Impact of predictive selection of LbCas12a CRISPR RNAs upon on‐ and off‐target editing rates in soybean

Clustered regularly interspaced short palindromic repeats (CRISPR) technology has revolutionized creating targeted genetic variation in crops. Although CRISPR enzymes have been reported to have high sequence-specificity, careful design of the editing reagents can also reduce unintended edits at highly homologous sites. This work details the first large-scale study of the heritability of on-target edits and the rate of edits at off-target sites in soybean (<i>Glycine max</i>), assaying ~700 T1 plants each resulting from transformation with LbCas12a constructs containing CRISPR RNAs (crRNAs) predicted to be either "unique" with no off-target sites or "promiscuous" with >10 potential off-targets in the soybean genome. Around 80% of the on-target edits observed in T0 plants were inherited in the T1 generation, and ~49% of the total observed on-target edits in T1 were not observed at T0, indicating continued activity of LbCas12a throughout the life cycle of the plant. In planta editing at off-target sites was observed for the Promiscuous but not the Unique crRNA. Examination of the edited off-target sites revealed that LbCas12a was highly tolerant to mismatches between the crRNA and target site in bases 21-23 relative to the start of the protospacer, but even a single mismatch in the first 20 nt drastically reduced the editing rate. In addition, edits at off-target sites have lower inheritance rates than on-target edits, suggesting that they occur later in the plant's lifecycle. Plants with a desired on-target edit and no off-target edits could be identified in the T1 generation for 100% of the T0 plants edited with the Unique crRNA compared with the 65% of T0 plants edited with the Promiscuous crRNA. This confirms that proper crRNA selection can reduce or eliminate off-target editing. Even when potential off-target sites are predicted, plants containing only the intended edits can still be identified and propagated.