IF=17! Major Publication: GeneRulor Team Develops Ultra-Sensitive and Specific Off-Target Detection Method AID-seq Published in Cell’s Flagship Sub-Journal Med

Background

The CRISPR-Cas9 system, derived from the adaptive immune systems of bacteria and archaea, has become a powerful gene editing tool due to its simplicity and efficiency. Since the first demonstration in 2013 that CRISPR/Cas9 can achieve precise DNA editing within cells, the field has experienced explosive growth. Compared with gene editing technologies such as ZFN and TALEN, CRISPR-Cas9 is simpler, more economical, and easier to operate, making it the most effective and accessible gene editing method to date.

Its applications span multiple areas:

• Gene editing: knock-out, knock-in, point mutations

• Gene therapy: antiviral therapies (HPV¹, HIV², HBV³, HSV), correction of inherited disorders such as β-thalassemia and Duchenne muscular dystrophy

• Cellular immunotherapy: precise CAR/TCR gene knock-in and checkpoint gene knockout in CAR-T, CAR-NK, TCR-T

• Functional gene screening: CRISPRi/a positive and negative selection

However, CRISPR-Cas9 has a critical drawback: off-target effects, where unintended DNA cleavage or editing occurs outside the intended target. Off-target events can lead to undesired gene mutations or even tumorigenesis, limiting the clinical application of CRISPR-based gene editing. Therefore, improving sgRNA specificity and reducing off-target activity is a major challenge for the next generation of CRISPR technologies.

Figure 1: Mechanism of Off-Target Effects (A) and Types of Mismatches (B)

CRISPR-Cas9 operates through the generation of DNA double-strand breaks (DSBs). During DSB repair, cells may introduce insertions or deletions (indels) or even larger chromosomal rearrangements. Consequently, genome-wide detection of off-target sites and mutation frequencies is critical, especially for therapeutic applications, as comprehensive off-target analysis helps avoid sgRNAs that could produce adverse clinical outcomes.

Existing Off-Target Detection Methods

Genome-wide off-target detection techniques can be broadly categorized into:

1. Cell-based in vivo methods: GUIDE-seq⁴, BLISS⁵, HTGTS⁶, DISCOVER-seq⁷

2. Cell-free in vitro methods: Digenome-seq⁸, SITE-seq⁹, CIRCLE-seq¹⁰, CHANGE-seq¹¹ (an improved version of CIRCLE-seq)

In Vivo Detection

Cell-based in vivo detection methods have limited applicability for hard-to-transfect cell lines.

• GUIDE-seq⁴ relies on the integration of dsODN tag sequences into the genome.

• HTGTS⁶ relies on chromosomal translocations.

Both are influenced by DNA repair specificity, timing, target locus, and cell cycle stage. Mutations at DSBs may be missed if dsODN fails to integrate or chromosomal translocation does not occur, especially for off-targets with <0.1% frequency.

• DISCOVER-seq⁷ uses ChIP-seq to detect the DSB repair factor MRE11 at Cas-induced cuts. However, Cas enzymes cut different sites asynchronously, and endogenous DSBs create false positives, reducing accuracy.

In Vitro Detection

Cell-free in vitro methods provide reproducibility and avoid transfection and repair biases. Higher concentrations of sgRNA and Cas enzymes facilitate detection of low-frequency off-targets. Key methods include:

1. Digenome-seq⁸: Cas enzyme cleaves target DNA in vitro; all free DNA ends are sequenced. Requires high sequencing depth and is limited in detecting low-frequency off-targets due to background noise.

2. SITE-seq⁹: Uses high-molecular-weight genomic DNA as substrate. Post-cleavage DNA is biotin-labeled, enriched, and sequenced. Reduces background DSBs but DNA extraction is challenging, potentially causing false positives.

3. CIRCLE-seq¹⁰: Circularizes DNA to remove background DSBs. Requires large amounts of DNA (~25 µg) and has limitations with primary or precious samples.

4. CHANGE-seq¹¹: Improved CIRCLE-seq using Tn5 library prep and USER enzyme to produce sticky ends. Still susceptible to false positives from free DSB ends.

Figure 2: Schematic of Common Gene Editing Off-Target Detection Methods¹²

Development of AID-seq

After years of effort, the GeneRulor team developed the ultra-sensitive and specific AID-seq in vitro off-target detection method, published on June 5, 2023, in Cell’s sub-journal Med¹³ (Massively parallel CRISPR off-target detection enables rapid off-target prediction model building).

AID-seq Workflow (Figure 3):

1. Genomic DNA is randomly fragmented and ligated to i7 hairpin adaptors, forming a “dumbbell” structure.

2. Free DNA is digested using a mixture of three exonucleases, repeated 2–4 times to minimize false positives from background DSBs.

3. Remaining dumbbell DNA is incubated with Cas9 or Cas12a RNP complexes; newly cut ends are ligated to biotinylated i5 adaptors.

4. Biotinylated DNA is enriched using streptavidin magnetic beads.

5. Nested PCR generates sequencing-ready libraries.

Figure 3: AID-seq Workflow

AID-seq Results

1. Sensitivity and Specificity

• Tested seven sgRNAs previously characterized by GUIDE-seq, SITE-seq, CIRCLE-seq, and CHANGE-seq, including VEGFA site 1–3, EMX1, FANCF, HEK293 site 4, PAPSS2.

• AID-seq detected more GUIDE-seq off-targets than SITE-seq, CIRCLE-seq, or CHANGE-seq.

• PR and ROC curves confirmed AID-seq achieves the highest AUPRC and AUROC, outperforming previous in vitro methods.

Figure 4: Sensitivity and Specificity Comparison

2. Cas12a Off-Target Detection

• Cas12a generates sticky ends and exhibits non-specific cleavage.

• AID-seq successfully detected off-target sites missed by GUIDE-seq and validated true positives at the cellular level.

  

Figure 5: AID-seq Detection of Cas12a Off-Targets

3. Multiplexed sgRNA Detection

• Pooled AID-seq detected multiple sgRNAs simultaneously (libraries of 16 and 66 sgRNAs).

• High reproducibility across biological replicates, retaining sensitivity comparable to single AID-seq.

  

Figure 6: Pooled AID-seq Measuring Multiple sgRNAs On- and Off-Target

4. Rapid Off-Target Data Acquisition and Model Building

• Pooled AID-seq characterized FrCas9 on 2,069 sgRNAs (~500 per library).

• On-target rate ≥95%, >50% of sgRNAs showed no off-targets.

• Established an accurate FrCas9 off-target prediction model (AUROC=0.97, AUPRC=0.29), available online: Cas-Designer FrCas9.

Figure 7: Rapid Off-Target Modeling for New CRISPR Variants Using Pooled AID-seq

Conclusions

AID-seq is a novel, accurate, and high-throughput method for detecting on- and off-target activity of various CRISPR systems (Cas9 and Cas12a). Advantages include:

• Broad applicability across species (human, animal, plant, bacteria, fungi) and cell types (cell lines, primary cells, non-dividing cells)

• Low DNA input requirement (2 µg), friendly for challenging samples

• Detects rare off-target events critical for gene therapy

• Pooled AID-seq enables high-throughput detection for hundreds of sgRNAs simultaneously

Applications:

• Selection of sgRNAs with highest efficiency and lowest off-target activity

• CRISPR library optimization to minimize library size and reduce confounding effects

• Characterization of newly discovered CRISPR systems, including cleavage efficiency, off-target effects, and PAM preferences, for building predictive models

GeneRulor has long been committed to gene editing off-target detection, and is the first company in China to offer these services. They have assisted numerous biotech companies with IND filings and provide multiple detection services:

• GUIDE-seq in vivo off-target detection

• AID-seq in vitro off-target detection

• Amplicon-based off-target validation

• PEM-seq chromosomal rearrangement detection

• Lentiviral integration site analysis

• Whole-genome off-target site detection

• RNA off-target detection

For inquiries or services, GeneRulor welcomes contact.