Multi-omics study identifies unique metabolic responses to high glycemic diet in liver and retina

High glycemic index (HGI) diet induces hyperglycemia, an etiologic risk factor in major diseases affecting multiple organ systems. We hypothesized that adaptation to dietary glycemic exposure is tissue-specific and depends on the length of intake. We therefore evaluated transcriptomic responses to HGI diet after one or 12-months in the liver and retina of C57BL/6J wild-type mice. In the liver, the genes associated with inflammation and fatty acid metabolism were altered within one-month of HGI diet, whereas 12-month HGI diet-fed group showed dysregulated expression of multiple cytochrome P450 genes and overexpression of key lipogenic factors including Srebf1 and Elovl5. In contrast, retinal transcriptomes exhibited fewer HGI-related changes after one-month, and notable alterations in energy metabolism genes were apparent after 12-months of HGI diet. Fatty acid profiles of liver samples corroborated transcriptomic trends and revealed elevated levels of monounsaturated fatty acid in the HGI group. Conversely, saturated fatty acids and polyunsaturated fatty acids were lower in livers of HGI fed animals, regardless of length of exposure. HGI also resulted in a significant increase in blood low-density lipoprotein (LDL) and overall cholesterol levels. We also uncovered diet-aging interactions affecting expression of mitochondrial oxidative phosphorylation genes in the liver, and inherited disease-associated genes in the retina. Our findings provide new insights into tissue-specific adaptive mechanisms, focusing on the liver and the retina, to dietary hyperglycemia. Male C57BL/6J mice at either 6 or 12 months (M) of age were fed a high glycemic index (HGI) or low glycemic index (LGI) diet for 1 or 12 months, respectively. The LGI diet consisted of 70% amylose and 30% amylopectin and the HGI diet consisted of 100% amylopectin. Mice were pair-fed to ensure equal consumption in both HGI and LGI diet groups. At experimental endpoint, mice were fasted for 6 hours and euthanized. Tissues were collected and snap frozen in liquid nitrogen. Total RNA was isolated from samples using a RNeasy Mini Kit according to the manufacturer’s instructions (Qiagen). Whole retinas or 100mg of liver tissue were homogenized and RNA-seq libraries were prepared using a TruSeq Stranded mRNA Sample Prep Kit (Illumina). Libraries were pair-end sequenced to 101 bases on an Illumina HiSeq 2500 system. Raw fastq files from the Illumina HiSeq 2500 System were quality checked and aligned to the v104 Mus musculus Ensembl gene annotation. The gene counts matrix was then normalized and expressed in counts per million (CPM) using the edgeR package from Bioconductor. To remove lowly expressed genes, a filter was used to select for genes with an average expression of greater than 1 CPM in any experimental group. Due to sequencing liver and retina samples on the same lane, 29 highly expressed retinal genes, which comprised the highest loading score values from principal component analysis, were removed from the liver samples. Differential expression analysis comparing HGI relative to LGI groups was conducted using the voom function on the normalized, filtered counts and fitted to a linear model with the limma Bioconductor package.