Reduced central and peripheral inflammatory responses and increased mitochondrial activity contribute to diet-induced obesity resistance in WSB/EiJ mice

Energy imbalance due to excess of calories is considered to be a major player in the current worldwide obesity pandemic and could be accompanied by systemic and central inflammation and mitochondrial dysfunctions. This hypothesis was tested by comparing the wild-derived diet-induced obesity- (DIO-) resistant mouse strain WSB/EiJ to the obesity-prone C57BL/6J strain. We analysed circulating and hypothalamic markers of inflammatory status and hypothalamic mitochondrial activity in both strains exposed to high-fat diet (HFD). We further analysed the regulations of hypothalamic genes involved in inflammation and mitochondrial pathways by high throughput microfluidic qPCR on RNA extracted from laser micro-dissected arcuate (ARC) and paraventricular (PVN) hypothalamic nuclei. HFD induced increased body weight gain, circulating levels of leptin, cholesterol, HDL and LDL in C57BL/6J whereas WSB/EiJ mice displayed a lower inflammatory status, both peripherally (lower levels of circulating cytokines) and centrally (less activated microglia in the hypothalamus) as well as more reactive mitochondria in the hypothalamus. The gene expression data analysis allowed identifying strain-specific hypothalamic metabolic pathways involved in the respective responses to HFD. Our results point to the involvement of hypothalamic inflammatory and mitochondrial pathways as key factors in the control of energy homeostasis and the resistance to DIO.


Bioinformatic data analysis
A custom R tool was constructed to perform the principal component analysis (PCA) using R packages: the FactoMineR (version 1.35) and missMDA (version 1.11). The PCA analysis was done starting from the dCT data for the 86 genes expression data measured by microfluidic qPCR, and the variance in the top two principal components is shown in the plots (Dim1, Dim2). Each dot represents an individual sample. Each ellipsoid represents a group strain-diet.

Figure S1. Principal component analysis (PCA) discriminates WSB/EiJ and C57BL/6J
strains on the basis of hypothalamic gene expression. Genes representative of inflammation and mitochondrial pathways were computed in a PCA analysis to verify whether the response to HFD differed between WSB/EiJ and C57BL/6J mice in ARC (upper panel) and PVN (lower panel). The individual factor maps revealed a clear discrimination between both strains around Dimension 2, especially in PVN. However, the response to HFD could not really be identified on the basis of PCA analysis. The contribution of each gene along PCA dimensions 1 and 2 is represented in Table S1 for ARC and Table S2 for PVN. Figure S2. Gene expression data for genes significantly more expressed in the ARC of WSB/EiJ mice compared to C57BL/6J mice: Data were collected by microfluidic qRT-PCR from microdissected ARC of twelve week-old C57BL/6J (red) and WSB/EiJ (green) mice either maintained under control feeding (CTRL), or challenged with HFD for the last 3 days of the period (3D) or during the whole 8-week experiment duration (8WK). A: genes more expressed in the WSB/EiJ, related to mitochondria. B: genes more expressed in the WSB/EiJ, related to inflammation. Gene expression data were analyzed using non-parametric two-way Anova with permutations to test the effects of diet, strain and their interaction. Post Hoc tests results are indicated on the graph (*, P≤0.05; **, P≤0.01; ***, P≤0.001. n=5-7); FC: relative fold-change expression compared to C57BL/6J control group.
. Figure S3. Gene expression data for genes significantly more expressed in the PVN of WSB/EiJ mice compared to C57BL/6J mice: Data were collected by microfluidic qRT-PCR from microdissected PVN of twelve week-old C57BL/6J (red) and WSB/EiJ (green) mice either maintained under control feeding (CTRL), or challenged with HFD for the last 3 days of the period (3D) or during the whole 8-week experiment duration (8WK). A: genes more expressed in the WSB/EiJ, related to mitochondria. B: genes more expressed in the WSB/EiJ, related to inflammation. Gene expression data were analyzed using non-parametric two-way Anova with permutations to test the effects of diet, strain and their interaction. Post Hoc tests results are indicated on the graph (*, P≤0.05; **, P≤0.01; ***, P≤0.001. n=5-7); FC: relative fold-change expression compared to C57BL/6J control group.
. Figure S4. Gene expression data for genes significantly less expressed in the ARC and PVN of WSB/EiJ mice compared to C57BL/6J mice Data were collected by microfluidic qRT-PCR from microdissected ARC and PVN of twelve week-old C57BL/6J (red) and WSB/EiJ (green) mice either maintained under control feeding (CTRL), or challenged with HFD for the last 3 days of the period (3D) or during the whole 8-week experiment duration (8WK). A: genes less expressed in the ARC of WSB/EiJ, related to mitochondria. B: genes less expressed in the ARC of WSB/EiJ, related to inflammation. C: genes less expressed in the PVN of WSB/EiJ, related to mitochondria. D: genes less expressed in the PVN of WSB/EiJ, related to inflammation. Gene expression data were analyzed using non-parametric two-way Anova with permutations to test the effects of diet, strain and their interaction. Post Hoc tests results are indicated on the graph (*, P≤0.05; **, P≤0.01; ***, P≤0.001. n=5-7); FC: relative fold-change expression compared to C57BL/6J control group.   Table S1. Summary table of the statistics regarding PCA and gene expression analyses for the ARC; PCA (yellow tables): list of genes showing a significant contribution to PCA dimension 2 (which discriminates C57BL/6J and WSB/EiJ mice ( Figure S1)), classify according to the pval of their correlation to the dimension 2 (the data positively correlated to dim 2 are given at the top of the table, the data negatively correlated to dim 2 are given at the bottom of the table). The genes were classified after the PCA accordingly to their involvement in the inflammatory, mitochondrial or both pathways (using Kegg or Genomatix tools). Finally, the significant contributions to dimension 2 of the Strain*Diet interaction, the C57BL/6J 3d-HFD group and the WSB/EiJ 8w-HFD group are also provided (bottom table).

Microfluidics qPCR data statistical analyses (non-parametric two-way ANOVA with permutations) (blue table):
The effects of Strain, Diet and Strain*Diet were statistically tested for each gene from qPCR data (NS (not significant): p-Value ≥ 0.05).