Soybean (Glycine max (L.) Merr.) is among the most important crops in the world for oil and proteins (Zhang et al., 2022). Given that weeds are a major challenge to soybean cultivation, non-transgenic herbicide-tolerant soybean varieties are in great demand, particularly in places where growing transgenic soybean is prohibited. The clustered regularly interspaced short palindromic repeats (CRISPR)-Cas gene editing systems, including base editors built from Cas9 and Cas12 systems, are revolutionary tools in plant breeding, including the breeding for herbicide-tolerant crops (Sun et al., 2016; Wang et al., 2019; Zhang et al., 2023). Although Cas9 base editor systems have been successfully applied in soybean (Huang et al., 2023; Niu et al., 2023), the application of Cas12 base editors in soybean breeding has not been reported.

Acetolactate synthase (ALS), a crucial enzyme in the biosynthesis of branched-chain amino acids, is the target of several important herbicides (Powles and Yu, 2010). Variants of ALS that harbours certain point mutations can confer tolerance to a major group of commercial herbicides. A previous study showed that a variant bearing the P197S substitution in AtALS exhibited herbicide resistance in Arabidopsis (Chen et al., 2017). The soybean genome contains four GmALS paralogues (GmALS1-4). Notably, the widely planted non-transgenic herbicide-resistant soybean harbours a single-point mutation P178S in GmALS1, equivalent to P197S in AtALS (Walter et al., 2014). Unfortunately, amino acid substitution mutation in individual ALS genes cannot avoid herbicide injuries (Walter et al., 2014 and this study). Therefore, we explored the use of amino acid substitution mutations in multiple ALS genes to produce soybeans with sufficient herbicide resistance for effective weed control. Among the four GmALS genes, GmALS1 and GmALS3 share 80.6% and 79.0% amino acid sequence identities with AtALS respectively. We selected GmALS1 and GmALS3 as the targets for generation of herbicide-resistant soybean by converting C to T at codons P178 of GmALS1 and P172 of GmALS3 using the BE4max-dCas12-SF01 cytosine base editor system, which was constructed through introducing the E844A mutation in the conserved active site of Cas12-SF01 (Duan et al., 2024) and fusing it to the deaminase Anc689 APOBEC. The cytosine base editor contained a CRISPR RNA (crRNA) transcription box driven by the GmU6 promoter, and the cassette Anc689 APOBEC-dCas12-SF01-UGI (AncBE4max) driven by the SlEF1a promoter (Niu et al., 2023). The crRNA1 and crRNA2 targeted GmALS1 and GmALS3 respectively (Figure 1a).

We introduced the base editor into the elite soybean cultivar ‘Xudou 18’ through Agrobacterium tumefaciens-mediated transformation. The base-editing events at the crRNA target sites showed an overall 2.16% editing efficacy (with 9216 explants, we obtained 416 independent T₀ transformants and 9 of them carried gene-editing events (4 for als1, 3 for als3 and 2 for als1/als3, respectively)). Among the plants harbouring C–T substitutions, we successfully obtained ALS1^P178S, ALS3^P172S and ALS1^P178S ALS3^P172S mutants (referred to as als1, als3 and als1/als3, respectively), and potential off-target sites were examined and no editing events were detected. We obtained transgene-free stable homozygous als1, als3 and als1/als3 plants in the T₁ generation.

We evaluated the herbicide resistance of wild-type (WT) and als1/als3 plants grown in a greenhouse by spraying the plants with different concentrations of flucarbazone sodium at the V2 stage. Seven days after the herbicide application, the WT plants were severely damaged while the als1/als3 plants showed no damage (Figure 1c). We repeated the experiment in the following year and included the als1 and als3. The als1 and als3 mutations both conferred some level of herbicide resistance, whereas the als1/als3 showed no symptoms of herbicide damage (Figure 1d). These results showed that als1 and als3 mutations have a synergistic effect that strongly promotes herbicide resistance. We subsequently examined herbicide resistance of the als1/als3 and WT plants grown in the field. The als1/als3 plants showed no symptoms of herbicide damage while the WT plants were severely damaged by the herbicide (Figure 1e). The herbicide-treated als1/als3 plants showed no differences in plant height, internode number, pod number, branching and hundred-grain weight from those of the non-treated WT plants (Figure 1f–k). Furthermore, the productivity of the als1/als3 plants showed no obvious change compared with that of the WT in field tests (4 m²/plot). The yield of the als1/als3 was not reduced under flucarbazone sodium dosages of 280 and 420 mg L⁻¹ compared with 3125 kg/ha for the non-treated WT (Figure 1l).

Inappropriate dosages (over 2× (120 mg L⁻¹) flucarbazone sodium) or uneven application of herbicides by farmers in the field could easily exceed the resistance threshold of crops (Walter et al., 2014). Improvement of the herbicide tolerance to meet the requirements of practical breeding applications can overcome these problems. In this study, we successfully established a BE4max editor system that efficiently introduced C-to-T conversions in soybean and generated the als1/als3 mutant, which showed tolerance to 6× (420 mg L⁻¹) flucarbazone sodium herbicide (Figure 1d). The identification of the als1/als3 with 6× flucarbazone sodium herbicide resistance shows the effectiveness of our strategy for breeding herbicide-tolerant soybeans for real-world applications.