TABLE 2.
Experimental methods for the detection of off-target editing by CRISPR-Cas9.
| Category | Method | Description | Advantages | Disadvantages |
|---|---|---|---|---|
| in vitro | Digenome-seq Kim et al. (2015) | gDNA is digested with RNPs and subjected to WGS. Cut sites are identified bioinformatically as sites that share the exact same sequence at one end of the sequencing read | High sensitivity | High false positive rate due to lack of chromatin accessibility context, expensive due to reliance on WGS |
| DIG-seq Kim and Kim, (2018) | Cell-free chromatin is subjected to Digenome-seq | Accounts for the chromatin context and hence has a higher validation rate than Digenome-seq | Relatively expensive due to the continued dependence on WGS | |
| nDigenome-seq Kim et al. (2020) | gDNA is digested by Cas9 nickase followed by WGS. Nick sites are identified as sites with both staggered and straight read alignments | High sensitivity, genome-wide profiling of DNA SSBs induced by nickases | Lacking in cellular context. Indirect method for profiling the genome-wide specificity of prime and base editors using Cas9 nickases | |
| Extru-seq Kwon et al. (2023) | Cells are pre-incubated with RNPs, then passed through an extruder to lyse the cells and bring the RNPs and gDNA in proximity. Unrepaired cut sites are identified by WGS and the Digenome-seq algorithm | High validation rate, easily adaptable to different primary cells | Difficult to identify SVs, costly | |
| SITE-seq Cameron et al. (2017) | gDNA is digested with RNPs and cut sites are labelled with biotinylated primers and enriched using streptavidin beads and sequenced. Cut sites are identified by read pileup | Less expensive as enrichment strategy enables shallower sequencing (∼0.62–2.46 million reads) | Low validation rate due to lack of chromatin context | |
| EndoV-seq Liang et al. (2019) | In vitro cleavage of inosine, the nucleoside intermediate that is created by ABEs, by endonuclease V (EndoV) followed by WGS | Sensitivity comparable to Digenome-seq. Multiplexed analysis of ABE-sgRNA complexes | Lacks nuclear/chromatin context Analysis limited to ABEs | |
| CIRCLE-seq Tsai et al. (2017) | gDNA is fragmented by sonication, circularized, and incubated with RNPs. Only circles containing nuclease digestion sites are linearized and used to create a sequencing library | Less expensive as enrichment strategy enables shallower sequencing (4–5 million reads) | Very high input requirement (∼25 μg DNA) | |
| CHANGE-seq Lazzarotto et al. (2020) | Similar to CIRCLE-seq, but uses enzymatic fragmentation instead of sonication to fragment gDNA | Lower input requirement than CIRCLE-seq | Lack of chromatin context | |
| UDiTaS Giannoukos et al. (2018) | Detects DSBs by using universal adapters and anchored primers to analyze repair outcomes after nuclease cleavage | Can detect translocations, inversions, and large deletions using short-read sequencing | Requires a priori knowledge for target enrichment | |
| Cell-based | Whole Genome Sequencing (WGS) Smith et al. (2014), Veres et al. (2014), Iyer et al. (2015) | WGS on DNA extracted from cells treated with Cas9 and sgRNA | Detects several types of off-target edits including INDELs and SVs | Poor signal to noise ratio, limited sensitivity for rare variants, expensive due to need for high coverage (20–60X) |
| Integrase-Defective Lentiviral vector (IDLV) integration Gabriel et al. (2011), Wang et al. (2015) | Cells are transfected with Cas9 and sgRNA plasmids and transduced with an IDLV with a propensity to integrate near DSBs, tagging the nuclease generated cut sites with lentiviral sequences. The IDLV integrated sites are then enriched via linear amplification-mediated (LAM) PCR or non-restrictive LAM PCR using primers complementary to the IDLV sequences, followed by NGS | Applicable for a variety of nuclease platforms and cell types | Lower sensitivity (0.5%) and high false positive rate | |
| GUIDE-seq Tsai et al. (2015) | Enriches nuclease-induced DSBs by the insertion of a double stranded oligonucleotide (dsODN) with a known sequence. dsODN specific primers are used for enrichment followed by sequencing | High validation rate and high sensitivity, commonly used | dsODN are cytotoxic for some cell lines, this approach is not feasible in vivo, cannot detect SVs | |
| iGUIDE Nobles et al. (2019) | GUIDE-seq protocol with a longer dsODN and dedicated software package | Enables detection of mispriming events lowering the false positive rate | Not commonly used | |
| Tagmentation-based tag integration site sequencing (TTISS) Schmid-Burgk et al. (2020) | Similar to GUIDE-seq, but uses tagmentation to shear cell-derived gDNA and tag it with Illumina sequencing adaptors | Enables multiplexed screening of upto 60 sgRNAs, applicable for prime editors | Lower sensitivity with higher multiplexing | |
| Direct in situ breaks labelling, enrichment on streptavidin and next-generation sequencing (BLESS) Crosetto et al. (2013) | Cells are fixed to preserve DSBs, nuclei are isolated and the DSBs are blunted and ligated to a biotinylated linker. gDNA is then isolated and biotinylated sequences are enriched with streptavidin beads and sequenced | Nucleotide-resolution DSB mapping, applicable to tissues derived from in vivo studies | Only provides a snapshot of the DSBs present in the cells at the time of fixation, can miss DSBs unless a very large number of cells is profiled. Low signal to noise ratio due to the cell fixation and handling steps, requires high input DNA/cells. Centrifugation steps in the protocol can damage the chromatin and introduce spurious DSBs and are incompatible with smaller nuclei | |
| Breaks labelling in situ and sequencing (BLISS) Yan et al. (2017) | Cells/tissues are fixed and attached to glass slides, and DSBs are labelled with a dsODN with a T7 promoter that serves to amplify the DSB sequences by in vitro transcription | More sensitive than BLESS, amenable to multiplexing, lower input requirement than BLESS | Only provides a snapshot of the DSBs present in the cells at the time of fixation, can miss DSBs unless a very large number of cells is profiled | |
| Surveyor Guschin et al. (2010) | Target DNA from both mutant and wild-type reference DNA are amplified by PCR and hybridized; followed by treatment of annealed DNA with Surveyor endonuclease to cleave heteroduplexes and analysis of digested DNA products | Rapid, relatively simple and cost-effective method | Requires a priori knowledge. Lacking in single nucleotide resolution. Cannot discriminate between alleles. Preferentially identifies substitutions | |
| T7E1 Mashal et al. (1995) | Target DNA from both mutant and wild-type reference DNA are amplified by PCR and hybridized; followed by treatment of annealed DNA with T7E1 endonuclease to cleave heteroduplexes and analysis of digested DNA products | Rapid, relatively simple and cost-effective method | Requires a priori knowledge. Lacking in single nucleotide resolution. Cannot discriminate between alleles. Preferentially identifies insertions and deletions | |
| TIDE, TIDER Brinkman et al. (2014), Brinkman et al. (2018) | PCR amplification of candidate sites followed by Sanger sequencing and bioinformatics analysis to identify off-target events | Provides details about the indels and mutations generated. User-friendly interface. Very affordable | Low throughput. Requires a priori knowledge. Requires fine tuning of settings by the user | |
| LAM-HTGTS Frock et al. (2015) | Genome-wide detection of “prey” DSBs via their translocation to a fixed “bait” DSB in cultured mammalian cells | Very high sensitivity | High input requirement | |
| PE-tag Liang et al. (2023) | DNA tag integration at target site and off-target sites by prime editor, followed by tagmentation and tag-specific amplification | Rapid and sensitive approach for the genome-wide identification of prime editor activity and evaluation of safety | Sensitivity to an off-target site is limited to sequences that can be extended by the associated reverse transcriptase. Low sensitivity in vivo due to modest editing efficiencies of PEs | |
| Detect-seq Lei et al. (2021) | Chemical labeling of deoxyuridine and biotin pulldown of CBE-edited DNA followed by deep sequencing | Genome-wide identification of CBE-induced off-target sites | Analysis limited to tools that generate deoxyuridine as an editing intermediate | |
| CAST-seq Turchiano et al. (2021) | PCR amplification uses a “bait primer” binding to the on-target sequence, a “prey primer” that recognizes the linker sequence, and “decoy primers” that bind the target sequence to prevent on-target amplification. Further PCR amplifications are successful only if the binding sites of the decoy primers are lost because of translocations or large deletions at the on-target site | High sensitivity and quantitative measurement of chromosomal rearrangements Can be performed directly in the clinically relevant cell-type | Does not recognize off-target sites that are repaired exclusively by NHEJ, not always is possible to design effective bait and decoy primers for a desired locus | |
| SURRO-seq Pan et al. (2022) | Targeted in-cell capture of off-targets based on a pooled lentiviral library encoding a sgRNA and barcoded surrogate off-target sites | Higher scalability than previous targeted methods, e.g., TIDE | Targeted approach, requires pre-selection of candidate sites | |
| in vivo | DISCOVER-Seq Wienert et al. (2019) | A modified chromatin immunoprecipitation approach where DSBs are indirectly identified as sites bound by meiotic recombination 11 homolog 1 (MRE11), a DNA repair protein that is part of the MRE11-RAD50-NBS1 (MRN) complex that colocalizes to DSBs created by CRISPR-Cas before repair | Can be applied in vivo | Lower sensitivity (∼0.3%) and high false positive rate |
| GUIDE-Tag Liang et al. (2022) | Modification of GUIDE-seq where Cas9 protein is fused with monomeric streptavidin (mSA), which helps to improve the rate of incorporation into DSB sites of a biotinylated dsODN that is delivered separately | Can be applied in vivo, can identify SVs, gDNA library compatible with UdiTas for identifying SVs | Low insertion rate of the dsODN | |
| VIVO Akcakaya et al. (2018) | In vitro discovery of off-targets by CIRCLE-seq Tsai et al. (2017) followed by in vivo validation | High sensitivity and applicable to whole organisms | In vivo validation is restricted to a subset of candidates | |
| GOTI Zuo et al. (2019) | Editing of single blastomeres of two-cell mouse embryos and progeny cells are examined by WGS | Suitable for CRISPR-Cas9 and base editors. Only detects edits that are improperly repaired and transmitted to daughter cells, directly compares edited and non-edited cells with identical genetic backgrounds | Results are specific to the species in which it is performed. Very expensive method. Requires high level of technical skill and specific apparatus |