Detecting DNA double-stranded breaks in mammalian genomes by linear amplification-mediated high-throughput genome-wide translocation sequencing

Abstract

Unbiased, high-throughput assays for detecting and quantifying DNA double-stranded breaks (DSBs) across the genome in mammalian cells will facilitate basic studies of the mechanisms that generate and repair endogenous DSBs. They will also enable more applied studies, such as those to evaluate the on- and off-target activities of engineered nucleases. Here we describe a linear amplification-mediated high-throughput genome-wide sequencing (LAM-HTGTS) method for the detection of genome-wide 'prey' DSBs via their translocation in cultured mammalian cells to a fixed 'bait' DSB. Bait-prey junctions are cloned directly from isolated genomic DNA using LAM-PCR and unidirectionally ligated to bridge adapters; subsequent PCR steps amplify the single-stranded DNA junction library in preparation for Illumina Miseq paired-end sequencing. A custom bioinformatics pipeline identifies prey sequences that contribute to junctions and maps them across the genome. LAM-HTGTS differs from related approaches because it detects a wide range of broken end structures with nucleotide-level resolution. Familiarity with nucleic acid methods and next-generation sequencing analysis is necessary for library generation and data interpretation. LAM-HTGTS assays are sensitive, reproducible, relatively inexpensive, scalable and straightforward to implement with a turnaround time of <1 week.

Figure 1. Step-by-step overview of HTGTS methods

(a) The original HTGTS method requires end processing and adapter ligation of sheared genomic DNA fragments prior to PCR amplification, enrichment of biotinylated products, and further amplification steps to increase specificity and to label ends for Miseq sequencing. (b) LAM-HTGTS directly amplifies junctions from sheared genomic DNA using LAM-PCR followed by enrichment and bridge adapter ligation to allow for exponential amplification and Miseq labeling of enriched products.

Figure 2. Primer design and enzyme cutting site strategies for LAM-HTGTS

(a) Strategies for designing biotinylated primer (bio-primer), nested primer and choosing enzyme blocking site for LAM-HTGTS with a given bait DSB site. Note that primers can be placed downstream of the bait DSB site and enzyme blocking sites should be correspondingly upstream of the bait DSB—keeping the relative distances the same—to clone from the other side of the bait DSB. (b) The major events after induction of bait DSBs are uncut, perfect joins and small insertion/deletions (indels) around the bait DSB sites, which are greatly suppressed during the enzyme blocking steps. (c) The minor events after induction of bait DSBs are translocations between bait DSBs and genome-wide DSBs, which lose the enzyme blocking sites and can be readily amplified for Miseq sequencing. The final amplified products will contain the following sequence components in the order listed: Illumina P5-I5, nested primer (with barcode), bait, insertions if any, prey, adapter, Illumina P7-I7. The bait is composed of the nested primer binding sequence leading up to the targeted DSB site. The prey represents the unique genome alignment with the junction representing the resulting join between the bait and prey sequences.

Figure 3. Flow chart of bioinformatic pipeline for translocation junction identification

Multiple HTGTS libraries with different barcodes can be sequenced in the same Miseq flow cell. De-multiplexing separates sequencing reads for each library, followed by sequence read processing which takes into account bait, prey, and adapter alignments to optimally define the sequence read. Uniquely mapped bait-prey junctions are retained as filtered junctions while identical junctions are separated.

Figure 4. Representative smear of amplified and Illumina sequence tagged products

Sequences harboring Bait/Prey components can vary in size due to the combination of stochastic shearing of genomic DNA and the juxtaposition of Bait/Prey sequences. Products ranging from 500bp-1kb are excised and purified for Miseq sequencing. Smaller products may also contain sequences with relevant junction information but co-migrate with various artifactual poly-priming intermediates. M = Molecular weight ladder.

Figure 5. Universal bait detection of off-targets for designed VEGFA gRNA

(a) Circos plot (circos.ca) on custom log scale showing genome-wide profile of Cas9:SeC9-2 junctions in cycling v-Abl pro-B cells. Bin size is 5 Mb and 8751 unique junctions are shown from 2 independent libraries. Chromosomes are displayed as centromere to telomere in a clock-wise orientation. Blue lines link SeC9-2 off-targets to bait break-site while red lines link VEGFA off-targets to bait break-site. (b) List of identified off-targets for SeC9-2 or VEGFA. The off-targets were identified by MACS2 as described previously. v-Abl cells (3×10⁶) were nucleofected with a combined total of 3 μg plasmid DNA in SF solution using the DN-100 program (Lonza) and collected 48 hours post nucleofection.