CRISPR/Cas9-mediated precise genome modification by a long ssDNA template in zebrafish

Abstract

Background: Gene targeting by homology-directed repair (HDR) can precisely edit the genome and is a versatile tool for biomedical research. However, the efficiency of HDR-based modification is still low in many model organisms including zebrafish. Recently, long single-stranded DNA (lssDNA) molecules have been developed as efficient alternative donor templates to mediate HDR for the generation of conditional mouse alleles. Here we report a method, zLOST (zebrafish long single-stranded DNA template), which utilises HDR with a long single-stranded DNA template to produce more efficient and precise mutations in zebrafish.

Results: The efficiency of knock-ins was assessed by phenotypic rescue at the tyrosinase (tyr) locus and confirmed by sequencing. zLOST was found to be a successful optimised rescue strategy: using zLOST containing a tyr repair site, we restored pigmentation in at least one melanocyte in close to 98% of albino tyr^25del/25del embryos, although more than half of the larvae had only a small number of pigmented cells. Sequence analysis showed that there was precise HDR dependent repair of the tyr locus in these rescued pigmented embryos. Furthermore, quantification of zLOST knock-in efficiency at the rps14, nop56 and th loci by next generation sequencing demonstrated that zLOST showed a clear improvement. We utilised the HDR efficiency of zLOST to precisely model specific human disease mutations in zebrafish with ease. Finally, we determined that this method can achieve a germline transmission rate of up to 31.8%.

Conclusions: In summary, these results show that zLOST is a useful method of zebrafish genome editing, particularly for generating desired mutations by targeted DNA knock-in through HDR.

Keywords: CRISPR/Cas9; Disease modeling; Genome editing; Homology-directed repair; Long single-stranded DNA; Next-generation sequencing; Zebrafish.

Fig. 1

CRISPR-mediated tyr knockout to establish a visual knock-in assay. a Schematic illustration of CRISPR/Cas9-mediated gene editing of tyr. First to knock out (KO) gene function with a 25 nt deletion within the first exon and then knock-in (KI) rescue of this gene using a repair template. b Target sites and T7E1 assays of tyr and tyr^25del/25del loci. PAMs are marked with red. Ctl represents PCR products without T7E1 digestion. WT denotes PCR products from uninjected embryos with T7E1 digestion. Tyr or tyr^25del/25del denotes PCR products from injected embryos with T7E1 digestion. c T-cloning and Sanger sequencing identify tyr^25del/25del in F2 zebrafish. Upper row shows wild type (WT) sequence. Open reading frame codons are demarked in green frame. 25 bp deletion in homozygous tyr mutants is marked with blue in upper row, which leads to a frameshift mutation (marked with red in lower row). d Lateral views of larvae at 2 dpf (scale bar = 1 mm) and adult (scale bar = 10 mm): wild type (upper row) and tyr^25de/l25del (lower row).

Fig. 2

A genetic assay for comparing the efficiency of homology-directed repair using tyr mutant. a Table of template design schematics (left), attributes of the template (middle) and proportion of observed pigmented embryos in the tyr^25de/l25del model (right). Embryos are analysed at 2 dpf after co-injection of zCas9 mRNA, tyr^25de/l25del gRNA together with repair template. Number of embryos evaluated (n) exceeded 100 for each condition. b Phenotypic evaluation of embryos at 2 dpf into three groups according to number of pigmented cells: low rescue (1–20 pigmented cells), medium rescue (21–40 pigmented cells) and high rescue (more than 40 pigmented cells). Scale bar = 1 mm. c Statistics of HDR efficiency induced by different repair templates. zLOST: long single stranded template 299 bp, ssODN: single strand DNA oligonucleotides 105 bp, cdsDNA: circular double stranded DNAs 1527 bp (with two gRNA sites at both ends of the homologous arms), Ctl: without repair template. Number of embryos assessed (n) is shown for each group. X²-test (***p < 0.001). d Sequence analysis confirming that the larvae contained a correctly repaired tyr locus by zLOST. Correct insertion by HDR (green), PAM region (blue), target sites (underlined), Indels (red) are indicated.

Fig. 3

Zebrafish genome editing at three other target sites by zLOST a Restriction enzyme-based method design of three target sites. Target sequence (black), PAM region (blue), target modification sites (red), and restriction site (underlined) are indicated. b Restriction enzymes are used to digest the amplified region of the target genes. T = th, N = nop56, R = rps14. The “positive embryos” groups are highlighted by asterisk. c Sequencing results of the th, nop56 and rps14 loci. Patterns of DNA modification observed in independent embryos pool. Note: △1 and △2 mean the presence of additional undesirable mutations outside of the shown sequence window.

Fig. 4

NGS analysis of precise point mutation introduction to the genes th, nop56 and rps14 Total percentages of defined sequence reads classes at knock-in sites of th (a), nop56 (b) and rps14 (c) genes as engineered with three types of repair template (cdsDNA, ssODN and zLOST). All the reads are divided into four classes: WT, others, correct_HDR and incorrect_HDR. Incorrect_HDR indicates the reads containing target modification sites, but with extra undesirable amino acid changes. d Representative examples of different classes of th, nop56 and rps14 HDR knock-ins: Correct_HDR and Incorrect_HDR.

Fig. 5

zLOST enables mimicking of human disease related mutations in zebrafish Alignment of human patients and desired zebrafish mutations to model human Barber-Say syndrome (BSS) or Diamond-Blackfan anaemia (DBA), schematic outlines of the gene editing strategy and sequencing of the resulting twist2 and rpl18 zebrafish loci. a Diagram of the mutation associated with human BSS. The substituted target base is marked in red, which means a p.E78Q amino-acid change in the zebrafish homologue precisely mimics the p.E75Q mutation found in human patients. b and d Design principles of HDR templates that contain a non-synonymous mutation of the sequence close to the PAM site in addition to synonymous nucleotide changes that create a Coding-bar used for genotyping that utilizes a de novo endonuclease restriction site. Sequencing result at the twist2 and rpl18 zebrafish loci targeted by the zLOST system. The Coding-bar includes a restriction endonuclease (PflFI) site 5′-GACNNNGTC-3′ in twist2 and a restriction endonuclease (PvuI) site 5′-CGATCG-3′ in rpl18. c Diagram depicting the mutation associated with human DBA, mimicking the p.L51S mutation at rpl18 locus found in patients.