ToxSci Advance Access originally published online on April 9, 2007
Toxicological Sciences 2007 98(1):286-297; doi:10.1093/toxsci/kfm077
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Uniform Procedure of 1H NMR Analysis of Rat Urine and Toxicometabonomics Part II: Comparison of NMR Profiles for Classification of Hepatotoxicity
* Department of Pharmacology
Department of Medical Chemistry
Department of Toxicology and Drug Disposition, N.V. Organon, Molenstraat 110, 5340 BH Oss, The Netherlands
Department of Analytical Research, TNO Quality of Life, Utrechtseweg 48, PO Box 360, 3700 AJ Zeist, The Netherlands
1 To whom correspondence should be addressed. Fax: 31-412-663532. E-mail: willem.schoonen{at}organon.com.
Received October 23, 2006; accepted March 22, 2007
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONSUPPLEMENTARY DATAREFERENCES |
|---|
A procedure of nuclear magnetic resonance (NMR) urinalysis using pattern recognition is proposed for early detection of toxicity of investigational compounds in rats. The method is applied to detect toxicity upon administration of 13 toxic reference compounds and one nontoxic control compound (mianserine) in rats. The toxic compounds are expected to induce necrosis (bromobenzene, paracetamol, carbon tetrachloride, iproniazid, isoniazid, thioacetamide), cholestasis (
-naphthylisothiocyanate (ANIT), chlorpromazine, ethinylestradiol, methyltestosterone, ibuprofen), or steatosis (phenobarbital, tetracycline). Animals were treated daily for 2 or 4 days except for paracetamol and bromobenzene (1 and 2 days) and carbon tetrachloride (1 day only). Urine was collected 24 h after the first and second treatment. The animals were sacrificed 24 h after the last treatment, and NMR data were compared with liver histopathology as well as blood and urine biochemistry. Pathology and biochemistry showed marked toxicity in the liver at high doses of bromobenzene, paracetamol, carbon tetrachloride, ANIT, and ibuprofen. Thioacetamide and chlorpromazine showed less extensive changes, while the influences of iproniazid, isoniazid, phenobarbital, ethinylestradiol, and tetracycline on the toxic parameters were marginal or for methyltestosterone and mianserine negligible. NMR spectroscopy revealed significant changes upon dosing in 88 NMR biomarker signals preselected with the Procrustus Rotation method on principal component discriminant analysis (PCDA) plots. Further evaluation of the specific changes led to the identification of biomarker patterns for the specific types of liver toxicity. Comparison of our rat NMR PCDA data with histopathological changes reported in humans and/or rats suggests that rat NMR urinalysis can be used to predict hepatotoxicity. Key Words: metabonomics; urinalysis; hepatotoxicity; necrosis; cholestasis; steatosis; NMR.
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONSUPPLEMENTARY DATAREFERENCES |
|---|
Attrition in the early development stages of drug candidates is dominated by toxicological side effects (50% of cases, Caldwell et al., 2001
; Cuatrecasas, 2006; Nicholson et al., 2002; Schuster et al., 2005; Smith and Schmid, 2006). Success rates may be improved by screening methods for toxicology in early research phases. In a previous study, it was shown that urine NMR analysis had at least a four- to 16-fold higher sensitivity as compared to histopathology and clinical chemistry in detecting adaptive and/or toxic changes caused by bromobenzene and paracetamol treatment (Schoonen et al., 2007). The high sensitivity of NMR urinalysis and the available knowledge of other studies applying urine NMR spectroscopy to determine the effects of hepatotoxic compounds (Azmi et al., 2005; Beckwith-Hall et al., 1998; Bollard et al., 2005; Espina et al., 2001; Klenø et al., 2004; Lehmann-McKeeman and Car, 2004; Mortishire-Smith et al., 2004; Nicholson et al., 1999; Robertson et al., 2000; Shockcor and Holmes, 2002) led us to believe that a uniform NMR approach for hepatotoxicity is feasible (Corcoran, 2005; London and Houck, 2004; Robertson, 2005). The ultimate aim of our study is to obtain a better predictive tool for early toxicity profiling at early stages of the development process.
The goal of the current study was to assess whether different types of hepatotoxic effects in rats can be classified by NMR urinalysis. The classification should be related to the histopathological and clinical chemical changes caused by these drugs. Drug-induced histopathological changes in the liver may be divided into different classes like hepatitis, necrosis, hepatocellular vacuolation up to steatosis, biliary toxicity leading to cholestasis, hypertrophy, hyperplasia, and more (Gopinath et al., 1987). In the present study, the focus was on three classes of compounds causing necrosis, cholestasis, and steatosis.
For this study, 13 hepatotoxic compounds and one control compound were selected. Although classification into distinct pathological classes is difficult due to overlap and mixed pattern histopathological changes, such a classification was still made to the best of our abilities (Table 3 and Supplementary Table 1).
|
The induction of necrosis in hepatocytes may be due to reactive radicals, radical oxygen species, or hydrazines. Reactive radicals or reactive oxygen species are produced by bromobenzene, paracetamol, and carbon tetrachloride, whereas reactive hydrazines are produced by isoniazid and iproniazid (Nelson et al., 1976
, 1978; Vandenberghe, 1996).
Liver cholestasis is caused by a reduced bile flow and excretion of organic anions. The cholestatic compounds in this study are -naphthylisothiocyanate (ANIT), chlorpromazine, ethinylestradiol, and methyltestosterone. It is known that in humans very high levels of anabolic steroids and estrogens can induce cholestasis (Ishak and Zimmerman, 1987; Welder et al., 1995; Westaby et al., 1977). Examples of compounds where this phenomenon occurs are methyltestosterone, estradiol, and ethinylestradiol (Accanito et al., 1995, 1996, Frezza et al., 1988; Sánchez Pozzi et al., 2003; Simon et al., 1996). This cholestatic effect is also observed during pregnancy in case of very high estradiol plasma levels (Kreek et al., 1967; Reyes et al., 1981). Similar effects can be reproduced in rats with high doses of estrogens but not with anabolic androgens.
Liver steatosis may result from an increased synthesis of fatty acids or a decrease in the oxidation of fatty acids by mitochondria. This leads to the presence of small fatty vesicles filling the cytoplasm of the hepatocyte or large fat droplets in hepatocytes pushing the nucleus to the periphery of the cell. Phenobarbital and tetracycline are compounds that specifically induce steatosis (Vandenberghe, 1996), and both are included in this study.
The toxic effects of two other compounds, i.e., thioacetamide and ibuprofen, are less clear from literature data. Thioacetamide used in our experiments is known to cause liver necrosis at high dosages (Gupta, 1956; Roberts et al., 2000), but it may also lead to renal cortical toxicity (Holmes et al., 1998). Ibuprofen is a compound that may induce a particular type of cholestasis, which leads to vanishing bile ducts in humans (Lewis, 2000). Finally, mianserine was added to the compound selection because it has no hepatototoxic effects and serves as a negative control. The compound is a marketed antidepressant (Tolvon).
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONSUPPLEMENTARY DATAREFERENCES |
|---|
Experimental setup.
The studies described below are in compliance with the Dutch Act on Animal Experimentation. The studies were approved by the Animal Experimentation Committee.
An overview of the experimental setup and the collection of samples are given in Table 1. In short, the compounds were given in a fixed-dose design either as a single dose (carbon tetrachloride) or daily for 1 and 2 days (paracetamol and bromobenzene) and daily for 2 and 4 days (all other compounds).
|
The experimental protocol was as follows: Male Wistar rats (Harlan) of 6 weeks old (140–170 g) were placed in metabolism cages 1 day before treatment. Dosing details and administration route are given in Table 1. The vehicle was 10% gelatine/mannitol. The dosages were guided by rat toxicity data from the literature to show overt toxicity but to avoid lethal effects. All urine of the first 24 h and all urine of the second 24 h after the first dosage as well as the last 24 h of the 4 days treatment were collected. These urine samples were used for clinical chemistry and for NMR. The urine of the last 24 h of the 4 days treatment was collected for clinical chemistry only. At the end of the experiment, i.e., 24 h after the last administration, the rats were decapitated. Plasma samples of these rats were collected and used for blood clinical chemistry. Moreover, livers of these rats were dissected for histopathological evaluation.
Pathology and clinical chemistry.
Pathology and clinical chemistry were performed according to standard procedures. Samples of the livers were preserved in 10% buffered formalin. Subsequently, tissue samples were dehydrated and embedded in paraffin wax. The paraffin-embedded liver was sliced into 5-µm thick sections. These sections were stained with hematoxylin and eosin sections and used for histological examination by light microscopy. Blood analysis was focused on the measurement of the enzyme levels of aspartate aminotransferase (AST), alanine aminotransferase (ALT), triglycerides, and total bilirubin levels.
NMR sample preparation.
After collecting the urine samples, these samples were stored at 4°C on ice, centrifuged to remove solids, and filtered over cotton wool. Per sample, a volume of 1-ml urine was lyophilized overnight and dissolved in 1-ml 0.1M phosphate buffer (pH 7.0) using D2O as aqueous solvent. Trimethylsilyl-2,2,3,3-tetradeuteropropionic acid was added as an internal chemical-shift reference. NMR spectra were recorded within 24 up to 28 h after the collection of urine, as described in Schoonen et al. (2007).
NMR data analysis and principal component discriminant analysis.
The analysis of the NMR data was performed according to the procedure described by Schoonen et al. (2007). Data reduction was performed by eliminating peaks originating from the administered compound (parent and/or metabolites). Relevant biomarker signals were identified using the Procrustus Rotation method (Gower and Dijksterhuis, 2004; Héberger and Andrade, 2004).
Principal component discriminant analyses (PCDA) were performed on different subgroups of treated rats to study whether pathology class–specific effects could be identified. To this end, compounds belonging to one particular pathology group were clustered. Thus four groups were compared, i.e.,
- Control group, all control experiments.
- Necrosis group, including bromobenzene, paracetamol, CCl4, thioacetamide, isoniazid, and iproniazid.
- Cholestasis group, including ANIT, chlorpromazine, ethinylestradiol, and methyltestosterone.
- Steatosis group, including phenobarbital and tetracycline.
- Ibuprofen formed a fifth group based on its markedly different effects in overall analysis.
The following analyses were performed:
- PCDA on all rats at first and second day of treatment with all compounds. No clustering was used in this calculation.
- PCDA on all rats at first and second day of treatment with all compounds. No clustering was used in the calculation. The extreme outliers in calculation A, i.e., ibuprofen and isoniazid were omitted to enable a more detailed analysis of the other compounds.
- PCDA on rats at first and second day of treatment with all compounds belonging to one pathology group with the controls included. This was performed for each pathology group as well as for ibuprofen alone.
- PCDA on all rats at first and second day of treatment with all compounds, using the clustering as described under the third bullet.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONSUPPLEMENTARY DATAREFERENCES |
|---|
General
All animals survived the treatment with the various compounds during the experiment. Chlorpromazine at the given dose level induced narcotic effects, and rats were sacrificed 72 h after dosing.
Histopathology
Experimental data including a literature review are summarized in Table 3 and the Supplementary Table 1. A few details are given below.
Control groups.
No histopathological abnormalities were observed in the livers of the placebo samples. Mianserine, a control compound, did not show any effects on the liver.
Necrotic compounds.
Bromobenzene and paracetamol induced predominantly slight to severe centrilobular liver necrosis. Carbon tetrachloride predominantly induced moderate to marked hydropic degeneration in the periportal area. Iproniazid caused only slight transient periportal fine vacuolation. Isoniazid induced moderate hepatocellular rarefaction. With thioacetamide, moderate hepatocellular degeneration and single-cell necrosis in the centrilobular areas were observed.
Cholestatic compounds.
ANIT induced predominantly moderate mixed inflammatory cell infiltrates with moderate early fibrosis present in periportal areas. Additionally, degeneration of bile duct epithelium and early signs of proliferation of bile ducts were also observed. With ibuprofen, moderate centrilobular necrosis was observed in some animals, which was accompanied with moderate mixed inflammatory cell infiltrates. With ethinylestradiol, the livers of a few animals showed slight vacuolation after 4 days, while with chlorpromazine and methyltestosterone no liver histopathology was observed.
Steatotic compounds.
Phenobarbital showed only a weak periportal rarefaction. Tetracycline induced slight vacuolation of hepatocytes in a few animals.
Clinical Chemistry
For all compounds, the classical clinical biochemistry plasma and urine parameters were measured, but only statistical significant changes were observed with AST, ALT, bilirubin, and triglycerides. Data on clinical chemistry are presented in Supplementary Table 2.
|
Control compound.
Mianserine did not cause changes in these four parameters.
Necrotic compounds.
Plasma concentrations of AST and ALT were markedly increased with bromobenzene, paracetamol, and carbon tetrachloride in all treated groups. Carbon tetrachloride also increased the bilirubin level markedly. For iproniazid, no relevant or minor changes were observed. Thioacetamide increased AST and bilirubin levels slightly. Isoniazid did not change any of these plasma parameters.
Cholestatic compounds.
ANIT increased the levels of AST, ALT, bilirubin, and triglycerides in varying amounts. Chlorpromazine increased the levels of AST and bilirubin moderately. Ibuprofen increased AST and ALT levels markedly. Ethinylestradiol had only a moderate effect on triglycerides. Methyltestosterone did not induce any changes in these plasma parameters.
Steatotic compounds.
Phenobarbital and tetracycline did not influence these plasma parameters.
Pattern Recognition
The peak listings of all urine spectra were combined into one single data matrix. Initial data reduction resulted in a set of 174 peaks. This set was further reduced to a set of relevant biomarkers by applying the Procrustus Rotation method. The only difference with our previous study was that the Procrustus error minimum led us this time to a number of 88 biomarker signals (instead of 80 signals earlier). Analyses and (partial) assignments completely focussed on these 88 signals. For biomarker identification, an in-house database of 400 compounds for compound identification in rat urine was used together with an external database (http://www.liu.se/hu/mdl/main/) (Agar et al., 1991; Fan, 1996a,b; Govindaraju et al., 2000) and several literature comparisons (Bollard et al., 2005; Clayton et al., 2006; Robertson et al., 2000). The preliminary assignments given here should be used with caution because small chemical-shift changes due to pH differences might cause ambiguities in crowded regions.
Overall, 45% (40 peaks) of these 88 marker peaks are identical in both studies. This partial overlap is not unexpected since the necrotic compounds, bromobenzene, and paracetamol were also used in the first study (Schoonen et al., 2007). Additionally, it implies that a number of common endogenous metabolites are affected by necrosis, cholestasis, and steatosis. For a review of the assigned peaks used in the PCDA of both articles, we refer to Figure 2D in Schoonen et al. (2007). Most of the biomarker signals are similar to or belong to the following groups of compounds (Table 2, Bollard et al., 2005; Robertson et al., 2000):
- sugars: citrate, -ketoglutarate, succinate, and lactate.
- amino acids: taurine and tyrosine.
- some respiratory failure–dependent products: allantoin and formate.
- other identified metabolites: dimethylglycine and betaine both of the choline shunt, acetylcholine, dimethylamine, uridine, and hippurate.
|
In Figure 1A, the complete urine NMR PCDA is shown for all compounds after 1 and 2 days of treatment without further stratification of treatment groups. Moreover in Figure 1B, a 3D (three-dimensional) presentation is given for this analysis. In Figure 1C, ibuprofen and isoniazid were omitted from the calculation to enable a closer examination of the other compounds. The two PCDA calculations (A and B, see "Materials and Methods" section) have a very similar distribution of the data points in the 3D representation. In Figures 1A and 1B, all control samples in the five independent experiments cluster together in the control area and all other treatment groups are shown.
|
Ibuprofen treatment clearly discriminated itself from all the other samples in PCDA analysis both after 1 and 2 days of treatment. This difference might have been due to decreased creatinine levels in the urine as a result of kidney failure, as shown with biochemistry and NMR. Isoniazid, iproniazid, thioacetamide, and paracetamol treatments showed a clear separation after 1 and 2 days in both 2D and 3D PCDA plots. The direction of the change for these compounds is similar, but isoniazid shows a much larger difference than the other three compounds. The effect of bromobenzene was rather different from the necrotic compounds isoniazid, iproniazid, thioacetamide, and paracetamol as well as from ibuprofen. (Fig. 1B).
The differences between the other compounds were much smaller, but ANIT and chlorpromazine show a remarkably good clustering. Ethinylestradiol and methyltestosterone come in closer proximity to control samples as were mianserine and tetracycline. Phenobarbital, on the other hand, was clearly separated again from the control samples. Magnification of this part of the analysis (Fig. 1C) confirmed the segregation of thioacetamide, iproniazid, and paracetamol, for which the changes seen with thioacetamide were more outspoken than with paracetamol and iproniazid. Also, the effects of ANIT and chlorpromazine become more prominent after 2 days of treatment than after 1 day of treatment. Also, ethinylestradiol and methyltestosterone cluster into the same direction after 2 days of treatment, while after 1 day of treatment, the differences were hardly noticeable. In a similar way, differences can be observed with phenobarbital and tetracycline after 1 and 2 days of treatment. Remarkably, the very severe toxicity of carbon tetrachloride does not seem to have great impact on the urine profile. The patterns of the nontoxic compound mianserine were relatively similar to the control groups (Fig. 1C). Hence, no effect was observed with mianserine after 1 and 2 days of treatment, and only minor effects were observed with CCl4, methyltestosterone, and ethinylestradiol after 1 day of treatment.
In Figure 2A, the changes of the 88 biomarker signals upon isoniazid administration are given in a (differential) factor spectrum. In such a factor spectrum, the increases (negative signals) or decreases (positive signals) of each of the 88 biomarkers upon toxic compound administration are visualized. Significant signal reduction is observed for the following peaks corresponding to "metabolites 114 and 115" (2.16 and 2.17 ppm), acetylcholine (2.18 ppm), and "metabolites 127, 38, 133, and 139" (3.82, 3.92, 6.96, and 7.71 ppm).
In Figure 2B, the same factor analysis is given for the changes upon ibuprofen dosing. The largest signal increase is seen in the peaks belonging to "metabolites 103, 105, 106, 108, and 16" (1.09, 1.21, 1.23, 1.43, and 2.80 ppm).
Total Analysis Within Each Pathology Class
The overall results are reanalyzed using further stratification, i.e., clustering by pathology class (see "Materials and Methods" section, analysis C). Firstly, the PCDA was done for each individual pathology class separately, using the same 88 biomarker signals. The nontoxic (control) compound mianserine was withdrawn. Moreover, due to the large difference of the ibuprofen-treated groups in the overall PCDA analysis, ibuprofen treatment was analyzed separately from the cholestatic pathology group.
Necrotic compounds.
In the PCDA on all necrotic compounds (Fig. 3A), it is seen that the direction of change for isoniazid, iproniazid, and bromobenzene was the same (read the front side of the rectangle), while paracetamol and thioacetamide were slightly different in their orientation being more in the direction of the left side and/or back corner of the rectangle. Carbon tetrachloride also differed from control, but this difference was much smaller with respect to the other necrotic compounds, despite the more severe histopathological effects.
|
Cholestatic compounds.
The PCDA on cholestatic compounds is shown in Figure 3B. Ethinylestradiol and methyltestosterone only showed very weak changes as compared to control, while the changes after 2 days were somewhat larger than after 1 day. Chlorpromazine and ANIT clustered after 1 day of treatment into one direction and after 2 days of treatment in a slightly different direction. Although the direction between 1 and 2 days differed, the main orientation was more or less the same. Comparison of all 1-day treatments and all 2-day treatments showed a more similar pattern orientation. In case of the 1-day treatments it is more directed to the right, while it shifts more to the left for the 2-day treatment results.
Ibuprofen.
Ibuprofen showed an identical pattern for the 1- and 2-day treatments (Fig. 3D). As only three groups were compared here, this could not be visualized in a 3D plot.
Steatotic compounds.
Phenobarbital and tetracycline each clustered in different directions from control. Phenobarbital showed an orientation to the front of the rectangle, and tetracycline was more directed to the lower bottom corner of the rectangle. Overall, the main orientation was more to the front for these compounds with respect to the control (Fig. 3C).
Biomarker Patterns for Each Pathology Class
By making factor spectra from the PCDA, specific urine NMR biomarker changes for each of the specific pathology classes can be derived (Figs. 4A–D). These changes are listed below per pathology class.
|
Necrotic compounds.
Isoniazid and iproniazid caused the largest changes in the metabolites (Fig. 4A). The factor spectrum has been taken in the direction of isoniazid 2 and iproniazid 2 and showed that the peaks of "metabolites 111, 11, 39, 40, 133, 138, and 53" (1.85, 2.07, 3.94, 3.95, 6.96, 7.50, and 7.52 ppm) were increased more clearly than the others. A massive decrease was observed with peaks of succinate (2.42 ppm), dimethylamine (2.73 ppm),
-ketoglutarate (2.476, 3.010, and 3.025 ppm), and hippurate (7.63 ppm and 7.66 ppm).
Cholestatic compounds.
The spectrum of cholestasis differs sharply from that of necrosis and has a slight similarity with that of steatosis (Fig. 4C). The main signal increases in cholestasis were found for "metabolites 20, 39, 134, 136, and 137" (negative signals at 3.09, 3.94, 7.06, 7.41, and 7.45 ppm), while the main signal decreases were observed for -ketoglutarate (2.48, 3.01 ppm), citrate (2.69 ppm), dimethylamine (2.73 ppm), "metabolite 118" (2.87 ppm), taurine (3.27 and 3.44 ppm), creatinine (4.05 ppm), allantoin (5.39 ppm), "metabolites 46 and 135" (6.98 and 7.18 ppm), and hippurate (7.63 ppm and 7.66 ppm).
Ibuprofen.
The main decreases were found for ibuprofen with dimethylglycine (2.93 ppm), -ketoglutarate (2.48, 3.01, and 3.02 ppm), creatinine (4.05 ppm), allantoin (5.39 ppm), uridine (5.79 ppm), and formate (8.48 ppm), while the main increases were found with "metabolites 103, 109, 119, 31, and 130" (1.09, 1.43, 3.35, 3.57, and 4.14 ppm; Fig. 4D).
Steatotic compounds.
The largest increases with phenobarbital and tetracycline were found in the peaks for "metabolite 137" (7.45 ppm) and formate (8.48 ppm), and the largest decreases in peak height for succinate (2.42 ppm), citrate (2.69 ppm), -ketoglutarate (2.48 and 3.01 ppm), "metabolite 124" (3.60 ppm), allantoin (5.39 ppm), "metabolite 46" (6.98 ppm), and hippurate (7.63 ppm and 7.66 ppm) (Fig. 4B).
All pathologies.
Finally, the PCDA was recalculated again for all data simultaneously with exclusion of mianserine but by clustering into groups according to their pathology class. The result of this overall PCDA for all compounds is shown in Figures 5A and 5B in a 2D and 3D representation, respectively. In this analysis, there is a very good separation between the different pathology groups of the necrotic, cholestatic, and steatotic compounds. The positioning of ibuprofen is completely separated from the other pathologies. In this calculation, it is clearly shown that NMR urinalysis using PCDA can discriminate different mechanisms of pathology in early stages of treatment.
|
Classification Based on PCDA-NMR Compared with Histopathology and Clinical Chemistry
An overview of all the results is given in Table 3. For the necrotic compounds, the histopathological and biochemical effects of bromobenzene, paracetamol, carbon tetrachloride, and thioacetamide were more pronounced than for isoniazid and iproniazid.
However, in NMR urinalysis, these compounds show overall more comparable changes. For the cholestatic compounds, the toxicity for ANIT and chlorpromazine was well demonstrated by biochemistry as well as NMR, which cannot be concluded for ethinylestradiol and methyltestosterone. For the steatotic compounds, no effects on pathology and biochemistry were shown for phenobarbital and tetracycline, but small changes in NMR were detected for both compounds. Ibuprofen showed some changes with respect to histopathology, AST, and ALT plasma elevations. The toxicity of the compound can be classified from NMR changes as a mixed profile in between necrosis and cholestasis. Thus, it seems that NMR urinalysis is more powerful in the discrimination of pathology subtypes within 24 or 48 h of treatment for the given liver toxicities compared to histology and biochemical parameters.
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONSUPPLEMENTARY DATAREFERENCES |
|---|
This study on 13 hepatotoxic compounds shows some of the capabilities of toxicometabonomics by means of urine 1H NMR spectroscopy in combination with PCDA. The NMR urinalysis procedure not only detected toxicity of the 13 compounds but also the type of hepatotoxicity could be assessed and discriminated between necrosis, cholestasis, and steatosis. To our knowledge, this is the first report on classification of compounds based on their hepatotoxicity subclass using metabonomics, although ideas in this direction are mentioned earlier (Griffin, 2004
; Lindon et al., 2005, 2006; Robertson, 2005). The use of the PCDA method on peak listings (Vogels et al., 1996) instead of bucketing (Cloarec et al., 2005), peak alignment, and the focus on the 88 most relevant biomarker peaks by the Procrustus Rotation selection method have obviously contributed to an improved analysis. Furthermore, the validity of the PCDA outcome was supported by histopathology, clinical chemistry, as well as by the known toxicity mechanisms for these compounds as reported in literature.
When we compare the sensitivity and discriminatory power of the NMR PCDA method with histopathology and clinical chemistry, we observe that NMR PCDA is able to pick up toxicity at an earlier stage than using a classical approach. This is in line with a dose-dependency study for bromobenzene and paracetamol of Schoonen et al. (2007). Sensitivity with the NMR PCDA technique appeared to be at least four- up to 16-fold higher. Histopathology and clinical chemistry identified only 50% of the compounds as clearly necrotic (bromobenzene, paracetamol, carbon tetrachloride, thioacetamide) or cholestatic (ANIT and ibuprofen). Ibuprofen, a compound showing cholestasis in humans at high dosing, clustered according to the NMR PCDA procedure in between the cholestatic and necrotic compounds. This was confirmed with histopathology, as in two out of five animals, liver necrosis, instead of cholestasis, was identified. In addition, compounds with hardly any histopathological irregularities and biochemistry changes (iproniazid, isoniazid, chlorpromazine, ethinylestradiol, methyltestosterone, phenobarbital, and tetracycline) could easily be distinguished from control by NMR PCDA. Moreover, the negative control compound, mianserine, clustered with the control group. Therefore, the data obtained provides evidence that clustering into hepatotoxic classes with NMR PCDA is now possible.
The NMR PCDA technique can be developed further into a diagnostic screening tool when the segregation for each toxicity class is carefully validated. For such a validation, we suggest to expand the number of compounds per toxicity class to at least six. A leave-one-out cross-validation approach is required to establish a solid base for the screening of unknown compounds. The number of toxicity classes can be extended beyond the mentioned (hepato)toxic classes in our study. Additional advantages of this technique are that it is noninvasive and fast. The experiments for a complete NMR PCDA can be performed within 1–2 days. Of course, also other species than rats may be considered. Note that the reproducibility of proton NMR spectra also is highly dependent on food composition and the feeding patterns of test animals.
Some of the hepatotoxic compounds were also used in other NMR toxicity studies. A few of the changes in NMR signals found in this study were also demonstrated in other studies with bromobenzene (Heijne et al., 2005), ANIT (Amzi et al., 2005; Beckwitt-Hall et al., 1998; Robertson et al., 2000, Shockcor and Holmes, 2002), carbon tetrachloride (Robertson et al., 2000, Shockcor and Holmes, 2002), thioacetamide (Nicholson et al., 1999, Shockcor and Holmes, 2002), and hydrazine (Shockcor and Holmes, 2002). This was especially true for the metabolites dimethylglycine, hippurate, citrate, -ketoglutarate, and succinate.
It is tempting to speculate on the cellular metabolism pathways leading to the assigned biomarker signals. It should be recognized that although the power of the NMR pattern recognition is remarkable, NMR profiling combines complex multidimensional changes. Thus, the observations made in NMR cannot be ascribed merely to liver pathology, as other physiological processes of kidney, heart, and other organs may affect urine patterns. Nevertheless, looking into more detail to the biomarkers observed with NMR urinalysis, the following biochemical observations can be made. The citric acid cycle seems to be hampered during necrosis, cholestasis, and steatosis. The production of citrate, -ketoglutarate, and succinate was reduced in all three categories of pathology, and these organic acids may be more universal markers for cellular pathology. The reduced organic acid levels suggest that oxidative phosphorylation was inhibited and that anaerobic glycolysis was increased. A decrease in the formation of hippurate is another common marker for cellular liver pathology. Decreased oxidative phosphorylation will also lead to a decreased formation of hippurate in rat liver mitochondria. The hippurate formation depends on the metabolism of benzoate. Benzoate is found in blood as a product from benzoic acid, which is formed in the gut by bacterial conversion of quinic acid, taken up as a food additive from fruits and berries (Yavuz et al., 2005). Benzoate is metabolized by benzoyl-CoA synthetase into benzoyl-CoA requiring one molecule of both CoA-SH and ATP. For the next step, Acyl-CoA-glycine N-acetyltransferase is used for the final formation of hippurate from benzoyl-CoA. Hereby, one molecule of glycine is needed and CoA-SH is spliced of again. This whole process depends on a good supply of ATP via the oxidative phosphorylation (Gatley and Sherratt, 1977, Sprague et al., 2004). Thus, the reduced production of ATP may lead to a reduced formation of hippurate during necrosis, cholestasis, and steatosis as observed in the present study.
In conclusion, NMR urinalysis could clearly identify hepatotoxicity at an early stage and additionally distinguish between three major hepatotoxic modes of action. It proved to be more sensitive than classical histopathology and clinical chemistry. Since NMR pattern recognition is also a noninvasive and relatively fast method, it is a very promising and sensitive technique for early screening of investigational compounds (Corcoran, 2005; London and Houck, 2004; Robertson, 2005). Ultimately, the applicability will depend on large-scale validation studies of appropriate design to determine sensitivity and specificity. One of the major remaining challenge is discrimination of metabolic profiles reflecting physiological/adaptive responses and those directly related to pathology and toxicology.
|
SUPPLEMENTARY DATA |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONSUPPLEMENTARY DATAREFERENCES |
|---|
Supplementary data are available online at http://toxsci.oxfordjournals.org/.
|
ACKNOWLEDGMENTS |
|---|
We would like to thank Dr S.H. Moolenaar for her input into the scientific discussion and Ms N. Bisseling and Mr J. van Orsouw for the NMR graphics.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONSUPPLEMENTARY DATAREFERENCES |
|---|
Accanito L, Figuerora C, Pirazzo M, Solis N. Enhanced biliary excretion of canalicular membrane enzymes in estradiol-induced and obstructive cholestasis, and effects of different bile acids in the isolated perfused rat liver. J. Hepatol. (1995) 22:658–670.[CrossRef][Web of Science][Medline]
Accanito L, Pirazzo M, Solis N, Koenig CS, Vollrath V, Chianale J. Modulation of hepatic content and biliary excretion of P-glycoproteins in hepatocellular and obstructive cholestasis in the rat. J. Hepatol. (1996) 25:349–361.[CrossRef][Web of Science][Medline]
Agar NS, Rae CD, Chapman BE, Kuchel PW. 1H NMR spectroscopic survey of plasma and erythrocytes from selected marsupials and domestic animals of Australia. Comp. Biochem. Physiol. (1991) 99:575–597.[CrossRef][Medline]
Azmi J, Griffin JL, Shore RF, Holmes E, Nicholson JK. Chemometric analysis of biofluids following toxicant induced hepatotoxicity: A metabonomic approach to distinguish the effects of 1-naphthylisothiocyanate from its products. Xenobiotica (2005) 35:839–852.[CrossRef][Web of Science][Medline]
Beckwith-Hall BM, Nicholson JK, Nicholls AW, Foxall PJD, Lindon JC, Connor SC, Abdi M, Connelly J, Holmes E. Nuclear magnetic resonance spectroscopic and principal components analysis investigation into biochemical effects of three model hepatoxins. Chem. Res. Toxicol. (1998) 11:260–272.[CrossRef][Web of Science][Medline]
Bollard ME, Stanley EG, Lindon JC, Nicholson JK, Holmes E. NMR-based metabonomic approaches for evaluating physiological influences on biofluid composition. NMR Biomed. (2005) 18:143–162.[CrossRef][Web of Science][Medline]
Caldwell GW, Ritchie DM, Masucci JA, Hageman W, Yan Z. The new pre-clinical paradigm: Compound optimization in early and late phase drug discovery. Curr. Top. Med. Chem. (2001) 1:353–366.[CrossRef][Medline]
Clayton TA, Lindon JC, Cloarec O, Antti H, Charuel C, Hanton G, Provost J-P, Le Net J-L, Baker D, Walley RJ, et al. Pharmaco-metabonomic phenotyping and personalized drug treatment. Nature (2006) 440:1073–1077.[CrossRef][Medline]
Cloarec O, Dumas ME, Trygg J, Craig A, Barton RH, Lindon JC, Nicholson JK, Holmes E. Evaluation of the orthogonal projection on latent structure model limitations caused by chemical shift variability and improved visualisation of biomarker changes in 1NMR spectrocopic metabonomic studies. Anal. Chem. (2005) 77:517–526.[Medline]
Corcoran O. NMR spectroscopy as a versatile analytical platform for toxicology research. In: Handbook of Toxicogenomics: Strategies and Applications—Borlak J, ed. (2005) Weinheim: Wiley-VCH Verlag GmbH & Co KgaA. 163–183. chapter 8.
Cuatrecasas P. Drug discovery in jeopardy. J. Clin. Invest. (2006) 116:2837–2842.[CrossRef][Web of Science][Medline]
Espina JR, Shockcor JP, Herron WJ, Car BD, Contel NR, Ciaccio PJ, Lindon JC, Holmes E, Nicholson JK. Detection of in vivo biomarkers of phospholipidosis using NMR-based metabonomic approaches. Magn. Reson. Chem. (2001) 39:559–565.[CrossRef][Web of Science]
Fan T. Metabolite profiling by one- and two-dimensional NMR analysis of complex mixtures. Prog. NMR Spectr. (1996a) 28:161–219.
Fan T. Recent advances in profiling plant metabolites by multinuclear & multidimensional NMR. In: Nuclear Magnetic Resonance in Plant Biology—Shachar-Hill Y, Pfeffer PE, eds. (1996b) Rockville, MD: American Society of Plant Physiologists. 181–254.
Frezza M, Tritapepe R, Pozzato G, Di Padova C. Prevention of S-adenosylmethionine of estrogen-induced hepatobiliary toxicity in susceptible women. Am. J. Gastroenterol. (1988) 83:1098–1102.[Web of Science][Medline]
Gatley SJ, Sherratt HSA. The synthesis of hippurate from benzoate and glycine by rat liver mitochondria. Submitochondrial localization and kinetics. Biochem. J. (1977) 166:39–47.[Web of Science][Medline]
Gopinath C, Prentice DE, Lewis DJ. Atlas of Experimental Toxicological Pathology (1987) Lancaster: MTP Press Limited. 122–126.
Govindaraju V, Young K, Maudsley AA. Proton NMR chemical shifts and coupling constants for brain metabolites. NMR Biomed. (2000) 13:129–153.[CrossRef][Web of Science][Medline]
Gower JC, Dijksterhuis GB. Procrustus Problems (2004) New York: Oxford University Press 2004. Oxford statistical science series, 30.
Griffin JL. The potential of metabonomics in drug safety and toxicology. Drug Discov. Today Technol. (2004) 1:285–293.[CrossRef]
Gupta DN. Acute changes in the liver after administration of thioacetamide. J. Pathol. Bacteriol. (1956) 72:183–192.[CrossRef][Web of Science][Medline]
Héberger K, Andrade JM. Procrustus and pair-wise correlation: A parametric and a non-parametric method for variable selection. Croatica Chem. Acta (2004) 77:117–125.
Heijne WH, Lamers RJAN, van Bladeren PJ, Groten JP, van Nesselrooij JH, van Ommen B. Profiles of metabolites and gene expression in rats with chemically induced hepatic necrosis. Toxicol. Pathol. (2005) 33:425–433.[Web of Science][Medline]
Holmes E, Nicholls AW, Lindon JC, Ramos S, Spraul M, Neidig P, Connor SC, Connelly J, Damment SJP, Haselden J, Nicholson JK. Development of a model for classification of toxin-induced lesions using 1NMR spectroscopy of urine combined with pattern recognition. NMR Biomed. (1998) 11:235–244.[CrossRef][Web of Science][Medline]
Ishak KG, Zimmerman HJ. Hepatotoxic effects of the anabolic/androgenic steroids. Semin. Liver Dis. (1987) 7:230–236.[Web of Science][Medline]
Klenø TG, Kiehr B, Baunsgaard D, Sidelmann UG. Combination of ‘omics’ data to investigate the mechanism(s) of hydrazine-induced hepatotoxicity in rats and to identify potential biomarkers. Biomarkers (2004) 9:116–138.[CrossRef][Web of Science][Medline]
Kreek MJ, Weser E, Sleisenger MH, Jeffries GH. Idiopathic cholestasis of pregnancy. The response to challenge with the synthetic estrogen, ethinyl estradiol. N. Engl. J. Med. (1967) 277:1391–1395.[Web of Science][Medline]
Lehmann-McKeeman LD, Car BD. Metabonomics: Application in predictive and mechanistic toxicology. Toxicol. Pathol. (2004) 32(Suppl. 2):94–95.
Lewis JHL. Drug-induced liver disease. Gastroenterol. Clin. North Am. (2000) 84:1275–1311.
Lindon JC, Holmes E, Nicholson JK. Metabonomics techniques and applications to pharmaceutical research & development. Pharm. Res. (2006) 23:1075–1088.[CrossRef][Web of Science][Medline]
Lindon JC, Keun HC, Ebbels TMD, Pearce JKT, Holmes E, Nicholson JK. The consortium for metabonomic toxicology (COMET): Aims, activities and achievements. Pharmacogenomics (2005) 6:691–699.[CrossRef][Web of Science][Medline]
London RE, Houck DR. Introduction to metabolomics and metabolic profiling. In: Toxicogenomics: Principles and Applications—Hamadeh HK, Afshari CA, eds. (2004) New York: John Wiley & Sons, Inc. 299–340. chapter 14.
Mortishire-Smith RJ, Skiles GL, Lawrence JW, Spence S, Nicholls AW, Johnson BA, Nicholson JK. Use of metabonomics to identify impaired fatty acid metabolism as the mechanism of a drug-induced toxicity. Chem. Res. Toxicol. (2004) 17:165–173.[CrossRef][Web of Science][Medline]
Nelson SD, Mitchell JR, Snodgrass WR, Timbrell JA. Hepatotoxicity and metabolism of iproniazid and isopropylhydrazine. J. Pharmacol. Exp. Ther. (1978) 206:574–585.
Nelson SD, Mitchell JR, Timbrell JA, Snodgrass WR, Concoran GB. Isoniazid and iproniazid: Activation of metabolites to toxic intermediates in man and rat. Science (1976) 193:901–903.
Nicholson JK, Connelly J, Lindon JC, Holmes E. Metabonomics: A platform for studying drug toxicity and gene function. Nat. Rev. (2002) 1:153–161.
Nicholson JK, Lindon JC, Holmes E. Metabonomics: Understanding the metabolic response of living systems to pathophysiological stimuli via multivariate statistical analysis of biological NMR spectroscopic data. Xenobiotica (1999) 29:1181–1189.[CrossRef][Web of Science][Medline]
Reyes H, Ribalta J, Gonzalez MC, Segovia N, Oberhauser E. Sulfobromophtalein clearance tests before and after ethinyl estradiol administration, in women and men with familial history of intrahepatic cholestasis of pregnancy. Gastroenterology (1981) 81:226–231.[Web of Science][Medline]
Roberts SM, James RC, Franklin MR. Hepatotoxicity: Toxic effects on the liver. In: Principles of Toxicology—Williams PL, James RC, Roberts SM, eds. (2000) 2nd ed. New York: John Wiley & Sons, Inc. 111–128. chapter 5.
Robertson DG. Metabonomics in toxicology: A review. Toxicol. Sci. (2005) 85:809–822.
Robertson DG, Reily MD, Sigler RE, Wells DF, Paterson DA, Braden TK. Metabonomics: Evaluation of nuclear magnetic resonance (NMR) and pattern recognition technology for rapid in vivo screening of liver and kidney toxicants. Toxicol. Sci. (2000) 57:326–337.
Sánchez Pozzi EJ, Crozenci FA, Pellegrino JM, Catania VA, Luquita MG, Roma MG, Rodriguez Garay EA, Mottino AD. Ursodeoxycholate reduces ethinylestradiol glucuronidation in the rat: Role of prevention in estrogen-induced cholestasis. J. Pharmacol. Exp. Ther. (2003) 306:279–286.
Schoonen WGEJ, Kloks CPAM, Ploemen J-PHTM, Smit MJ, Zandberg P, Horbach GJ, Mellema JR, Thijssen-van Zuylen C, Tas AC, van Nesselrooij JHJ, et al. Sensitivity of 1H NMR analysis of rat urine in relation to toxicometabonomics. Part I: Dose dependent toxic effects of bromobenzene and paracetamol. Toxicol. Sci. (2007) doi:10.1093/toxsci/kfm076.
Schuster D, Laggner C, Langer T. Why drugs fail—a study on side effects in new chemical entities. Curr. Pharm. Des. (2005) 11:3545–3559.[CrossRef][Web of Science][Medline]
Shockcor JP, Holmes E. Metabonomic applications in toxicity screening and disease diagnosis. Curr. Top. Med. Chem. (2002) 2:35–51.[CrossRef][Medline]
Simon FR, Fortune J, Iwahashi M, Gartung C, Wolkoff A, Sutherland E. Ethinyl estradiol cholestasis involves alterations in expression of liver sinusoidal transporters. Am. J. Physiol. (1996) 271G:1043–1052.
Smith DA, Schmid EF. Drug withdrawals and the lessons within. Curr. Opin. Drug Discov. Devel. (2006) 9:38–46.[Web of Science][Medline]
Sprague CL, Phillips LA, Young KM, Elfarra AA. Species and tissue differences in the toxicity of 3-butene-1,2-diol in male Sprague-Dawley rats and B6C3F1 mice. Toxicol. Sci. (2004) 80:3–13.
Vandenberghe J. Hepatotoxicology: Mechanisms of liver toxicity and methodological aspects. In: Toxicology, Principles and Applications—Niesink RJM, de Vries J, Hollinger MA, eds. (1996) New York: CRC Press. 702–723. chapter 23.
Vogels JTWE, Tas AC, Venekamp J, van der Greef J. Partial linear fit: A new NMR spectroscopy preprocessing tool for pattern recognition applications. J. Chemometrics (1996) 10:425–438.[CrossRef]
Welder AA, Robertson JW, Melchert RB. Toxic effects of anabolic-androgenic steroids in primary rat hepatic cell cultures. J. Pharmacol. Toxicol. Methods (1995) 33:187–195.[CrossRef][Web of Science][Medline]
Westaby D, Ogle SJ, Paradinas FJ, Randell JB, Murray-Lyon IM. Liver damage from long-term methyltestosterone. Lancet (1977) 2:262–263.[Medline]
Yavuz A, Tetta C, Ersoy FF, D'Intini V, Ratanarat R, De Cal M, Bonello M, Bordoni V, Salvatori G, Andrikos E, et al. Uremic toxins: A new focus on an old subject. Semin. Dial. 18:203–211.
CiteULike 
Connotea 
Del.icio.us What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||