Genome Engineering with Cre-loxP

Twenty years ago Gu, Zou and Rajewsky published the first paper using the Cre-loxP system to study the immune response, “Independent control of immunoglobulin switch recombination at individual switch regions evidenced through Cre-loxP-mediated gene targeting” (1). That experimental strategy has enabled construction of clean mutations in essentially any gene of interest, in almost any cell type, and it has been fundamental to our current detailed understanding of immunity and development. It is especially interesting to look back at this achievement in light of the rapid recent progress in use of customized nucleases for targeted gene correction and gene disruption.

Gene disruptions — in contrast to deletions — had become common following publication of the classic homologous gene targeting paper by Thomas and Capecchi in 1987 (2). However, targeting constructs typically bore a selectable neo expression cassette that remained embedded in the engineered locus. This cassette could in principle (and occasionally did) corrupt expression of the targeted gene or function of its product. The paper by Gu et al. (1) posed the question of whether cis-elements regulate class switch recombination (CSR). Rigorous analysis clearly required a deletion mutation, because persistence of the promoter and enhancer driving neo transcription had clear potential to confound identification of cis-regulatory elements at IgH.

Enter Cre-loxP. Rajewsky’s laboratory (1) targeted the region of IgH spanning J_H-Eμ for deletion, using a targeting construct very much like those used for gene ablations, but distinguished by the presence of loxP sites flanking the region targeted for deletion and selectable markers. Expression of Cre was predicted to cause deletion of the region bounded by the loxP sites, and this did indeed occur, at a frequency of 2–4%. Three lines produced by deletional recombination were successfully used to generate mutant founder mice that transmitted the J_H-Eμ deletion through the germline; while a fourth gave rise to a single male chimeric mouse, which was sterile. Finally, the J_H-Eμ deletion had a clear effect on both CSR and B cell development. In heterozygous mice bearing the J_H-Eμ deletion on a single chromosome, VDJ recombination and efficient CSR at μ occurred on the wild-type chromosome but not on the chromosome bearing the deletion. Homozygous mutants lacked all B cells, reflecting failed VDJ recombination in the absence of Eμ.

Genome engineering can be accompanied by collateral damage, and the paper offered reassurance on this point. Cre expression appeared not to cause rampant genomewide instability, based on the observation that three of four ES lines tested were capable of germline transmission. In addition, the excised DNA fragment appeared not to reintegrate elsewhere in the genome, as assayed by southern blotting. Neither concern was without foundation. In subsequent experiments designed specifically to assay toxicity, high level Cre expression in spermatids was shown to abolish fertility (3). And later experiments found that reintegration of the fragment deleted by RAG1/2-mediated V(D)J recombination occurs a frequency of 1/50,000 events (4).

Cre is one member of a family of site-specific recombinases that cleave 34 bp target sequences (5). Cre derives from the coliphage P1, and its temperature optimum of 37°C is perfectly suited for genome engineering in murine cells. A related system, Flp-FRT, derived from S. cerevisiae, had also shown initial promise for genome engineering. Flp-FRT was used by Jung, Rajewsky and Radbruch to generate a targeted deletion 5′ of Sγ1 (6), reported a few months before the Gu et al. paper came out. However, Flp and has a 30°C optimum, making it less effective than Cre in mouse cells, and Cre came rapidly to dominate murine genetics (7). Flp-FRT has been improved over the years, and it now matches Cre-loxP in efficiency (5). These two site-specific recombination systems are now readily combined for sophisticated genome engineering applications.

Much of the power of the Cre-loxP system derives from the potential to generate conditional mutants. Conditional deletions can overcome the barrier that embryonic lethality otherwise poses to studying essential genes, as first shown by Rajewsky’s laboratory in experiments that analyzed T cell-restricted deletion of the repair polymerase, pol β (8). Conditional deletions have since been key to a vast number of studies that have elucidated the functions of specific genes and regulatory elements in the immune response and development. Inspired by the many successes, the International Knockout Mouse Consortium developed a repository of murine ES cell lines in which essentially every gene is flanked by loxP sites (or “floxed”) to enable studies of individual gene functions in specific tissues (9). However, conditional deletion of every gene in every cell type may not become a reality. Conditional deletions are only as good as the promoters that regulate Cre expression, and transgenic Cre driver lines have proven to be somewhat problematical, exhibiting expression outside the target tissue as well as inefficient cleavage leading to mosaicism (10).

Will Cre-loxP and related systems will continue to dominate? Or has genome engineering entered a new era, now that TALEN or CRISPR/Cas-based nucleases can be readily customized to cleave at essentially any genomic site (11, 12)? These new approaches are enabling genome engineering in organisms that were recalcitrant to Cre-loxP, like zebrafish (13), where it now takes only two weeks to generate a targeting TALEN and use it to create a modified embryo (14). TALEN and CRISPR/Cas-based approaches must surmount some of the same hurdles initially confronted by Cre-loxP or Flp-FRT, such as optimizing efficiency and minimizing collateral damage. Current technologies like high-throughput screens and deep sequencing will undoubtedly accelerate optimization of TALENs and CRISPR/Cas-based systems. Nonetheless, in the end these new approaches will not be judged not by how rapidly the genome was engineered but by the quality of the mutants generated. The Cre-loxP system has set a very high bar.