Double strand break repair by capture of retrotransposon sequences and reverse-transcribed spliced mRNA sequences in mouse zygotes

Abstract

The CRISPR/Cas system efficiently introduces double strand breaks (DSBs) at a genomic locus specified by a single guide RNA (sgRNA). The DSBs are subsequently repaired through non-homologous end joining (NHEJ) or homologous recombination (HR). Here, we demonstrate that DSBs introduced into mouse zygotes by the CRISPR/Cas system are repaired by the capture of DNA sequences deriving from retrotransposons, genomic DNA, mRNA and sgRNA. Among 93 mice analysed, 57 carried mutant alleles and 22 of them had long de novo insertion(s) at DSB-introduced sites; two were spliced mRNAs of Pcnt and Inadl without introns, indicating the involvement of reverse transcription (RT). Fifteen alleles included retrotransposons, mRNAs, and other sequences without evidence of RT. Two others were sgRNAs with one containing T7 promoter-derived sequence suggestive of a PCR product as its origin. In conclusion, RT-product-mediated DSB repair (RMDR) and non-RMDR repair were identified in the mouse zygote. We also confirmed that both RMDR and non-RMDR take place in CRISPR/Cas transfected NIH-3T3 cells. Finally, as two de novo MuERV-L insertions in C57BL/6 mice were shown to have characteristic features of RMDR in natural conditions, we hypothesize that RMDR contributes to the emergence of novel DNA sequences in the course of evolution.

Figure 1. CRISPR/Cas mediated gene manipulation.

(a) The pCAG-EGxxFP plasmid contains 5′ and 3′ EGFP fragments that share 482 bp under a ubiquitous CAG promoter. A 500-bp genomic fragment containing the sgRNA target sequence was placed between the EGFP fragments of the pCAG-EGxxFP plasmid. The resulting target plasmid was co-transfected with pX330 plasmids expressing the sgRNA and hCas9 into HEK293T cells. Once the target sequence was digested by sgRNA guided CAS9 endonuclease, homology dependent repair (HR: homologous recombination, or SSA: single-strand annealing) took place and reconstituted the EGFP expression cassette. MCS; multi cloning site. (b) The DSB efficiency was validated with the pCAG-EGxxFP system by observing EGFP fluorescence 48 hrs after the transfection (scale bar: 200 μm). The percentages of EGFP-positive cells are indicated. (c) Schematic representation of the positions of each sgRNA and primer to check the CRISPR/Cas mediated mutations (left side of (c)). Electrophoresis of the PCR products from each of the pX330 plasmid-injected mice (the right side of (c)). At least four PCR products (yellow arrowheads) were larger than expected (WT: red arrowheads) in the Peg10-ORF2-sgRNA-injected mice. One and two PCR products (yellow arrowheads) were larger than Cxx1a WT and Cxx1b WT, respectively, in the Cxx1a/b-sgRNA-injected pups. Three PCR products (yellow arrowheads) were larger than 1041 bp (Rgag1 WT) in the Rgag1-sgRNA-injected pups.

Figure 2. Structure of the captured retrotransposons associated with DSB repair.

De novo inserted retrotransposons at the Peg10-ORF2 (a–c), Cxx1a/b (d,e), and Rgag1 (f) loci were induced by pX330 injection into mouse zygotes. Both the post-integration site and pre-integration sequences (bottom of the panel) are shown. The nucleotide sequences that correspond to the single guide RNA sequence and the PAM sequences are shown in red and bold red characters, respectively. The black lines indicate the junction sites between pre- and post-integration sequences. The sequences in the blue boxes are overlapping microhomologies and are marked with black dotted lines. Short sequences of unknown origin are shown in green. Each insertion was truncated at both the 5′ and 3′ ends, but they demonstrated distinct features. These included the absence of LTRs and TSDs. (a) Together with MuERV-L, 50 bp of Peg10 cDNA sequence was inserted with 1-bp microhomology. (b) A truncated MaLR internal sequence was inserted with a 7-bp overlapping microhomology (a 2-bp mismatch). (c) A truncated MuERV-L Pol region was inserted with a 7-bp microhomology (1-bp mismatch) and 4-bp TSDs (yellow bars) ‘ttct’. (d) A truncated MuERV-L was inserted with 3-bp overlapping microhomology. (e) Multiple truncated retrotransposons were inserted with 1-5-bp microhomologies. (f) A truncated ORR1AI retrotransposon was inserted with a 6-bp microhomology (1-bp mismatch).

Figure 3. RT-product-mediated DSB repair (RMDR) and non-RT-product-mediated DSB repair (non-RMDR).

(a) A partial sequence of the processed Pcnt gene (6803–7599 bp: NM_001282992) in reverse orientation was inserted into the DSB site mediated by RMDR. The integrated Pcnt fragment skipping introns 30–32 was inserted with a 3-bp microhomology. (b) A partial sequence of the processed Inadl gene (1527–1836 bp: NM_001005784) and a DNA oligo (pink bar) including a T to G mutation with 40 bp out of 53 bp in the 5′ homology arm were inserted into the DSB site. The 3′ side of the DSB site was repaired by HR with the long homology of the DNA oligo and 5′ side of the DSB site repaired by capture of the Inadl gene fragment, skipping introns 11–14 (RMDR). (c) A partial sgRNA sequence with T7 promoter (green bar) was inserted with 3-bp and 1-bp microhomologies mediated by non-RMDR. (d) A partial sgRNA in reverse orientation was inserted with a 21-bp microhomology, including a 20-bp guide RNA sequence. (e) Distribution of 30 insertion sequences at CRISPR/Cas DSB sites in 20 mice in which the DSBs were repaired by the capture of long DNA sequences. Approximately 57% of the inserted sequences were derived from LTR retrotransposons. The 8 insertions correspond to the exon regions of 8 genes. Two insertion sequences correspond to multiple exons, skipping introns, demonstrating that they are derived from cDNA.

Figure 4. Possible mechanisms of RMDR, and RMDR under natural conditions.

DSBs (the orange triangles) induced by CRISPR/CAS (blue sphere) are shown undergoing repair by NHEJ (a), non-RT-product-mediated DSB repair (non-RMDR) (b), RMDR with pre-existing cDNA (c) and RMDR with direct RT (d). Most of the DSBs induced by CRISPR/Cas are repaired by NHEJ (a), while certain DSBs are repaired by the capture of other sequences (b–d). (c) A pre-existing cDNA (red bar) generated by RT anneals with both DNA ends of a DSB site, which is repaired (RMDR with double microhomologies). (d) mRNA anneals with one DSB end with microhomology, and cDNA is synthesized by RT machinery. Two murine-specific truncated MuERV-L insertions were identified by comparing rodent genomes. Pre-integration sequences (indicated in blue text) were deduced from other available rodent genomes. (e) Murine-specific truncated MuERV-L (52824976–52825534 bp: chromosome 9) was integrated with two 2-bp microhomologies at both the 5′ and 3′ ends. (f) C57BL/6 strain-specific truncated MuERV-L (191922408–191922914 bp: chromosome 1) was integrated with a 27-bp microhomology (6-bp mismatches and a 5-bp insertion).