Predicting chemoevasion potential of cancer cells from statistical chromatin packing behavior.

Abstract

In malignancy, genetic, metabolic, or environmental insults may remodel chromatin structure, potentially increasing the plasticity of aberrant cell state transitions. Observed alterations to the epigenetic landscape of chromatin include gain-of-function mutations of Polycomb repressor EZH2 that block differentiation. Nucleosome remodeling and higher-order chromatin organization may also increase malignancy, with abnormalities ranging from prevalent mutations of ATP-dependent chromatin remodelers to mixing between transcriptionally active and inactive compartments. However, a universal relationship between chromatin structure and the disease state does not yet exist. Here, using the first principles of physics, we demonstrate that chromatin packing, or the statistical distribution of chromatin density in 4D, is a genome-wide regulator of phenotypic plasticity. The nucleus is a highly crowded, heterogeneous environment where DNA, RNA, proteins, and other macromolecules exert excluded volume effects, altering the kinetics and efficiency of transcription reactions. As chromatin is the major crowder in the nucleus, characterizing the chromatin mass density distribution facilitates a more complete understanding of transcription processes in a realistic nuclear environment. Combining high-resolution chromatin scanning transmission electron microscopy (ChromSTEM) and live-cell, label-free Partial Wave Spectroscopy (PWS) along with polymer physics-based analysis, we identified the existence of spatially separable packing domains, around 80 nm in radius. We then employed our understanding of fundamental chromatin packing to build a chromatin-mediated model of transcription that incorporates two pertinent aspects of phenotypic plasticity: transcriptional malleability, the ability of a cell to upregulate certain genes within a critical time frame, and transcriptional heterogeneity, a cell population's range of functional states. Finally, we extended our transcription model to predict how the average chromatin packing behavior of a cell population contributes directly to cancer cell survival upon chemotherapy treatment. Experimental observations confirmed model predictions, underlying the importance of chromatin packing in chemoevasion processes.