Methylation pseudotime analysis for label-free profiling of the temporal chromatin landscape with long-read and single-molecule resolution [ATAC-seq]

Faithful epigenetic inheritance across cell divisions is essential to maintaining cell identity and involves numerous epigenetic modifications, whose roles in coordinating chromatin architecture are less understood. Technological approaches to temporally order epigenetic modifications throughout the cell cycle often face limitations in sequence resolution and rely on potentially damaging mitotic labeling or conversion steps. Herein, we present Methylation Pseudotime Analysis Through read-level Heterogeneity (MPATH), a label- and conversion-free method to infer post-replication DNA strand maturity from methylation patterns across single molecules. We use MPATH to temporally order hydroxymethylation throughout mitotic inheritance revealing, for the first time, that CpGs within cis-regulatory elements undergo transitions between methylation states at sub-cell-cycle timescales. When applied to long reads generated by NOMe-seq, MPATH uncovered relationships between nucleosome occupancy and DNA maturity. Finally, extension of MPATH to phased reads reveals allele-specific trends in pseudotime distribution associated with X chromosome inactivation. Our findings suggest that when coupled with multimodal sequencing strategies, MPATH could provide valuable insights into chromatin restoration dynamics. Overall design: HUES64 hESCs were subjected to several different experimental conditions prior to DNA extraction for long-read sequencing. In the first experiment, hESCs were treated with 2µg/ml nocodazole for 16h. BrdU pulse-chase experiments were conducted on HUES64 hESCs to investigate DNA methylation patterns during chromatin maturation. For the pulse optimization, cells were subjected to 50 µM BrdU for durations of 10 minutes, 20 minutes, 30 minutes, and 1 hour, followed by genomic DNA collection at 0h (nascent chromatin) and 16h (mature chromatin). Mock controls were included for both 0h and 16h timepoints. Optimized labeling conditions included a 500 µM BrdU pulse (10 minute duration) with 0h and 16h chase points. Unlabeled HUES64 samples were also generated (sequenced to a total of 62x coverage across 2 technical replicates). All long-read data was acquired using ONT sequencing platforms. In a separate sequencing experiment, hESCs were subjected to Repli-ATAC sequencing in order to temporally profile chromatin accessibility across replication-associated timescales.