Heritable targeted gene disruption in zebrafish using designed zinc-finger nucleases

Abstract

We describe the use of zinc-finger nucleases (ZFNs) for somatic and germline disruption of genes in zebrafish (Danio rerio), in which targeted mutagenesis was previously intractable. ZFNs induce a targeted double-strand break in the genome that is repaired to generate small insertions and deletions. We designed ZFNs targeting the zebrafish golden and no tail/Brachyury (ntl) genes and developed a budding yeast-based assay to identify the most active ZFNs for use in vivo. Injection of ZFN-encoding mRNA into one-cell embryos yielded a high percentage of animals carrying distinct mutations at the ZFN-specified position and exhibiting expected loss-of-function phenotypes. Over half the ZFN mRNA-injected founder animals transmitted disrupted ntl alleles at frequencies averaging 20%. The frequency and precision of gene-disruption events observed suggest that this approach should be applicable to any loci in zebrafish or in other organisms that allow mRNA delivery into the fertilized egg.

Figure 1. Yeast-based system of identification of maximally active ZFNs

(A) Components of yeast-based chromosomal reporter system: expression vectors for the ZFNs (top), target gene integrated into the chromosome at the HO locus (middle). After a double-strand break (DSB) is induced at the target by a pair of ZFNs, it is processed via single-strand annealing (SSA) to repair the MEL1 reporter gene, the activity of which can be rapidly assayed in liquid culture. (B) Outline of procedure for ZFN activity measurement in yeast. (C) Reporter gene restoration is dependent on a precisely targeted DSB. A reporter gene was engineered as in Fig. 1A, except it carried a recognition site of the HO endonuclease. Mel1 activity is only observed following induction of expression of the endonuclease (“+gal”), and of its target in the reporter locus (bottom row). (D) Exceedingly low frequency of spontaneous MEL1 gene repair. Essentially no Mel1 activity is observed when HO expression is not induced (left panel), with the exception of very low-frequency spontaneous MEL1 restoration events (right panel), visualized as small sectors of blue cells within otherwise white colonies. As shown in the middle, induction of HO endonuclease expression converts the overwhelming majority of the cells in the sample to a MEL1 state.

Figure 2. Injection of golden ZFN-encoding mRNA into zebrafish embryos induces targeted loss-of-function mutations in somatic cells

(A) Twenty one pairs of high-fidelity, obligate-heterodimer ZFNs (Supplementary Table 1) targeting the gol locus were screened for activity in yeast using the assay shown in Fig. 1, with the exception that Mel1 activity was assayed in the yeast growth medium, rather than relying on the number of MEL1-positive colonies. For each ZFN pair, activity in the medium of the Mel1 enzyme is listed in mU, as detected after induction by galactose. (B) Schematic of the gol locus indicating recognition sites of three ZFN pairs chosen for further testing. ZFN pair 1 recognizes a site within Exon 4 upstream of the loss-of-function gol^b1 allele, whereas ZFN pairs 14 and 15 both recognize a site within the last exon upstream of coding sequence for two putative C-terminal transmembrane helices. (C) Injection of ZFN pair 14 mRNA (5 ng) into gol^b1 heterozygous 1-cell embryos induces somatic loss-of-function mutations in the wild-type gol allele. The 2 dpf embryo on the top left shows no evidence of loss of heterozygosity; the eyes are uniformly and darkly pigmented like wild-type embryos. Single or multiple patches of unpigmented cells can be observed in gol ZFN-injected embryos (remaining three panels), and sometimes mosaic patches were seen in both eyes. Except for these pigment clones, the embryos shown were phenotypically normal. (D) When PCR products from pooled individuals were sequenced, deletions and insertions typical of NHEJ were observed. In some cases, these are in frame and in other cases are frameshifts.

Figure 3. Injection of no tail ZFN-encoding mRNA into zebrafish embryos induces targeted loss-of-function mutations in somatic cells

(A) Five high-fidelity, obligate-heterodimer ZFN pairs (Supplementary Table 2) targeting the ntl locus were screened for activity using the yeast-based assay shown in Fig. 1. See Fig. 2A legend for details. (B) Schematic of the ntl locus indicating the recognition site of the ZFN pairs chosen for further testing. ZFN pairs 2 and 3 recognize the same site within Exon 2 just upstream of the site of the loss-of-function ntl^b195 insertion allele. (C) Injection of 5 ng ntl ZFN pair 2 mRNA into 1-cell ntl^b195 heterozygous embryos induces mutations in the wild-type ntl allele. Compared to uninjected embryos at 1 dpf (top panel), ZFN-injected embryos lack a notochord and tail like null mutants, or have patchy notochords and forked tails (inset), like hypomorphic ntl mutants. (D) Graph showing the percentage of ntl^b195 heterozygous embryos displaying wild-type or ntl mutant phenotypes after injection of 5 ng mRNA encoding either ZFN pair 2 or ZFN pair 3. A small percentage of injected embryos were defective (“unscored”). Approximately one-quarter of the ntl-like embryos were slightly more necrotic than the typical ntl mutant (4/17 for ZFN pair 2 and 5/18 for ZFN pair 3). (E) When PCR products from pooled ntl-like individuals were sequenced, deletions and insertions typical of NHEJ were observed. In some cases, these are in frame and in other cases cause frameshifts.

Figure 4. Injection of no tail ZFN-encoding mRNA in wild-type embryos creates novel ntl mutations that are transmitted through the germline

(A) Some juvenile fish derived from wild-type embryos injected with mRNA encoding ntl-targeting ZFNs (both conventional and high-fidelity FokI domains, the latter shown here) show posterior truncations. Right two panels show posteriorly truncated juveniles; left panels are normal-appearing siblings. (B) ntl phenotypes observed in progeny of ZFN-injected founder animals in complementation crosses at 2 dpf. Wild-type embryos injected with mRNA encoding high-fidelity, ntl-targeting ZFNs were grown to adulthood and eggs from founder females were fertilized in vitro with sperm from a ntl^b195 heterozygous male. Representative progeny from this complementation test with Founder female A are shown in the right panel. (C) Novel ntl alleles (7 total) carried by founders that gave phenotypically ntl progeny in complementation cross (see also Table 3). Founder A was derived from a wild-type embryo injected with ZFN pair 2 (5 ng mRNA), and founders B through D from wild-type embryos injected with ZFN pair 3 (5 ng mRNA).