Junguk Hur

Junguk Hur, Ph.D.
10

Ph.D. Program
Assistant Professor
University of North Dakota

Chair

Dissertation Title

Integration of Text Mining with Systems Biology Provides New Insight into the Pathogenesis of Diabetic Neuropathy

Research Interest

Diabetic neuropathy (DN) is the most common complication of diabetes affecting approximately 60% of all diabetic patients leading to significant mortality, morbidity, and poor quality of life. Though more than 50% of patients with DN develop substantial nerve damage prior to noticeable symptoms, no biomarkers for predicting the onset or progression of DN are currently available. Here we present a biomarker discovery platform integrating literature mining and a systems biology approach to identify potential DN biomarkers. A web-based target identification and functional analysis tool, SciMiner (http://jdrf.neurology.med.umich.edu/SciMiner), was developed that identifies targets using a context specific analysis of MEDLINE abstracts and full texts. A comprehensive list of 1,026 targets from diabetes and reactive oxygen species (ROS) related literature was compiled by SciMiner. The expression levels of nine genes, selected from the over-represented ROS-diabetes targets, were measured in the dorsal root ganglia (DRG) of diabetic and non-diabetic DBA/2J mice. Eight genes exhibited significant differential expression and the directions of expression change in six of those genes paralleled enhanced oxidative stress in the DRG, suggesting the involvement of ROS related targets in DN. A microarray analysis was also performed on sural nerve biopsies from two DN patient groups with fast or slow DN progression to identify gene expression profiles related to DN progression. In the fast progressing DN, defense response and inflammatory response related genes were up-regulated, while lipid metabolic process and peroxisome proliferator-activated receptor (PPAR) signaling pathway related genes were down-regulated. We also developed mRNA expression signatures that predict DN progression in humans with a high prediction accuracy. Ridge-regression based predictive models with 14 genes achieved a prediction accuracy of 92% (correct prediction of 11 out of 12 patients). Our results identifying the unique gene signatures of progressive DN and compiling ROS-diabetes targets can facilitate the development of new mechanism-based therapies and predictive biomarkers of DN.

Current Placement

University of North Dakota