This site is a work in progress. In its current form it provides a basic idea for the method we have developed to screen for new genes involved in any process of interest in zebrafish. In time it will be expanded to serve as a repository for any useful CRISPR information in the hopes of helping other labs use this technology because its just so powerful.

The CRISPR technology allows for unprecedented ease in modifying the genomes of a broad variety of organisms. CRISPR targets the genome using a single guide RNA (gRNA) and the Cas9 enzyme. Cas9 then cuts the DNA and the cell repairs this break using non-homologous end joining (NHEJ). NHEJ is a messy process that frequently causes insertions and deletions (InDels). This can be harnessed to create mutations in any gene of interest.

InDels in zebrafish can be created easily by injecting a gRNA with cas9 RNA. The simplicity of cloning (protocols) has made this the emerging reverse genetics approach of choice. Some of the first efforts in zebrafish found that known mutant phenotypes could be phenocopied in injected embryos.
We reasoned that with a sufficiently high mutation rate we could quickly screen large numbers of genes for involvement in a process of interest.

We verified that we could knockout genes of interest in injected embryos by examining gjd1a. In a previous forward genetic screen we identified gjd1a (cx34a) as being required for electrical synapse formation. We created a gRNA targeting gjd1a and tested guide and cas9 concentrations across many orders of magnitude for efficiency. We found that we could phenocopy the mutant and that increasing concentration greatly increased mutation efficiency. At its best, every CRISPR injected embryo lost electrical synapses, with the phenotypes ranging from complete loss (75% of embryos), to half of synapses lost (15% of embryos), to many of the synapses lost (10%). The variability is due to the fact that each embryos is mosaic for CRISPR induced mutations, thus each animal has a unique genotype. The very high percentages of phenotypically mutant embryos in the injections made it clear we could screen for new genes of interest involved in synaptogenesis.

Our goal was to screen large numbers of genes for electrical synapse phenotypes. While injections could be done one at a time, we wanted to test whether we could multiplex multiple gRNAs in a single injection and still efficiently see phenotypes and InDels in the genome. We multiplexed gjd1a and seven other gRNAs in a single injection and found that embryos still efficiently lost electrical synapses. Moreover, embryos also had efficient cutting of the genome at the all targeted loci. This suggested we could quickly screen multiple targets by pooling gRNAs. We designed 48 gRNAs to screen for synapse phenotypes.

While pooling reduces the number of injections needed to screen multiple genes, a "pool hit" hides the individual gene involved. Because of this we plated targets in a 48-well format, and pooled across each column and row, resulting in 14 pools. The strategy was that any "column hit" would also get a "row hit", and the intersection would reveal the position of the likely gene. Our targets were genes biochemically linked to electrical synapses or synaptic cell adhesion molecules important for chemical synapses but with no known role in electrical synaptogenesis. Both the column and row injections had 3 positive hits, creating 9 intersections. We injected these 9 individually (as well as the others within each pool) and found that 3 genes within our screen had phenotypes. The first, gjd1a, was our positive control. The screen also found that gjd2a, and tjp1b caused a loss of synapses in injected embryos. We have confirmed that gjd2a is required by raising mutant animals, and tjp1b mutants are currently growing.  The CRISPR screening strategy can be used efficiently to identify new genes involved in a process of interest.

The positive hits we found in the screen were exciting, but we wondered how many of the genes we had actually screened. In other words, were we missing important genes because the CRISPRs were not cutting the genome at a high enough frequency. We used next generation sequencing to assay the InDel efficiency at each target site. The mutation efficiency varied from 85% to 15% of alleles carrying an InDel, depending on the target site (targets). However, we found synapse phenotypes in genes that had 20% cutting efficiency, suggesting we had screened most, if not all, of the genes in our screen. We found that 5 genes failed to cut efficiently, with InDel frequencies as all of the off target sites that we sequenced. These failures were either errors in guide creation or guides that did not cut. The sequencing makes us confident that we screened ~90% of our targets. Our results show that CRISPR screening is a powerful strategy that provides an efficient and cost effective method to screen for genes of interest.

While we used the electrical synapse phenotype to develop the method, we and others have phenocopied other types of mutations in injected embryos. In our hands we have phenocopied mutants with defects in neuronal migration, fin formation, ciliogenesis, and pigmentation. We have additionally phenocopied known mutants, and identified new genes, involved in convergent extension. Thus the strategy can be applied in many labs interested in wide ranging processes.