Base editing in rice using nuclease‐deactivated CRISPR/Cas‐SF01

Adenine and cytosine base editing using dCas-SF01 and the 35S-CmYLCV-U6 composite promoter successfully introduced targeted base substitutions at multiple loci in rice, with average editing efficiency of 33.3%. Using the protospacer adjacent motif (PAM)-relaxed SF01-IKRR variant enabled base editing using 5'-NTN PAMs in rice. Cas12i3 was demonstrated genome editing activity in rice and broomcorn millet (Bai et al., 2024; Lv et al., 2024), then its engineered variant Cas12i3-5M (also termed as Cas-SF01, S7R/D233R/D267R/N369R/S433R) achieved editing efficiency of 75% and 90% in rice (Duan et al., 2024) and wheat (Wang et al., 2025), respectively. Since the efficiency of Cas12i3-derived base editors (BEs) remains low and no nickase variant is currently available, we introduced high-catalytic variants Cas12i3-3M (S7R/D233R/D267R) and Cas-SF01 into our previous iABE844 (TadA8e–dCas12i3-WT), which consists of a nuclease-deactivated Cas12i3 (E844A) and an adenine deaminase TadA8e (Lv et al., 2024). The resulting adenine base editors (ABEs) TadA8e–dCas12i3-3M and TadA8e–dSF01 (Figure 1A) were evaluated in rice protoplasts for editing three endogenous target sites: Phytoene Desaturase (OsPDS)-sg1, OsPDS-sg2, and NARROW LEAF1 (OsNAL1)-sg1. Amplicon deep sequencing revealed that TadA8e–dSF01 increased the total editing efficiency by 6.8-fold (OsPDS-sg1), 6.2-fold (OsPDS-sg2), and 2.4-fold (OsNAL1-sg1), with an average improvement of 5.1-fold compared to that of TadA8e–dCas12i3-WT (Figure 1B). When considering the most edited base site, the editing efficiency was enhanced by 28.7-fold (OsPDS-sg1), 11.1-fold (OsPDS-sg2), and 2.7-fold (OsNAL1-sg1), averaging 14.2-fold (Figure S1). However, TadA8e–dCas12i3-3M only showed limited improvement over TadA8e–dCas12i3-WT (Figures 1B, S1). The TadA8e–dSF01 was thus selected for further optimization. Optimized dSF01-BEs improve base editing efficiency and generate herbicide-resistant germplasm in rice (A, C, E) Schematic diagrams of the adenine base editors (ABEs) and cytidine base editors (CBEs). (B, D, F) Total editing frequencies in rice protoplast. Data are represented as means ± standard deviation (SD) of three biological replicates. P values are obtained using the two-tailed Student's t-test. ns, no significant difference; *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. (G) Summary of the editing efficiencies in transgenic rice. Ho, homozygous; Bi, bi-allelic; He, heterozygous. (H–J) Wild-type (WT) plants, homozygous (Ho) and heterozygous (He) mutants of OsACCase–I1879V in T0 generation were treated with 0.082 g/L Sethoxydim (Nabu) (H) or 0.108 g/L Haloxyfop-R-methyl (Gallant) (I). Phenotypes were recorded after 21 d. Scale bar, 5 cm. (J) Representative Sanger sequencing chromatograms of homozygous and heterozygous mutations of OsACCase–I1879V. The PAMs are marked with black boxes, and the coding sequences of I1879V are marked in a red box. The expression level of BEs is one of the major bottlenecks limiting editing efficiency. We then replaced the OsU6 promoter in the TadA8e–dSF01 with a high-expression 35S-CmYLCV-U6 composite promoter (U6com), resulting in U6com/TadA8e–dSF01 (Figure 1C). This new editor, together with OsU6/TadA8e–dSF01 and the widely used OsU6/TadA8e–nCas9 (D10A), was evaluated in rice protoplasts for editing five endogenous target sites: OsPDS-sg1, OsPDS-sg2, OsPDS-sg3, OsNAL1-sg1, and OsNAL1-sg2. Owing to the distinct PAM motifs, sgRNA architectures and editing window between Cas-SF01 and SpCas9, we designed guide sequences such that the editing windows of both editors overlapped as closely as possible to enable meaningful comparisons. Amplicon deep sequencing revealed that U6com/TadA8e–dSF01 achieved editing efficiencies ranging from 8.9% to 18.0% (average 13.2%) across the five target sites, representing a 1.6-fold improvement over OsU6/TadA8e–dSF01 (5.5%–12.7%, average 8.4%). However, its editing efficiency remained lower than OsU6/TadA8e–nCas9 (4.2%–49.0%, average 31.7%) (Figure 1D). We also found that U6com/TadA8e–dSF01 exhibited an editing window spanning positions 7–19, with highly preferred editing occurring at positions 8–16, counting the base adjacent to the PAM as position 1 (Figure S2). We subsequently developed dSF01-derived cytidine base editors (CBEs) by using the cytidine deaminase evoAPOBEC1, resulting in U6com/evoAPOBEC1–dSF01 and OsU6/evoAPOBEC1–dSF01 (Figure 1E). These constructs and the widely used OsU6/evoAPOBEC1–nCas9 (D10A) were used to target five endogenous sites in rice protoplasts: OsPDS-sg1, OsPDS-sg3, OsNAL1-sg1, Acetolactate Synthase (OsALS)-sg1, and Slender Rice 1 (OsSLR1)-sg1. Our results showed that U6com/evoAPOBEC1–dSF01 exhibited editing efficiencies ranging from 22.0% to 26.5% (average 23.9%) across the five target sites, representing a 1.3-fold improvement over OsU6/evoAPOBEC1–dSF01 (13.8%–22.6%, average 18.8%) (Figure 1F). OsU6/evoAPOBEC1–nCas9 maintained robust activity at most of the target sites (20.3%–63.2%, average 44.3%), with the exception of OsPDS-sg1 (0.5%), likely due to reduced Cas9 binding affinity for the non-canonical AAG PAM (Figures 1F, S3). Further analysis of the editing window revealed that U6com/evoAPOBEC1–dSF01 edits within positions 7–22, with a strong preference for positions 9–18 (Figure S3). We further evaluated our optimized U6com/TadA8e–dSF01 and U6com/evoAPOBEC1-dSF01 in transgenic rice. Ten target sites across five endogenous gene loci were selected for editing: Acetyl-CoA Carboxylase (OsACCase), OsPDS, OsNAL1, OsSLR1, and OsALS (Tables S1, S2). The U6com/TadA8e–dSF01 achieved high editing efficiencies of 85.3%–85.7% at two OsACCase target sites. At the OsNAL1 locus, 14 out of 49 transgenic lines were identified as edited mutants (28.6%). In contrast, editing frequencies were lower at two OsPDS target sites (9.7%–9.8%). Similarly, the editing efficiencies of U6com/evoAPOBEC1–dSF01 varied across targets: 15.5%–43.9% at two OsACCase targets, 4.1% at OsSLR1-sg2, 22.9% at OsSLR1-sg3, and 27.9% at OsALS-sg1 (Figures 1G, S4; Table S1). Collectively, these results demonstrate that the editing outcome of dSF01-BEs is affected by target site context. Cas-SF01 was recently artificially engineered to generate an IKRR variant (T208I/D293K/V460R/E494R), which enables efficient editing at the non-canonical 5′-NTN PAM (Chen et al., 2026). In transgenic rice, we found that U6com/TadA8e–dSF01(IKRR) and U6com/evoAPOBEC1–dSF01(IKRR) exhibit editing efficiencies up to 73.2% and 15.0% at target sites with relaxed 5′-NTN PAMs, respectively. However, no editing activity was detected at half of the tested target sites (Figures 1G, S4; Table S1). These results indicate that dSF01(IKRR)-BEs also display target site dependency. Interestingly, no detectable byproducts were observed across all editing events (Table S1), a notable distinction from the editing outcomes typically associated with nCas9-BEs. To evaluate potential off-target editing, we analyzed 16 mutants generated from editing at OsACC-sg1, OsACC-sg2, OsSLR1-sg3, and OsALS-sg1, respectively. No mutations were found from the top five potential sgRNA-dependent off-target loci (Table S3), consistent with the previous studies of Cas-SF01 (Duan et al., 2024; Zhang et al., 2026). Collectively, these results confirm that dSF01-BEs achieve high product purity and editing precision. Following editing at OsACC-sg2 by U6com/TadA8e–dSF01, ATA to GTA/GTG substitutions were introduced into the OsACCase codon sequences and resulted in I1879V, a well-characterized resistant variant conferring tolerance to ACCase-inhibiting herbicides Sethoxydim and Haloxyfop-R-methyl. In T0 generation, both homozygous and heterozygous OsACCase–I1879V mutants exhibited reliable resistance after treating by Sethoxydim and Haloxyfop-R-methyl (Figure 1H–J). Collectively, our optimized U6com/TadA8e–dSF01 and U6com/evoAPOBEC1–dSF01 achieved base editing efficiencies of 4.1% to 85.7% (average 33.3%), and their IKRR variants also presented editing activity using non-canonical 5′-NTN PAMs in transgenic rice. Recently, efficient dSF01-BEs have also been developed in rice by either fusing with VP64 and HMG‑D or dimerizing TadA8e (Liu et al., 2026; Zhang et al., 2026). All these results highlight the prospect of dSF01-BEs for editing in crops. However, while Cas-SF01 achieved an average gene-knockout efficiency of 75% in rice (Duan et al., 2024), its lower base editing efficiency may stem from the lack of a nickase variant, which facilitates non‑edited strand repair and increases the edited product. Recently, a noncanonical TSminiCBE introduces nicks in the non‑target strand to preferentially promote editing on the target strand, illustrating a novel strategy for improving Cas12‑derived BEs (Song et al., 2025). Moreover, further improvements include testing more active deaminases, optimized NLS, and flexible linkers, as reported in optimization of Cas12a‑ and Cas12i‑derived BEs (Cheng et al., 2023; He et al., 2025). This study was supported by the Agriculture Science and Technology Major Project, and Nanfan special project of CAAS (YBXM2514) to M.W.; National Key R&D Program of China (2021YFA1300404), Guangdong S&T Program (2024B1111130001), Shenzhen Science and Technology Program (KQTD20240729102038044), and National Natural Science Foundation of China (32188102) to J.-K.Z. The authors declare no conflicts of interest. M.W. and J.-K.Z. conceived of and designed the research. M.J., C.W., and H.W. conducted the transient transformations. M.J., Y.X., L.X., and C.Z. conducted the stable transformations in rice. M.W. and M.J. analyzed the data and prepared the manuscript. Additional Supporting Information may be found online in the supporting information tab for this article: http://onlinelibrary.wiley.com/doi/10.1111/jipb.70265/suppinfo Figure S1. dCas12i3-5M (dSF01) improves A-to-G base editing efficiency in rice protoplasts. Figure S2. The 35S-CmYLCV-U6 composite promoter increases editing efficiency of dSF01-ABE in rice protoplasts. Figure S3. The 35S-CmYLCV-U6 composite promoter increases editing efficiency of dSF01-CBE in rice protoplasts. Figure S4. Representative Sanger sequencing chromatograms of different mutations derived from base editing in rice transgenic lines. Table S1. Summary of the editing efficiencies of dSF01-BEs in rice transgenic lines. Table S2. Target sequences in rice stable transformations. Table S3. Analysis of potential off-target effects. Table S4. The primers used in this study Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.