Abstract 232: High Glucose Attenuates In Vitro Tube Growth of Microvascular Endothelial Cells

Abstract

Aim: To investigate how high glucose affects the ability of microvascular endothelial cells to form tube-like structures in vitro and to develop open source software that automates the analysis of tube formation assays. Background: Over 300 million people worldwide suffer from diabetes. Patients with diabetes experience impaired angiogenesis. Tube formation assays (TFAs) mimic in vivo angiogenesis. During a TFA, endothelial cells that are placed onto a matrigel coated surface arrange into tube-like structures. Quantification of TFAs is difficult; automation typically requires a fluorescent stain and a specific software package. Manual quantification is tedious and subject to inter-user variability. In this study, we describe how high glucose affects the ability of rat microvascular endothelial cells (RMVECs) to form tube-like structures and the development of an open source tool called “Pipeline” that analyzes brightfield TFA images quantifying the total tube length and number of branchpoints. Methods: RMVECs were cultured in media with normal (5.6 mM) or high glucose (25 mM) for one and two week intervals. RMVECs were added to four-chamber slides coated with growth factor reduced Matrigel. After 24 and 48 hours, brightfield images were taken of the TFA. Images were analyzed by manually tracing the total tube length in MetaMorph image processing software or automatically using Pipeline. Results from automatic and manual analysis were compared. The algorithm was validated by measuring the total length of images generated in silico . The code was compiled in C and will run on any computer with the freely available MATLAB R2012a runtime environment. Results: A significant decrease in the total tube length was found when comparing groups treated with high glucose for 2 weeks versus 1 weeks (20.3 +/- 1.4mm vs 25.2 +/- 1.3mm, p<0.05). Additionally, cells in high glucose for 2 weeks versus 1 week were found to have significantly less branch points (68 +/- 6.1 vs 90.9 +/- 5.8, p<0.05). When validating the Pipeline algorithm against images generated in silico (at multiple orientations), Pipeline was able to detect total image length within 1% (9216μm vs 9348 +/- 98μm). Analysis using pipeline was found to be 25X faster than manual analysis (7 minutes versus 180 minutes).