NOVEMBER 2003 VOLUME 16, NUMBER 11 © Copyright 2003 by the American Chemical Society
Communications NMR Spectroscopic-Based Metabonomic Studies of Urinary Metabolite Variation in Acclimatizing Germ-Free Rats Andrew W. Nicholls,*,†,‡ Russell J. Mortishire-Smith,§ and Jeremy K. Nicholson† Biological Chemistry, Faculty of Medicine, Imperial College London, Sir Alexander Fleming Building, South Kensington, London SW7 2AZ, United Kingdom, and Merck Sharp and Dohme Research Laboratories, Neuroscience Research Centre, Terlings Park, Eastwick Road, Harlow, Essex CM20 2QR, United Kingdom Received February 17, 2003
Understanding metabolic variation in “normal” animals is critical to the evaluation of druginduced metabolic perturbation related to toxicity or pharmacology. NMR spectroscopic-based metabonomic methods were used to evaluate the acclimatization pathways of germ-free (axenic) rats to standard laboratory conditions concomitant with the associated development of gut microfloral communities. Urine samples from male Fischer 344 germ-free rats were collected over 21 days following introduction to a standard laboratory environment and analyzed using NMR spectroscopy. NMR spectra were data-reduced and analyzed using principal component analysis to visualize the changes in the host metabolic trajectory over the course of the study. At days 2 and 6 of the acclimatization process, there were marked episodes of glycosuria. In comparison to the concentrations in the 0-6 h samples, there was a reduction in the level of the tricarboxylic acid cycle intermediates (citrate, 2-oxoglutarate, and succinate) from 6 h to day 6, after which there was a sustained increase until the end of the study. The concentrations of hippurate and trimethylamine N-oxide increased over the course of the study in comparison to the levels at 0-6 h, with the most pronounced increase in the former between days 17 and 21. Phenylacetylglycine levels increased after 6 h whereas 3-hydroxypropionic acid was observed at day 12 and increased up to day 17. By day 21, the urinary metabolic profile was within the control range when compared to historical data, implying the establishment of a stable gut microflora. Although the metabolic alterations caused by the microbial alterations were not as substantial as those from metabolic dysfunction, their presence does have an effect on the interpretation of the profiles, the state of the animal, and the mechanism for the cause of such alterations. Furthermore, the use of oral drug delivery will have an effect on the microbial state, not only as a direct influence of the drug but also from it’s associated vehicle. Such effects are likely to be observed particularly in the area of preclinical investigation where the data from these studies are of particular relevance.
Introduction Recent research has led to the development of several novel methodologies for the characterization of biological * To whom correspondence should be addressed. Tel: +44 207 594 3079. Fax: +44 207 594 6818. E-mail:
[email protected]. † Imperial College London. ‡ Current address: Metabometrix Ltd., RSM, Prince Consort Road, London, SW7 2BP, U.K. § Merck Sharp and Dohme Research Laboratories.
variability following drug-induced toxicity. One of these tools, metabonomics (1, 2), uses information rich analytical methods such as high-resolution NMR spectroscopy coupled with multivariate statistical methods for the analysis of biological fluids and intact tissue samples. Metabonomics has been applied to the study of metabolic effects of drugs and toxins in vivo and is defined as “the measurement of the multiparametric metabolic responses of biological systems to pathophysiological processes or
10.1021/tx0340293 CCC: $25.00 © 2003 American Chemical Society Published on Web 10/25/2003
1396 Chem. Res. Toxicol., Vol. 16, No. 11, 2003
genetic modification” (1). Such metabonomic studies are complementary to proteomic, transcriptomic, and genomic tools and provide a useful representation of the integrated metabolic state of the whole animal. NMR spectroscopy, in particular 1H NMR spectroscopy, has been used for many years to study the toxicity of xenobiotics (3-6). In such studies, 1H NMR spectroscopy has been highly effective in the identification of novel biomarkers that may be indicative of specific organ damage including creatine as a marker of testicular toxicity (7-9) and PAG1 as a marker of phospholipidosis (10). 1H NMR spectra of biofluids acquired on high-field NMR spectrometers (>500 MHz 1H observation frequency) are complex and potentially contain structural and quantitative information on many hundreds of endogenous metabolites. To effectively examine 1H NMR spectra containing such large quantities of information, data reduction and multivariate statistical analysis techniques, such as PCA and more sophisticated supervised tools, have been applied (11-13). PCA is used to calculate a new smaller set of orthogonal variables from a linear combination of a large set of correlated variables, while still maintaining the maximum level of variability from the original data. This permits the simple visualization of separation or clustering between samples, caused by compound-induced metabolic perturbations (5, 6, 14) or genetic variability and metabolic phenotypes (15) using two- or three-dimensional (2D, 3D) plots of the principal components (PCs, scores). The weightings (loadings) given to each variable in calculating the PCs allow for the identification of those variables of greatest influence to the separation/clustering and, hence, the isolation of biomarkers of toxicity or disease states (16). Such multivariate statistical methods provide a more efficient and rigorous examination of NMR spectral data than visual interpretation and have been applied to studies into druginduced toxicity (5, 6, 10, 16), genetic variation (15), and normal physiological variation (17). It is now accepted that the health and integrated metabolic state of a mammal can be highly dependent on the composition of its endogenous gut microflora, the gut microbiolome as it has recently been termed (18). Alterations to this transgenomic metabolic axis can result in major metabolic and even disease consequences for the host. Given that many oral drugs, including those not primarily in an antibiotic class, may modulate or reselect the gut microflora, many metabolic, and consequently proteomic and gene expression changes, may result from treatments that are unrelated to the mechanism of drug action or toxicity. Even drugs given by nonoral routes could generate biliary excreted metabolites that have the potential to affect the microflora. Phipps and co-workers have shown that alterations to the diet of an animal had marked effects on urinary composition, highlighting the variation in excretion of hippurate and chlorogenic acid metabolites, which are primarily of microbial origin (19). Previous work using germ-free rats has shown that following fecal inoculation for 2 weeks, there was an increase in the rate of urinary excretion of benzoic acid, 1Abbreviations: PCA, principal component analysis; FID, free induction decay; TCA, tricarboxylic acid cycle; PAG, phenylacetylglycine; TSP, sodium 3-trimethylsilyl-(2,2,3,3-2H4)-1-propionate; TOCSY, total correlation spectroscopy; TMAO, trimethylamine-N-oxide; 3-HPPA, 3-hydroxypropionic acid; 4-HPPA, 4-hydroxypropionic acid; TMA, trimethylamine.
Communications
phenylacetic acid, 3-hydroxyphenylacetic acid, and 4-hydroxyphenylacetic acid (20). The present study was undertaken to investigate the changes with time to the host urinary profile from male germ-free rats over a 21 day period during acclimatization to the “normal” animal house environment with concomitant exposure to a range of commensal gut enterobacteria and potential pathogens. These changes reflect both the progressive colonization/redistribution of gut microflora and the changing health of the animal in relation to its enterobacterial load.
Materials and Methods Sample Collection and Preparation. All experiments were conducted according to the U.K. Animals (Scientific Procedures) Act 1986 and its associated amendments. Fischer 344 male germ-free rats (n ) 3, 234-257 g, BIBRA, U.K.) were bred in a sterile facility and maintained on irradiated food (Rat and Mouse No. 3 breeding diet, SDS Ltd., Witham, Essex, U.K.) and irradiated water supplemented with vitamin K. In the sterile environment, the animals were housed communally in solid bottom cages, and this was maintained once removed into the normal environment, with the exception of periods of urine collection, when each animal was individually housed in an all glass metabolism cage. At all times in the course of the study, food (Rat and Mouse No. 1 maintenance diet, SDS Ltd., Witham, Essex, U.K.) and water (supplemented with vitamin K for the first 7 days) were provided ad libitum. Urine was collected onto dry ice over 0-6 h and 6-24 h on day 1 and for 24 h periods on days 2, 4, 6, 9, 12, 15, 17, and 21. Samples were stored at -20 °C until urinalysis. No urine samples were obtained prior to environmental exposure due to the lack of sterile apparatus for urine collection. The urine samples for NMR spectroscopy were made up from a 2:1 mixture of urine to phosphate buffer (0.2 M Na2HPO4/0.2 M NaH2PO4, pH 7.4). Aliquots of the resulting mixture (500 µL) were placed in 5 mm NMR tubes to which 50 µL of a solution of TSP in D2O was added (final concentration ) 1mM). The D2O plus TSP addition provided both a chemical shift reference (δ 0.0) and a field frequency lock signal. NMR Spectroscopic Analysis. One-dimensional (1D) 1H NMR spectra were acquired at 500.13 MHz on a Bruker DRX500 spectrometer using a standard presaturation pulse sequence for water suppression (21). NMR spectra were acquired using 64 scans into 64k points with a spectral width of 10080.6 Hz, an acquisition time of 3.25 s, and thus a total pulse recycle delay of 6.25 s. The FIDs were multiplied by an exponential function corresponding to a 0.3 Hz line broadening prior to Fourier transformation (FT). Two-dimensional NMR spectroscopic experiments were acquired on selected samples to aid identification of the low molecular weight components. To confirm the identification of endogenous metabolites, 1H-1H TOCSY spectra (22) were acquired for selected samples using an MLEV-17 spinlock sequence of 60 ms duration, with 128 scans collected into 2k data points per increment for 128 increments and F1 and F2 spectral widths of 7183 Hz. The FIDs were multiplied by a shifted sine-bell squared function in both dimensions prior to FT. Analysis of NMR Spectral Data. All NMR spectra were phased and baseline corrected in XWINNMR (Bruker Analytik GmbH, Rheinstetten, Germany) and data-reduced using AMIX (Bruker Analytik GmbH) to integrated regions 0.04 ppm wide corresponding to the region δ 10.0 to 0.2 ppm. The region δ 6.04.5 in the urine spectra was set to zero integral value for the purposes of pattern recognition analysis to remove the variability in presaturation of the water resonance and crossrelaxation effects on the urea signal. The regions attributed to TMAO were combined to create a single variable to remove chemical shift variation induced by intersample pH variation. Each data point was normalized to the sum of its row (i.e., to the total integral for each NMR spectrum) to compensate for
Communications
Chem. Res. Toxicol., Vol. 16, No. 11, 2003 1397
variation in urine volumes, and the values of all variables were mean-centered prior to PCA using SIMCA-P (Umetrics Inc., Umeå, Sweden). Scores plots of the PCs were constructed to visualize any inherent separation of the urine samples with time, and from the values of the PC loadings (which indicated the importance of each variable to the separation), the NMR spectral regions and endogenous metabolites were identified. The means for PC1 and PC2 were calculated and plotted to show the metabolic trajectory with time following removal from the sterile environment. On the basis of the multivariate statistical analysis and visual inspection of the NMR spectra, the major metabolic markers were identified and quantified at each time point from the integrals of the NMR resonances of the metabolites. Each metabolite was quantified to the resonance for TSP and adjusted for dilution effects, variations in urinary volume, and the length of collection times. The logarithm of each concentration was calculated, and the mean data at each time point were compared using a one way ANOVA. Where a significant difference between the mean values was found, Dunnett’s test was applied to identify the time points of significant variation (GraphPad InStat version 3.01, GraphPad Software, San Diego, CA). The data from the current study were compared to a set of control data from the literature (5) to determine the similarity of each time period to a conventional control sample. The mean values for the first two PCs were calculated and plotted for each control group (n ) 200), and the 95% confidence interval was calculated from the standard deviation.
Results Shown in Figure 1 are a series of 500 MHz 1H NMR spectra of whole rat urine samples from one animal at selected time points. Resonance assignments were based on chemical shifts, coupling constants, and relative intensities as found in previous reports (6, 21, 23) and from analysis of 1H-1H TOCSY NMR experiments. Time “zero” represents the point of removal from the sterile environment (no samples were available immediately prior to environmental exposure). The NMR spectral region δ 9.5-5.8 has been reported to contain resonances from several dietary derived constituents, including metabolites of chlorogenic acid (19). Expansions of this region from the spectra in Figure 1 are shown in Figure 2, and numerous resonances were noted to vary with time including 3-HPPA, 4-HPPA, and hippurate. Each was observed at distinct time points, with 4-HPPA appearing in the urine at 6-24 h and 3-HPPA appearing by day 9. Neither of these metabolites, both of which derive from chlorogenic acid, were observed in the urine samples taken on day 21. At this time point, hippurate, which was not formed from chlorogenic acid, now dominated the aromatic region of the NMR spectrum. The 3-HPPA and hippurate resonances were quantified and compared to the concentrations in the 6 h samples (Figure 3). The concentration of 3-HPPA was significantly different (p < 0.01) on days 12, 15, and 17, while although hippurate was detected in one of the day 15 samples, it was only significantly elevated (p < 0.01) on day 21. From comparison of the chemical shifts of TMAO (δ3.29, s) and creatinine (δ4.09, s; δ 3.05, s) to literature data (24), the urines were determined to be acidic at pH 5.8. The NMR spectra of the 0-6 h urine samples showed many similarities to the NMR spectra of whole urine from conventional animals, in particular those resonances in the aliphatic region. The resonances of greatest intensity observed in the NMR spectra were assigned to the TCA cycle intermediates citrate, 2-oxoglutarate, and succinate. All three of these compounds were observed to decrease
in concentration from 6 h to 6 days, with a subsequent increase back to initial levels by the end of the study. By statistical comparison to the initial concentration at 6 h, only the succinate increase on day 21 was considered significant (p < 0.01). Both R- and β-glucose anomeric protons were visible in the 0-6 h samples (δ 5.22 and δ 4.65, respectively). A significant increase (p < 0.01) in the concentration of glucose was noted at 48 h and on day 6, with increases also observed on days 9, 15, 17 (p < 0.05), and 21 (p < 0.01). The resonances for taurine (δ 3.42, t; δ 3.26, t) were identified in the NMR spectra of the 0-6 h sample. Taurine was not observed in any of the subsequent time periods due to extensive overlap from the resonances of other endogenous components. The concentration of TMAO was observed to increase on days 4 and 6 and to decrease on days 9 and 12. However, this increase was not statistically significant (p < 0.05) until day 17 after which the TMAO concentration remained elevated until the end of the study. The 6-24 h samples were observed to have an increase in intensity for a singlet at δ 8.45 and the resonances at δ 7.45 and δ 7.38, which were assigned to formate and PAG, respectively. The concentration of PAG although maximal at 24 h after initial exposure remained significantly increased (p < 0.01) until day 21. The concentration of urinary formate was significantly different from the 6 h samples at 24 h (p < 0.01) and at 48 h (p < 0.05). Typical NMR spectra of urine from conventional animals contain resonances from hippurate, and the NMR spectra of the day 21 urines were observed to be visually of greatest similarity to published data (5). The chemical shifts for the observed metabolites are listed in Table 1. Figure 4 shows the trajectory plot of the PC1 vs PC2 scores for each time period following exposure (panel A) and the corresponding loadings plot (panel B). The trajectory of the samples showed that throughout the study period there was a gradual change in the NMR spectral profile of urine with two periods of notable variation. After 6 h, there was a movement to the lefthand side of the plot with a maximum shift reached by 48 h. Following this, there was a shift back to the center of the plot by the fourth day, with a subsequent return to the 48 h position at day 6. After this, the remaining time points were observed to map to the center of the plot. From examination of the loadings plot (Figure 4B), this shift at 48 h and 6 days was attributed to a large increase in the concentration of glucose and a decrease in the concentration of citrate due to the relative positions of the NMR spectral regions in the loadings plot. An increase in intensity was indicated in those NMR spectral regions in a similar position in the loadings plot to the positions of the samples in the scores plot, i.e., to the left. Conversely, those NMR spectral regions that mapped to the opposite side of the loadings plot as compared to the mapping of the samples in the scores plot, i.e., the right, were decreased. These alterations were confirmed from analysis of the corresponding NMR spectra. The shift over the 21 day period was largely influenced by a reduction in the level of 2-oxoglutarate and the increase in both hippurate and TMAO based on the loadings in Figure 4B and the corresponding NMR spectra. The NMR spectra of the day 21 samples were visually very similar to standard spectra of whole rat urine as previously published (5). These data from the current study were also compared using pattern recognition to a set of urine samples from a group of conventional control
1398 Chem. Res. Toxicol., Vol. 16, No. 11, 2003
Communications
Figure 1. Stack plot of 500 MHz 1H NMR spectra of whole rat urine at selected time points. The assignments were based on literature data and 2D NMR experiments.
Communications
Chem. Res. Toxicol., Vol. 16, No. 11, 2003 1399
Figure 2. Stack plot of expansions (δ 9.5-5.5) from the 500 MHz 1H NMR spectra of whole rat urine at selected time points. The assignments were based on literature data and 2D NMR experiments.
1400 Chem. Res. Toxicol., Vol. 16, No. 11, 2003
Communications
Figure 3. Bar chart of the mean concentration and standard deviations of the main endogenous metabolites identified from the NMR spectra at each time period. Statistically significant alterations in concentration were determined by comparison to the 0-6 h samples; *p < 0.05, **p < 0.01.
rats (Han Wistar and Sprague-Dawley). Initial PCA of the combined data indicated that the pH effect on the chemical shift of TMAO lead to a separation of the data. Furthermore, the urinary concentration for the TCA intermediates citrate, 2-oxoglutarate, and succinate have been reported previously to vary between rat strains (5). Therefore, the chemical shift regions attributed to the TCA intermediates were excluded from the analysis while the two regions for TMAO (δ 3.30 and δ 3.26) were combined into one region. The scores plot of PC1 vs PC2 is shown in Figure 5A and indicated that the samples from day 21 mapped within the confidence interval for both of the literature control groups (p < 0.05). By day 21, the acclimatization of the animals was sufficient for them to be considered as conventional control animals. The major metabolite influencing the mapping of the day 21 samples with the control data was hippurate, as determined from the loadings plot shown in Figure 5B. The large spread of the day 17 data arose from an increased level of hippurate in one of the three animals. The separation of the 48 h and day 6 samples from the rest of the data remained due to increases in the glucose resonances.
Discussion To understand the alterations observed in 1H NMR spectra following either environmental, toxicological, or genetically induced metabolic change, it is important to identify other potentially important contributions to the overall variability. In this study, we have shown that variation, probably induced from gut flora changes, occurs to the urinary metabolic profile following the exposure of a germ-free rat to a normal environment.
Table 1. Assignments of the Key Endogenous Components in Urine from Germ-Free Ratsa endogenous metabolite alanine allantoin citrate creatine creatinine dimethylglycine (DMG) formate fumarate R-glucose β-glucose hippurate 3-HPPA
4-HPPA lactate 1-methylnicotinic acid 2-oxoglutarate PAG succinate taurine threonine TMAO uracil
1H
chemical shifts (δ), multiplicity, and assignment
3.79 (q, CH), 1.48 (d, CH3) 5.40 (s CH) 2.72 (d, 1/2-CH2), 2.56 (d, 1/2-CH2) 3.94 (s, CH2), 3.04 (s, CH3) 4.05 (s, CH2), 3.05 (s, CH3) 2.93 (s, CH3), 3.72 (s, CH2) 8.45, (s, CH) 6.53 (s, CH) 5.22 (d, CH), 3.79 (dd, H6), 3.86 (dd, H5), 3.73 (t, H3), 3.53 (dd, H2), 3.41 (t, H4) 4.65 (d, CH), 3.91 (dd, H6), 3.48 (m, H3,H5), 3.41 (dd, H4), 3.25 (t, H2) 7.83 (d, H2 and H6), 7.64 (t, H4), 7.55 (t, H3 and H5), 3.97 (d, CH2) 7.27 (t, CH), 6.92 (dd, CH), 6.80 (s, CH), 6.76 (d, CH), 2.84 (t, CH2), 2.48 (t, CH2) 7.18 (d, CH), 6.85 (d, CH), 2.81 (t, CH2), 2.45 (t, CH2) 4.11 (q, CH), 1.33 (d, CH3) 9.28 (s, H2), 8.97 (d, H6), 8.90 (d, H4), 8.19 (t, H5), 4.49 (s, CH3) 3.01 (t, CH2), 2.45 (t, CH2) 7.42 (t, CH), 7.35 (t, CH), 7.35 (d, CH), 3.75 (d, CH2), 3.67 (s, CH2) 2.42 (s, CH2) 3.43 (t, CH2), 3.26 (t, CH2) 1.34 (d, CH3) 3.26 (s, CH3) 7.54 (d, CH), 5.81 (d, CH)
a Key: s, singlet; d, doublet; dd, doublet of doublets; t, triplet; q, quartet; qu, quintet; m, multiplet.
Communications
Chem. Res. Toxicol., Vol. 16, No. 11, 2003 1401
Figure 4. (A) Plot of PC1 vs PC2 for the PC scores of the three animals at each time period. Each animal is denoted by a different symbol. The arrow shows the trajectory formed from the mean of each time period. (B) Plot of the eigenvector loadings for PC1 and PC2 corresponding to the mean scores in the trajectory plot. Key: 6 h, 0-6 h; 24 h, 6-24 h; 48 h, 24-48 h; 4 d, 4 days; 6 d, 6 days; 9 d, 9 days; 12 d, 12 days; 15 d, 15 days; 17 d, 17 days; and 21 d, 21 days.
This represents an extreme example of such variation, and the effects that intestinal microflora can have on the urinary profile can be marked. If the experimental procedure either directly or indirectly affects these microbial species, then any analyses must consider such
extracorporal influences on the resulting urinary composition. Variation Due to Infection. From direct visual inspection of the NMR spectra for each of the time periods, it was clear that a distinct series of events had
1402 Chem. Res. Toxicol., Vol. 16, No. 11, 2003
Communications
Figure 5. (A) Scores plot of PC1 vs PC2 for the scores data for each time point in the current study and the mean scores data for conventional control data from a previously published study (5) with the NMR regions attributed to TMAO combined and those for the TCA intermediates excluded from analysis. The confidence interval ellipse (p < 0.05) is shown for the two conventional data groups. The data from the 21 day time point are mapped within the confidence intervals (p < 0.05) for the conventional control data. (B) Plot of the eigenvector loadings for PC1 and PC2 corresponding to the scores plot in panel A. Key: 6 h, 0-6 h; 24 h, 6-24 h; 48 h, 24-48 h; 4 d, 4 days; 6 d, 6 days; 9 d, 9 days; 12 d, 12 days; 15 d, 15 days; 17 d, 17 days; 21 d, 21 days; Scon, control SpragueDawley; and Hcon, control Han Wistar.
Communications
occurred, including two periods of glycosuria, coinciding with a low concentration of the TCA cycle intermediates. These effects have been previously reported following the dosing of known nephrotoxic compounds, but in such cases, substantial organic and amino aciduria was also observed (25). Therefore, it is most likely that the observed effects resulted from a functional variation in the mammalian biochemistry rather than a structural variation in the kidney. Because the germ-free animals had not been previously exposed to any pathological microbial species, there is a possibility of an infectionderived biochemical response. Further work involving histopathological data would be required to confirm the exact origin of this glycosuria. Variation Due to Microbial Colonization. We observed in this study an increase in 3-HPPA between days 12 and 17, which by day 21, had been replaced by an elevation in hippurate. Variations to the urinary concentration of both of these metabolites have been reported when gut flora were introduced to germ-free animals via fecally contaminated food (20). Studies have also shown that the concentration of these two metabolites can vary in a manner dependent on the diet of the animal, and this has been suggested to be linked to changes in the distribution of intestinal flora (19). Because in the current work the diet was kept the same for the whole study, it would seem most likely that the source of hippurate/3-HPPA variation was colonization and subsequent redistribution of the gut flora. PAG was noted at 6 h, with a maximum concentration at 24-48,h and a sustained elevation throughout the remainder of the study. The concentration of PAG in the NMR spectra of whole urine has been reported to be increased in cases of drug-induced phospholipidosis (10), but the maximum concentration measured in the present study was comparable to the predose concentrations in this previous work. It was likely that the extent of increase in PAG concentration observed in the current work could be attributed to variability in the gut flora. However, a higher concentration of PAG may be indicative of either drug-induced toxicity or a drug effect on the gut microflora. Studies are ongoing to determine the quantitative contributions to PAG excretion from both mammalian and microbial sources. Although a chemical shift variation for TMAO was noted in the current study, it was still apparent that the urinary concentration of this metabolite had increased throughout the course of the study. TMAO can either come from the oxidation of TMA, which is derived from dietary constituents, or directly from intestinal bacteria (26). No alterations were made to the diet throughout the course of the study so the elevated excretion of TMAO may have been due to an increased supply from intestinal bacteria with little or no mammalian metabolic removal. An increase in urinary TMAO has been observed in previous studies, most commonly in cases of drug-induced nephrotoxicity (27, 28), and in the current work, two metabolic episodes were observed that were attributed to metabolic dysfunction in the kidney. Further studies are in progress to characterize the variation in TMAO concentration caused by microbial supply and renal release following metabolic dysfunction. Analysis of the NMR spectra of the day 21 samples showed an almost identical metabolic profile to that from literature data for a conventional control animal (24). Using PCA, the day 21 samples were shown to map in
Chem. Res. Toxicol., Vol. 16, No. 11, 2003 1403
the same scores space as a set of literature control data. This analysis also showed that the difference between the literature control group and the 0-6 h samples was not as great as that for the 48 h and day 6 samples. Therefore, although variability in the gut microflora distribution did lead to separation from control, the changes resulting from a direct alteration of the mammalian biochemistry separated to a greater extent. However, because gut microflora variability may occur within a study and between studies, it is still important to determine quantitatively the extent that this will have on the definition of statistical control space. Further work is needed to more precisely define the range of variation in urinary biochemistry resulting from alterations to the intestinal microflora so that subtle drug-induced alterations to mammalian biochemistry can be more accurately identified. In studies of drug-induced metabolic dysfunction, it is imperative to allow a sufficiently long acclimatization period to provide for adaptation to microbial variability in the location of the study. Insufficient acclimatization time may lead to alterations in urinary profiles that could be ascribed to a change in the animal’s environment and hence the microbial composition in the gut. Because the drug and/or its vehicle may adversely or beneficially affect the gut flora, alterations to the urinary constituents identified in the current work need to be viewed as arising from changes not limited to those from the host animal. This study has shown that a minimum of 21 days is required to allow a germ-free animal to acclimatize to a normal environment. However, the differences between the urine of these germ-free animals when compared to control were not as great as the variation induced by mammalian metabolic dysfunction. We have shown that the use of NMR spectroscopy and multivariate statistical methods can provide a powerful tool for identification of metabolic variations from both endogenous mammalian and symbiotic microbial sources.
Acknowledgment. We acknowledge the support of Merck Sharp and Dohme for funding, Dr. Brian Lake and Dr. J. C. Philips at BIBRA International Ltd. for the collection of samples, and the ULIRS NMR spectroscopy service at Birkbeck College, University of London, for use of the Bruker 500 MHz NMR spectrometer.
References (1) 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. (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 Discovery 1, 153-161. (3) Nicholson, J. K., Timbrell, J. A., and Sadler, P. J. (1985) Proton NMR spectra of urine as indicators of renal damage. Mercuryinduced nephrotoxicity in rats. Mol. Pharmacol. 27, 644-651. (4) Nicholson, J. K., and Wilson, I. D. (1989) High-resolution proton magnetic resonance spectroscopy of biological fluids. Prog. NMR Spectrosc. 21, 449-501. (5) 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. (6) Nicholls, A. W., Holmes, E., Lindon, J. C., Farrant, R. D., Haselden, J. N., Damment, S. J. P., and Nicholson, J. K. (2001)
1404 Chem. Res. Toxicol., Vol. 16, No. 11, 2003
(7) (8)
(9)
(10)
(11)
(12)
(13) (14)
(15)
(16)
Metabonomic investigations into hydrazine toxicity in the rat: induction of 2-amino acidipic acid. Chem. Res. Toxicol. 14, 975987. Gray, J. A., Nicholson, J. K., and Timbrell, J. A. (1986) Creatinuria as an early indicator of cadmium induced testicular damage. Hum. Toxicol. 5, 402-403. Nicholson, J. K., Higham, D. P., Timbrell, J. A., and Sadler, P. J. (1989) Quantitative high-resolution 1H NMR urinalysis studies on the biochemical effects of cadmium in the rat. Mol. Pharmacol. 36, 398-404. Gray, J. A., Nicholson, J. K., Creasy, D. M., and Timbrell, J. A. (1990) Studies on the relationship between testicular toxicity and urinary and plasma creatine concentration. Arch. Toxicol. 64, 443-450. Nicholls, A. W., Nicholson, J. K., Haselden, J. N., and Waterfield, C. J. (2000) A metabonomic approach to the investigation of druginduced phospholipidosis: a NMR spectroscopy and pattern recognition study. Biomarkers 5, 410-423. Gartland, K. P., Beddell, C. R., Lindon, J. C., and Nicholson, J. K. (1991) Application of pattern recognition methods to the analysis and classification of toxicological data derived from proton nuclear magnetic resonance spectroscopy of urine. Mol. Pharmacol. 39, 629-642. Holmes, E., Nicholls, A. W., Lindon, J. C., Ramos, S., Spraul, M., Neidig, P., Connor, S. C., Connelly, J., Damment, S. J. P., Haselden, J. N., and Nicholson, J. K. (1998) Development of a model for classification of toxin-induced lesion using 1H NMR spectroscopy of urine combined with pattern-recognition. NMR Biomed. 11, 235-244. 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. Holmes, E., Nicholson, J. K, Nicholls, A. W., Lindon, J. C., Connor, S. C., Polley, S., and Connelly, J. (1998) The identification of novel biomarkers of renal toxicity using automatic data reduction techniques and PCA of proton NMR spectra of urine. Chemom. Intell. Lab. Syst. 44, 245-255. Gavaghan, C. L., Holmes, E., Lenz, E., Wilson, I. D., and Nicholson, J. K. (2000) An NMR-based metabonomic approach to investigate the biochemical consequences of genetic strain differences: application to the C57BL10J and Alpk: ApfCD mouse. FEBS Lett. 484, 169-174. 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.
Communications (17) 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 (1)H NMR spectroscopy of urine and pattern recognition. Anal. Biochem. 295, 194202. (18) Gilmore, M. S., and Ferreti, J. J. (2003) The thin line between gut commensal and pathogen. Science 299, 1999-2002. (19) Phipps, A. N., Stewart, J., Wright, B., and Wilson, I. D. (1998). Effect of diet of the urinary excretion of hippuric acid and other dietary-derived aromatics in rat. A complex interaction between diet, gut microflora and substrate specificity. Xenobiotica 28, 527537. (20) Goodwin, B. L., Ruthven, C. R. J., and Sandler, M. (1994) Gut flora and the origin of some urinary aromatic phenolic compounds. Biochem. Pharmacol. 47, 2294-2297. (21) 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. (22) Bax, A., and Davis, D. G. (1985) MLEV-17-based two-dimensional homonuclear magnetization transfer spectroscopy. J. Magn. Reson. 65, 355-360. (23) Lindon, J. C., Nicholson, J. K., and Everett, J. R. (1999) NMR spectroscopy of biofluids. Annu. Rep. NMR Spectrosc. 38, 188. (24) Zuppi, C., Messana, I., Forni, F., Rossi, C., Pennacchietti, L., Ferrari, F., and Giardina, B. (1997) 1H NMR spectra of normal urines: Reference ranges of major metabolites. Clin. Chem. Acta 265, 85-97. (25) Anthony, M. L., Sweatman, B. C., Beddell, C. R., Lindon, J. C., and Nicholson, J. K. (1993) Pattern recognition classification of the site of nephrotoxicity based on metabolic data derived from proton nuclear magnetic resonance spectra of urine. Mol. Pharmacol. 46, 199-211. (26) Smith, J. L., Wishnok, J. S., and Deen, W. M. (1994) Metabolism and excretion of methylamines in rats. Toxicol. Appl. Pharmacol. 125, 296-308. (27) Holmes, E., Bonner, F. W., and Nicholson, J. K. (1997) 1H NMR spectroscopic and histopathological studies on propyleneimineinduced renal papillary necrosis in the rat and the multimammate desert mouse (Mastomys natalensis). Comp. Biochem. Physiol., Part C: Pharmacol. Toxicol. Endocrinol. 116, 125-134. (28) Feng, J., Li, X., Pei, F., Chen, X., Li, S., and Nie, Y. (2002) 1H NMR analysis for metabolites in serum and urine from rats administrated chronically with La(NO3)3. Anal. Biochem. 301, 1-7.
TX0340293
