High-throughput gene targeting and phenotyping in zebrafish using CRISPR/Cas9

Abstract

The use of CRISPR/Cas9 as a genome-editing tool in various model organisms has radically changed targeted mutagenesis. Here, we present a high-throughput targeted mutagenesis pipeline using CRISPR/Cas9 technology in zebrafish that will make possible both saturation mutagenesis of the genome and large-scale phenotyping efforts. We describe a cloning-free single-guide RNA (sgRNA) synthesis, coupled with streamlined mutant identification methods utilizing fluorescent PCR and multiplexed, high-throughput sequencing. We report germline transmission data from 162 loci targeting 83 genes in the zebrafish genome, in which we obtained a 99% success rate for generating mutations and an average germline transmission rate of 28%. We verified 678 unique alleles from 58 genes by high-throughput sequencing. We demonstrate that our method can be used for efficient multiplexed gene targeting. We also demonstrate that phenotyping can be done in the F1 generation by inbreeding two injected founder fish, significantly reducing animal husbandry and time. This study compares germline transmission data from CRISPR/Cas9 with those of TALENs and ZFNs and shows that efficiency of CRISPR/Cas9 is sixfold more efficient than other techniques. We show that the majority of published "rules" for efficient sgRNA design do not effectively predict germline transmission rates in zebrafish, with the exception of a GG or GA dinucleotide genomic match at the 5' end of the sgRNA. Finally, we show that predicted off-target mutagenesis is of low concern for in vivo genetic studies.

Figure 1.

Overview of mutagenesis and phenotyping strategies. (A) Single-guide RNA (sgRNA) was synthesized from a template that is generated by annealing and extending oligo A and B. Oligo B is generic and is common in all reactions, while Oligo A contains a T7 promoter, 20-nt target sequence, and another 20 nt overlapping the chimeric gRNA core sequence. Two sgRNAs targeting a single gene were co-injected along with Cas9mRNA into either the yolk or the cell of one-cell stage embryos. The injected embryos were raised to generate the founder fish. (B) The founder fish were then outcrossed to wild type to generate heterozygous F₁ fish. The mutant fish were identified by fluorescence PCR and sequencing. The siblings carrying mutations were then crossed to generate F₂ progeny, and phenotype-genotype correlations were done using the F₂ embryos. (C) Alternatively, the founder fish were inbred and phenotyping was performed in the F₁ generation, and the embryos were genotyped by fluorescence PCR or sequencing. (D) Phenotypes can also be observed in the injected embryos from 0 to 5 d, although off-target effects are more common with this approach. In order to score phenotypes in injected embryos, the sgRNA and Cas9 must be injected in the cell instead of the yolk to achieve maximum efficiency.

Figure 2.

Overview of the mutant identification strategies. Mutant alleles were identified by fluorescence PCR and high-throughput sequencing. Gene-specific primers were designed to amplify the regions around the target site (200–300 bp). The forward primer contains an M13-forward sequence on the 5′ end, and the reverse primer contains a pig-tail sequence to reduce size heterogeneity PCR artifacts. Depending on the application, one of two additional primers was added. (Option A) For fluorescent PCR, an M13-forward oligo with the fluorophore FAM attached was added. The resulting primers were run on an ABI 3100 or 3730 along with a size standard to obtain the amplicon sizes accurate to base-pair resolution. The size of the peak relative to the wild-type peak control determines the nature and length of the mutant. (Option B) When amplicons were sequenced, a third primer containing M13-forward sequence with a 6-bp barcode attached to the 5′ end was used. A unique barcode was assigned to each embryo from the same target. The amplicons were sequenced on an Illumina MiSeq platform with 300-bp paired-end reads. The sequences were processed using ampliconDIVider for the identification of insertions and deletions induced by CRISPR/Cas9 (http://research.nhgri.nih.gov/CRISPRz).

Figure 3.

Summary and comparison of CRISPR/Cas9 to ZFN- or TALEN-induced mutations. (A) Summary of mutant identification data compiled from CRISPR/Cas9, TALENs, and ZFNs. The mutants were identified using fluorescent PCR and the aggregate data from each technique is shown. (B) Comparison of the types of mutations detected by sequencing. The mutations were classified in three different groups: deletions, insertions, and complex (at least one deletion and at least one insertion detected within ±30 bp of the 3′ end of the sgRNA or ZFN/TALEN target).

Figure 4.

Distribution of CRISPR/Cas9-induced mutations in the germline. CRISPR/Cas9 induced more deletions than insertions. Approximately two-thirds of the deletions and insertions were frameshift mutations and most likely caused gene truncations. (B) Mutations were classified according to their insertion or deletion size as determined by fluorescent PCR data. Left bars (dark gray) indicate the deletion range and right bars (light gray) indicate insertion sizes. The bars on each extreme of the x-axis indicate the total number of mutations (deletions or insertions) that are >20 bp in size. Insertions and deletions >20 bp ranged to the largest detected size of a 182-bp deletion and 52-bp insertion.

Figure 5.

Heritable chromosomal deletions induced by CRISPR/Cas9. (A) Summary of detected large deletions between two target sites. Large deletions were identified in man2a1, mgat1b, man2b1, and mgat3a. (B) Graphical representation of the man2a1 locus and the sequence of the deletion interval. (C–E) Sequence of the deleted interval for mgat1b, man2b1, and mgat3a. The target sites are marked in gray, while the PAM sites are underlined. The sequences were obtained from the PCR products across the two target sites. The top strand represents the wild-type sequence and the bottom strand represents the junction sequences between two target sites. The size of the deletion is indicated on the right side.

Figure 6.

fgfr4^−/− phenotype identified by inbreeding injected founder fish. Two single-guide RNAs (sgRNAs) targeting two different exons of the fgfr4 gene were co-injected with Cas9mRNA into wild-type embryos and raised to adults to generate founder fish (F₀). Six pairs of founder fish (F₀) were inbred and scored for mutant phenotypes. One of the six pairs showed multiple embryos with the same phenotype observable at 48 h post-fertilization. (A) Wild-type embryo at 48 h post-fertilization. White arrow indicates a normally sized hindbrain, white arrowhead indicates a normal heart chamber, and black arrow represents a normal tail length. (B) An fgfr4 compound heterozygous mutation displaying multiple phenotypes. White arrow indicates a reduced hindbrain region, white arrowhead points to a heart edema, and black arrow shows the shortened body axis. All phenotypes were only found in the compound heterozygous mutant embryos.