CRISPR-Cpf1 mediates efficient homology-directed repair and temperature-controlled genome editing

Delivering Cpf1 as a ribonucleoprotein complex pre-assembled with its crRNA produces dramatically more genome editing in zebrafish embryos than injecting Cpf1 mRNA, and the limiting factor is guide stability rather than protein production. AsCpf1 and LbCpf1 mRNAs are efficiently translated with protein detectable by western blot, yet crRNAs injected alongside them are rapidly degraded, whereas pre-assembling Cpf1 with its crRNA significantly increases crRNA half-life in vivo — the protein itself shields the guide from degradation.

Structured precursor crRNAs increase neither mutagenic activity nor crRNA stability when Cpf1 is delivered as mRNA in zebrafish embryos, despite Cpf1's ability to process pre-crRNAs into mature guides.

LbCpf1 and AsCpf1 ribonucleoproteins both cleave DNA in vitro, but only LbCpf1 drives efficient mutagenesis in zebrafish and Xenopus tropicalis, where AsCpf1 shows very low activity. The difference carries into the germline: incrossing LbCpf1-injected zebrafish yields roughly 88% homozygous loss-of-function mutants for slc45a2 (albino) and roughly 99% for tyr, while AsCpf1-injected fish give a low rate of homozygous mutants.

Individual crRNAs targeting the same locus differ substantially in LbCpf1-mediated mutagenic activity in zebrafish, and co-injecting several crRNAs against one locus produces no synergy — the activity of a pool reflects its single most efficient crRNA.

LbCpf1 delivered as a ribonucleoprotein complex in zebrafish produces very high on-target mutagenesis with no detectable mutagenesis at predicted off-target sites bearing up to five mismatches, for two highly efficient crRNAs tested.

Temperature modulates Cpf1 activity, and does so far more strongly for AsCpf1 than for LbCpf1. AsCpf1 is less active than LbCpf1 at 25 °C and 28 °C while the two cleave comparably at 37 °C in vitro, which accounts for AsCpf1's poor performance in zebrafish, Xenopus tropicalis, Drosophila and plants, all of which develop below 28 °C. A brief post-injection incubation at 34 °C significantly increases AsCpf1-mediated editing of tyr and slc45a2 in zebrafish, and raises LbCpf1 activity to about 70% albino-like and about 100% tyrosinase-like mutant embryos. The effect is specific to Cpf1: raising temperature does not change the activity of SpCas9-sgRNA ribonucleoprotein complexes either in vitro or in vivo.

Temperature acts on Cpf1's competence to access or unwind genomic DNA rather than on its catalytic chemistry. Targeting catalytically dead SpCas9 to a site 29–32 nt from a Cpf1 target — the proxy-CRISPR approach, which opens the flanking DNA — enhances both LbCpf1 and AsCpf1 activity at 28 °C, reproducing at low temperature what raising the temperature achieves.

Because the Cpf1-crRNA complex remains competent well after delivery, the timing of a temperature shift sets when mutagenesis occurs, giving post-translational on-off control over editing without changing the injected reagents. Longer 34 °C incubations increase the number of mutant cells and lift AsCpf1 efficiency in zebrafish to the level LbCpf1 reaches at 28 °C, and shifting embryos to 34 °C during either 8–24 hpf or 24–48 hpf increases both phenotype severity and the extent of mutant cells relative to a constant 28 °C.

LbCpf1 improves homology-directed repair in zebrafish over SpCas9, raising HDR up to about 4-fold at two of four loci tested and matching SpCas9 at the other two. The mechanistic reason is where each nuclease cuts: Cpf1 breaks DNA roughly 18 nt from its PAM, so the indels it creates leave the seed sequence pairing intact and the site can be cleaved repeatedly, widening the window for donor-templated repair, whereas SpCas9-induced indels create seed mismatches between target and sgRNA that block recurrent cleavage.

The optimal single-stranded DNA donor geometry for Cpf1-mediated homology-directed repair differs from SpCas9's. Across 13 ssDNA donors tested at four zebrafish loci, Cpf1 gives its highest HDR rate when the donor is complementary to the target strand and carries the longer homology arm on the PAM-proximal side.