Mutagenized populations carry a large number of potentially deleterious recessive mutations that can be discovered based on the phenotypes of homozygous mutants (forward genetics) or by detecting them directly in the genomic DNA of heterozygous or homozygous individuals, irrespective of any phenotypes they may cause (reverse genetics). In the latter case, the challenge is to identify rare mutations within a large population of non-mutant individuals. This process, called ‘TILLING’ has been applied to a range of plant and animal species [1–11]. The idea of screening mutagenized populations of zebrafish by the polymerase chain reaction (PCR) for mutations in genes of interest was first proposed for mutagens that cause deletions [12, 13]; however, the challenges of working with large deficiencies limited this approach. A key advance was made by McCallum et al. [14], who used automated denaturing high-performance liquid chromatography to detect heteroduplexes resulting from rare single nucleotide polymorphisms (SNPs) in pooled ethyl methanesulfonate (EMS)-mutagenized Arabidopsis populations. They coined the term ‘TILLING’ to describe this general approach. Soon after, the same group introduced the use of the Cel1 enzyme isolated from bulk celery to detect polymorphisms in these populations [15]. Cel1 is a plant-specific extracellular glycoprotein that cleaves heteroduplex DNA at all possible single nucleotide mismatches [16]. Using fluorescently labelled primers to detect Cel1 cleavage products on a LiCor acrylamide slab gel, Colbert et al. [15] showed that Cel1 could detect a heterozygous SNP in pools of eight mutagenized plants. This pooling strategy allowed a 1 kb fragment to be screened in a library of 3000 mutagenized plants in three 96-well LiCor runs—a throughput of one gene per day.

The general approach for TILLING in zebrafish is simple: genomic DNA from a large library of ENU-mutagenized zebrafish is screened for rare mutations in genes of interest either using Cel1 on pools of fish or by resequencing individual fish. Mutations are then recovered by out-crossing the single identified carrier. The key components of a successful TILLING project are (i) a large, well-mutagenized library of fish, (ii) an efficient screening method and (iii) a near-perfect ability to recover valuable mutations once they have been found.

The single major drawback of TILLING as a method for reverse genetics is that a mutation of interest that is identified in any given F1 fish is only one of many heterozygous mutations in its genome. By multiplying the estimated size of the zebrafish genome (1.45 × 10³ Mb) by the mutation rate detected by TILLING (2–4 mutations per megabase), an estimate of the number of ENU-induced mutations per F1 fish is ~3–6000. As a result, a mutant phenotype observed in a clutch of eggs from a pair of F2 fish both heterozygous for a mutation identified by TILLING is likely to be due to another ENU-induced mutation(s). Since TILLING mutants are generated by the same mutagenesis regime used for forward genetic screens, the criteria for attributing a phenotype to a mutation identified by TILLING should ultimately be the same as for attributing a mutant phenotype to a mutation identified through a positional cloning approach: minimally, co-segregation of the phenotype and the mutant genotype in a large number of fish; preferably, phenotypic rescue with the wild-type gene; and ideally, a second non-complementing allele from a different F1 male. When TILLING identifies two deleterious alleles in a single gene, both alleles can be recovered so that the mutant phenotype can be studied unambiguously in compound heterozygotes. Where genomic deletions in the region of interest are available, the TILLING mutant can be placed over this deficiency. Indeed, an unanticipated use of TILLING has been to generate non-complementing alleles in genes identified by positional cloning for which only one allele was originally identified in a forward genetic screen.

TILLING is the most established method of reverse genetics available in the zebrafish; however, new methodologies, reviewed in this volume, are becoming available. Insertional mutagenesis projects using pseudotyped retroviruses have produced large libraries of F1 fish carrying as many as 25 random heterozygous insertions that can be sequenced and mapped to produce an indexed library of mutant loci recoverable from cryopreserved sperm [32] (see chapter by S. Burgess). Because retroviruses tend to integrate into the first introns of genes or into exons, many of these integrations are mutagenic as determined by a strong reduction in RNA levels. Zinc-finger nucleases, which can be designed to deliver frameshift mutations at defined target sites in the genome, have also recently been applied to zebrafish [33, 34] (see chapter by S. Amacher). To date, TILLING efforts have focused on identifying loss-of-function mutations, in the form of nonsense or splice alleles; however, as these other methods for generating loss-of-function mutations become more accessible and affordable, the unique ability of TILLING to generate allelic series, including a range of hypomorphic and even temperature-sensitive alleles caused by missense mutations, will become increasingly valuable. To this end, we and others continue to explore ways to improve the effectiveness and efficiency of TILLING methodologies.