Efficient genome-wide mutagenesis of zebrafish genes by retroviral insertions

Abstract

Using a combination of techniques we developed, we infected zebrafish embryos using pseudotyped retroviruses and mapped the genomic locations of the proviral integrations in the F(1) offspring of the infected fish. From F(1) fish, we obtained 2,045 sequences representing 933 unique retroviral integrations. A total of 599 were mappable to the current genomic assembly (Zv6), and 233 of the integrations landed within genes. By inbreeding fish carrying proviral integrations in 25 different genes, we were able to demonstrate that in approximately 50% of the gene "hits," the mRNA transcript levels were reduced by >/=70%, with the highest probability for mutation occurring if the integration was in an exon or first intron. Based on these data, the mutagenic frequency for the retrovirus is nearly one in five integrations. In addition, a strong mutagenic effect is seen when murine leukemia virus integrates specifically in the first intron of genes but not in other introns. Three of 19 gene inactivation events had embryonic defects. Using the strategy we outlined, it is possible to identify 1 mutagenic event for every 30 sequencing reactions done on the F(1) fish. This is a 20- to 30-fold increase in efficiency when compared with the current resequencing approach [targeting induced local lesions in genomes (TILLING)] used in zebrafish for identifying mutations in genes. Combining this increase in efficiency with cryopreservation of sperm samples from the F(1) fish, it is now possible to create a stable resource that contains mutations in every known zebrafish gene.

Fig. 1.

Construction of a zebrafish retroviral insertional mutant library. Pipeline for constructing the zebrafish insertional mutant library: (i) infect zebrafish embryos with pseudotyped MLV virus at the 1,000- to 2,000-cell stage; (ii) raise the injected founder fish with high infection rates, determined by qPCR; (iii) inbreed or outcross founders (depending on the infection rates of the founder fish) and raise the F₁ fish; (iv) cryopreserve sperm samples from the F₁ fish and perform LM-PCR followed by shotgun cloning, sequencing, and mapping the integrations in the corresponding sperm samples.

Fig. 2.

Germ-line transmission rates tested in selected founders. Seven founders with an average EAV of 63 integrations per cell were randomly selected and outcrossed with WT fish to generate F₁ families, OF1–OF7. Six F₁ fish were randomly selected from each family, and the copy number of proviral integrations in each F₁ fish was determined by quantitative PCR. Variations in germ-line transmission of proviral insertions can be seen between different founders (e.g., all selected F₁ fish from family OF6 have >10 copies of proviral insertions per cell, whereas from OF5, only two F₁ fish have more than five copies of proviral insertions per cell) and between the F₁ siblings of each founder (e.g., the copy numbers of proviral insertions in selected F₁ fish from OF1 range between 0.7 and 16.6 copies per cell). Overall, 32 of 42, or 76% of randomly selected F₁ fish harbor more than five proviral insertions per cell with an average value of 10.4 ± 5.0 copies per cell.

Fig. 3.

Summary of 933 integrations. (A) BLAST analysis of 933 proviral integrations; 599 of 933 sequences could be mapped in the zebrafish genome based on the latest genome assembly Zv6. (B) Distribution of 599 mapped proviral integrations across chromosomes. (C) Distribution of 225 proviral integrations landed in introns; 59% (132/225) of integrations in introns landed in the first intron.

Fig. 4.

Quantitative RT-PCR analysis of RNA transcript levels in integration homozygotes. (A) Twenty-five integrations landing within genes in either an intron or an exon or in the near upstream or downstream proximity were inbred. Total RNA from 24 embryos of each inbreed was isolated and the genotype was determined. The RNA transcript levels between three integration homozygotes (−/−) and three siblings with the WT background (+/+) were then compared by qRT-PCR. Of 25 gene hits tested, 11 of 25 show significant reduction (>70%) in RNA levels; 8 of 25 show >90% reduction. There is a strong trend for integrations in the first intron to be mutagenic; 8 of 10 cases with integrations in the first intron showed >70% reduction in gene expression; 6 of 10 showed >90% reduction.