CRISPR-based genome editing through the lens of DNA repair

Abstract

Genome editing technologies operate by inducing site-specific DNA perturbations that are resolved by cellular DNA repair pathways. Products of genome editors include DNA breaks generated by CRISPR-associated nucleases, base modifications induced by base editors, DNA flaps created by prime editors, and integration intermediates formed by site-specific recombinases and transposases associated with CRISPR systems. Here, we discuss the cellular processes that repair CRISPR-generated DNA lesions and describe strategies to obtain desirable genomic changes through modulation of DNA repair pathways. Advances in our understanding of the DNA repair circuitry, in conjunction with the rapid development of innovative genome editing technologies, promise to greatly enhance our ability to improve food production, combat environmental pollution, develop cell-based therapies, and cure genetic and infectious diseases.

Figure 1.. Site-specific DNA perturbations induced by CRISPR-based genome editing technologies and DNA repair processes that resolve them

Schematic of the five major CRISPR-based genome editing technologies and six site-specific DNA lesions/structures generated by them. Enzymatic activities of genome editors generating the depicted DNA lesions/structures, DNA repair processes involved in their resolution, and outcomes of the repair events are also illustrated. BER (base excision repair), HDR (homology-directed repair), MMR (mismatch repair), SSBR (single-strand break repair) and TLS (translesion synthesis).

Figure 2.. Repair of DSBs induced by site-specific DNA nucleases and strategies to stimulate precision genome editing

DNA double-strand breaks generated by engineered or programmable nucleases can either be repaired by end joining (left) or HDR mechanisms (right). While end joining can occur throughout the cell cycle, HDR is confined to the S and G2 phases of the cell cycle when a sister chromatid is available for recombination (top left). The choice between end joining and HDR is regulated by DNA end resection. DSB ends that require minimal or no end processing can be ligated by NHEJ through the activities of Ku, DNAPKcs, XLF, XRCC4 and LIG4 (1–2). Short-range end resection catalyzed by MRE11-RAD50-NBS1 (MRN) in complex with CtIP can expose regions of microhomology, which can undergo annealing and end joining by MMEJ (3–5). MMEJ is mediated by Polθ, which extends the annealed ends, followed by the removal of 5’ flaps by FEN1 and DNA ligation by LIG1/3 (5). End joining-mediated genome editing creates substitutions and small indel mutations, and can also generate large insertions in the presence of linearized dsDNA donors (bottom left, in green). End joining events can also result in the formation of large deletions and chromosomal rearrangements (bottom left, in red). Long-range resection is catalyzed by EXO1 or by DNA2 in complex with BLM or WRN and is inhibited by 53BP1 bound to histones with H4K20me2 (me) and H2AK15ub (ub) marks (top right). Long-range resection results in the generation of 3’ ssDNA tails that initiate HDR events (6–7). Annealing of homologous sequences by RAD52, followed by excision of 3’ flaps by XPF-ERCC1 and DNA ligation by LIG1/3 promotes the joining of DSB ends by SSA. SSA causes deletions and potential genomic rearrangements (bottom right, in red). In the presence of ssODNs or dsDNA donors, SSTR or dsDNA donor-dependent HDR (dsDNA HDR) promote the generation of precise DNA substitutions, deletions, insertions and complex mutations (bottom right, in green). While SSTR is mediated by RAD52-dependent annealing of 3’ ssDNA tails to DNA donors, followed by templated DNA synthesis, dsDNA HDR is catalyzed by the recombinase RAD51, which promotes the invasion of 3’ ssDNA tails into homologous sequences of DNA donors to initiate DNA synthesis. Precision genome editing can be stimulated by end joining inhibitors or activators of end resection and HDR. Enhanced HDR can also be obtained by alteration of the cell cycle (top left) and chromatin structure (top right), modulation of the activity of DNA repair factors, expression of engineered DNA repair variants, fusion of DNA repair proteins to Cas nucleases, and modification of DNA donor molecules. HDR modulators are highlighted in blue (see also Table 1). While largely error-free, SSTR and dsDNA HDR can also cause deletions, insertions, point mutations and genomic rearrangements when conducted inaccurately (bottom right, in red). These events can occur in a manner dependent or independent from DNA donors.

Figure 3.. Repair of CRISPR-induced SSBs for precision genome editing

Multiple RAD51-dependent and -independent cellular mechanisms act at SSBs generated by nCas9 to promote the installment of genomic changes. In the presence of dsDNA donors, dsDNA donor-dependent HDR (dsDNA HDR) can lead to BRCA2- and RAD51-mediated invasion of the 3’ end of the nicked genomic DNA strand into the donor DNA, followed by templated DNA synthesis (1). ssODNs complementary to the nicked genomic DNA strand can instead serve as a template for DNA synthesis following RAD52-dependent ssODN annealing to the nicked strand through a SSTR-like process (2). Displacement of the genomic DNA strand from dsDNA or ssDNA donors and its reannealing to the parental strand can lead to the generation of a 5’ flap and the formation of heteroduplex DNA containing a mismatch between the edited and the parental sequence (3–4). Flap excision by 5’ flap endonucleases, followed by DNA ligation and DNA replication can then lead to the incorporation of the desired change (5–7). Annealing of ssODNs complementary to the non-nicked DNA strand through single-strand DNA incorporation (ssDI) can result in the formation 5’ and 3’ flap structures and the generation of heteroduplex DNA with a mismatch (10). Incorporation of the desired change can then occur upon flap excision by 3’ and 5’ endonucleases, DNA ligation and DNA replication (10–13). Recognition of mismatches between the edited and parental strand by the MMR machinery can cause EXO1-dependent degradation of the edited strand, restoring the original DNA sequence (8–9 and 14–15). Religation of the nCas9-induced nick can also result in the restoration of the original DNA sequence (16). Alternatively, DNA end processing and gap filling by SSB repair (SSBR) can lead to error-free repair or formation of indels and substitutions (17–18). Unrepaired SSBs can be converted into DSBs during DNA replication, resulting in the collapse of replication forks and their subsequent repair by either error-free HDR or error-prone DSB repair (DSBR), which can cause indels, substitutions and chromosomal aberrations (19–20).

Figure 4.. Repair of site-specific base lesions generated by canonical base editors

Generation of site-specific base transitions by the canonical base editors ABE and CBE. Traditional ABE is constituted of a fusion of nCas9 to the TadA deoxyadenosine deaminase. ABE promotes the deamination of adenines located within the ssDNA of the R-loop generated upon gRNA pairing to the targeted DNA strand (1). Adenine deamination leads to the formation of inosine and the generation of a I:T mismatch, which can be recognized by the MMR machinery (1). nCas9-dependent nicking of the non-edited strand can promote its degradation by EXO1 and its subsequent resynthesis using the edited strand as a template, resulting in the incorporation of a cytosine opposite to inosine (2–3). Alternatively, replacement of the T with a C can occur upon DNA synthesis initiated by the 3’ end of the nicked non-edited strand, followed by strand displacement, excision of the resulting 5’ flap and DNA ligation (4–5). These events can occur in a MMR-dependent manner or result from long-patch repair of the nCas9-induced SSB. T to C substitution can also occur following religation of the nicked strand and DNA replication (not shown). Subsequent DNA replication or BER can lead to the replacement of the I with a G and the generation of a A:T->G:C transition (6). Distinct from ABE, CBE is traditionally constituted of a fusion of nCas9 to the cytidine deaminase APOBEC1/3, which catalyzes the deamination of cytosines into uracils within the ssDNA of the gRNA-containing R-loop (7). CBE inhibits uracil excision by UNG through a UGI peptide fused to nCas9, resulting in the persistence of a G:U mismatch, which can be recognized by MMR proteins (7). Similar to ABE, CBE then catalyzes the nicking of the non-edited strand, favoring its degradation or its displacement and excision, followed by the replacement of the G with an A upon DNA synthesis (7–11). G to A substitution can also occur upon DNA replication (not shown). Subsequent DNA replication or BER can lead to the replacement of the U with a T and the generation of a C:G->T:A transition (12).

Figure 5.. Repair of site-specific base lesions generated by non-canonical base editors

Modification of dsDNA and generation of transversion mutations or predictable deletions by non-canonical base editors (DddA, CGBE and AFID system). dsDNA-dependent base editing in the mitochondrial genome can be obtained using a split DddA deaminase fused to TALE arrays, which provide site-specific DNA binding. Reconstituted DddA deaminates cytosines located at the TALE DNA binding site, generating uracil and inhibiting its excision through a UGI peptide fused to DddA (1). The resulting G:U mismatch can then be resolved through DNA replication and BER events, leading to the generation of a C:G->T:A transition (2–3). CGBE enzymes are derived from CBE, whereby the UGI peptide has been removed or replaced with the uracil DNA glycosylase UNG or UdgX. Similar to CBE, CGBE promotes site-specific cytosine deamination and nicks the non-edited strand (4). However, in the case of CGBE, the uracil is then excised by uracil DNA glycosylases, resulting in the formation of an abasic site (5). The non-edited strand can then be degraded by EXO1 and undergo resynthesis by the replicative DNA polymerases (6–7). Upon encountering the abasic site, the replicative DNA polymerases are replaced by translesion DNA polymerases, such as REV1, which can insert a cytosine opposite to the abasic site (7). Insertion of a C opposite to the abasic site can also occur upon DNA synthesis initiated by the 3’ end of the nicked non-edited strand, followed by the bypass of the abasic site by translesion DNA polymerases and the excision of the 5’ flap resulting from strand displacement DNA synthesis (8–9). DNA replication or abasic site repair by AP lyases and other BER enzymes can then lead to the replacement of the abasic site with a G, resulting in a C:G->G:C transversion (10). The generation of transversion mutations is stimulated by DNA repair effectors fused to CGBE (6–10). AFID systems are based on a fusion of Cas9 to APOBEC3 and a bacterial uracil DNA glycosylase. Similar to CGBE, AFID systems promote cytosine deamination into uracil, which is then excised by the uracil DNA glycosylase, generating an abasic site (11–12). However, distinct from CGBE, AFID systems introduce a DSB near the PAM sequence and express an AP lyase, which introduces a nick at the abasic site, resulting in the generation of a DNA end with a 5’ ssDNA tail (13–14). Degradation of the 5’ ssDNA tail and DSB repair by NHEJ results in predictable deletions spanning from the deaminated cytosine to the site of the Cas9-induced DSB (15).

Figure 6.. Repair of DNA lesions generated by prime editing

A) Modules that constitute prime editors. Prime editors consist of the fusion of an engineered reverse transcriptase (RT) to nCas9. The reverse transcriptase utilizes an RNA:DNA heteroduplex formed upon the annealing of the nicked non-target DNA strand to a primer binding site (PBS) in the pegRNA. The 3’ end of the nicked DNA strand is then extended by the reverse transcriptase using the pegRNA sequence containing the desired edits as template (RT template). B) Processing of DNA intermediates generated by prime editors. Prime editor-mediated reverse transcription leads to the generation of a 3’ flap containing the edit, in equilibrium with a 5’ flap that does not contain the edit (1). Formation of the 5’ flap is accompanied by the generation of heteroduplex DNA containing a mismatch between the edited strand and the parental strand (1). In PE2 prime editing systems, excision of the 5’ flap by FEN1 can be followed by resolution of the mismatch upon ligation of the nicked strand and DNA replication, leading to the incorporation of the desired change (2–3). Recognition of the mismatch by the MMR machinery can lead to MLH1- and EXO1-dependent degradation of the nicked edited strand, restoring the original sequence (4–5). In PE3 prime editing systems, introduction of a nick on the 5’ side of the mismatch using a second gRNA can engage the MMR machinery on the non-edited strand and promote its degradation by EXO1, resulting in the incorporation of the desired change following DNA synthesis (6–9). Incorporation of the edits of interest in PE3 systems can also occur upon DNA synthesis initiated by the 3’ end of the nicked non-edited strand, followed by strand displacement and excision of the resulting 5’ flap (10–11). Engagement of the MMR machinery on the nicked edited strand can lead to the restoration of the original sequence also in PE3 systems (12, 4–5). To suppress MMR-mediated restoration of the original sequence, the PE4 and PE5 prime editing systems combine respectively PE2- and PE3-based approaches with the transient expression of a dominant negative MLH1 protein (MLH1dn) (4). Excision of the 3’ flap containing the edited sequence results in MMR-independent restoration of the original sequence (13).

Figure 7.. Processing of DNA intermediates induced by paired prime editors to generate precise substitutions, insertions and/or deletions

A) Design of paired prime editing strategies. A pair of prime editors is targeted to opposite DNA strands to generate two 3’ flaps with complementary sequences and desired modifications. B) Mechanisms for the generation of precise substitutions, insertions and/or deletions using paired prime editors. Substitutions (left) can be obtained using paired prime editors that generate two complementary 3’ flaps containing the desired edit(s) (green) (1). Annealing of the two edited 3’ flaps and the two corresponding 5’ flaps containing the parental DNA sequence, followed by excision of the annealed 5’ flaps and ligation of the nicked strands, can then lead to the incorporation of the desired substitution(s) (2–3). Precise insertions and deletions (middle) can be induced using paired prime editors that generate two 3’ flaps that contain complementary (purple) and non-complementary (lighter and darker purple) sequences of the desired insert adjacent to the genomic sequence to be replaced (4). Annealing of the two edited 3’ flaps and the two parental 5’ flaps, followed by excision of the annealed 5’ flaps, filling of ssDNA gaps within the insert sequence and DNA ligation, can lead to the replacement of the genomic sequence of interest with the desired insert (5–6). Deletions (right) can be obtained using paired prime editors that generate 3’ flaps (blue and red) complementary to genomic sequences upstream and downstream of the generated flaps (blue and red) (7). Annealing of the 3’ flaps with their complementary genomic sequences, followed by excision of the corresponding annealed 5’ flaps and ligation of the nicked strands, can result in the deletion of the genomic sequence located between the two 3’ flaps (8–9). Paired prime editors can be utilized for the insertion of serine recombinase sites (e.g., attP or attB sites), enabling site-specific integration of gene-size fragments upon transfection of donor gene constructs and expression of serine recombinases (e.g., Bxb1) (10). Site-specific integration can also be obtained using a prime editor fused to Bxb1 (not shown).