Files
Download Full Text (887 KB)
Abstract
How well can we predict efflux by ATP-binding cassette G2?
It is estimated that there will be about 1.6 million new cases of cancer and half a million cancer deaths in the US during 2015.ATP-binding cassette (ABC) efflux transporters such as ABCG2 are overexpressed in chemotherapy-resistant cancer cells. Anticancer drugs are prone to efflux by these transporters. Being able to identify drugs that are effluxed is of great interest in drug discovery.The current arsenal of methods used to detect efflux are not easily adaptable to high throughput formats and are limited in scope, making experimental analysis an expensive prospect. Hence, computational analysis of efflux is of interest. We accumulated a dataset of ABCG2 substrates and non-substrates which contains 179 substrates and 110 non-substrates. This dataset forms the basis for all studies reported herein. We attempted to identify descriptors capable of segregating substrates and non-substrates, ending up with Log P, Polar Volume, Atom Count, Radius of Gyration, Binding Energy, Length, and Width. They had significant differential distribution that could be used to build mathematical models to fulfil our goals. A statistical learning method that creates non-linear models called Support Vector Machine generated the best predictive models. This model demonstrated between 75-80% accuracy in identifying substrates and non-substrates. Importantly our model suggests mechanistic details of the efflux process; previous reports have suggested that Arg482 in ABCG2 is critical for transport of substrates. In conclusion, we were able to address efflux of chemotherapeutic agents by the ABCG2 efflux pump using a mathematical modeling approach.
Publication Date
2015
Subject Major(s)
Biology
Keywords
ABCG2, BCRP, Substrate Specificity, Modeling, R482, Discriminant Analysis
Disciplines
Integrative Biology | Molecular Biology
Current Academic Year
Senior
Faculty Advisor/Mentor
Glen E. Kellogg
Rights
© The Author(s)