Machine learning-assisted Repli-Histo labeling reveals distinct transcription-dependent constraints on chromatin motion in living cells

Genomic DNA is wrapped around core histones to form nucleosomes, which are organized in cells from euchromatin to heterochromatin with distinct genome functions. Although transcription is known to shape chromatin behavior in live cells, it remains unclear how different transcription systems act across chromatin classes and nuclear subcompartments. We develop machine learning-assisted Repli-Histo labeling to classify euchromatin and heterochromatin classes (classes IA, IB, II, and III) and combine it with single-nucleosome imaging in living human cells. Nucleosome motion is progressively constrained from euchromatin to heterochromatin. RNA polymerase II inhibition by THZ1, DRB, or α-amanitin increases nucleosome motion in euchromatic classes IA and IB and in heterochromatin around nucleoli, but not at the nuclear periphery. In contrast, RNA polymerase I inhibition by CX-5461 selectively increases nucleosome motion in Class III heterochromatin around nucleoli. Our study reveals that Pol II and Pol I transcription shape chromatin behavior in distinct chromatin classes and nuclear subcompartments.