Comparative analysis of novel noninvasive renal biomarkers and metabonomic changes in a rat model of gentamicin nephrotoxicity.

The kidney is one of the main target organs of xenobiotic induced toxicity. However, minor effects on renal function are difficult to detect due to the functional reserve of the kidney. So far, the most commonly used clinical markers of renal injury in routine toxicity studies remain blood urea nitrogen (BUN) and serum creatinine, but they suffer from lack of sensitivity, revealing kidney damage not until 70–80% of the renal epithelial mass has been lost. Because of these limitations, there is a need to establish more sensitive and reliable, preferably noninvasive markers, which may be used to detect toxicant induced kidney injury and monitor renal function during drug development and chemical safety testing. Over the last years, a range of candidate biomarkers for kidney injury have emerged primarily from toxicogenomic studies, and several of these markers have also been shown to be released into urine in response to kidney damage, highlighting their potential to serve as noninvasive markers for nephrotoxicity detection. These include neutrophil gelatinase–associated lipocalin (NGAL, lipocalin-2), which has been shown to be rapidly induced and secreted into urine in a range of preclinical and clinical studies on acute kidney injury (Bennett et al., 2008; Mishra et al., 2004, 2006; Nickolas et al., 2008; Wheeler et al., 2008), clusterin, a secreted glycoprotein synthesized in response to tubular injury (Hidaka et al., 2002; Ishii et al., 2007; Kharasch et al., 2006; Yang et al., 2007), and kidney injury molecule (Kim-1), which has been demonstrated to be suitable for early prediction of graft loss in renal transplant recipients, adverse clinical outcome in patients with acute renal failure, and drug/chemical induced proximal tubule damage (Liangos et al., 2007; Prozialeck et al., 2007; van Timmeren et al., 2007; Zhou et al., 2008). Similarly, elevated levels of tissue inhibitor of metalloproteinases 1 (Timp-1) have been observed in models of kidney injury and in urine of patients with renal disease as compared to healthy controls (Chromek et al., 2003; Horstrup et al., 2002; Wasilewska and Zoch-Zwierz, 2008). In addition to these protein markers, metabonomic approaches, that is, the multicomponent analysis of the biochemical composition of body fluids combined with statistical models, are increasingly being used in drug and chemical safety assessment to diagnose or predict toxicity (Lindon et al., 2007). The principal analytical technique for global metabolic profiling has long been high field proton nuclear magnetic resonance (1H-NMR) spectroscopy, and numerous studies indicate that 1H-NMR based metabonomics can discriminate healthy and diseased, control and treated animals/individuals (Lienemann et al., 2008; Odunsi et al., 2005; Wei et al., 2008). However, 1H-NMR analysis is restricted to a limited number of high-concentration metabolites. An alternative approach is to use mass spectrometry (MS) based metabonomic techniques, which require separation of individual metabolites by either liquid chromatography (LC) or gas chromatography (GC), but offer greater sensitivity as compared to 1H-NMR. Thus, GC-MS or LC-MS not only enable detection of low-concentration metabolites, but also hold promise for biomarker identification. However, it is important to recognize that 1H-NMR spectroscopy and mass spectrometry are complementary tools and that a combination of methods provides wider coverage of metabolites and thus a much more comprehensive picture of the metabolome than any single technique by itself. Utilizing a combination of 1H-NMR, GC-MS, and LC-MS techniques, Atherton et al. (2006) were able to identify metabolic perturbations in the PPAR-α null mutant mouse liver as compared with wild-type mice, ranging from decreased glucose and choline (1H-NMR) to increased steric acid, cholesterol and pentadecanoic acid (GC-MS). Similarly, plasma analysis using all three analytical platforms provided a more comprehensive metabolite profile of normal and Zucker (fa/fa) obese rats than any methodology would have on its own. For instance, GC-MS revealed an increase in arachidonic acid and tocopherol, whereas a rise in taurocholate in Zucker rats was detected using ultra performance LC-MS (Williams et al., 2006). The aim of this study was to compare the sensitivity of a combined 1H-NMR and GC-MS metabonomics approach and a set of novel urinary protein markers for early detection of chemically induced nephrotoxicity. The overall study design and choice of model compound, that is, the aminoglycoside antibiotic gentamicin, was based on a study by Lenz et al. (2005) in which LC-MS analysis of metabolic changes in rat urine revealed altered excretion of xanthurenic acid, kynurenic acid and various sulfate conjugates in addition to increased glucose, lactate and decreased betaine zand trimethylamine-N-oxide detected by 1H-NMR, but an additional low dose and quantitative assessment of the major urinary metabolites detected by 1H-NMR were included in our study to allow a comprehensive assessment of the sensitivity/specificity of metabolic changes and candidate urinary protein markers along with traditional markers of nephrotoxicity.