Introduction

Cas12a (Cpf1), an endonuclease from Class 2 Type V clustered regularly interspaced short palindromic repeat (CRISPR)/CRISPR-associated nuclease (Cas) systems, has several unique attributes as a genome-editing tool (Makarova et al. 2020; Zetsche et al. 2015). First, unlike Class 2 Type II CRISPR/Cas systems, Cas12a nucleases are complexed with a single CRISPR RNA (crRNA) that does not require a trans-activating CRISPR RNA (tracrRNA) (Zetsche et al. 2015). The crRNA of Cas12a (∼42 nt) is shorter than the single guide RNA (sgRNA) of Cas9 (∼100 nt), making it more amenable to chemical synthesis and easier to deliver with other CRISPR reagents into cells as ribonucleoprotein (RNP) complexes (Banakar et al. 2020; Su et al. 2023). Second, identified Cas12a nucleases, such as LbCas12a from Lachnospiraceae bacterium and AsCas12a from Acidaminococcus sp., typically recognize a T-rich protospacer adjacent motif (PAM) sequence at the 5' end of the protospacer (Zetsche et al. 2015). This attribute facilitates targeting AT-rich genomic regions that would be challenging to target with the most commonly used SpCas9 from Streptococcus pyogenes, which prefers G-rich PAMs (Zetsche et al. 2015). Third, Cas12a induces double-strand breaks (DSBs) with 5' overhangs that are distal from its PAM sequence (Zetsche et al. 2015). Since Cas12a-induced mutations are distal from the PAM, and mismatches distal from the PAM between the guide sequence of the crRNA and its target site in genomic DNA are more frequently tolerated by Cas9 or Cas12, Cas12a may potentially repeatedly induce DSBs even after the initial target site has been altered via imprecise DNA repair. Consequently, mutations induced by Cas12a are biased toward larger deletions compared to the mutations typically produced by Cas9 (Zetsche et al. 2015). In addition, repeated DNA cleavage facilitates releasing homology-directed repair (HDR) donors from genomic DNA and prolonging the time-window of HDR prior to repair via the non-homologous end joining (NHEJ) pathway. Finally, Cas12a proteins can process precursor crRNA (pre-crRNA) arrays, which can be harnessed for multiplex genome editing from a single transcript (Zetsche et al. 2017). Thus, Cas12a-induced genome editing could be a valuable alternative to Cas9.

Like Cas9, Cas12a has been employed for genome editing in plants (Bernabe-Orts et al. 2019; Lee et al. 2019; Li et al. 2018, 2021), and the ttLbCas12a Ultra sequence was codon-optimized with monocot plants, we generated another version of ttLbCas12a Ultra that harbored the same NLS sequences as in PEmax (Chen et al. 2021) and was codon-optimized for dicot species, which we named ttLbCas12a Ultra V2 (Fig. 3C; Table S6). The optimized NLS sequences were composed of a bipartite SV40 NLS at both terminals and an additional c-Myc NLS at C-terminal. Indeed, we obtained much more edited plants when expressing ttLbCas12a Ultra V2 than with ttLbCas12a Ultra, indicating that editing efficiency can be further improved in Arabidopsis by optimizing NLS sequences and the codon usage of the nuclease gene (Fig. 3D; Table S6).

Mutations in T1 plants are heritable to the subsequent generation

The mCherry cassette in the Cas12a vectors provided an effective strategy for reliably isolating Cas12-free Arabidopsis plants (Gao et al. 2016). Indeed, seeds harboring the transgene with the mCherry cassette display strong red fluorescence, while the absence of red fluorescence indicates segregation of the T-DNA, resulting in T-DNA-free seeds (Gao et al. 2016). Before analyzing the heritability of each mutation, we first isolated T-DNA-free seeds, based on the lack of red fluorescence.

To test whether mutations at the ECA3-1 and ECA3-2 target sites are heritable, we isolated T-DNA-free T2 seeds from six and nine T1 biallelic mutant plants, respectively. Since Cas12a-induced mutations at the ECA3-1 and ECA3-2 target sites disrupted the BglII and EcoRV restriction enzyme sites, respectively, we employed restriction fragment length polymorphism (RFLP) analysis to identify mutations. RFLP analysis indicated that all T-DNA-free T2 plants harbor homozygous or biallelic mutations, as evidenced by the lack of DNA cleavage (Fig. S1, S2; Table S7). These results demonstrate that mutations in ECA3 present in the T1 plants are heritable.

To provide evidence that mutations at the GL2 target are heritable, we isolated 30 T-DNA-free T2 seeds from the only identified T1 biallelic mutant plant, which surprisingly did not exhibit a glabrous phenotype (Table S7). High-throughput sequencing analysis of the resulting 30 T2 plants showed that they carry homozygous or biallelic mutations identical to the two mutations detected in the T1 plant (Table S8). We observed that six plants homozygous for an 8-bp deletion at the target site exhibit the glabrous phenotype (Fig. 4A), whereas the remaining six plants homozygous for a 21-bp insertion and 18 plants with biallelic mutations did not show a glabrous phenotype (Table S8). These results indicate that the 21-bp insertion, which leads to an insertion of seven amino acids, does not affect GL2 function, explaining the lack of a glabrous phenotype of the original T1 plant. These results do, however, demonstrate that the introduced mutations in GL2 are heritable.

Fig. 4
figure 4

Phenotypes of T-DNA-free homozygous gl2 or tt4 mutants derived from T1 plants harboring ttLbCas12a Ultra and the U6-tRNA cassette. A Glabrous phenotype of a T-DNA-free T2 homozygous gl2 mutant (right) compared to the wild type (WT; left). B Yellow seed coat phenotype of a T-DNA-free T3 homozygous tt4 mutant (right) compared to the WT (left)

Full size image

We performed a similar analysis of mutations at the TT4 target site; to this end, we isolated T-DNA-free T2 seeds from seven T1 homozygous or biallelic mutant plants. All T2 seeds were yellow rather than dark brown (Fig. 4B and Table S8), a phenotype typical of tt4 mutants (Malzahn et al. 2019). We conclude that all T2 progeny are homozygous or biallelic tt4 mutants, demonstrating that mutations present in T1 plants at the TT4 target site are heritable.

We also isolated T-DNA-free T2 seeds from ten (GL1-1) and five (GL1-2) T1 phenotypically homozygous or biallelic mutant plants in which GL1 was targeted for editing (Table S10). We determined that all T2 plants are completely glabrous (Table S10). Sanger sequencing of the target site in individual T2 plants confirmed that they harbor homozygous or biallelic mutations in GL1 (Tables S11, S12). These results demonstrate that mutations present in T1 plants at the GL1 locus are heritable.

Off-target mutations were not detected in mutant plants

High editing efficiency usually means high off-target mutagenesis. We selected the most efficient two targets, GL1-1 and GL1-2, to analyze off-target mutations. We searched potential off-target sites of the GL1-1 target and obtained two harboring 3 or 4 mismatches in At5G40330 or At1G22640, respectively (Table S13). We also obtained an off-target site with 5 mismatches of the GL1-2 target (Table S13). We amplified PCR fragments spanning off-target sites from 36 gl1-1 and 36 gl1-2 T1 mutant plants that harbor homozygous or biallelic mutations and analyzed mutations by Sanger sequencing. We detected no off-target mutations (Table S13), indicating that careful selection of targets will be able to avoid off-target mutagenesis induced by highly efficient LbCas12a variants.

Discussion

Temperature sensitivity and insufficient cleavage activity of Cas12a nucleases limit editing efficiency of targets in plants and many efforts have been made to improve these restrictions (Guo et al. 2022; Huang et al. 2022; Lee et al. 2019; Liu et al. 2019a; Ma et al. 2022; Zhang et al. 2023, 2021). Two LbCas12a Ultra variants have previously been reported: one harbors a single E795L mutation (Zhang et al. 2021) and another harbors the two mutations N527R and E795L (Huang et al. 2022). In this report, we combined the D156R mutation from the low-temperature-tolerant variant and the E795L mutation to greatly enhance editing efficiency of LbCas12a (Fig. 1). Our results are consistent with a previous report that introducing mutations from a low-temperature-tolerant variant into a LbCas12a Ultra variant greatly enhanced editing efficiency of LbCas12a (Huang et al. 2022). The main difference with the previous report is that we used the variant harboring the single E795L mutation, whereas Huang et al. used the variant harboring the two mutations N527R and E795L. However, although E795L enhanced the editing efficiency, the mutation N527R showed the detrimental effect on LbCas12a activity, and the variant harboring the two mutations failed to show improved editing efficiency in human cells (Zhang et al. 2023). In the future, it will be interesting to compare the ttLbCas12a Ultra variant generated in this report to other highly active variants, such as Cas12a-Plus (Huang et al. 2022), hyper-Cas12a (Guo et al. 2022), iCas12a (Ma et al. 2022), and LbCas12a-RRV (Zhang et al. 2023) for use in plants. These comparisons will help generate more Cas12a variants with higher editing efficiency.

Since processing of tRNA-cRNAs or HH-crRNAs is not complete (Gao and Zhao 2014; ** of targeted mutations. Mol Plant 8:1431–1433. https://doi.org/10.1016/j.molp.2015.05.009 " href="/article/10.1007/s42994-024-00144-w#ref-CR17" id="ref-link-section-d68206792e1757">2015).