When we talk about CRISPR applications, one negative always comes up: the low editing efficiency of homology-directed repair (HDR). Compared to the random process of non-homologous end joining, HDR occurs at a relatively low frequency, and in nondividing cells, this pathway is further downregulated. Like all CRISPR applications that use wild-type Cas9, editing by HDR also has some potential for off-target cleavage even when gRNAs are well designed. Rather than try to improve HDR, Addgene depositor David Liu’s lab created new Cas9 fusion proteins that act as “single base editors.” These fusions contain dCas9 or Cas9 nickase and the rat cytidine deaminase APOBEC1, which can convert cytosine to uracil without cutting DNA. Uracil is subsequently converted to thymine through DNA replication or repair. Komor et al. estimate that hundreds of genetic diseases could be good targets for base editing therapy, not to mention the potential basic and preclinical research applications. Read on to learn about this new way to make point mutations using CRISPR without double-stranded breaks.
dCas9 Fusion Proteins
When the CRISPR revolution first began, we were most excited about direct DNA target cleavage. This application is still important, but catalytically dead dCas9 is almost equally as valuable as a targeting scaffold. While dCas9 can’t cleave DNA, it can target a region of DNA with equal specificity to wild-type Cas9. dCas9 fusions have been used to regulate promoter activity, make epigenetic modifications, visualize genomic loci in living cells, and more.
A cytidine deaminase fusion to dCas9 makes a lot of sense when you take a closer look at the literature. Early work by Tsai et al. on CRISPR nickases showed that nickases could induce C->T mutations at a low frequency. They hypothesized that cytidine deaminases like APOBEC were the cause of these mutations, not Cas9. APOBEC and other DNA cytidine deaminases can only edit single-stranded DNA - a perfect scenario for Cas9 given that the target DNA and gRNA interact to displace a small portion of the nontargeted DNA strand. This specificity narrows down the number of cytosines potentially accessible to a dCas9-APOBEC fusion protein.
dCas9/Nickase Base Editing
- Cas9 fused to a cytidine deaminase
- A gRNA targeting Cas9 to a specific locus
- A target cytosine at position 4-8 in the non-targeted strand
Komor et al. created a first-generation base editor (BE1) using the rat cytidine deaminase APOBEC1 connected to dCas9 via a 16 base XTEN linker. To determine the editing capabilities of this protein, they incubated the purified base editor with DNA substrates containing Cs in positions 1-13 of the 20 base targeting sequence. After deamination, they used the uracil-specific USER enzyme to cleave deaminated DNA and determine deamination efficiency. BE1 displayed an activity window of approximately 5 nucleotides, from target positions 4 to 8. Across multiple targets, the average in vitro editing efficiency was 44%, with cytidines preceded by a T or a C edited at the highest rate. These editing rates are particularly impressive considering that when you modify only one strand of DNA, the maximum editing efficiency is 50%.
Once Komor et al. moved into a cell culture model, they noticed that editing efficiency dropped from 44% to 0.8-7.7%, likely due to base excision repair (BER) removing uracils from the DNA. Their second-generation BE2 system fuses an inhibitor of this process to dCas9, raising editing efficiency three-fold to a maximum of ~20%. For BE1 and BE2, indel formation is very low (<0.1%) since the DNA is not directly cleaved.
To increase base editing efficiency beyond 50%, you’d need a way to copy the edits into the opposite strand of DNA. To do so, Komor et al. turned dCas9 back into a nickase to simulate mismatch repair. BE3 nicks the unmodified DNA strand so that it appears “newly synthesized” to the cell. Thus, the cell repairs the DNA using the U-containing strand as a template, copying the base edit.
The BE3 system increased editing frequency to above 30% for a variety of targets in human cell lines, with an average indel frequency of only 1.1%. These numbers are a vast improvement over Cas9-mediated HDR for the loci tested; average HDR-mediated editing frequency was only 0.5%, and more indels were observed than point modifications. CRISPR base editing persists through multiple cell divisions, indicating that this method produces stable edits. Komor et al. also tested HDR and BE3 head-to-head for disease-relevant mutations in APOE4 and p53. Cas9-mediated HDR efficiency was very low (0.1-0.3%) or undetectable, but base editing efficiency ranged from 58-75% for APOE4 to 3-8% for p53.
Komor et al. also examined whether base editors can induce off-target mutations. Using previously characterized gRNAs with known off-target sequences and analyzing 10 off-target sequences per gRNA, they observed cytidine conversion in positions 4-8 for a subset of those off-targets. Unsurprisingly, the frequency of off-target deamination was highest at the sites where wild-type Cas9 induces the most off-target mutations, as previously determined using GUIDE-seq. The good news is that cytidines surrounding the gRNA sequences do not appear to be subject to off-target editing, indicating that off-target edits come from Cas9 activity, not APOBEC1 activity. Future base editors made using high-fidelity Cas9s could ameliorate the problem of off-target editing.
Comparison of the Gene Editing Technologies Studied in Komor et al.
|Cas9||WT or Nickase||dCas9||dCas9||Nickase|
|Mutation Site||~10-100 bp from PAM (higher efficiency at shorter distances)||C in position 4-8 of targeting sequence on free strand||C in position 4-8 of targeting sequence on free strand||C in position 4-8 of targeting sequence on free strand|
|gRNA||Single (or paired with nickase)||Single||Single||Single|
|Repair Template||ssDNA or dsDNA; homology arm length varies with edit size||None||None||None|
|Edit Types||Insertions, deletions, point mutations||Point mutations (C->T or G->A)||Point mutations (C->T or G->A)||Point mutations (C->T or G->A)|
|Frequency of NHEJ at Target Locus||Indels may be more common at target site than HDR repair||Almost zero||Almost zero||Lower than HDR|
|Types of Off-target Effects||Indels at off-target loci||Conversion of other Cs in editing window and at off-target sites||Conversion of other Cs in editing window and at off-target sites||Conversion of other Cs in editing window and at off-target sites|
The Future of Base Editing
The work described in Komor et al. is incredibly creative and exciting, as it removes the need for DNA cleavage and repair according to a donor template. To really bring base editing to primetime, a few issues need to be worked out. The first, most obvious problem is the need for enzymes that can make conversions other than C->T, or for the opposite strand, G->A. Based on data from the NCBI ClinVar database, Komor et al. estimate that about 900 clinically relevant C->T or G->A variants are properly positioned near an NGG-PAM and could be edited using these first-generation editors. However, these SNPs represent a small fraction of the estimated 12,000 pathogenic SNPs in ClinVar, many more of which should be targetable with other base editing technologies. Luckily, given the number of DNA-modifying enzymes currently known, it seems like only a matter of time before new base editors are developed.
The second potential hurdle for base editing is the small editing window available. For optimal editing, the targeted base should be at position 4-8 of the targeting sequence. Since PAM requirements limit the number of targetable sequences, using other Cas9 variants should expand targeting capacity. Other CRISPR enzymes like Cpf1 could also be adapted for base editing. Having more options for gRNAs would allow researchers to test for optimal gRNA on-target activity, likely increasing the frequency of base editing. This aspect will be especially important should base editing be used in clinical applications.
Recent work by Nishida et al. has further confirmed the utility of base editing while demonstrating the benefits of base editing variants. Nishida et al. created the Target-AID base editor using a cytidine deaminase from sea lamprey fused to Cas9 nickase. Target-AID acts similarly but not identically to BE3, modifying a 3-5 base window 18 bases upstream of the PAM instead of the 5 base window 15 bases upstream of the PAM seen with BE3. The authors suggest that the specific deaminase or the method of attachment to Cas9 may impact base editing action and efficiency.
Base editing efficiency will also likely increase as CRISPR delivery methods improve. In particular, ribonucleoprotein delivery may be a good option for base editing, as shorter BE expression could lower the editing frequency for other, non-preferred cytosines in the editing window, or for cytosines at off-target loci.
Base editing is the latest triumph for the CRISPR field, but it’s important to remember that this advance builds on previous work. Structural and functional annotation of Cas9 has given researchers the knowledge to improve Cas9 beyond its original capabilities, as previously seen in the engineering of high fidelity Cas9s. New dCas9 fusions are popping up all the time, combining the benefits of CRISPR targeting with many new applications.
1. Komor, Alexis C., Yongjoo B. Kim, Michael S. Packer, John A. Zuris & David R. Liu. “Programmable editing of a target base in genomic DNA without double-stranded DNA cleavage.” Nature (2016). doi:10.1038/nature17946 Epub 20 April 2016. PubMed PMID: 27096365.
- Find plasmids from this publication at Addgene.
2. Nishida, Keiji, et al. “Targeted nucleotide editing using hybrid prokaryotic and vertebrate adaptive immune systems.” Science (2016). doi:10.1126/science.aaf8729 Epub 4 August 2016. PubMed PMID: 27492474.
- Find plasmids from this publication at Addgene.
- Find plasmids from this publication at Addgene.
4. Tsai, Shengdar Q, et al. “GUIDE-seq enables genome-wide profiling of off-target cleavage by CRISPR–Cas nucleases.” Nat Biotechnol. 33(2):187–197 (2015). PubMed PMID: 25513782. PubMed Central PMCID: PMC4320685.
Additional Resources on the Addgene Blog
- CRISPR 101: Homology-Directed Repair
- CRISPR 101: Non-Homologous End Joining
- Learn about Other CRISPR Tools like Cpf1, eSpCas9, and SpCas9-HF1
Additional Resources on Addgene.org
- Catch Up on Your CRISPR Background with Our CRISPR Guide
- Browse CRISPR Plasmids
- Find Validated gRNAs