Genome-wide analysis of replication timing by next-generation sequencing with E/L Repli-seq

Abstract

This protocol is an extension to: Nat. Protoc. 6, 870-895 (2014); doi:10.1038/nprot.2011.328; published online 02 June 2011Cycling cells duplicate their DNA content during S phase, following a defined program called replication timing (RT). Early- and late-replicating regions differ in terms of mutation rates, transcriptional activity, chromatin marks and subnuclear position. Moreover, RT is regulated during development and is altered in diseases. Here, we describe E/L Repli-seq, an extension of our Repli-chip protocol. E/L Repli-seq is a rapid, robust and relatively inexpensive protocol for analyzing RT by next-generation sequencing (NGS), allowing genome-wide assessment of how cellular processes are linked to RT. Briefly, cells are pulse-labeled with BrdU, and early and late S-phase fractions are sorted by flow cytometry. Labeled nascent DNA is immunoprecipitated from both fractions and sequenced. Data processing leads to a single bedGraph file containing the ratio of nascent DNA from early versus late S-phase fractions. The results are comparable to those of Repli-chip, with the additional benefits of genome-wide sequence information and an increased dynamic range. We also provide computational pipelines for downstream analyses, for parsing phased genomes using single-nucleotide polymorphisms (SNPs) to analyze RT allelic asynchrony, and for direct comparison to Repli-chip data. This protocol can be performed in up to 3 d before sequencing, and requires basic cellular and molecular biology skills, as well as a basic understanding of Unix and R.

Figure 1:. Overview of repli-seq protocol and analysis.

Cultured cells are pulse-labeled with BrdU, fixed and sorted into early and late S phase fractions depending on DNA content by flow cytometry. DNA from both cell fractions is purified, fragmented and adaptors are ligated. BrdU labeled DNA is immunoprecipitated, indexed and amplified. The indexed samples are pooled and sequenced. Sequenced reads are aligned on the reference genome and the normalized coverage is calculated for each cell fraction. Log ratio of early to late coverage is calculated and data are normalized and smoothed.

Figure 2:. Quantile normalization and Loess smoothing allow comparison between samples.

A.: Replication timing (RT) profiles of two technical repli-seq replicates (F121–9 mouse ESC, mapped on mm10) before and after each normalization step. Data are visualized using IGV. B.: Correlation between log ratio early / late at indicated step of normalization of 50kb windows along the genome of samples in A. R = Pearson correlation coefficient,

Figure 3:. Repli-chip and repli-seq give highly similar replication timing profiles at genome-wide level.

A. B.: Replication timing (RT) profile on a portion of chr1 of 46C mouse ESC (mm10) (A) and human lymphoblastoid cells (hg38) (B), visualized using IGV. RT is defined as the log2 ratio early fraction over late fraction (reads number is normalized on number of mapped reads for repli-seq). C. D.: correlation between average RT on 50kb windows on the whole genome in 46C mouse ESC (C) and human lymphoblastoid cells (D). Because repli-chip and repli-seq have different dynamic value ranges, data are scaled in R using the scale function, prior to visualisation. R = Pearson correlation coefficient.

Figure 4:. Repli-seq allows the discrimination between haplotypes.

A.: Comparison of RT in F121–9 mouse hybrid ES cells (M. musculus / M. castaneus hybrid cells). The mapping on the two genomes has been performed as described in Supplementary Method V, then log ratio datasets have been quantile normalized and Loess smoothed. Replication timing in 50kb windows for each allele and each replicate has been plotted using R. B.: Comparison of RT for three homologous regions in the same cells. Data are visualized using IGV. R = Pearson correlation coefficient.