Rapid reverse genetic screening using CRISPR in zebrafish

Abstract

Identifying genes involved in biological processes is critical for understanding the molecular building blocks of life. We used engineered CRISPR (clustered regularly interspaced short palindromic repeats) to efficiently mutate specific loci in zebrafish (Danio rerio) and screen for genes involved in vertebrate biological processes. We found that increasing CRISPR efficiency by injecting optimized amounts of Cas9-encoding mRNA and multiplexing single guide RNAs (sgRNAs) allowed for phenocopy of known mutants across many phenotypes in embryos. We performed a proof-of-concept screen in which we used intersecting, multiplexed pool injections to examine 48 loci and identified two new genes involved in electrical-synapse formation. By deep sequencing target loci, we found that 90% of the genes were effectively screened. We conclude that CRISPR can be used as a powerful reverse genetic screening strategy in vivo in a vertebrate system.

Figure 1

Optimizing multiplexed CRISPR screening for electrical synapses. (a) Model of the Mauthner (M) circuit. Each M cell body resides in the hindbrain and sends an axon into the spinal cord making electrical synapses (Cx36) with repeating Commissural Local (CoLo) neurons. (b–d) Images are 15 µm dorsal view projections of two spinal cord segments at 5 dpf. Anterior is to the left. Scale bar = 10 µm. Larvae are stained for Connexin36 (Cx36, white) and neurofilaments (RMO44, red) to mark neuronal processes, including M and CoLo. Individual Cx36 channels are shown in neighboring panels. The Cx36 staining found at M/CoLo synapses (b) is lost in gjd1a^fh436 (8 bp deletion) mutant animals (c) and the phenotype can be recapitulated in injected CRISPR F₀ embryos (d, 100/1200 pg sgRNA/cas9). (e) Model of gjd1a exon1 and CRISPR targets. Arrows denote primers used for qPCR analysis of mutational efficiency. Underlined sequence denotes the NGG motif used by Cas9. (f,g) Quantitation of mosaic electrical synapse loss (f) and mutational efficiency (g) in injected embryos for each gjd1a target injected at 100/1200 pg sgRNA/cas9. The “ratio eSyn defect” is the proportion of electrical synapses missing from at least 30 sampled per animal. Mutational efficiency was assessed by qPCR and was done in triplicate. (h,i) Quantitation of mosaic electrical synapse loss (h) and toxicity (i) seen in injected embryos for gjd1a Target1 multiplexed with five other sgRNAs that have no effect on electrical synapses (gfp, slc24a5, pk1b, sox10, hmcn1) with 1200 pg of Cas9-encoding mRNA. “Toxicity” encompasses embryo death, edema, localized cell death, and general developmental defects. In f and h, N > 24 embryos for each bar. In g, each bar represents 5 embryos pooled in 3 replicates, error bars denote s.e.m. In i, N > 85 embryos for each point.

Figure 2

CRISPR screening identifies new genes required for electrical synaptogenesis. (a) Layout of target genes for multiplexed row and column injections. Genes were arranged such that gene family members were targeted in either a row or column. Grey rows/columns denote multiplexed injections that caused a loss of electrical synapses. The dark-grey genes highlight the nine intersecting targets most likely to be causative for synapse loss. sgRNAs were designed against Short (S), Long (L), and if no unique site could be found, Both (B), isoforms of several genes. (b) Quantitation of mosaic electrical synapse loss in injected embryos for multiplexed sgRNAs of the noted row or column. (c) Quantitation of mosaic electrical synapse loss in injected embryos for each of the intersecting sgRNAs identified individually injected at 100/1200 pg sgRNA/cas9. (d–h) Images are oriented, imaged, and stained as in Figure 1. The Cx36 staining found at M/CoLo synapses (d) is lost in injected CRISPR F₀ embryos targeting gjd2a or tjp1b (e,f). Homozygous mutant gjd2a^fh437 (5 bp deletion) and trans-heterozygous tjp1b (10 bp deletion / 5 bp deletion) animals confirm each gene is required for electrical synapse formation (g,h). (i) Quantitation of InDel frequency at all target loci in the CRISPR screen based on deep sequencing. Each unfilled box represents an individual screen target. Filled black boxes represent the predicted off target loci for the phenotypically positive hits identified. In b and c, N > 24 embryos for each bar. In i, each point represents >136,000 reads.