon, J. C., Connor, S. C., Abdi, M., Connelly, J., and Holmes, E. (1998) Nuclear magnetic resonance spectroscopic and principal components analysis investigations into biochemical effects of three model hepatotoxins. Chem. Res. Toxicol. 11, 260272. Millis, K., Mass, W. E., Cory, D., and Singer, S. (1997) Gradient high-resolution magic angle spinning NMR spectroscopy of human adipocyte tissue. Magn. Reson. Med. 38, 399-403. Weybright, P., Millis, K., Campbell, N., Cory, D. G., and Singer, S. (1998) Gradient high-resolution, magic angle spinning 1H nuclear magnetic resonance spectroscopy of intact cells. Magn. Reson. Med. 39, 337-344. Tomlins, A., Foxall, P. J. D., Lindon, J. C., Lynch, M. J., Spraul, M., Everett, J. R., and Nicholson, J. K. (1998) High-resolution magic angle spinning 1H NMR analysis of intact prostatic hyperplastic and tumour tissues. Anal. Commun. 35, 113-115. Cheng, L. L., Ma, M. J., Becerra, L., Ptak, T., Lackner, A., and Gonzalez, R. R. (1997) Quantitative neuropathology by highresolution magic angle spinning proton magnetic resonance spectroscopy. Proc. Natl. Acad. Sci. U.S.A. 94, 6408-6413. Waters, N. J., Holmes, E., Williams, A., Waterfield, C., Farrant, R. D., Nicholson, J. K. (2001) NMR and pattern recognition studies on the time-related metabolic effects of R-naphthylisothiocyanate on liver, urine and plasma in the rat: an integrative metabonomic approach. Chem. Res. Toxicol. 14, 1401-1412. Boyd, E. M., and Bereczky, G. M. (1966) Liver necrosis from paracetamol. Br. J. Pharmacol. 26, 606-614. Davidson, D. G., and Eastham, W. N. (1966) Acute liver necrosis following overdose of paracetamol. Br. Med. J. 5512, 497-499. Davis, M., Simmons, C. J., Garrison, N. G., and Williams, R. (1976) Paracetamol overdose in man: relationship between pattern of urinary metabolites and severity of liver damage. Q. J. Med. 178, 181-191. Kleinman, J. G., Breitenfield, R. V., and Roth, D. A., (1980) Acute renal failure associated with acetaminophen ingestion: report of a case and review of the literature. Clin. Nephrol. 14, 201-205. McMurtry, R. J., Snodgrass, W. R., and Mitchell, J. R. (1978) Renal necrosis, glutathione depletion and covalent binding after acetaminophen. Toxicol. Appl. Pharmacol. 46, 87-100. Shaka, A. J., Barker, P. B., and Freeman, R. (1985) Computeroptimized decoupling scheme for wideband applications and lowlevel operation. J. Magn. Reson. 64, 547-552.

Chem. Res. Toxicol., Vol. 16, No. 3, 2003 303 (20) Muller, L. (1979) Sensitivity enhanced detection of weak nuclei using heteronuclear multiple quantum coherence. J. Am. Chem. Soc. 101, 4481-4484. (21) Bax, A., and Davis, D. G. (1985) MLEV-17 based two-dimensional homonuclear magnetisation transfer spectroscopy. J. Magn. Reson. 65, 355-360. (22) Larijani, B., Poccia, D. L., and Dickinson, L. C. (2000) Phospholipid identification and quantification of membrane vesicle subfractions by 31P-1H two-dimensional nuclear magnetic resonance. Lipids 35, 1289-1297. (23) Wold, S., Antti, H., Lindgren, F., and Ohman, J. (1998) Orthogonal signal correction of near-infrared spectra. Chemom. Intell. Lab. Syst. 44, 175-185. (24) Ruepp, S. U., Tonge, R. P., Shaw, J., Wallis, N., and Pognan, F. (2002) Genomics and proteomics analysis of acetaminophen toxicity in mouse liver. Toxicol. Sci. 65, 135-150. (25) Tonge, R., Shaw, J., Middleton, B., Rowlinson, R., Rayner, S., Young, J., Pognan, F., Hawkins, E., Currie, I., and Davison, M. (2001) Validation and development of fluorescence two-dimensional differential gel electrophoresis proteomics technology. Proteomics 1, 377-396. (26) Casu, M., Anderson, G. J., Choi, G., and Gibbons, W. A. (1991) NMR lipid-profiles of cells, tissues and body fluids. Magn. Reson. Chem. 29, 594-602. (27) Raffelt, K., Moka, D., Suellentrop, F., Dietlein, M., Hahn, J., and Schicha, H. (2000) Systemic alterations in phospholipid concentrations of blood plasma in patients with thyroid carcinoma: an in vitro 31P high-resolution NMR study. NMR Biomed. 13, 8-13. (28) Nicholson, J. K., Foxall, P. J. D., Spraul, M., Farrant, R. D., and Lindon, J. C. (1995) 750 MHz 1H and 1H-13C NMR spectroscopy of human blood plasma. Anal. Chem. 67, 793-811. (29) Wong, L. T., Whitehouse, L. W., Solomonraj, G., and Paul, C. J. (1981) Pathways of disposition of acetaminophen conjugates in the mouse. Toxicol. Lett. 9, 145-151. (30) Karbowski, M., Kurono, C., Wozniak, M., Ostrowski, M., Teranishi, M., Nishizawa, Y., Usukura, J., Soji, T., and Wakabayashi, T, (1999) Free radical-induced megamitochondria formation and apoptosis. Free Radical Biol. Med. 26, 396-409. (31) Amimoto, T., Matsura, T., Koyama, S.-Y., Nakanishi, T., Yamada, K., and Kajiyama, G. (1995) Acetaminophen-induced hepatic injury in mice: The role of lipid peroxidation and effects of pretreatment with coenzyme Q10 and R-tocopherol. Free Radical Biol. Med. 19, 169-176.

TX0256127

An Integrated Metabonomic Investigation of Acetaminophen Toxicity in

Recommend Documents