Metabolomics and diabetes: analytical and computational approaches

Abstract

Diabetes is characterized by altered metabolism of key molecules and regulatory pathways. The phenotypic expression of diabetes and associated complications encompasses complex interactions between genetic, environmental, and tissue-specific factors that require an integrated understanding of perturbations in the network of genes, proteins, and metabolites. Metabolomics attempts to systematically identify and quantitate small molecule metabolites from biological systems. The recent rapid development of a variety of analytical platforms based on mass spectrometry and nuclear magnetic resonance have enabled identification of complex metabolic phenotypes. Continued development of bioinformatics and analytical strategies has facilitated the discovery of causal links in understanding the pathophysiology of diabetes and its complications. Here, we summarize the metabolomics workflow, including analytical, statistical, and computational tools, highlight recent applications of metabolomics in diabetes research, and discuss the challenges in the field.

Figure 1

Summary of the metabolomics workflow.

Figure 2

MFA of isotope tracers into glyceraldehyde 3-phosphate (G3P). A: Schematic depicting isotope incorporation into G3P using either U-[¹³C₆]glucose or U-[¹³C₃]lactate. With each isotope-labeled substance, G3P derived from glycolysis or gluconeogenesis, respectively, would have a mass shift of +3 due to all three carbons incorporating the ¹³C label. Comparison of percent incorporation following addition of U-[¹³C₆]glucose or U-[¹³C₃]lactate would allow for the determination of how much G3P is derived from each pathway. Characterization of each metabolite in the pathway (G6P, F6P, FBP, TCA cycle metabolites, etc.) could help identify blockages in each metabolic pathway. B: MS/MS spectrum of [¹²C]G3P (top panel) and [¹³C₃]G3P (bottom panel) in the liver following treatment with U-[¹³C₆]glucose. The area of G3P m+3 (bottom panel) divided by the sum of the total, following correction for naturally occurring ¹³C isotopes, gives the percent of G3P derived from glycolysis following the addition of U-[¹³C₆]glucose.

Figure 3

Comparison of imputation methods for missing values. Methionine concentrations in urine were determined by GC-MS, and 5 out of 27 control subjects had values below the limit of detection. Data were log₂ transformed and analyzed using different imputation methods for the missing values (three nearest neighbors [KNN 3], five nearest neighbors [KNN 5], metabolite mean value [Mean Imp], or metabolite median value [Median Imp]). The median and variance for each imputation method is shown. KNN 3 had the smallest effect on data distribution. +, outlier.

Figure 4

MetScape network for valine, leucine, and isoleucine degradation pathway. A: The metabolites are shown as pink hexagons. The metabolites that were experimentally measured by Wang et al. (45) are shown in red. Green border shows significant metabolites. Gene expression data from Mootha et al. (46) were superimposed on the metabolic network. Gene nodes are blue; the size of the node represents the direction of the change. Dark blue color is reserved for genes from enriched pathways. Gray nodes represent the reactions, and green nodes are enzymes. B: A zoomed in view of the same network, where MetDisease was used to annotate the metabolites with MeSH disease terms. The lower part of the figure shows the portion of MeSH tree. When diabetes mellitus is selected, the related metabolites (in this case valine) are selected. Additional information can be obtained by right-clicking on metabolite node. The insert on the right shows the list of publications that support the connection between the metabolite and the MeSH term.