Single cell high content textural image analysis of dynamic epigenetic signatures as a response to defined microenvironmental parameters.

摘要

Currently, quantifiable investigations of the epigenome require cell lysis and are population based, prohibiting direct investigations of intact intranuclear structural organization and introducing noise into data obtained from inherently heterogeneous stem cell populations. To address this, we have developed and employed a single-cell high-content image informatics framework to capture organizational signatures of epigenetic signaling components from images of cellular nuclei obtained via superresolution nanoscopy. High dimensional quantitative texture descriptors of the organizational dynamics of key posttranslational modifications to core histone proteins were imaged in different human stem cell systems using time-gated stimulated emission depletion confocal nanoscopy. Influential texture descriptors were identified, validated at the nanoscale using immuno-gold electron microscopy, and organizational sub-classifiers were generated from this bioimage informatics data representing a range of "open" versus "closed" chromatin states. When applied to growth factor-induced lineage differentiation of human mesenchymal stem cells, the organizational classifiers showed a clear evolution with temporal cell state, which was more sensitive than the conventional mass spectrometry-based quantitation of the relative abundance of these PTMs. When a range of stem cell phenotypes sharing common DNA sequences were imaged, clear sub-classifiers emerged correlating with the divergent phenotypes for undifferentiated, adipogenic, and osteogenic hMSCs, as well as for human foreskin fibroblasts, induced pluripotent stem cells, neural stem cells, and reprogrammed neurons. Thus, high content bioimage informatics reflective of chromatin organization yields a higher order organizational signature corresponding to an epigenetic "activity" state.;To elucidate the influence of biophysical factors on stem cell epigenetic states, these imaging-based organizational classifiers were tested on human mesenchymal stem cells exposed to physically constraining cues, and successfully predicted the early differentiation toward adipogenic hMSCs on hydrogel substrates with spatially graded mechanical stiffness, as well as osteogenic hMSCs on soft-lithographed, graded nanotopographies. In summary, in contrast to the traditional reductionist, population-level readouts in epigenomics, the approach outlined in this thesis offers a more integrated, single-cell, organizational index of emergent stem cell activity in response to defined environmental cues, and can be applied for the screening of discrete microenvironmental properties for the enhancement of stem cell behavioral control and facilitated integration in regenerative medicine applications.