Abstract: FR-PO997

Transcriptome Signatures for Dietary Fructose Induced Changes in the Proximal Tubule

Session Information

Category: Bioengineering and Informatics

  • 101 Bioengineering and Informatics


  • Hopfer, Ulrich, Case Western Reserve University, Cleveland, Ohio, United States
  • Gonzalez-Vicente, Agustin, Case Western Reserve University, Cleveland, Ohio, United States
  • Garvin, Jeffrey L., Case Western Reserve University, Cleveland, Ohio, United States

Fructose consumption has been associated with renal dysfunction and salt-sensitive hypertension. Proximal tubules reabsorb and metabolize fructose.


To study the effect of fructose on genetic programs in proximal tubules, rats were given a 20% fructose (FRU) or water for 7 days. We used 4 groups of rats fed either chow 1 (normal salt) or 2 (high salt) +/- FRU. Each group contained 4 rats. The exposure time to FRU was short enough so that the rats showed no signs of metabolic syndrome. Total RNA from the superficial kidney cortex was analyzed using the Affymetrix microarray RaGene-2_0-st. Differential expression (DE) analysis of FRU vs. water on chow 1 or 2, was carried out using 3 different statistical methods: 1) Univariate ANOVA (Affymetrix TAC software), 2) Bayesian (BAMarray), and 3) Multivariate Characteristic Direction (CD). CD analysis of DE includes also information from the co-variance matrix between pairs of groups and calculates a vector of co-regulated genes ranked on a quantitative relative scale.


Cosine similarity comparison of the FRU/water vectors from the 2 different chows showed greater similarity than FRU/water vector under either chow versus chow-alone vector. Intersection of the high ranking genes from the two FRU/water vectors (under diet 1 or 2) yielded 139 FRU-specific genes. Only 6 of these genes were also identified as DE genes by both univariate methods. They include the 2 most highly ranked CD genes as well as some lower ranked ones, suggesting that the univariate methods miss many DE genes because of high FRU-unrelated variance. DE genes identified by CD include G6pc, Slc5a8, Tkfc, Slc2a5, Khk, which directly transport or metabolize fructose in proximal tubules. Enrichment for GO:Biological Processes identified “response to nutrient” (GO:0007584), “cellular response to fructose stimulus” (GO:0071332) and “response to fructose” (GO:0009750). Enrichment for GO:Cellular Constituents identified 11 mitochondrial enzymes, suggesting amino acid catabolism, cataplerosis, ketogenesis, and fatty acid synthesis. DE of the cytosolic enzymes Pck1 and G6pc in FRU regardless of chow, suggest increases in gluconeogenesis and glyceroneogenesis.


In summary, the CD method outperforms commonly used univariate methods and provides a meaningful transcriptome signature of FRU in proximal tubules.


  • Other NIH Support