Description
Multivariate Image
Analysis of Epigenetic Landscape at Single Cell Level Identifies Signature of Glioma
Stem Cell Differentiation
Chen Farhy1, Dmitry Myshkin1, Santosh Hariharan2,
David W. Andrews2 and Alexey V. Terskikh1
1Sanford Burnham Prebys Medical Discovery
Institute, La Jolla, CA, USA; 2Sunnybrook Research Institute,
University of Toronto, ON, Canada
Treatment of Glioblastoma
multiform (GBM) inevitably results in tumor recurrence due to a highly
resistant subpopulation within the GBM mass known as glioma stem cells (GSCs). Failure
of cytotoxic drugs to improve clinical outcome led to development of approaches
which attempt to induce GSCs differentiation to less tumorigenic cell fates. However,
the comprehensive transcriptomic and epigenetic profiling methods required to
monitor GCS differentiation are not applicable for the high-throughput
screening.
We developed a novel
approach, which profiles chromatin organization alterations at single cell
level using automated imaging and multiparametric analyses. Microscopic Imaging
of Epigenetic Landscape (MIEL) captures the nuclear staining texture obtained
by immunofluorescence for histone and DNA modifications to derive textural
features. Multivariate analysis and supervised learning models assess the
similarity of textures between cell populations and identify classifiers which
accurately distinguish them.
Using MIEL we defined the
epigenetic signatures of various cell types enabling identification of each
cell type with high accuracy. Employing drugs targeting multiple molecular
mechanisms as well as treatments known to induce GSC differentiation we
categorized the textural alterations associated with specific epigenetics
changes to be used in high-throughput screening. Finally, the texture features
were used in pseudotemporal ordering algorithms to reveal at single cell level,
the dynamics of epigenetic changes during GCS differentiation and identified
differences between several GSCs lines.
Our results demonstrate
that by providing a snapshot of the entire epigenetic landscape, MIEL allows
screening for compounds inducing transient and poorly defined cellular states as
well as conducting phenotypic screens.