Variation in Mutation Spectra Among CRISPR/Cas9 Mutagenized Poplars.

In an effort to produce reliably contained transgenic trees, we used the CRISPR/Cas9 system to alter three genes expected to be required for normal flowering in poplar (genus ). We designed synthetic guide RNAs (sgRNAs) to target the poplar homolog of the floral meristem identity gene, (), and the two poplar orthologs of the floral organ identity gene (). We generated 557 transgenic events with sgRNA(s) and the Cas9 transgene and 49 events with Cas9 but no sgRNA, and analyzed all events by Sanger Sequencing of both alleles. Out of the 684 amplicons from events with sgRNAs, 474 had mutations in both alleles (77.5%). We sequenced both paralogs for 71 events in INRA clone 717-1B4 and 22 events in INRA clone 353-53, and found that 67 (94.4%) and 21 (95.5%) were double locus knockouts. Due partly to a single nucleotide polymorphism (SNP) present in the target region, one sgRNA targeting the paralogs was found to be completely inactive by itself (0%) but showed some activity in generating deletions when used in a construct with a second sgRNA (10.3-24.5%). Small insertion/deletion (indel) mutations were prevalent among mutated alleles of events with only one sgRNA (ranging from 94.3 to 99.1%), while large deletions were prevalent among alleles with two active sgRNAs (mean proportion of mutated alleles was 22.6% for small indels vs. 77.4% for large indels). For both and , each individual sgRNA-gene combination had a unique mutation spectrum ( < 0.001). An -sgRNA construct with two sgRNAs had similar mutation spectra among two poplar clones ( > 0.05), however, a -sgRNA construct with a single sgRNA gave significantly different mutation spectra among the same two clones ( < 0.001). The 49 empty vector control events had no mutations in either allele, and 310 potential "off-target" sequences also had no mutations in 58 transgenic events studied. CRISPR/Cas9 is a very powerful and precise system for generating loss-of-function mutations in poplars, and should be effective for generating reliably infertile trees that may promote regulatory, market, or public acceptance of genetic engineering technology.