Figure 2

Discriminant analysis clearly segregates dietary treatment groups in a mouse model of pre-existing obesity (Cohort 2). In a pre-existing obesity mouse model (control DIO diet for 12 weeks), a PLS-DA model using 61 plasma lipid metabolites, physiologic measurements, and mRNA abundance in retroperitoneal adipose tissue (A-), liver (L-), and nodose ganglia (N-) illustrates differential distribution between dietary treatment groups. A plots the latent variable 1 (LV1; X axis) and 2 (LV2; Y axis) scores for each mouse fed 0.5% calcium (Control; pink) and 1.5% calcium (HighCa; yellow) or 1.5% calcium + nonfat dry milk (NFDM; green), while B plots the variable weights in these axes. Variables were grouped and color coded based on a hierarchical cluster analysis. Selected variables from clusters and lipid classes driving group segregation are labeled (for a complete list of variables, see Additional file 3: Tables S3 and S4). Plasma fatty acid diols cluster together (green cluster) and are reduced in NFDM-fed mice relative to Control and HighCa-fed mice. Conversely, monoacylglycerols cluster with a variety of adipose, liver and nodose ganglia gene expression levels (purple cluster), and are elevated in the NFDM-fed mice.