Pharmacometabonomic Characterization of Xenobiotic and

Aug 9, 2012 - Jeannine S. McCune , Meagan J. Bemer , Janel Long-Boyle ... Dhungana , Robert Clark , Wimal Pathmasiri , Susan McRitchie , Susan Sumner...
0 downloads 0 Views 3MB Size
Article pubs.acs.org/jpr

Pharmacometabonomic Characterization of Xenobiotic and Endogenous Metabolic Phenotypes That Account for Inter-individual Variation in Isoniazid-Induced Toxicological Response Katharine Cunningham,† Sandrine P. Claus,‡ John C. Lindon,† Elaine Holmes,† Jeremy R. Everett,§ Jeremy K. Nicholson,*,† and Muireann Coen*,† †

Biomolecular Medicine, Department of Surgery and Cancer, Faculty of Medicine, Imperial College London, Sir Alexander Fleming Building, South Kensington, London SW7 2AZ, U.K. ‡ Department of Food and Nutritional Sciences, The University of Reading, Whiteknights, PO Box 226, Reading RG6 6AP, U.K. § Pharmaceutical, Chemical and Environmental Sciences, School of Science, University of Greenwich at Medway, Central Avenue, Chatham Maritime, Kent ME4 4TB, U.K. S Supporting Information *

ABSTRACT: An NMR-based pharmacometabonomic approach was applied to investigate inter-animal variation in response to isoniazid (INH; 200 and 400 mg/kg) in male Sprague−Dawley rats, alongside complementary clinical chemistry and histopathological analysis. Marked inter-animal variability in central nervous system (CNS) toxicity was identified following administration of a high dose of INH, which enabled characterization of CNS responders and CNS non-responders. High-resolution post-dose urinary 1H NMR spectra were modeled both by their xenobiotic and endogenous metabolic information sets, enabling simultaneous identification of the differential metabolic fate of INH and its associated endogenous metabolic consequences in CNS responders and CNS non-responders. A characteristic xenobiotic metabolic profile was observed for CNS responders, which revealed higher urinary levels of pyruvate isonicotinylhydrazone and β-glucosyl isonicotinylhydrazide and lower levels of acetylisoniazid compared to CNS non-responders. This suggested that the capacity for acetylation of INH was lower in CNS responders, leading to increased metabolism via conjugation with pyruvate and glucose. In addition, the endogenous metabolic profile of CNS responders revealed higher urinary levels of lactate and glucose, in comparison to CNS non-responders. Pharmacometabonomic analysis of the pre-dose 1H NMR urinary spectra identified a metabolic signature that correlated with the development of INH-induced adverse CNS effects and may represent a means of predicting adverse events and acetylation capacity when challenged with high dose INH. Given the widespread use of INH for the treatment of tuberculosis, this pharmacometabonomic screening approach may have translational potential for patient stratification to minimize adverse events. KEYWORDS: isoniazid, pharmacometabonomics, inter-individual variability, NMR spectroscopy



INTRODUCTION

non-enzymatic reactions with the endogenous keto acids pyruvate and 2-oxoglutarate to form pyruvate isonicotinylhydrazone (INH-PA, VII) and 2-oxoglutarate isonicotinylhydrazone (INH-KA, VIII), respectively.7 INH-PA is formed more readily than INH-KA,8 with the extent of hydrazone formation dependent on the availability of INH.9 INH also forms a Schiffbase complex (isoniazidyl-pyridoxal complex, INH-PY, IX) with pyridoxal-5-phosphate10,11 and has been shown to conjugate with glucose to form β-glucosyl isonicotinylhydrazide (INHGLC, X) in dogs.12

Isoniazid (INH, I) is a widely used antitubercular drug for treatment of both active and latent tuberculosis (TB) and as a prophylactic therapy. The metabolic fate of INH has been welldescribed1,2 and is summarized in Figure 1. Acetylisoniazid (AcINH, II) is a major urinary metabolite of INH, which is generated on acetylation of INH by the polymorphic enzyme N-acetyltransferase-2 (NAT-2).1,3,4 AcINH is hydrolyzed to acetylhydrazine (AcHz, III) and isonicotinic acid (INA, IV), and AcHz undergoes further acetylation to produce diacetylhydrazine (DiAcHz, V).5,6 INA is also formed, along with hydrazine, via direct hydrolysis of INH. INA undergoes enzymatic conjugation with glycine resulting in the formation of isonicotinylglycine (INA-GLY, VI). INH also undergoes © XXXX American Chemical Society

Received: May 9, 2012

A

dx.doi.org/10.1021/pr300430u | J. Proteome Res. XXXX, XXX, XXX−XXX

Journal of Proteome Research

Article

Figure 1. Scheme of proposed metabolism of INH (adapted from Timbrell et al.1). Key: AcHz, acetylhydrazine; AcINH, acetylisoniazid; DiAcHz, diacetylhydrazine; INA, isonicotinic acid; INA-GLY, isonicotinylglycine; INH, isoniazid; INH-GLC, β-glucosyl isonicotinylhydrazide; INH-KA, 2oxoglutarate isonicotinylhydrazone; INH-PA, pyruvate isonicotinylhydrazone; INH-PY, isoniazidyl-pyridoxal complex; NAT2, N-acetyltransferase-2.

population studies that have identified a higher incidence of INH-induced hepatotoxicity in slow acetylators.20−22 Acetylation is also implicated in the development of acute INHinduced neurotoxicity, with non-acetylating species, such as dogs, subject to more severe neurological toxicity.23 While histopathology and clinical chemistry remain “gold standards”, there is a need to develop novel and complementary methods that provide additional mechanistic insight into toxic response. Metabonomics is defined as “the quantitative measurement of the dynamic multiparametric metabolic response of living systems to pathophysiological stimuli or genetic modification”.24,25 Metabolic phenotypes, or “metabotypes”,26 generated from biofluids or tissues reflect the complex interplay between genetic background, environmental factors such as diet, nutritional status, and lifestyle, and other extra-genomic influences such as gut microbiotal composition.24,25,27 Metabonomic approaches have proved valuable in the field of toxicology to elucidate the endogenous metabolic consequences of xenobiotic administration, together with the metabolic fate of xenobiotics.28,29 Metabonomic studies that have applied high-resolution NMR spectroscopic techniques to profile biofluids and tissues, combined with multivariate statistical modeling, have been shown to be useful in the study of mechanisms of toxicity30−33 and in determining differential metabolic phenotypes relevant to variability in toxic outcomes.34 In addition, metabonomic approaches enable metabolic changes to be followed across time through the generation of metabolic trajectories,35 showing the onset,

Despite the widespread use of INH, it is associated with a number of adverse effects when administered in isolation or as part of combination therapies, ranging from relatively mild reactions, such as skin rash, to more severe incidents of hepatotoxicity, peripheral neuropathy, and central nervous system (CNS) effects.1,13 INH-induced hepatotoxicity has been linked to the presence of the metabolite acetylhydrazine, which produces a reactive acylating intermediate that covalently binds to hepatic proteins leading to hepatic necrosis in preclinical models and humans.1,3 The primary site of acute INH-induced toxicity in human and animal models is the CNS.2,14,15 The development of INHinduced neurotoxicity is linked to the depletion of the cellular reserve of pyridoxal, which subsequently leads to a depletion of GABA levels in the brain, resulting in excessive CNS stimulation and seizures.10,16 In the clinic, pyridoxine (vitamin B6) is co-administered with INH to counteract the INHinduced depletion of the pyridoxal pool.17 The importance of pyridoxal in the prevention of INH-induced neurotoxicity has been demonstrated in a rat model of vitamin B6-deficient rats that were more susceptible to INH-induced neurotoxicity.18 Genetic polymorphism of NAT-2 in man has been shown to give rise to slow, intermediate, and fast acetylators.19 The slow acetylator phenotype has been shown to be important in the development of INH-induced peripheral neuropathies.2,13 It is also believed that slow acetylators are at greatest risk of development of hepatotoxicity due to their relative inability to detoxify AcHz to DiAcHz; this is supported by numerous B

dx.doi.org/10.1021/pr300430u | J. Proteome Res. XXXX, XXX, XXX−XXX

Journal of Proteome Research

Article

approximately 2 h post-treatment and were euthanized. Postdose urine was available from 3 of the 4 rats that developed an adverse CNS response, whereas post-dose plasma was not collected.

progression, and where applicable, recovery from toxic insult.36,37 Pharmacometabonomics involves the prediction of the outcome of a xenobiotic intervention in an individual based on a statistical model of preintervention metabolite signatures.38,39 Early pharmacometabonomic studies proved successful for the prediction of inter-individual response to acetaminophen (APAP) in a preclinical model, 40 and subsequent work in man has also enabled prediction of the metabolic fate of APAP.41,42 In the clinical oncology setting, recent pharmacometabonomic applications enabled the successful prediction of the severity of toxic response to capecitabine therapy in patients with inoperable colorectal cancer43 and the progression of breast cancer in response to chemotherapy.44 In the present study, we have applied an NMR-based metabonomic strategy to understand inter-animal variability in toxic response to INH, specifically between responders that developed adverse CNS effects and non-responders that displayed no adverse toxic response. This approach enabled characterization of variable response metabotypes based on both the xenobiotic metabolic profile and the associated endogenous metabolic perturbations. In addition, pharmacometabonomic modeling of the pre-dose urinary metabolic profiles enabled pretreatment classification of the post-treatment adverse CNS outcome.



Clinical Chemistry and Histopathology

Urine parameters (total volume, pH, bilirubin, ketone substances, glucose, hemoglobin, and protein concentration) were recorded using an automatic test-strip reader (Clinitek 200, Bayer Diagnostics). Plasma was analyzed for urea, creatinine, glucose, cholesterol, triglycerides, albumin, total protein, total bilirubin, alkaline phosphatase (ALP), aspartate aminotransferase (AST), alanine aminotransferase (ALT), 5′nucleotidase (5-NU), γ-glutamyl transferase (GGT), and bile acids using an AU600 multiparametric clinical analyzer (Olympus) following the standard operating procedures of the laboratory. Significant changes in these parameters from control activity or levels were determined using a Student’s two-tailed t test (with a significance threshold of p < 0.05). Tissue from the left lobe of the liver, kidney, and brain were sampled at termination (scheduled at 48 and 168 h posttreatment; the CNS responders were sampled within 2 h of dosing) and fixed in a 10% formalin solution. Representative samples were processed routinely in an automatic tissue processor, embedded in paraffin, sectioned at 4−6 μm, stained with hematoxylin and eosin (H&E), and examined histologically. Periodic acid Schiff staining was performed on additional samples of the left liver lobe to further characterize hepatic glycogen content.

MATERIAL AND METHODS

Animal Handling and Sample Collection

1

All animal manipulations were conducted by Pfizer Global R&D, Amboise, France, in accordance with the relevant national requirements and local guidelines. Male Sprague− Dawley rats (7 weeks of age, approximately 250 g, n = 30, from Charles River, France) were allocated randomly to three dose groups (n = 10 per group) and administered, via intraperitoneal (i.p.) injection, a single dose of either vehicle (0.9% saline) or INH in vehicle at dose levels of 200 or 400 mg/kg (dose volume 10 mL/kg). Animals were housed in individual metabolic cages with free access to water and a standardized diet (diet A04C, Usine d’Alimentation Rationnelle, Villemoisoon-sur-Orge, France). Temperature (20 ± 2 °C) and relative humidity (60 ± 20%) were maintained, with a 12 h light/dark cycle throughout the study. The planned protocol was that half of the animals from each group were euthanized via CO2 anesthesia at 48 h post-treatment, and the remainder at 168 h post-treatment for removal of target organs for histopathological examination. Urine samples were collected into ice-cooled vessels containing 0.1 mL sodium azide (100 mg/mL) over the following time periods: −48 to −41 h; −24 to −17 h; 0 to 7 h; 7 to 24 h; 24 to 31 h; 48 to 55 h; 72 to 79 h; 96 to 103 h; 120 to 127 h; and 144 to 151 h. All urine samples were centrifuged at 9447g for 10 min at room temperature to remove particulate matter, divided into aliquots, and stored at −20 °C. Blood was sampled prior to animals being placed in metabolic cages and prior to planned termination, i.e., at 24 and 168 h post-dose, into commercially available plastic tubes containing lithium heparin, as an anticoagulant, from which plasma samples were isolated by centrifugation. Clinical observation was carried out once a day over the periods −48 to 0 h and 48 to 168 h, and twice daily over the period 0 to 48 h. Four animals in the high dose group displayed clinical signs of adverse CNS effects at

H NMR Spectroscopy of Rat Urine

Urine samples were thawed, vortexed, and allowed to stand for 10 min at room temperature prior to mixing urine (400 μL) with phosphate buffer (200 μL, 0.2 M, in 90:10 H2O/D2O, pH 7.4, containing 3 mM 3-trimethylsilyl-[2,2,3,3-2H4]-propionic acid sodium salt (TSP, Sigma Aldrich) and 3 mM sodium azide (Sigma Aldrich). The urine−buffer mixtures were centrifuged at 9447g for 10 min at room temperature. Supernatants (550 μL) were transferred to 5 mm NMR tubes (Norell Standard Series HP-507) for NMR analysis. The D2O provided a field frequency lock and TSP a chemical shift reference (1H, δ 0.0). One-dimensional (1D) 1H NMR spectra were acquired on an NMR spectrometer (Bruker Biospin, Rheinstetten, Germany), operating at 14.1 T (600.13 MHz 1H frequency) and at a temperature of 300 K. A standard one-dimensional solvent suppression pulse sequence of the form [relaxation delay−90° pulse−3 μs delay−90° pulse−mixing time−90° pulse−acquire FID] was applied.45 For each sample, 196 transients were collected into 64 K data points, using a spectral width of 12019.2 Hz with a relaxation delay of 4 s, an acquisition time of 2.72 s, and a mixing time of 100 ms. Presaturation of the water resonance occurred during the relaxation delay and the mixing time. A line-broadening factor of 0.3 Hz was applied to all spectra prior to Fourier transformation (FT). The assignment of NMR peaks was facilitated by the acquisition of a range of two-dimensional (2D) homo- and heteronuclear NMR spectra. 1H−1H correlation spectroscopy (COSY),46 1H−1H total correlation spectroscopy (TOCSY),47 1H-J-resolved (JRES),48 1H−13C heteronuclear single quantum coherence (HSQC),49 and 1 H−13C heteronuclear multiple bond correlation (HMBC)50 were acquired for selected samples using standard pulse sequences supplied by the instrument manufacturer. C

dx.doi.org/10.1021/pr300430u | J. Proteome Res. XXXX, XXX, XXX−XXX

Journal of Proteome Research

Article

Figure 2. Scheme summarizing the approach of (a) separating the 1H NMR urinary spectral profiles into both endogenous and xenobiotic metabolic profiles, with (b) these metabotypes treated separately for subsequent multivariate statistical modeling.

NMR Spectral Data Processing

remaining urinary spectral regions represented an endogenous metabolic data set (Figure 2a). The scheme in Figure 2 illustrates the approach of separating the xenobiotic and endogenous metabolic spectral signatures to enable subsequent multivariate statistical modeling of these distinct metabolic phenotypes reflecting both the metabolic fate of an administered drug and its endogenous consequences. Both data sets were normalized separately using probabilistic quotient normalization to correct for variations in urinary volume and concentration.52 Multivariate statistical analysis (principal components analysis (PCA), partial least-squares discriminant analysis (PLS-DA), and orthogonal partial leastsquares discriminant analysis (O-PLS-DA)) was carried out in MATLAB.53−55 PCA models were computed using scaling to unit variance for the separate endogenous and xenobiotic metabolic profiles (Figure 2b). Metabolic trajectory plots were

1

The H NMR spectra were phased, baseline corrected, and referenced to TSP at δ 0.0 automatically using an in-house routine (Dr. T. Ebbels and Dr. H. Keun). Full resolution 1H NMR spectra were imported into MATLAB (version 7.6, The MathWorks, Natick, USA) using MetaSpectra 4.1.1 (an inhouse routine written by Dr. O. Cloarec). The regions corresponding to water (δ 4.70−4.85), urea (δ 5.50−6.10), and TSP (δ −0.20−0.20) were removed. The spectra were aligned using the recursive segment-wise peak alignment (RSPA) method to account for any chemical shift variations induced by differences in pH between samples.51 The urinary spectral regions containing INH-related resonances (δ 2.14− 2.24, 2.37−2.39, 2.47−2.52, 2.78−2.92, 4.24−4.29, 7.67−7.95, and 8.60−8.82) were separated from the remaining spectral regions to form a xenobiotic spectral data set (Figure 2a). The D

dx.doi.org/10.1021/pr300430u | J. Proteome Res. XXXX, XXX, XXX−XXX

Journal of Proteome Research

Article

Figure 3. Representative 600 MHz 1H NMR spectrum of rat urine collected 0−7 h after administration of INH (400 mg/kg), showing simultaneous assignment of endogenous and xenobiotic metabolites. Key: see Table 1 and DMA, dimethylamine; DMG, dimethylglycine; MA, methylamine; TMAO, trimethylamine-N-oxide; U1/2, unassigned INH-related metabolites.

predictive ability). Relative levels of key xenobiotic metabolites, as identified as discriminatory from loadings plots, were calculated by integration of their NMR spectral resonances, with integrals corrected for the number of protons giving rise to a resonance (absolute excretion values were not determined).

calculated using the mean PCA score (PC1 and PC2) values of the control, low dose, and high dose urinary spectra at each time for each data set separately (Figure 2b). An O-PLS-DA model was constructed to investigate the statistical relationship between the pre-dose NMR spectral data (X) and the post-dose development of adverse CNS effects (Y, with CNS responders classed as group 1; CNS non-responders were classed as group 0). To test the validity of all multivariate statistical models, 7fold cross-validation was used, whereby a seventh of the data were left out of the model and then predicted back in, repeating the process until all of the data had been excluded at least once to calculate the Q2Y value (a measure of the cross-validated

Identification of INH Urinary Metabolites

The NMR resonances of INH-related metabolites were assigned using 1D and 2D NMR spectra of standard compounds and incubations of standards. Five millimolar solutions of INH (Sigma Aldrich), INA (Sigma Aldrich), AcHz (AcHz, Sigma Aldrich), AcINH (synthesized by Pfizer Global R E

dx.doi.org/10.1021/pr300430u | J. Proteome Res. XXXX, XXX, XXX−XXX

Journal of Proteome Research

Article

Table 1. 1H NMR Spectral Assignment of INH and INH-Related Metabolites metabolitea

1

H NMR shifts (δ), 3J coupling constants

isoniazid (INH, I)

8.685 7.705 8.619 7.748 8.733 7.812 2.150 8.751 7.803 2.194 8.741 7.813 2.220 8.751 7.804 2.813 2.494 8.762 7.908 2.887 2.559 2.813 2.494 8.693 7.723 4.263 3.409 3.744 8.689 7.791 3.983

isonicotinic acid (INA, IV) acetylisoniazid (AcINH, II)

pyruvate isonicotinylhydrazone (INH-PA A, VII)

pyruvate isonicotinylhydrazone (INH-PA B, VII)

2-oxoglutarate isonicotinylhydrazone (INH-KA A, VIII)

2-oxoglutarate isonicotinylhydrazone (INH-KA B, VIII)

β-glucosyl isonicotinylhydrazide (INH-GLC, X)

isonicotinylglycine (INA-GLY, VI)

a b

(d, 2H, H2 + H6, J = 6.2 Hz) (d, 2H, H3 + H5, J = 6.2 Hz) (d, 2H, H2 + H6, J = 6.1 Hz) (d, 2H, H3 + H5, J = 6.1 Hz) (d, 2H, H2 + H6, J = 6.3 Hz) (d, 2H, H3 + H5, J = 6.3 Hz) (s, COCH3) (d, 2H, H2 + H6, J = 6.2 Hz) (d, 2H, H3 + H5, J = 6.2 Hz) (s, CH3) (d, 2H, H2 + H6, J = 6.1 Hz) (d, 2H, H3 + H5, J = 6.1 Hz) (s, CH3) (d, 2H, H2 + H6, J = 6.0 Hz) (d, 2H, H3 + H5, J = 6.0 Hz) (t, 2H, CH2 J = 7.6 Hz) (t, 2H, CH2, J = 7.6 Hz) (d, 2H, H2 + H6, J = 5.8 Hz) (d, 2H, H3 + H5, J = 5.8 Hz) (t, 2H, J = 6.2 Hz) (t, 2H, J = 6.2 Hz) (t, 2H, J = 7.6 Hz) (t, 2H, CH2, J = 7.6 Hz) (d, 2H, H2 + H6, J = 6.0 Hz) (d, 2H, H3 + H5, J = 6.0 Hz) (d, 1H, H1′, J = 8.9 Hz) (dd, 1H, H2′); 3.575 (t, 1H, H3′); 3.467 (dd, 1H, H4′); (dd, 1H, H5′); 3.925 (dd, 2H, H6′)b (d, 2H, H2 + H6, J = 6.1 Hz) (d, 2H, H3 + H5, J = 6.1 Hz) (s, 2H, CH2)

A/B refer to the isomers (syn and anti) of INH-PA and INH-KA, representing the major and minor forms. Structures are shown in Figure 1. Accurate J-coupling constants were not available for H2′−H6′ due to peak overlap.

and/or buffer was acquired, the second compound (100 μL, 5 mM in 0.2 M phosphate buffer) was added, and a series of 1D 1 H NMR spectra were acquired over a period of approximately 6 h (acquisition parameters as above; incubation time subject to variation). Visual comparisons and “spike-in” experiments were carried out to identify and confirm the presence of reaction products in the post-treatment urinary spectra. Additional 2D homo- and heteronuclear NMR spectra were acquired as necessary, as previously described.46−50

& D), and DiAcHz (Lancaster Synthesis) were prepared in 0.2 M phosphate buffer (as above) and analyzed by 1D 1H NMR using the standard 1D solvent suppression sequence described above.45 Typically, 128 transients were collected into 64 K data points using a spectral width of 12019.2 Hz and a relaxation delay of 2 s. A line-broadening factor of 0.3 Hz was applied to all spectra prior to FT. Additional solutions of pyruvate (Sigma Aldrich), 2oxoglutarate (Sigma Aldrich), pyridoxal (Sigma Aldrich), pyridoxal-5-phosphate (Sigma Aldrich), glycine (Sigma Aldrich), and glucose (Sigma Aldrich) were prepared (5 mM in 0.2 M phosphate buffer). Twenty-four hour incubations at 37 °C of INH (500 μL, 5 mM in 0.2 M phosphate buffer) with pyridoxal-5-phosphate, pyruvate, glycine, glucose, and 2oxoglutarate (all 500 μL, 5 mM in 0.2 M phosphate buffer) were carried out, and all reaction products (5 mm NMR tube, 600 μL volume) analyzed using the standard 1D 1H NMR solvent suppression sequence described above.45 β-Glucosyl isonicotinylhydrazide (INH-GLC) was prepared as described previously56 by heating INH and glucose (both 5 mM solutions in methanol, 5 mL of each) at 90 °C for 2 h. The product was reconstituted in phosphate buffer and characterized by 1D 1H NMR as described above. Additional time-course incubation reactions were carried out to follow the formation of the above INH hydrazone and hydrazide metabolites: an initial spectrum of INH (500 μL, 5 mM in 0.2 M phosphate buffer) in urine



RESULTS

NMR Spectroscopic Identification of Urinary INH Metabolites

The urinary INH metabolites identified by 1H NMR spectroscopy were acetylisoniazid (AcINH, II), isonicotinic acid (INA, IV), pyruvate isonicotinylhydrazone (INH-PA, VII), 2oxoglutarate isonicotinylhydrazone (INH-KA, VIII), β-glucosyl isonicotinylhydrazide (INH-GLC, X), isonicotinylglycine (INA-GLY, VI), and unchanged INH. A representative 600 MHz 1H NMR spectrum of urine collected 0−7 h after administration of 400 mg/kg INH (high dose) is shown in Figure 3 to highlight the simultaneous assignment of diverse INH metabolites and endogenous metabolites. The use of authentic compounds enabled the assignment of urinary spectral resonances arising from INH, AcINH, and INA. INH undergoes non-enzymatic reactions with the endogenous ketoF

dx.doi.org/10.1021/pr300430u | J. Proteome Res. XXXX, XXX, XXX−XXX

Journal of Proteome Research

Article

Figure 4. PC1 versus PC2 scores trajectory plot showing the mean points of the individual PC scores for control, low dose, and high dose urinary 1H NMR spectra across the time course from 24 h pre-dose to 55 h post-dose for (a) endogenous metabolic data set and (b) xenobiotic metabolic data set. The high dose animals were classified as CNS responders and CNS non-responders. The numbers refer to the sampling time. The error bars indicate the standard deviation. Key: (black) control; (green) low dose; (blue) CNS non-responders (high dose); (red) CNS responders (high dose).

acids, pyruvate and 2-oxoglutarate, to form INH-PA and INHKA, respectively (Figure 1). Spectroscopic characterization of a standard solution of both INH-PA and INH-KA led to the identification of the syn and anti isomers of these metabolites, which were also detectable in the post-treatment urinary spectra. Variable-temperature NMR experiments revealed the coalescence of the resonances from the syn and anti isomers of INH-PA and INH-KA with increasing temperature, indicative

of exchange between the isomers. The ratio of the two isomers of both INH-PA and INH-KA was approximately 2:1, determined from the relative ratio of the integral for the respective NMR resonances in the post-dose urinary spectra, where A refers to the major form and B to the minor form. 1D and 2D NMR spectroscopic analysis of the reaction products of INH and glucose enabled characterization of the corresponding urinary metabolite as β-glucosyl isonicotinylhydrazide (INHG

dx.doi.org/10.1021/pr300430u | J. Proteome Res. XXXX, XXX, XXX−XXX

Journal of Proteome Research

Article

Figure 5. PCA model derived from the xenobiotic spectral data following administration of INH (high dose, 400 mg/kg) at 0−7 h, representing CNS responders and CNS non-responders. (a) PC1 versus PC2 scores plot. (b) Loadings plot corresponding to PC1. The color scale corresponds to the correlation of determination (r2) of the variables. (c) Ratio of AcINH to INH, INH-PA to AcINH and of INH-GLC to AcINH present in the urinary spectra collected 0−7 h after administration of INH at 200 and 400 mg/kg. Ratios were calculated from integrals of these spectral resonances in individual animals and the mean ± standard deviation calculated for each group. Significant changes in the ratios observed in the high dose CNS responders relative to CNS non-responders were determined using a Student’s two-tailed t test. Full integral values for all INH metabolites are presented in Figure S1 in Supporting Information. Key: (blue) CNS non-responders; (red) CNS responders; (yellow) CNS non-responder B (this individual exhibited no overt clinical signs of an adverse CNS effect but shared some of the metabolic features of the CNS responders). * p < 0.05. ** p < 0.01.

addition, histopathological assessment of the brain and kidney tissues revealed no microscopic abnormalities. The variability in response to INH was characterized via PCA analysis of urinary NMR spectra following separation of the NMR spectral profiles into distinct endogenous and xenobiotic metabolic data sets as schematically illustrated in Figure 2. PCA analysis of the 1H NMR spectra of urine from both the low dose and high dose cohorts and the computation of a metabolic trajectory from the mean PC score values enabled visualization of INH-induced perturbations in the endogenous (Figure 4a) and xenobiotic (Figure 4b) data sets across time (from 24 h pre-dose to 55 h post-dose). The low dose endogenous metabolic trajectory showed clear deviation from the control endogenous metabolic space over the collection period 0−7 h post-dose, followed by a return to the control endogenous space over the collection period 7−31 h post-dose (Figure 4a). The endogenous metabolic trajectory was indicative of INH-induced homeostatic perturbation, which was most marked from 0 to 7 h post-dose, followed by recovery of metabolic homeostasis by 31 h post-dose. In comparison, the movement observed in the low dose xenobiotic trajectory reflected the presence of INH and its metabolites at 0−7 and 7−24 h post-treatment followed by complete clearance of INHrelated metabolites from 24 h onward (Figure 4b). The high dose CNS non-responders exhibited an endogenous metabolic trajectory similar to that of the low dose animals but experienced a greater displacement from the control endogenous metabolic space at 0−7 h post-dose with recovery over the period 24−31 h post-dose (Figure 4a). The

GLC). Spectroscopic characterization of the products of the reaction of INH with glycine revealed resonances due to INAGLY, which were also observed in the 0−7 h post-dose urinary spectra. NMR spectral assignments for INH and its related metabolites are provided in Table 1. Characterization of the products from the reaction of pyridoxal and pyridoxal-5phosphate with INH indicated that the isoniazidyl−pyridoxal complex (Schiff base) was not detected in the post-dose urine by NMR spectroscopy, which was consistent with earlier reports of the urinary metabolites of INH in a rat model. Inter-animal Variation in Response to INH Exposure

Clinical observations of adverse CNS effects following administration of the high dose of INH enabled classification of “CNS responders” (4 of 10 rats). The remainder of the high dose treated cohort displayed no clinical signs of INH-induced CNS effects (6 of 10 rats) and were classified as “CNS nonresponders”. Statistically significant reductions in the plasma activity of AST and ALT were reported 24 h post-dose for the low dose cohort and high dose CNS non-responders, relative to control activity. No clinical chemistry data were available for the CNS responders. Histopathological assessment of the liver revealed hepatocellular glycogen depletion in 3 of the 4 CNS responders (representing a sampling point within 2 h postdosing) and in one CNS non-responder (later classified as CNS non-responder B). There was no histopathological evidence to suggest the presence of a hepatic necrotic lesion in the CNS non-responders or CNS responders post-treatment. In H

dx.doi.org/10.1021/pr300430u | J. Proteome Res. XXXX, XXX, XXX−XXX

Journal of Proteome Research

Article

Figure 6. PCA model derived from the endogenous spectral data set following administration of INH (high dose, 400 mg/kg) at 0−7 h, representing CNS responders and CNS non-responders. (a) PC1 versus PC2 scores plot. (b) Loadings plot corresponding to PC1. The color scale corresponds to the coefficient of determination of the variables (r2). Key: (blue) CNS non-responders; (red) CNS responders; (yellow) CNS non-responder B (this individual exhibited no overt clinical signs of an adverse CNS effect but shared some of the metabolic features of the CNS responders); CRN, creatinine.

of INH-PA to AcINH was significantly elevated in the CNS responders relative to the CNS non-responders (a ratio of 3.26 compared to 0.65, respectively, p < 0.01). In addition, the postdose ratio of INH-GLC to AcINH was significantly elevated in the CNS responders relative to the CNS non-responders (a ratio of 5.75 compared to 0.31, respectively, p < 0.05). NMR spectral integral values for the resonances of INH and its metabolites for both low and high dose cohorts at 0−7 h and 7−24 h post-dose are presented in the Supporting Information, Figure S1. It was notable that the urinary levels of AcINH over the period 0−7 h post-dose were comparable in the low dose cohort (200 mg/kg) and high dose CNS non-responders (400 mg/kg) (Supporting Information, Figure S1). The endogenous metabolic phenotype reflective of differential CNS response to INH was assessed through PCA analysis of the endogenous metabolic data set for CNS responders and CNS non-responders at 0−7 h post-treatment (Figure 6). A PCA scores plot revealed separation of CNS responders and non-responders along PC1 (Figure 6a) with significant inter-animal variation in INH-induced endogenous metabolic response. The corresponding loadings plot indicated the presence of elevated levels of glucose and lactate, coupled with low levels of creatinine in the urine of CNS responders relative to CNS non-responders (Figure 6b). CNS nonresponder B, the animal that displayed a unique INH metabolic profile, i.e., elevated urinary levels of INH-PA and INH-GLC compared to the remaining CNS non-responders, was clustered with the CNS non-responders on the basis of the PCA scores plot of the endogenous data set (labeled CNS non-responder B). The corresponding loadings plot showed that this animal had elevated post-dose urinary levels of glucose relative to the non-responders, but no increase in urinary lactate was observed.

PCA trajectory representing the xenobiotic metabolic profile for the high dose CNS non-responders exhibited a more prolonged perturbation relative to the low dose cohort (Figure 4b). The endogenous and xenobiotic metabolic trajectories for the high dose CNS responders exhibited greater displacement compared to the high dose CNS non-responders at 0−7 h postdose (Figure 4a and b). The differential metabotypes of the high dose CNS responders and non-responders were further explored through PCA analysis of the high-dose endogenous and xenobiotic data sets at the 0−7 h post-treatment timepoint. The PC1 versus PC2 scores plot from PCA analysis of the 0−7 h post-dose urinary xenobiotic metabolic data set showed separation of the CNS responders and non-responders along PC1 (Figure 5a). The corresponding PCA loadings plot indicated that elevated levels of INH-PA and INH-GLC, coupled with low levels of AcINH, were seen in the urinary spectra of CNS responders in comparison to CNS nonresponders (Figure 5b). One animal displayed no clinical signs of CNS toxicity, i.e., was classified as a CNS non-responder, but was clustered with the metabolic profiles of the CNS responders on the PCA scores plot (CNS non-responder B). The urinary NMR spectrum from this animal showed high post-dose levels of INH-PA and INH-GLC, but no significant reduction in levels of AcINH as was seen in the remaining animals classified as CNS responders, which partially explains its anomalous position in the PCA scores plot. To further characterize the CNS responder and CNS nonresponder metabolic phenotypes, integrals of the INH metabolite spectral resonances were calculated from the 1H NMR 0−7 h post-dose urinary spectra. A number of peak ratios were calculated to express the relative urinary levels of the key INH metabolites (Figure 5c); CNS non-responder B was excluded from the CNS non-responder group and treated separately (see Supporting Information, Figure S1). Within the high dose group at 0−7 h post-dose, the CNS responders had significantly lower levels of AcINH relative to parent INH when compared to the CNS non-responders (a ratio of 0.37 compared to 0.70 respectively, p < 0.05). The post-dose ratio

Pre-dose Classification of Post-dose Variability in Toxic Response to INH

O-PLS-DA was used to identify the relationship between the pre-dose urinary metabolic profiles and the presence or absence of an adverse CNS toxic outcome. An O-PLS-DA model was constructed to explore discrimination of the pre-dose urinary I

dx.doi.org/10.1021/pr300430u | J. Proteome Res. XXXX, XXX, XXX−XXX

Journal of Proteome Research

Article

Figure 7. O-PLS-DA pairwise model of pre-dose urinary 1H NMR spectra, comparing high dose CNS responders (n = 4) with CNS non-responders (n = 5, with anomalous CNS non-responder B excluded). (a) O-PLS-DA scores plot (cross-validated scores versus class) showing separation between CNS responders (red) and CNS non-responders (blue). (b, d) Spectral regions of the O-PLS-DA loadings coefficient plot, showing metabolites responsible for discrimination between CNS responders and CNS non-responders. The color scale corresponds to the correlation coefficient of the variables. The upper section of the plot represents metabolites that were elevated in the urinary profiles of CNS responders. (c, e) Corresponding spectral regions color-coded according to degree of post-dose response (red, CNS responders; blue, CNS non-responders). (f, g) Predose spectral regions for low dose individuals color coded according to post-dose urinary levels of AcINH (red, AcINH integral value of less than 3 × 108; blue, AcINH integral value of greater than 3.5 × 108; NMR spectral integrals; arbitrary units; see Figure S1 in Supporting Information. Model statistics: Q2Y = 0.34; R2Y = 0.76.

and non-responders (Figure 7a). The corresponding loadings plot indicated that the discrimination was due to elevated levels of a number of urinary spectral resonances in the CNS responders, specifically resonances at δ 5.11 (d), 6.65 (dd), and 6.72 (s) (Figure 7b, d). Visual comparison of spectra confirmed

spectral profiles of CNS responders and CNS non-responders (with CNS non-responder B excluded). The cross-validated scores for this model, which was characterized by a Q2Y of 0.34 and R2Y of 0.76, showed clear separation between the pre-dose urinary spectra derived from the high dose CNS responders J

dx.doi.org/10.1021/pr300430u | J. Proteome Res. XXXX, XXX, XXX−XXX

Journal of Proteome Research

Article

urinary levels of AcINH were excreted by both the low dose animals and the high dose non-responders (0−7 h). In contrast, the high dose CNS responders exhibited significantly lower urinary levels of AcINH than the high dose CNS nonresponders suggesting that acetylation capacity may have been lower in those animals that experienced INH-induced effects on the CNS. A greater proportion of INH would thus be expected to be metabolized via other routes. Evidence to support this was provided from observation that significantly elevated levels of the INH hydrazones and hydrazides (i.e., INH-PA, INHGLC) were excreted by the CNS responders suggesting enhanced metabolism of INH via conjugation with pyruvate and glucose. In humans, severe, acute INH-induced neurotoxicity is characterized by a clinical triad of refractory seizures, metabolic acidosis with an elevated anion gap, and coma, with symptoms generally observed between 45 min and 2 h post-exposure.14,62 The neurotoxicity latency period is consistent with previous animals studies2 and with the clinical signs of the adverse CNS effect observed within 2 h post-treatment in this study. Discrimination of high dose CNS responders from CNS nonresponders was also determined from modeling of the distinct endogenous metabolic profiles. The CNS responders excreted significantly elevated urinary levels of glucose and lactate. INHinduced neurotoxicity in man has been associated with the clinical presence of hyperglycemia, hypokalemia, glucosuria, and ketonuria.14,63 Metabolic acidosis following INH-induced neurotoxicity in man is associated with elevated urinary lactate, the most likely cause being tissue hypoxia, induced by seizures, and inhibition of lactate dehydrogenase, responsible for the conversion of lactate to pyruvate.14,62−64 Pharmacometabonomics has proved beneficial in the prediction of inter-individual variation in animal and human models of xenobiotic interventions and was applied here to classify toxic outcome on the basis of pre-dose urinary metabolic profiles. An O-PLS-DA model identified a positive relationship between pre-dose levels of a urinary metabolite/ metabolites and the development of INH-induced adverse effects on the CNS, providing further support for the value of the pharmacometabonomic concept. Corroborating support was provided by a second O-PLS-based regression model that identified a positive relationship between the same pre-dose urinary metabolite/metabolites and the post-dose ratio of INHPA to AcINH. In addition, an elevation in this predictive metabolic signature was detected in 3 of 10 animals administered a low dose of INH, corresponding to those rats with the lowest post-dose levels of AcINH. We hypothesize that this pre-dose metabolic signature may be predictive of the capacity for acetylation of INH in this preclinical model.

increased intensities of these resonances in the pre-dose urinary spectra of CNS responders (Figure 7c, e). An O-PLSregression model (Q2Y of 0.71 and R2Y of 0.95) constructed with the post-dose ratio of INH-PA to AcINH identified the same discriminatory spectral resonances (Supporting Information, Figure S2). Variable intensities of these signals were also observed in the pre-dose urinary spectra of the low dose cohort (Figure 7f, g). Interestingly, the highest levels of these pre-dose signals corresponded with those animals with the lowest levels of AcINH in the 0−7 h post-dose urine following the low-dose INH treatment. A range of 2D NMR spectroscopic experiments were applied for structural identification of these urinary NMR signals. 1H−1H TOCSY showed that the aromatic resonances at δ 6.65, and 6.72 were coupled to a resonance at δ 7.26; this pattern is typical of a 1,2,4-trisubstituted phenolic moiety. In addition, the resonance at δ 5.11 was found to be coupled to a resonance at δ 3.61 and is typical of an ether glucuronide moiety. We have thus tentatively assigned these NMR signals to a phenolic ether-glucuronide, although further experiments are necessary to elucidate fully and confirm an exact molecular structure.



DISCUSSION We have applied a dual metabonomic and pharmacometabonomic strategy for exploration of INH-induced toxicity in a rat model by considering the systemic response to INH treatment through analysis of distinct xenobiotic and endogenous metabolic phenotypes. We have characterized CNS responders (40% of cohort) and CNS non-responders (60% of cohort) based on the presence or absence of unexpected adverse neuropathic effects. Histopathological assessment revealed glycogen depletion in 3 of 4 CNS responders, which suggested a rapid utilization of hepatic glycogen stores in response to INH-induced CNS toxicity. There was no histopathological evidence of hepatic necrosis nor a hepatic lesion in CNS nonresponders. Reductions in the plasma activity of ALT and AST were observed in the low dose and high dose CNS nonresponders at 24 h post-treatment, which supports previous findings.57,58 The transaminase enzymes require the cofactor pyridoxal-5-phosphate, which is known to be sequestered by INH,14,15 and by hydrazine, a known metabolite of INH.59,60 1 H NMR spectroscopic analysis enabled the simultaneous identification of numerous endogenous metabolites and INH urinary metabolites, including AcINH, INA, INH-KA (syn and anti isomers), INH-PA (syn and anti isomers), INH-GLC, INAGLY, and unchanged INH. This work represents the first example of the application of 1H NMR spectroscopy to simultaneously identify multiple INH urinary metabolites, an outcome with clear translational potential to the clinic. PCA analysis of the urinary NMR spectral profiles enabled discrimination of both the endogenous and xenobiotic metabolic profiles of CNS responders and CNS nonresponders following INH administration. INH-induced CNS responder effects were characterized by statistically significantly elevated urinary levels of the INH metabolites INH-PA and INH-GLC, accompanied by lower levels of AcINH, a metabolite produced via acetylation of INH. Acetylation has been identified as a contributing factor in the development of INH-induced hepatotoxicity20−22 and peripheral neuropathy.13 Further, acetylation of INH has been shown to be saturated at high doses of INH in both human and rat models.1,6,9,61 In the current study, possible saturation of acetylation was observed following administration of INH at 200 mg/kg as comparable



CONCLUSION We have characterized the xenobiotic and endogenous metabolic phenotypes associated with inter-animal variability in toxicity outcomes for a preclinical model of INH treatment. The urinary metabolic fate of INH was characterized by NMR spectroscopic analysis and was shown to differ with respect to acetylation capacity and the production of hydrazone and hydrazide metabonates in responders that developed adverse effects on the CNS. The endogenous consequences of INH administration were also simultaneously identified, highlighting the power of this multivariate approach for assessment of a systems level response to a toxin. The application of pharmacometabonomics enabled identification of a pretreatK

dx.doi.org/10.1021/pr300430u | J. Proteome Res. XXXX, XXX, XXX−XXX

Journal of Proteome Research

Article

(10) Wiegand, R. G. The formation of pyridoxal and pyridoxal 5phosphate hydrazones1. J. Am. Chem. Soc. 1956, 78 (20), 5307−5309. (11) Davison, A. N. Pyridoxal phosphate as coenzyme of diamine oxidase. Biochem. J. 1956, 64 (3), 546−8. (12) Peters, J. H.; Hayes, V. E. Comparative studies on the metabolism of isoniazid and isoniazid hydroazones in the dog. Arch. Int. Pharmacodyn. Ther. 1966, 159 (2), 328−39. (13) Devadatta, S.; Gangadharam, P. R.; Andrews, R. H.; Fox, W.; Ramakrishnan, C. V.; Selkon, J. B.; Velu, S. Peripheral neuritis due to isoniazid. Bull. World Health Org. 1960, 23, 587−98. (14) Shah, B. R.; Santucci, K.; Sinert, R.; Steiner, P. Acute isoniazid neurotoxicity in an urban hospital. Pediatrics 1995, 95 (5), 700−4. (15) Cash, J. M.; Zawada, E. T., Jr. Isoniazid overdose. Successful treatment with pyridoxine and hemodialysis. West. J. Med. 1991, 155 (6), 644−6. (16) Wood, J. D.; Peesker, S. J. The effect on GABA metabolism in brain of isonicotinic acid hydrazide and pyridoxine as a fuction of time after administration. J. Neurochem. 1972, 19 (6), 1527−37. (17) Snider, D. E., Jr. Pyridoxine supplementation during isoniazid therapy. Tubercle 1980, 61 (4), 191−6. (18) Boone, I. U.; Magee, M.; Turney, D. F. Metabolism of C14labeled isoniazid in vitamin B6-deficient rats. J. Biol. Chem. 1956, 221 (2), 781−9. (19) Parkin, D. P.; Vandenplas, S.; Botha, F. J.; Vandenplas, M. L.; Seifart, H. I.; van Helden, P. D.; van der Walt, B. J.; Donald, P. R.; van Jaarsveld, P. P. Trimodality of isoniazid elimination: phenotype and genotype in patients with tuberculosis. Am. J. Respir. Crit. Care Med. 1997, 155 (5), 1717−22. (20) Sunahara, S.; Uranom; Ogawam. Genetical and geographic studies on isoniazid inactivation. Science 1961, 134 (3489), 1530−1. (21) Cascorbi, I.; Drakoulis, N.; Brockmoller, J.; Maurer, A.; Sperling, K.; Roots, I. Arylamine N-acetyltransferase (NAT2) mutations and their allelic linkage in unrelated Caucasian individuals: correlation with phenotypic activity. Am. J. Hum. Genet. 1995, 57 (3), 581−92. (22) Hiratsuka, M.; Kishikawa, Y.; Takekuma, Y.; Matsuura, M.; Narahara, K.; Inoue, T.; Hamdy, S. I.; Endo, N.; Goto, J.; Mizugaki, M. Genotyping of the N-acetyltransferase 2 polymorphism in the prediction of adverse drug reactions to isoniazid in Japanese patients. Drug Metab. Pharmacokinet. 2002, 17 (4), 357−62. (23) Yard, A. S.; McKennis, H., Jr. Aspects of the metabolism of isoniazid and acetylisoniazid in the human and the dog. J. Med. Pharm. Chem. 1962, 52, 196−203. (24) Nicholson, J. K.; Lindon, J. C.; Holmes, E. ’Metabonomics’: understanding the metabolic responses of living systems to pathophysiological stimuli via multivariate statistical analysis of biological NMR spectroscopic data. Xenobiotica 1999, 29 (11), 1181−9. (25) Lindon, J. C.; Nicholson, J. K.; Holmes, E.; Everett, J. R. Metabonomics: Metabolic processes studied by NMR spectroscopy of biofluids. Concepts Magn. Reson. 2000, 12 (5), 289−320. (26) Gavaghan, C. L.; Holmes, E.; Lenz, E.; Wilson, I. D.; Nicholson, J. K. An NMR-based metabonomic approach to investigate the biochemical consequences of genetic strain differences: application to the C57BL10J and Alpk: ApfCD mouse. FEBS Lett. 2000, 484 (3), 169−174. (27) Nicholson, J. K.; Wilson, I. D. Opinion: understanding ’global’ systems biology: metabonomics and the continuum of metabolism. Nat. Rev. Drug Discovery 2003, 2 (8), 668−76. (28) Lindon, J. C.; Holmes, E.; Nicholson, J. K. Toxicological applications of magnetic resonance. Prog. Nucl. Magn. Reson. Spectrosc. 2004, 45 (1−2), 109−143. (29) Coen, M.; Holmes, E.; Lindon, J. C.; Nicholson, J. K. NMRbased metabolic profiling and metabonomic approaches to problems in molecular toxicology. Chem. Res. Toxicol. 2008, 21 (1), 9−27. (30) Mortishire-Smith, R. J.; Skiles, G. L.; Lawrence, J. W.; Spence, S.; Nicholls, A. W.; Johnson, B. A.; Nicholson, J. K. Use of metabonomics to identify impaired fatty acid metabolism as the mechanism of a drug-induced toxicity. Chem. Res. Toxicol. 2004, 17 (2), 165−73.

ment metabolic signature that was reflective of the differential metabolic fate of INH and the subsequent development of INH-induced adverse CNS effects. Such an approach is widely applicable to the clinical setting to explore the relationship between adverse drug reactions, the metabolic fate of pharmaceuticals and their impact on metabolic homeostasis. This metabonomic strategy may lead to novel means to determine the metabolic basis for adverse drug reactions and ultimately to enable prediction of susceptibility to adverse drug reactions.



ASSOCIATED CONTENT

S Supporting Information *

Integral values for INH-related peaks in urine samples collected at 0−7 and 7−24 h post-dose (Figure S1) and O-PLSregression model of pre-dose urine and post-dose ratio of INHPA to AcINH (Figure S2). This material is available free of charge via the Internet at http://pubs.acs.org.



AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected], [email protected]. Notes

The authors declare no competing financial interest.



ACKNOWLEDGMENTS This work was funded by a BBSRC CASE studentship in collaboration with Pfizer. The MRC Integrative Toxicology Training Partnership (ITTP) is acknowledged for financial support to M.C. We thank Dr Claude Charuel and the technical staff of Pfizer Global R&D in Amboise, France, for their assistance in performing the animal-related work. We thank Dr. O. Cloarec for assistance with validation of statistical models and Dr. T. A. Clayton for valuable contributions to the study design and for assistance in data analysis and interpretation.



REFERENCES

(1) Timbrell, J. A.; Mitchell, J. R.; Snodgrass, W. R.; Nelson, S. D. Isoniazid hepatoxicity: the relationship between covalent binding and metabolism in vivo. J. Pharmacol. Exp. Ther. 1980, 213 (2), 364−9. (2) Preziosi, P. Isoniazid: metabolic aspects and toxicological correlates. Curr. Drug Metab. 2007, 8 (8), 839−51. (3) Nelson, S. D.; Mitchell, J. R.; Timbrell, J. A.; Snodgrass, W. R.; Corcoran, G. B., 3rd Isoniazid and iproniazid: activation of metabolites to toxic intermediates in man and rat. Science 1976, 193 (4256), 901− 3. (4) Timbrell, J. A. The role of metabolism in the hepatotoxicity of isoniazid and iproniazid. Drug Metab. Rev. 1979, 10 (1), 125−47. (5) Mitchell, J. R.; Zimmerman, H. J.; Ishak, K. G.; Thorgeirsson, U. P.; Timbrell, J. A.; Snodgrass, W. R.; Nelson, S. D. Isoniazid liver injury: clinical spectrum, pathology, and probable pathogenesis. Ann. Intern. Med. 1976, 84 (2), 181−92. (6) Ellard, G. A.; Gammon, P. T. Pharmacokinetics of isoniazid metabolism in man. J. Pharmacokinet. Biopharm. 1976, 4 (2), 83−113. (7) Zamboni, V.; Defranceschi, A. Identification of isonicotinoylhydrazones of pyruvic and alpha-ketoglutaric acid in rat urine after treatment with isonicotinic acid hydrazide (isoniazid). Biochim. Biophys. Acta 1954, 14 (3), 430−2. (8) Colvin, L. B. Metabolic fate of hydrazines and hydrazides. J. Pharm. Sci. 1969, 58 (12), 1433−43. (9) Peters, J. H.; Miller, K. S.; Brown, P. Studies on the metabolic basis for the genetically determined capacities for isoniazid inactivation in man. J. Pharmacol. Exp. Ther. 1965, 150 (2), 298−304. L

dx.doi.org/10.1021/pr300430u | J. Proteome Res. XXXX, XXX, XXX−XXX

Journal of Proteome Research

Article

(31) Connor, S. C.; Hodson, M. P.; Ringeissen, S.; Sweatman, B. C.; McGill, P. J.; Waterfield, C. J.; Haselden, J. N. Development of a multivariate statistical model to predict peroxisome proliferation in the rat, based on urinary 1H-NMR spectral patterns. Biomarkers 2004, 9 (4−5), 364−85. (32) Coen, M.; Hong, Y. S.; Clayton, T. A.; Rohde, C. M.; Pearce, J. T.; Reily, M. D.; Robertson, D. G.; Holmes, E.; Lindon, J. C.; Nicholson, J. K. The mechanism of galactosamine toxicity revisited; a metabonomic study. J. Proteome Res. 2007, 6 (7), 2711−9. (33) Shipkova, P.; Vassallo, J. D.; Aranibar, N.; Hnatyshyn, S.; Zhang, H.; Clayton, T. A.; Cantor, G. H.; Sanders, M.; Coen, M.; Lindon, J. C.; Holmes, E.; Nicholson, J. K.; Lehman-McKeeman, L. Urinary metabolites of 2-bromoethanamine identified by stable isotope labelling: evidence for carbamoylation and glutathione conjugation. Xenobiotica 2010, 41 (2), 144−54. (34) Coen, M.; Want, E. J.; Clayton, T. A.; Rhode, C. M.; Hong, Y. S.; Keun, H. C.; Cantor, G. H.; Metz, A. L.; Robertson, D. G.; Reily, M. D.; Holmes, E.; Lindon, J. C.; Nicholson, J. K. Mechanistic aspects and novel biomarkers of responder and non-responder phenotypes in galactosamine-induced hepatitis. J. Proteome Re.s 2009, 8 (11), 5175− 87. (35) Holmes, E.; Bonner, F. W.; Sweatman, B. C.; Lindon, J. C.; Beddell, C. R.; Rahr, E.; Nicholson, J. K. Nuclear-magnetic-resonance spectroscopy and pattern-recognition analysis of the biochemical processes associated with the progression of and recovery from nephrotoxic lesions in the rat induced by mercury(II) chloride and 2bromoethanamine. Mol. Pharmacol. 1992, 42 (5), 922−930. (36) Yap, I. K.; Clayton, T. A.; Tang, H.; Everett, J. R.; Hanton, G.; Provost, J. P.; Le Net, J. L.; Charuel, C.; Lindon, J. C.; Nicholson, J. K. An integrated metabonomic approach to describe temporal metabolic disregulation induced in the rat by the model hepatotoxin allyl formate. J. Proteome Res. 2006, 5 (10), 2675−84. (37) Bohus, E.; Coen, M.; Keun, H. C.; Ebbels, T. M.; Beckonert, O.; Lindon, J. C.; Holmes, E.; Noszal, B.; Nicholson, J. K. Temporal metabonomic modeling of L-arginine-induced exocrine pancreatitis. J. Proteome Res. 2008, 7 (10), 4435−45. (38) Nicholson, J. K. Global systems biology, personalized medicine and molecular epidemiology. Mol. Syst. Biol. 2006, 2, 52. (39) Wilson, I. D. Drugs, bugs, and personalized medicine: pharmacometabonomics enters the ring. Proc. Natl. Acad. Sci. U.S.A. 2009, 106 (34), 14187−8. (40) Clayton, T. A.; Lindon, J. C.; Cloarec, O.; Antti, H.; Charuel, C.; Hanton, G.; Provost, J. P.; Le Net, J. L.; Baker, D.; Walley, R. J.; Everett, J. R.; Nicholson, J. K. Pharmaco-metabonomic phenotyping and personalized drug treatment. Nature 2006, 440 (7087), 1073−7. (41) Clayton, T. A.; Baker, D.; Lindon, J. C.; Everett, J. R.; Nicholson, J. K. Pharmacometabonomic identification of a significant host-microbiome metabolic interaction affecting human drug metabolism. Proc. Natl. Acad. Sci. U.S.A. 2009, 106 (34), 14728−33. (42) Winnike, J. H.; Li, Z.; Wright, F. A.; Macdonald, J. M.; O’Connell, T. M.; Watkins, P. B. Use of pharmaco-metabonomics for early prediction of acetaminophen-induced hepatotoxicity in humans. Clin. Pharmacol. Ther. 2010, 88 (1), 45−51. (43) Backshall, A.; Sharma, R.; Clarke, S. J.; Keun, H. C. Pharmacometabonomic profiling as a predictor of toxicity in patients with inoperable colorectal cancer treated with capecitabine. Clin. Cancer Res. 2011, 17 (9), 3019−28. (44) Stebbing, J.; Sharma, A.; North, B.; Athersuch, T. J.; Zebrowski, A.; Pchejetski, D.; Coombes, R. C.; Nicholson, J. K.; Keun, H. C. A metabolic phenotyping approach to understanding relationships between metabolic syndrome and breast tumour responses to chemotherapy. Ann. Oncol. 2012, 23, 860−6. (45) Beckonert, O.; Keun, H. C.; Ebbels, T. M.; Bundy, J.; Holmes, E.; Lindon, J. C.; Nicholson, J. K. Metabolic profiling, metabolomic and metabonomic procedures for NMR spectroscopy of urine, plasma, serum and tissue extracts. Nat. Protoc. 2007, 2 (11), 2692−703. (46) Aue, W. P.; Bartholdi, E.; Ernst, R. R. Two-dimensional spectroscopy. Application to nuclear magnetic resonance. J. Chem. Phys. 1976, 64 (5), 2229−46.

(47) Braunschweiler, L.; Ernst, R. R. Coherence transfer by isotropic mixing - application to proton correlation spectroscopy. J. Magn. Reson. 1983, 53 (3), 521−528. (48) Aue, W. P.; Karhan, J.; Ernst, R. R. Homonuclear broad-band decoupling and 2-dimensional J-resolved NMR-spectroscopy. J. Chem. Phys. 1976, 64 (10), 4226−4227. (49) Bodenhausen, G.; Ruben, D. J. Natural abundance nitrogen-15 NMR by enhanced heteronuclear spectroscopy. Chem. Phys. Lett. 1980, 69, 185. (50) Bax, A.; Summers, M. F. Proton and carbon-13 assignments from sensitivity-enhanced detection of heteronuclear multiple-bond connectivity by 2D multiple quantum NMR. J. Am. Chem. Soc. 1986, 108, 2093−2094. (51) Veselkov, K. A.; Lindon, J. C.; Ebbels, T. M.; Crockford, D.; Volynkin, V. V.; Holmes, E.; Davies, D. B.; Nicholson, J. K. Recursive segment-wise peak alignment of biological 1h NMR spectra for improved metabolic biomarker recovery. Anal. Chem. 2009, 81 (1), 56−66. (52) Dieterle, F.; Ross, A.; Schlotterbeck, G.; Senn, H. Probabilistic quotient normalization as robust method to account for dilution of complex biological mixtures. Application in 1H NMR metabonomics. Anal. Chem. 2006, 78 (13), 4281−90. (53) Trygg, J.; Holmes, E.; Lundstedt, T. Chemometrics in metabonomics. J. Proteome Res. 2007, 6 (2), 469−79. (54) Trygg, J.; Wold, S. O2-PLS, a two-block (X-Y) latent variable regression (LVR) method with an integral OSC filter. J. Chemom. 2003, 17 (1), 53−64. (55) Cloarec, O.; Dumas, M. E.; Trygg, J.; Craig, A.; Barton, R. H.; Lindon, J. C.; Nicholson, J. K.; Holmes, E. Evaluation of the orthogonal projection on latent structure model limitations caused by chemical shift variability and improved visualization of biomarker changes in 1H NMR spectroscopic metabonomic studies. Anal. Chem. 2005, 77 (2), 517−26. (56) Fox, H. H. Synthetic tuberculostats. VI. Some sugar derivatives of isonicotinylhydrazine. J. Org. Chem. 1953, 18 (8), 990−993. (57) Sumner, S. J.; Burgess, J. P.; Snyder, R. W.; Popp, J. A.; Fennell, T. R. Metabolomics of urine for the assessment of microvesicular lipid accumulation in the liver following isoniazid exposure. Metabolomics 2010, 6 (2), 238−249. (58) O’Brien, P. J.; Slaughter, M. R.; Polley, S. R.; Kramer, K. Advantages of glutamate dehydrogenase as a blood biomarker of acute hepatic injury in rats. Lab. Anim. 2002, 36 (3), 313−21. (59) Cornish, H. H. The role of vitamin B6 in the toxicity of hydrazines. Ann. N.Y. Acad. Sci. 1969, 166 (1), 136−45. (60) Lightcap, E. S.; Silverman, R. B. Slow-binding inhibition of gamma-aminobutyric acid aminotransferase by hydrazine analogues. J. Med. Chem. 1996, 39 (3), 686−94. (61) Timbrell, J. A.; Wright, J. M. Studies on the effects of isoniazid on acetylhydrazine metabolism in vivo and in vitro. Drug Metab. Dispos. 1979, 7 (4), 237−40. (62) Okutur, S. K. Acute isoniazid intoxication: Convulsion, rhabdomyolysis and metabolic acidosis. Turk. J. Med. Sci. 2006, 36 (6), 397. (63) Alvarez, F. G.; Guntupalli, K. K. Isoniazid overdose: four case reports and review of the literature. Intensive Care Med. 1995, 21 (8), 641−4. (64) Chin, L.; Sievers, M. L.; Herrier, R. N.; Picchioni, A. L. Convulsions as the etiology of lactic acidosis in acute isoniazid toxicity in dogs. Toxicol. Appl. Pharmacol. 1979, 49 (2), 377−84.

M

dx.doi.org/10.1021/pr300430u | J. Proteome Res. XXXX, XXX, XXX−XXX