A systems biology analysis of the unique and overlapping transcriptional responses to caloric restriction and dietary methionine restriction in rats

Abstract

Dietary methionine restriction (MR) and calorie restriction (CR) each improve metabolic health and extend life span. We used comprehensive transcriptome profiling and systems biology analysis to interrogate the unique and overlapping molecular responses in rats provided these dietary regimens for 20 mo after weaning. Microarray analysis was conducted on inguinal white adipose (IWAT), brown adipose tissue (BAT), liver, and skeletal muscle. Compared to controls, CR-induced transcriptomic responses (absolute fold change ≥1.5 and P≤0.05) were comparable in IWAT, BAT, and liver (~800 genes). MR-induced effects were largely restricted to IWAT and liver (~2400 genes). Pathway enrichment and gene-coexpression analyses showed that induction of fatty acid synthesis in IWAT was common to CR and MR, whereas immunity and proinflammatory signaling pathways were specifically down-regulated in MR-treated IWAT and liver (FDR≤0.07-0.3). BAT demonstrated consistent down-regulation of PPAR-signaling under CR and MR, whereas muscle was largely unaffected. Interactome analysis identified CR-specific down-regulation of cytoskeletal matrix components in IWAT and MR-specific up-regulation of ribosomal genes in liver (FDR≤0.001). Transcriptomic down-regulation of inflammation genes by MR in IWAT was consistent with upstream inhibition of STAT3. Together, these results provide an integrated picture of the breadth of transcriptional responses to MR and CR among key metabolic tissues.

Keywords: amino acid sensing; animal models; insulin sensitivity; obesity.

Figure 1.

Comparative analysis of differentially expressed genes using Venn diagrams. Differentially expressed genes (defined as genes with an absolute fold change ≥1.5-fold and P≤0.05) were compared among the 4 tissues studied and for each of the treatments (CR, MR). Number of differentially expressed genes for each comparison is indicated, including the number of overlapping genes for any pairwise or higher-order comparisons.

Figure 2.

Two-way hierarchical clustering of treatments vs. KEGG pathways. Pathway enrichment was determined by GSEA, and pathways with ≤30% FDR were included. Pathways are colored according to whether they were down-regulated (blue), up-regulated (red), or unchanged (white) in the treated groups (MR, CR) compared to control.

Figure 3.

KEGG functional profiles and interaction modules, illustrating transcriptomic interactions in IWAT from MR group (A) and CR group (B) animals. Functional themes, represented by the differentially expressed gene-enriched KEGG pathways, were correlated in a proximity interaction network driven by the expression similarity of their annotated genes. KEGG categories are indicated by nodes, and the strength of the proximity between nodes is depicted by the edges in the correlation network. Nodes are colored red or green to represent overall up- or down-regulation, respectively, of the transcripts underlying the nodes. Edges represented by solid lines indicate stronger proximities compared to dashed lines. The network was further clustered to identify topologically proximal transcriptional nodes based on the number of common coexpressed transcripts shared by the nodes.

Figure 4.

Functional network analysis on differentially expressed genes. Differentially expressed genes were mapped to preexisting interaction networks to generate subnetworks via the ReactomFI plug-in in Cytoscape. Down-regulated nodes (genes) are shown in green, up-regulated nodes in red. Edges are represented variously (arrow, solid line, dashed line) to depict different types of interactions among the nodes. Subnetworks were further clustered to identify substructures. Clustered subnetworks are represented in a circular layout to better visualize the individual clusters and the relationships between clusters. Identities of all genes comprising all clusters are detailed in Ancillary Table A3 (http://www.pbrc.edu/docs/AncillaryTable3.xlsx). A) Network derived from IWAT in MR animals. From the 8 clusters containing ≥5 genes, the cluster enriched in MR down-regulated chemokine genes are highlighted by the red box. Individual components of the chemokine cluster are shown in expanded view. B) Network derived from IWAT in CR animals, showing a highly interconnected module of structural genes. C) Network clusters derived from liver in MR animals. Network clustering identified a distinct MR up-regulated module consisting of ribosomal genes (red box). Gene contents of this module are shown in expanded view.

Figure 5.

Two-way clustering of the subcategories of the inflammatory response biofunction in MR- and CR-treated samples. A) Inflammatory response subfunctions with absolute z score ≥ 2.0 in ≥1 treatment and containing 20–200 genes were considered for clustering. Treatments are clustered in columns and functions clustered in rows. Activation or inhibition of a function is shown in shades of red and green, respectively (based on z scores in IPA). Black indicates the absence of a function from a comparison. No biofunction subcategory was found to pass the filter for CR-treated BAT, IWAT, muscle or MR-treated muscle samples. B) Clustering of inflammation-related subfunctions (20–200 genes) in MR-treated IWAT, based on the similarity in their gene contents. Clusters are differentially colored; number of differentially expressed genes in each subfunction is indicated in parentheses. C) Distribution of the most frequently occurring genes across the inflammation subfunctions. Genes present in ≥20 subfunctions are shown in columns and subfunctions in rows. Red indicates presence and blue indicates absence of a gene subfunction.

Figure 6.

Analysis of upstream regulators. A) Two-way hierarchical clustering of treatments and upstream regulators that are predicted to be activated or inhibited. Analysis was restricted to transcription factors that show significant evidence of activation or inhibition (z score ≥2.0 or ≤−2.0, respectively) based on the direction of differential expression of the genes regulated by them. Treatments are clustered in columns and individual regulators in rows. Regulators are colored according to whether they were inhibited (green), activated (red), or unchanged (black). B) Expression of STAT3 target genes in IWAT from MR animals. Average log gene expression is plotted on the x axis and log fold change (MR/control) is plotted on the y axis. Genes in red are predicted to be activated and those in blue to be inhibited by STAT3. Open circles represent genes for which reliable evidence for activation/inhibition is not available in the IPA knowledge base.