The mutagenic chain reaction: A method for converting heterozygous to homozygous mutations

It is often desirable to generate recessive loss-of-function mutations in emergent model organisms; however, identifying such mutations in the heterozygous condition is challenging. Taking advantage of the CRISPR/Cas9 genome-editing method (1, 2), we have developed a strategy to convert a Drosophila heterozygous recessive mutation into a homozygous condition manifesting a mutant phenotype. We reasoned that autocatalytic insertional mutants could be generated with a construct having three components: (i) A Cas9 gene (expressed in both somatic and germline cells), (ii) a guide RNA (gRNA) targeted to a genomic sequence of interest, and (iii) homology arms flanking the Cas9-gRNA cassettes that match the two genomic sequences immediately adjacent to either side of the target cut site (Fig. 1A). In such a tripartite construct, Cas9 should cleave the genomic target at the site determined by the gRNA (Fig. 1A) and then insert the Cas9-gRNA cassette into that locus via homology-directed repair (HDR) (Fig. 1, B and C). Cas9 and the gRNA produced from the insertion allele should then cleave the opposing allele (Fig. 1D), followed by HDR-driven propagation of the Cas9-gRNA cassette to the companion chromosome (Fig. 1, E and F). We refer to this trans-acting mutagenesis scheme as a mutagenic chain reaction (MCR).

We expected that autocatalytic allelic conversion by MCR should be very efficient in both somatic and germline precursor cells, given the high frequency and specificity of mutagenesis (3) and efficacy of homology-based integration (4) mediated by separate genome-encoded Cas9 and gRNA genes observed in previous studies. We tested this prediction in D. melanogaster with the use of a characterized efficient target sequence (y1) (5) in the X-linked yellow (y) locus as the gRNA target and a vasa-Cas9 transgene as a source of Cas9 (Fig. 2C) because it is expressed in both germline and somatic cells (4). As the defining element of our MCR scheme, we also included two homology arms, ~1 kb each, flanking the central elements (Fig. 2C) that precisely abut the gRNA-directed cut site. Wild-type (y+) embryos were injected with the y-MCR element (see supplementary materials), and emerging F₀ flies were crossed to a y+ stock. According to Mendelian inheritance, all F₁ female progeny of such a cross should have a y+ phenotype (i.e., F₁ females inherit a y+ allele from their wild-type parent).

From two independent F₀ male (♂) × y+ female (♀) crosses and 7 F₀♀ × y+♂ crosses, we recovered y– F₁♀ progeny, which should not happen according to Mendelian inheritance of a recessive allele. Six such y^MCR F₁♀ were crossed individually to y+♂, resulting in 95 to 100% (average = 97%) of their F₂ progeny exhibiting a full-bodied y– phenotype (Fig. 2, E and G, and table S1), in contrast to the expected rate of 50% (i.e., only in males). We similarly tested MCR transmission via the germline in two y– F₁♂ recovered from an F₀♀ cross that also yielded y– female siblings. These y– F₁♂ were considered candidates for carrying the y-MCR construct and were crossed to y+ females. All but one of their F₂ female progeny had a full-bodied y– phenotype (Fig. 2, E and F). Occasionally among y^MCR F₂♀ we also recovered mosaics (~4%) with a few small y+ patches as well as a lone example of a 50% chimeric female (Fig. 2H), and in two instances, we recovered y+ male progeny from a y^MCR F₁♀ mother (Fig. 2E and table S1). These infrequent examples of imperfect y-MCR transmission indicate that although HDR is highly efficient at this locus in both somatic and germline lineages, the target occasionally evades conversion.

Polymerase chain reaction (PCR) analysis of the y locus in individual y– F₁ progeny confirmed the precise gRNA- and HDR-directed genomic insertion of the y-MCR construct in all flies giving rise to y– female F₂ progeny (Fig. 2D). Males carried only this single allele, as expected, whereas females in addition possessed a band corresponding to the size of the wild-type y locus (Fig. 2D, lane 4), which varied in intensity between individuals, indicating that females were mosaic for MCR conversion. The left and right y-MCR PCR junction fragments were sequenced from y– F₁ progeny from five independent F₀ parents. All had the precise expected HDR-driven insertion of the y-MCR element into the chromosomal y locus. In addition, sequence analysis of a rare nonconverted y+ allele recovered in a male offspring from a y^MCR F₁♀ (Fig. 2E) revealed a single-nucleotide change at the gRNA cut site (resulting in a T→I substitution), which most likely resulted from nonhomologous end-joining repair, as well as an in-frame insertion-deletion (indel) in a y+♀ sibling of this male (fig. S1 and table S1). The high recovery rate of full-bodied y– F₁ and F₂ female progeny from single parents containing a y^MCR allele detectable by PCR indicates that the conversion process is remarkably efficient in both somatic and germline lineages. Phenotypic evidence of mosaicism in a small percentage of MCR-carrying females and the presence of y locus–derived PCR products of wild-type size in all tested y– F₁ females suggest that females may all be mosaic to varying degrees. In summary, both genetic and molecular data reveal that the y-MCR element efficiently drives allelic conversion in somatic and germline lineages.

MCR technology should be applicable to different model systems and a broad array of situations, such as enabling mutant F₁ screens in pioneer organisms, accelerating genetic manipulations and genome engineering, providing a potent gene drive system for delivery of transgenes in disease vector or pest populations, and potentially serving as a disease-specific delivery system for gene therapy strategies. We provide an example in this study of an MCR element causing a viable insertional mutation within the coding region of a gene. It should also be possible, however, to efficiently generate viable deletions of coding or noncoding DNA by including two gRNAs in the MCR construct targeting separated sequences and appropriate flanking homology arms. Using the simple core elements tested in this study, MCR is applicable to generating homozygous viable mutations, creating regulatory mutations of essential genes, or targeting other nonessential sequences. The method may also be adaptable to targeting essential genes if an in-frame recoded gRNA-resistant copy of the gene providing sufficient activity to support survival is included.

In addition to these positive applications of MCR technology, we are also keenly aware of the substantial risks associated with this highly invasive method. Failure to take stringent precautions could lead to the unintentional release of MCR organisms into the environment. The supplementary material includes a stringent, institutionally approved barrier containment protocol that we developed and are currently adhering to for MCR experiments. Since this study was submitted for publication, a preprint has been posted on the bioRxiv web server showing that a split Cas9-gRNA gene drive system efficiently biases inheritance in yeast (6). The split system was used to avoid accidental escape of the gene drives. The use of a similar strategy in future MCR organisms would reduce, but not eliminate, risks associated with accidental release. We therefore concur with others (7, 8) that a dialogue on this topic should become an immediate high-priority issue. Perhaps, by analogy to the famous Asilomar meeting of 1975 that assessed the risks of recombinant DNA technology, a similar conference could be convened to consider biosafety measures and institutional policies appropriate for limiting the risk of engaging in MCR research while affording workable opportunities for positive applications of this concept.