NMR-Based Metabonomic Studies on the Biochemical Effects of

The scores plots generated from Pareto-scaled data and mean-centered ..... associated with the progression and recovery from nephrotoxic lesions in th...
0 downloads 0 Views 145KB Size
Download PDF
1136

Chem. Res. Toxicol. 2002, 15, 1136-1141

NMR-Based Metabonomic Studies on the Biochemical Effects of Commonly Used Drug Carrier Vehicles in the Rat Bridgette M. Beckwith-Hall,*,† Elaine Holmes,† John C. Lindon,† John Gounarides,‡ Alison Vickers,‡ Michael Shapiro,‡ and Jeremy K. Nicholson† Biological Chemistry, Biomedical Sciences Division, Imperial College of Science, Technology and Medicine, Sir Alexander Fleming Building, South Kensington, London SW7 2AZ, U.K., Novartis Pharmaceuticals Corporation, Core Technologies, 556 Morris Avenue, Summit, New Jersey 07901-1398 Received March 11, 2002

The biochemical effects of a series of commonly used drug carrier vehicles were investigated using 1H NMR spectroscopic and pattern recognition based metabonomic analysis. Animals were treated by oral gavage with six dosage vehicles: 0.5% (w/v) sodium carboxymethylcellulose/ 0.2% (v/v)tween; microemulsion (consisting of propylene glycol, ethanol, cremophor, and corn oil glycerides); labrafil [consisting of poly(ethylene glycol) 300 esterified with oleic acid] (30%)/ corn oil (70%); 0.1 M sodium phosphate buffered water; poly(ethylene glycol) 300 and 0.5% methocel. Urine samples (n ) 7) collected over a 96 h period post administration were analyzed using 600 MHz 1H NMR spectroscopy, and principal components analysis of the spectral data was used to analyze these data. Of the six vehicles studied, three (labrafil/corn oil, PEG 300 and microemulsion) gave rise to strong vehicle-related signals in the 1H NMR spectra of urine and were, therefore, deemed to be less suitable for NMR-based toxicity studies. To investigate any biochemical consequences of vehicle dosing, PCA was used to analyze spectral regions that did not contain vehicle-related signals, i.e., the NMR-detectable endogenous metabolite profile. PEG 300 and labrafil/corn oil induced changes in the biochemical composition of urine including increased concentrations of dicarboxylic acids, creatinine, taurine, and sugars, indicating that these vehicles were bioactive in their own right and that this might confound interpretation of biochemical effects of weakly toxic drugs dosed in these carriers. This study shows the importance of selecting appropriate vehicles for NMR-based metabonomic studies with a view to minimizing the possibility of vehicle resonances obscuring endogenous compound peaks. Furthermore, we have shown that at least two of the commonly used drug carrier vehicles caused metabolic perturbations in the urine profile. These alterations in the biochemical profile reflect vehicle-induced changes in the physiological status of the organism that may obscure the pharmacologic or toxicologic effects of drugs.

Introduction 1

H NMR spectroscopy of biofluids results in the generation of endogenous metabolite profiles that alter characteristically in response to changes in physiological status, toxic insult, or disease processes (1-4). Moreover, signals relating to xenobiotics such as drugs or drug carrier vehicles can also be readily detected by this technique and can act as interferences to the measurement of endogenous metabolites as a biomarker generation tool (1, 5, 6). The usefulness of 1H NMR spectroscopy can be increased by the application of multivariate statistical analysis including pattern recognition (PR)1 methods that allow sample classification and novel biomarker combinations to be identified (2, 7-11). The * To whom correspondence should be addressed. † Imperial College of Science, Technology and Medicine. ‡ Novartis Pharmaceuticals Corporation. 1 Abbreviations: CO, corn oil; CMC, carboxymethylcellulose; DCV, drug carrier vehicle; FID, free induction decay; NMR, nuclear magnetic resonance; PR, pattern recognition; PEG, poly(ethylene glycol); PCA, principal components analysis; TSP, sodium 3-trimethylsilyl [2,2,3,32H ] propionate. 4

measurement of 1H NMR spectroscopy of such biofluids offers an efficient and nondestructive means of investigating variation in animal populations as a wide range of biochemical components can be monitored directly (2). This has led to the concept of metabonomics defined as “the quantitative measurement of the dynamic multiparametric metabolic response of living systems to pathophysiological stimuli or genetic modification” (10). An important requirement in animal-based toxicology is the minimization of variation in control animals relative to the changes induced by the putative toxin. However, various physiological factors can affect the metabolic composition of biological matrixes such as urine, and these include diet, age, state of health, diurnal cycles, stress, genetic drift, and strain differences, and thus it is necessary to characterize these differences in order to distinguish between physiological and pathophysiological responses to a xenobiotic in animal models (12-15). Furthermore, in NMR-based metabolic studies, it is preferable that the vehicle itself should give negligible contribution to NMR spectra of the biofluids under study, and that the compounds present in the vehicle

10.1021/tx020020+ CCC: $22.00 © 2002 American Chemical Society Published on Web 08/09/2002

Metabolic Effects of Drug Carrier Vehicles

Chem. Res. Toxicol., Vol. 15, No. 9, 2002 1137

Table 1. Drug Carrier Vehicles Investigated and Their Identity Codes, Corresponding to Figures 2 and 3

a

drug carrier vehicle (DCV)a

abbreviation

identifier on PC plots

class

0.1 M sodium phosphate buffered water 0.5% (w/v) sodium carboxymethylcellulose/0.2% (v/v) tween (comprising polyoxyethylene sorbitan monooleate) 0.5% methocel (comprising hydroxypropyl methylcellulose) microemulsion (composition:propylene glycol USP, ethanol USP, cremophor,b corn oil glycerides) labrafil (comprising PEG esterified with oleic acid) (30%)/ corn oil (70%) poly(ethylene glycol) 300

buffered water CMC/tween

O b

1 1

methocel microemulsion

4 2

1 4

labrafil/CO

0

4

PEG 300

9

4

DCVs obtained from Sigma-Aldrich, St. Louis, MO. b Cremophor: poly(ethylene glycol) mono(hexadecyl/octadecyl) ether.

should minimally affect in vivo metabolic pathways in order to avoid misleading metabolic conclusions in terms of toxicological significance. In previous 1H NMR spectroscopic studies of toxicity, drug carrier vehicles that were predominantly physiologically inert were used such as 0.9% sodium chloride together with others such as corn oil, which is presumed to be inert in terms of pathway balance (11). However, because of compound class variations in the physicochemical properties in novel therapeutic agents, it is often necessary to use a range of drug carrier vehicles so that solubility or suspension of the drug is possible, a desired route of administration can be utilized, and pharmacokinetic information can be obtained. The present study was designed to investigate the suitability of drug carrier vehicles for NMR-PR studies and to evaluate any metabolic perturbations resulting from the administration of commonly used drug carrier vehicles. This work is of direct relevance to the study design of toxicological studies in which metabonomic methods are currently being employed in the pharmaceutical industry.

Materials and Methods Animal Experiments. Control urine samples were collected from male HsdBrl:WH rats (n ) 40) ca. 10 weeks old and grouped according to weight (approximately 160-200 g). All animals were acclimatized in grid-based plastic cages prior to group allocation and were placed in metabolism cages for a 6-day period. Animals were dosed daily (n ) 5 per group) with one of six drug carrier vehicles (DCVs) as listed in Table 1. All vehicles were dosed p.o. at a volume of 10 mL/kg. Urine samples were collected from each animal 24 h prior to dosing and at 0-8 h, 8-24 h, 24-32 h, 32-48 h, 48-72 h, and 72-96 h after the initial dose, 0 h being defined as the initiation of daily dosing. Animals were allowed free access to food (Certified Rodent Diet no. 5002, PMI Feeds, Richmond, IN) and water throughout the study. The inside of the metabolism cage funnel was washed down daily and rinsed with distilled water to remove any contaminating hair, feed, feces, or residual urine from the separation cone in order to minimize the possibility of urinary bacterial contamination due to retention within the cage system. Buffering and Preparation of Urine Samples for NMR Analysis. Samples were made up from 600 µL of urine mixed with 300 µL of 100 mM sodium phosphate buffer (pH 7.4 made up in D2O) and 90 µL of sodium 3-trimethylsilyl [2,2,3,3-2H4] propionate (TSP) solution in D2O. The buffer solution was used in order to narrow the pH range of the samples, the TSP was added to act as an internal chemical shift reference (δ1H 0.0) and the D2O provided an NMR lock signal for the NMR spectrometer. The solution was left to stand for 10 min, and then the buffered urine was centrifuged at 10 000 rpm for 10 min to remove precipitates. A 700 µL aliquot of the resulting supernatant was then placed into vials immediately prior to NMR spectroscopic analysis.

600 MHz 1H NMR Spectroscopic Analysis of Rat Urine. NMR measurements were made on a Bruker AMX600 spectrometer operating at 600.13 MHz 1H frequency at ambient probe temperature (298 ( 1 K) using the BEST (Bruker Efficient Sample Transfer; Bruker Analytische GmbH, Rheinstetten, Germany) flow probe. For each sample, 32 free induction decays (FIDs) were collected into 65 536 data points using a spectral width of 12 019 Hz and an acquisition time of 2.72 s. An additional relaxation delay (RD) of 2.0 s was added between pulses to allow T1 relaxation. 1H NMR spectra were acquired with suppression of the water resonance using a standard pulse presaturation sequence comprising (RD-90°-t1-90 o-tm-90 o-acquire FID) where t1 is a short delay of typically 3 µs and tm is a mixing time (150 ms). Irradiation of the water signal was achieved during the tm and the RD. The summed FIDs were multiplied by an exponential weighting function corresponding to a line broadening of 0.3 Hz prior to Fourier transformation. Metabolite assignments have been made on the basis of previous literature data (16-18). 1H

Data Reduction and Principal Components Analysis of NMR Spectra. Each spectrum was corrected for phase and baseline distortions using XWINNMR (version 2.1, Bruker Analytische Germany) referenced to TSP (δ 0.0) and reduced to 256 integrated regions of equal width (0.04 ppm) corresponding to the δ region 0.2-10 using AMIX (version 2.5, Bruker Analytische GmbH, Rheinstetten Germany). The region between δ1H 6.2 and 4.52 was set to zero integral in order remove effects of variation in the suppression of the water resonance and the effects of variation in the urea signal caused by partial saturation via solvent exchanging protons. These data were collected into a single Excel (Microsoft, Excel 97, SR-2) data table where each row described the integral descriptors for an NMR spectrum. Two data sets were generated, the first was normalized to unit area so as to partially remove concentration differences between dilute and concentrated urine samples. The second data set excluded integral regions containing drug carrier-related resonances (the region δ1H 3.44-4.00 was removed). Multivariate analysis was performed using the software package SIMCA (version 8, Umetrics AB, Umeå, Sweden). The data-reduced NMR spectra of the urines generated for each of the six different dosing vehicles in all of the data sets were analyzed using PCA (19, 20) on both mean-centered data and Pareto scaled data. Mean centering involves the subtraction of the mean value of a descriptor from all values of that descriptor so that the mean for each variable is 0. Pareto scaling also mean-centers these data, but each data value is given a variance numerically equal to 1/xstandard deviation such that descriptors based on lower concentration metabolites in biofluids are given a relatively higher weighting than is the case using only mean-centered analysis. For all data sets, two-dimensional PC scores plots were constructed in order to establish the presence of any dose-related patterns or clusters in the data. It was also possible at this stage to detect the presence of “outliers” (either by a large distance to model in the outlier diagnostic plot or by isolated mapping in the scores plot).

1138

Chem. Res. Toxicol., Vol. 15, No. 9, 2002

Figure 1. 600 MHz 1H NMR spectra of urine (δ 0.5-4.5) 0-8 h post-administration of several drug carrier vehicles: (a) buffered water, (b) microemulsion, (c) labrafil/CO, and (d) PEG 300.

Results and Discussion 1H

NMR Spectroscopy of Urine from Drug Carrier Vehicle Dosed Rats. The 1H NMR spectra of the urine samples obtained from animals treated with the six dosing vehicles were inspected and compared visually. Of the six drug carrier vehicles studied, three (buffered water, CMC/tween and methocel) induced few differences in the urinary biochemical profile after administration. Buffered water is an uncomplicated DCV and could be considered the “gold standard”. Buffered water did not give rise to any NMR detectable signals; neither did CMC/tween nor methocel, both of which have complex organic matrixes (Table 1). Moreover, the similarities in predose and postdose profiles would indicate that these vehicles exerted little metabolic effect on the animals. However, the spectra acquired after administration of labrafil/CO, microemulsion, and PEG 300 gave rise to strong signals from the dosing vehicle components (Figure 1), which were apparent in the samples collected from all post-dose sampling periods (0-96 h p.d). Additionally spectra obtained from rats dosed with these vehicles showed subtle fluctuations in endogenous urinary components. The drug carrier related resonances in the urine spectra have been summarized in Table 2. PCA of All Control 1H NMR Spectral Data. PC maps were generated from the spectra of all urine samples. The scores plots generated from Pareto-scaled data and mean-centered data were similar. The scores plot of PC1 versus PC2 for mean-centered data (Figure 2) shows that the populations of urine samples corresponding to each drug carrier vehicle were overlapped, except in the cases of microemulsion, labrafil/CO, and PEG 300 (2, 0, and 9, respectively). Samples generated

Beckwith-Hall et al.

following the administration of these three vehicles each formed a distinct and isolated cluster, indicating substantial perturbations in the urine profiles from basal state. Labrafil/CO and PEG 300 were clearly separated from the main sample cluster in the first PC. Examination of the PC loadings showed that this separation was predominantly influenced by the presence of large resonances dominating the region δ1H 3.44-4.00. Microemulsion samples mapped separately in both PC1 and PC2, reflecting the fact that a high proportion of the NMR spectrum was affected by vehicle resonances. In such instances, where vehicles are present in high concentrations in the urine and give rise to NMR peaks, subtle endogenous variations in control samples may be obscured. Thus, due to the difference in magnitude between endogenous and vehicle-derived spectral effects, samples comprising the groups for the remaining three vehicles formed a tight cluster (Figure 2). Resonances originating from microemulsion were present in 25% of the integrated spectral regions, while reonances from PEG 300 and labrafil/CO both influenced 7% of the total spectral integral regions (Table 2). Many of the spectral areas influenced by vehicle resonances also contain endogenous resonances of diagnostic importance. For example, previous studies have shown that renal cortical toxicity can be characterized by a combination of increased urinary excretion of glucose together with organic and amino acids (21). However, since microemulsion resonances are overlapped with signals from lactate, glucose, alanine, and other amino acids, the diagnostic capability for renal cortical toxicity would be severely compromised when using this vehicle. Consequently, microemulsion can be regarded as a generally unsuitable dosing vehicle for NMR-based metabolic studies. Although both PEG 300 and labrafil/CO also produced 1H NMR visible resonances, signals were contained within the narrow shift region corresponding to δ 3.44-4.00. Accordingly, this region was omitted from the data set allowing unbiased interrogation of the endogenous urinary biochemical profile. PCA of Control 1H NMR Spectral Data Following Exclusion of Microemulsion and Vehicle Metabolites. Having excluded the set of microemulsion samples and having removed any remaining spectral regions pertaining to PEG 300 and labrafil/CO metabolites, the mean-centered scores plot (PC1 versus PC2) showed improved dispersion of samples (Figure 3a). Clusters of both labrafil/CO and PEG 300 samples mapped more closely to the other vehicle data but remained largely separated from the main cluster from 8 h postdose onward. Clearer separation still of the labrafil/CO and PEG 300 samples was noted with Pareto scaling (Figure 3b). Both vehicles separated from the main cluster in PC1 but were further subdivided in PC2. Theoretically, healthy control animals should have a similar urinary biochemical profile and hence the coordinates of these samples should occupy a similar position in a PR map based on 1 H NMR spectra (22). Indeed, the large degree of overlap observed in the mean-centered and Pareto-scaled PC plots of samples relating to buffered water, CMC/tween, and methocel (Figure 3, panels a and b) indicates the existence of a high degree of commonality between spectra and would suggest that there are no major perturbations to the metabolites that are present in the urine after administration of these three DCVs. Within a control population of animals, the spectral profile may

Metabolic Effects of Drug Carrier Vehicles

Chem. Res. Toxicol., Vol. 15, No. 9, 2002 1139

Table 2. Visible Drug Carrier Vehicle Resonances in 1H NMR Spectra of Urine drug carrier vehicle (DCV) labrafil/CO PEG 300 microemulsion

chemical shift (δ1H) and multiplicity

dosing component

3.64(m), 3.71(s), 3.74(m) 3.52(m), 3.59(m), 3.64(m), 3.71(m), 3.74(m), 3.83(m), 3.85(m) 1.18(t), 3.63(q) 1.14(d), 3.44(m), 3.55(m), 3.88(m) 3.70(s) 0.87(m), 1.32(m), 1.52(m), 1.56(m), 1.75(m), 2.18(m), 2.23(m), 3.64(m), 3.74(m)

vehicle signals vehicle signals ethanol propylene glycol cremophore vehicle resonances including triacylglyceride peaks

Figure 2. Plot of PC1 vs PC2 scores based on mean-centered data from 1H NMR spectra of whole urine samples obtained prior to, and following the administration of six drug carrier vehicles. Key: (O) buffered water; (b) CMC/tween; (4) methocel; (2) microemulsion; (0) labrafil/CO; (9) PEG 300.

be complicated by the presence of general biochemical effects and physiological variance. The common mapping position of CMC/tween, buffered water, and methocel discounts physiological variation being the cause of the separation of the labrafil/CO and PEG 300 samples and would suggest that there are perturbations in the levels of detectable endogenous metabolites present in the 1H NMR spectra arising as a consequence of administration of these two DCVs. Identification of Endogenous Metabolites Altered Following the Administration of PEG 300 and Labrafil/CO. Examination of the PC loadings showed the regions of variance within the NMR spectra. This gave a list of potential biomarkers for PEG 300- and labrafil/CO-induced alterations to the urinary biochemical profile, which is shown in Table 3. Representative spectra obtained from rats dosed with these vehicles are compared to a spectrum from the buffered water vehicle (Figure 4). The most significant metabolites distinguishing between these two vehicles included, for PEG300, reduced citrate and 2-oxoglutarate levels, elevated creatinine and taurine levels, an altered N-acetyl glycoprotein region, and the detection of several metabolites present at lower concentrations, which are marked on the spectrum generated from a 48-72 h urine sample. In the case of labrafil/CO, the predominant metabolite differences were relatively higher concentrations of taurine, longchain dicarboxylic acids, and creatinine, while changes in metabolites present at lower levels also contributed to the discrimination of this vehicle group and are marked in the corresponding spectrum. Although several of the labrafil/CO- and PEG 300induced urinary perturbations were characteristic of

Figure 3. Plot of PC scores from 1H NMR spectra of whole urine samples having removed microemulsion and the spectral regions containing drug carrier vehicle based on (a) meancentered data (PC1 vs PC2) and (b) Pareto data (PC1 vs PC2). Key: (O) buffered water; (b) CMC/tween; (4) methocel; (2) microemulsion; (0) labrafil/CO; (9) PEG 300.

other xenobiotic induced urinary effects (23, 24), the suite of urinary perturbations observed with both vehicles was unique, reflected by their distinct mapping positions. PEG is a substance of known low toxic potential with an LD50 in the mouse and rat for PEG 400 by p.o. administration of 26 and 30 mL/kg respectively (25). Here the similar substance PEG 300, and labrafil (which comprises of PEG esterified with oleic acid) were used at a dose volume of 10 mL/kg and with daily dosing it is quite probable that these levels could cause biochemical changes to be detected. Removal of interfering metabolite resonances was easily achieved by excluding the relevant integral regions from the data set. However, several points must be considered in the analysis of such a data set. As already

1140

Chem. Res. Toxicol., Vol. 15, No. 9, 2002

Beckwith-Hall et al.

Table 3. PCA-Detected 1H NMR Spectral Regions That Cause Separation of the PEG 300 and Labrafil/CO Vehiclesa relative changes occurring after PEG 300 administration

relative changes occurring after labrafil/CO administration

spectral region* (δ1H)

assignment

effect

spectral region* (δ1H)

assignment

effect

2.68 2.56 3.02 2.46 4.06 3.06 2.42 2.06 3.34 4.10 4.02 4.14 3.22 3.30 3.38 3.42 4.18 2.14 2.10 0.94

citrate citrate 2-oxoglutarate 2-oxoglutarate creatinine creatinine 2-oxoglutarate N-acetylglycoproteins Ub lactate Ub lactate choline related metabolite taurine Ub taurine Ub Ub N-acetylglycoproteins isoleucine

+ + + + +

3.22 1.30 2.18 1.58 4.22 0.86 1.54 4.02 3.30 4.10 2.74 4.06 3.34 0.90 4.46 3.06 3.18 3.26 2.34 2.30

taurine dicarboxylic acids/lactate dicarboxylic acids dicarboxylic acids sugar anomeric proton CH3s of organic/amino acids dicarboxylic acids creatinine taurine lactate citrate creatinine Ub CH3s of organic/amino acids Ub creatinine phenylalanine taurine/TMAO CH2s of organic/ amino acids

+ + + + + + + + + + +

+ + + + + + + + +

+ + + + + + +

a (*) Mid spectral integral region, i.e., 2.68 represents δ1H 2.66 to 2.70; (+) an increased concentration; (-) a decreased concentration in urine (as determined by examination of the PC loadings); (U) an unidentified metabolite. b Not assigned at 600 MHz even with the use of 2D NMR methods because of low signal intensity (1/200th to 1/1100th of total intensity).

signals of diagnostic importance. Also, the future determination of metabolites present at lower concentrations in the samples may be complicated by the fact that the vehicle remains in the biofluid sample. For example, although two-dimensional correlation spectra were acquired for the purpose of metabolite identification, the strong resonances deriving from PEG 300 and labrafil/ CO presented a marked dynamic range problem masking some of the signals from low concentration endogenous metabolites and distorting the spectral baselines.

Figure 4. 600 MHz 1H NMR spectra (δ 0.5-4.5) obtained from (a) buffered water, (b) labrafil/CO, and (c) PEG 300 48-72 h post-administration showing the metabolic response induced by labrafil/CO and PEG 300. An asterix indicates regions of the spectrum that are highlighted as significantly different by examination of the PC loadings but which remain unassigned at 600 MHz even with the use of 2D NMR methods because of low signal intensity.

mentioned, the regions omitted due to the obscuring vehicle resonances may contain endogenous metabolite

Categorization of Drug Carrier Vehicles for Metabonomic Studies. In this study, a statistical model to define biofluid normality in control vehicle-dosed animals based on the urinary excretion patterns has shown that microemulsion is a generally unsuitable dosing vehicle due to its metabolite signals being widely distributed across the NMR spectrum. Vehicle resonances from both PEG 300 and labrafil/CO dominated a discrete region of the NMR spectrum, yet analysis of purely endogenous metabolites indicated that repeated administration of both altered the 1H NMR detectable metabolic profile. While buffered water, CMC/tween and methocel did not give rise to any NMR detectable signals and the metabolic profile was not disturbed with repeated dosing. To this end, it may be possible to begin to classify DCVs for the purpose of NMR studies into four subtypes according to whether they introduce unwanted signals into the spectra (analytical interference) or whether they cause metabolic perturbations, viz., (Class 1) no NMR detectable signals and no changes to the endogenous metabolite profile (most acceptable); (Class 2) no NMR detectable signals of vehicle but with alterations observed to the endogenous metabolite profile; (Class 3) NMR detectable signals but no changes to the endogenous metabolite profile, and (Class 4) both NMR detectable signals of vehicle and a perturbed endogenous metabolite profile (least acceptable). This study investigated six vehicles, which fell into either Class 1 or Class 4 and will aid in the choice of solubilizing agents.

Metabolic Effects of Drug Carrier Vehicles

Chem. Res. Toxicol., Vol. 15, No. 9, 2002 1141

Conclusions To improve the reliability of detecting specific toxinrelated alterations (particularly subtle alterations) in the biochemical profile of urine, it is necessary to define the extent of normal physiological variance within a control population and to establish the potential adverse effects of vehicle administration. Furthermore, analysis of factors that are influencing any variation should be undertaken. An important facet of high data density metabolic analysis is that it is not essential to understand fully the complex molecular differences that underlie the spectral features associated with these vehicle-induced effects to be able to classify samples. Further analysis of the molecular basis of the spectral differences would give greater insight into the mechanistic processes involved. As the use and magnitude of toxicological databases increase, it becomes ever more important to control all sources of avoidable variation within the data set. This study established that some vehicles were more suitable than others in terms of NMR analysis, preferably those without extensive resonances in the 1H NMR spectrum. Moreover, certain drug carrier vehicles exhibited evidence of bioactivity, thereby interfering with the derivation of potential diagnostic inferences from toxicological or clinical studies.

References (1) Nicholson, J. K., and Wilson, I. D. (1989) High-resolution proton magnetic resonance spectroscopy of biological fluids. Prog. Nucl. Magn. Reson. Spectrosc. 21, 449-501. (2) 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 Discov. 1, 153-161. (3) Holmes, E., Bonner, F. W., Sweatman, B. C., Lindon, J. C., Beddell, C. R., Rahr, E., and Nicholson, J. K. (1992) Nuclear magnetic resonance spectroscopy and pattern recognition investigations into the biochemical processes associated with the progression and recovery from nephrotoxic lesions in the rat induced by mercury II chloride and 2-bromoethanamine. Mol. Pharmacol. 42, 922-930. (4) Maxwell, R. J., Martinez-Perez, I., Cerdan, S., Cabanas, M. E., Arus, C., Moreno, A., Capdevila, A., Ferrer, E., Bartomeus, F., Aparicio, A., Conesa, G., Roda, J. M., Carcellar, F., Pascual, J., Howells, S. L., Mazucco, R., and Griffiths, J. R. (1998) Pattern recognition analysis from perchloric acid extracts of human brain tumour biopsies. Magn. Reson. Med. 39, 869-877 (5) Connor, S. C., Everett, J. R., Jennings, K. R., Nicholson, J. K., and Woodnut, G. (1994). High resolution 1H NMR studies of the metabolism and excretion of ampicillin in rats and amoxycillin in rats and man. J. Pharm. Pharmacol. 46, 128-134. (6) Shockcor, J. P., Unger, S. E., Wilson, I. D., Foxall, P. J., Nicholson, J. K., and Lindon, J. C. (1996) Combined HPLC, NMR spectroscopy and ion-trap mass spectrometry with application to the detection and characterization of xenobiotic and endogenous metabolites in human urine. Anal. Chem. 68, 4431-4435. (7) Holmes, E., Nicholls, A. W., London, J. C., Ramos, S., Spraul, M., Neidig, P., Connor, S. C., Connelly, J., and Nicholson, J. K. (1998) Development of a model for classification of toxin-induced lesions using 1H NMR spectroscopy of urine combined with pattern recognition. NMR Biomed. 11, 235-244. (8) Robertson, D. G., Reily, M. D., Sigler, R. E., Wells, D. F., Paterson, D. A., and Braden, T. K. (2000) Metabonomics: Evaluation of nuclear magnetic resonance (NMR) and pattern recognition

(9)

(10)

(11)

(12)

(13)

(14)

(15) (16) (17) (18) (19) (20) (21)

(22)

(23)

(24)

(25)

technology for rapid in vivo screening of liver and kidney toxicants. Toxicol. Sci. 57, 326-337. Holmes, E., Nicholls, A. W., Lindon, J. C., Connor, S. C., Connelly, J. C., Haselden, J. N., Damment, S. J. P., Spraul, M., Neidig, P., and Nicholson, J. K. (2000) Chemometric models for toxicity classification based on NMR spectra of biofluids. Chem. Res. Toxicol. 13, 471-478. 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, 1181-118. Beckwith-Hall, B. M., Nicholson, J. K., Nicholls, A. W., Foxall, P. J. D., Lindon, J. C., Connor, S. C., Abdi, M., Connelly, J. C., 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. Bollard, M. E., Holmes, E., Lindon, J. C., Mitchell, S. C., Branstetter, D., Zhang, W., and Nicholson, J. K. (2001) Investigations into biochemical changes due to diurnal variation and estrus cycle in female rats using high resolution 1H NMR spectroscopy of urine and pattern recognition. Anal. Biochem. 295, 194-202. Gavaghan, C. L., Nicholson, J. K., Connor, S. C., Wilson, I. D., Wright, B., and Holmes, E. (2001) HPLC-NMR spectroscopic and chemometric studies on metabolic variation in Sprague Dawley rats. Anal. Biochem. 291, 245-252. Phipps, A. N., Stewart, J., Wright, B., and Wilson, I. D. (1998) Effect of diet on the urinary excretion of hippuric acid and other dietary-derived aromatics in the rat. A complex interaction between diet, gut microflora and substrate specificity. Xenobiotica 28 (5), 527-537. Bell, J. D., Sadler, P. J., Morris, V. C., and Levander, O. A. (1991) Effect of aging and diet on proton NMR spectra of rat urine. Magn. Reson. Med. 17, 414-422. Nicholson, J. K., Foxall, P. J. D., Spraul, M., Farant, R. D., and Lindon, J. C. (1995) 750 MHz 1H and 1H-13C NMR spectroscopy of human blood plasma. Anal. Chem. 67 (5), 793-811. Gartland, K. P. R., Bonner, F. W., and Nicholson, J. K. (1989) Investigations into the biochemical effects of region-specific nephrotoxins. Mol. Pharm. 35, 242-250. Fan, T. W. M. Metabolite profiling by one- and two-dimensional NMR analysis of complex mixtures (1996) Prog. Nucl. Magn. Reson. Spectrosc. 28, 161-219. Beebe, K. R., Pell, R. J., and Seahsholt, M B. (1998) Chemometrics: A Practical Guide, John Wiley & Sons, New York. Jurs, P. C. (1986) Pattern recognition used to investigate multivariate data in analytical chemistry. Science 232, 1219-1224. Anthony, M. L., Beddell, C. R., Lindon, J. C., and Nicholson, J. K. (1994) Studies on the comparative toxicity of S-(1,2-dichlorovinyl)-l-cysteine, S-(1,2-dichlorovinyl)-homocysteine and 1,1,2trichloro-3,3,3-trifluoro-1-propene in the Fischer 344 rat. Arch. Toxicol. 69, 99-110. Holmes, E., Nicholson, J. K., and Tranter, G. (2001) Metabonomic classification of genetic variations in toxicological and metabolic responses using probabilistic neural networks. Chem. Res. Toxicol. 48, 182-191. Waterfield, C. J., Turton, J. A., Scales, M. D. C., and Timbrell, J. A. (1993) Investigations into the effects of various hepatotoxic compounds on urinary and liver taurine levels in rats. Arch. Toxicol. 67, 244-254. Montgomery, J. A., Mamer, O. A., and Colle, E. (1989) Profile in altered metabolism. IV-Induction of acute aciduria following 2-octymoic acid administration to the rat. Biomed. Environ. Mass. Spectrom. 18 (6), 416-423. Bartsch, W., Sponer, G., Dietmann, K., and Fuchs, G. (1976) Acute toxicity of various solvents in the mouse and rat. LD50 of ethanol, diethylacetamide, dimethylformamide, dimethylsulfoxide, glycerine, N-methylpyrrolidone, polyethylene glycol 400 and 1,2propanediol and Tween 20. Arzneimittelforschung 26 (8), 15811583.

TX020020+