Gas chromatography-mass spectrometry-based untargeted metabolomic analysis of organ tissue, plasma and urine samples from a Rat Model of polycystic kidney disease

Introduction: A predisposition to nephronophthisis (NPHP) is inherited and typically presents with cysts in the kidney and liver, leading to end-stage kidney disease. The mechanisms underlying onset and progression of cyst growth remain unknown and detection of NPHP and other polycystic kidney diseases (PKDs) is not sensitive or specific. For these reasons, management and treatment are limited to renal replacement therapy and transplantation. The LPK rat phenotype has been characterized and classified as a model of the PKD, NPHP9, caused by mutation of the nek8 gene. Using this model, the aim of this collection of studies was to use a GC-MS-based untargeted metabolomic analysis to determine key biochemical changes in kidney and liver tissue of the LPK rat and to investigate biomarkers in the blood plasma and urine. Furthermore, the study determined whether sample derivatisation could be streamlined in an automated process to improve reproducibility and investigated the use of BSTFA as a derivatisation reagent, as an alternative to MSTFA. Methods: Following a pilot study using 4 LPK and 4 Lewis controls, 11 LPK and 11 Lewis age- and sex-matched control animals aged 5 to 16 weeks were used. Blood and urine were sampled weekly and organs harvested at the conclusion of the study. Metabolites were extracted with methanol and water containing 13C6-sorbitol (IS) and derivatised with methoxyamine-HCl and MSTFA. A Shimadzu QP2010 Ultra GC-MS was used for sample analysis, and for data analysis, AnalyzerPro, The Unscrambler X and SPSS were used. Features were matched to an in-house library of metabolites and the NIST mass spectral database. Results: For metabolomic analysis of the kidney and liver tissue, principal component analysis (PCA) distinguished signal corrected metabolite profiles from Lewis and LPK rats iv for kidney (PC-1 77%) and liver (PC-1 46%) tissue. In kidney tissue, 122 metabolites were found to be significantly different between the LPK and Lewis strains and five biochemical pathways showed three or more significantly altered metabolites: transcription/translation, arginine and proline metabolism, alpha-linolenic and linoleic acid metabolism, the citric acid cycle and the urea cycle. In the liver, 30 metabolites were found to be significantly different. Urine and plasma metabolites were tested for age (Kruskal-Wallis; p < 0.05) and strain effects (Mann-Whitney U-test; p < 0.05). Fifty-nine putatively identified metabolites from the LPK plasma and urine were found to be significantly different from Lewis controls. These results were concomitant with data from kidney and liver tissue analyses. The results of these studies validate and complement the current literature and are consistent with suggestions relating to the pathobiology of PKD. Most notably, myo-inositol was suggested as an early marker of renal dysfunction in PKD. Derivatised metabolite responses were highly variable throughout the ten analytical batches of urine samples compared to the 12 batches of plasma samples, even for test mixtures, which were not affected by sample concentration or matrix. Derivatization reagent and protocol are key factors affecting the reproducibility and intensity of individual urinary metabolites, so we tested both BSTFA as an alternate to MSTFA and the use of automated protocols (batch and in-time) using a CTC CombiPAL auto sampler. Of 249 features detected in rat urine, 40 features were significantly different (p < 0.05) based upon reagent and 154 features were significantly different (p < 0.05) based upon protocol. The overall reproducibility of the methods was similar, although highly feature dependent