A common feature in most cancers is both inter- (between patients) and intra- (within a patient) tumor heterogeneity. An important step toward improving treatment strategies and enabling personal care is mapping how these types of heterogeneity impact clinical phenotypes, especially among deadly tumors such as glioblastoma multiform (GBM). Recent studies have identified instances where the three-dimensional folding of chromatin into DNA loops is associated with inter-tumor heterogeneity. Presently, intra-tumor DNA looping variability has not been measured though this is likely responsible for single-cell transcriptional differences observed within patient tumors. To identify DNA loops genome wide, many chromatin conformation capture (3C)-derived assays have been developed. However, reliably using DNA loops to uncover tumor heterogeneity is hindered by two key deficiencies. First, a direct comparison of 3C-derived techniques has not been conducted to assess assay- specific biases in identifying inter-tumor variable DNA loops. Second, each of these approaches requires millions of cells to infer chromatin structure, obscuring differences at the single-cell level. Here, I propose methodological advances to address these two deficiencies through computational approaches that will elucidate the role of DNA looping in inter- and intra- tumor heterogeneity in GBM.
In Aim 1, I will use data generated in my sponsor's lab for three different 3C-derived methods mapping DNA loops in isocitrate dehydrogenase (IDH) mutant and wildtype glioblastoma cell lines. I will identify biases specific to each assay and determine differential loops associated with the IDH mutation. This work will be critical for developing future computational techniques for identifying important DNA loops. Moreover, this analysis will reveal the epigenetic effects of the IDH mutation, which is prevalent in GBM and other cancers (e.g. acute myeloid leukemia). Results from this aim will be broadly applicable to bioinformatics researchers developing tools for DNA looping data as well as cancer biologists seeking to understand the IDH mutation.
In Aim 2, I propose to resolve single-cell differences in the same glioblastoma cell lines to infer patterns of chromatin loop variability within individual tumors. Specifically, I will build a computational framework integrating DNA loops nominated by bulk populations with single-cell chromatin accessibility (scATAC-seq) data. I will work with the inventor of the scATAC-seq technology to develop a sensitive, zero-inflated model to identify chromatin loops that are variable within individual GBM tumor models. The research results from this proposal will yield critical insights into chromatin biology associated with tumor heterogeneity of GBM and other cancers, which will motivate future therapeutic development strategies.
Inside every human cell, nearly six feet of DNA must be folded into a three-dimensional structure that is altogether thinner than the width of a human hair. Strikingly, abnormal DNA folding has become a signature of many human diseases, including glioblastoma multiform, but the overall importance of changing three- dimensional DNA in cancer is poorly understood. This project provides key bioinformatics and computational advancements in an effort to understand differences in the three-dimensional genome both within and between glioblastoma tumors. !
