One-step generation of prime-edited transgene-free rice

Dear Editor,

Genome editing tools are leading a revolution in plant breeding. In particular, prime editors (PEs) can install all types of base changes and small insertions/deletions at precise positions in plant genomes (Anzalone et al., 2019). PEs are by far the most powerful approach for improving traits conferred by gain-of-function point mutations. Early versions of PEs suffered from low editing efficiency, but the latest PEs can perform edits at a much higher efficiency thanks to the extensive efforts of researchers from around the world. Most modifications to improve PE efficiency have focused on the optimization of PE protein components and structure, increasing the expression and stability of pegRNA and redirecting the endogenous repair mechanism to favor PE edits (Chen and Liu, 2023).

Currently, nearly all PE edits in plants are made by generating transgenic plants that stably overexpress PE components and then selecting transgene-free PE-edited offspring at the T1 generation. The one-step acquisition of transgene-free PE-edited T0 plants can accelerate the breeding process and is a requirement for the editing of vegetatively propagated crops. One-step transgene-free gene-edited plants have been obtained by two methods, delivery of ribonucleoprotein (RNP) and transient expression from mRNA or DNA. Delivery of RNP and mRNA to plant cells via gold particles or PEG-mediated transfection is less feasible than Agrobacterium-mediated transient expression of T-DNA (Liu et al., 2019). For this reason, Agrobacterium-mediated transient gene knockout and base editing have been developed to obtain transgene-free gene-edited T0 plants (Zhang et al., 2019b; Veillet et al., 2019; Huang et al., 2023). However, Agrobacterium-mediated transient prime editing has not yet been realized in plants.

Because the W548M mutation of the rice ALS gene can confer resistance to a wide spectrum of ALS-inhibiting herbicides (Chen et al., 2021), we proposed that OsALS–W548M could serve as a potential selection marker for transient prime editing of other genes of interest. We therefore constructed PE–ALS (performing the OsALS–W548M edit) and PE–ALS–EPSPS (performing both the OsALS–W548M and EPSPS–TAP/IVS edits) using PE3max as a backbone (Supplemental Figure 1A; Supplemental Table 1). These two PE vectors were transformed into the same quantity of rice calli (∼7.5 mL) via Agrobacterium (strain LBA4404/pVS1-VIR2) (Zhang et al., 2019a). After co-culture in the dark for 3 days, rice calli were transferred to NB medium supplemented with 500 mg L⁻¹ cefalexin for 7 days of recovery in the dark. Next, the calli were transferred to NB medium supplemented with both 500 mg L⁻¹ cefalexin and 5 mg L⁻¹ pyroxsulam for 14 days of selection in the dark. Pyroxsulam is an ALS-inhibiting herbicide, a member of the triazolopyrimidine family. Pyroxsulam-resistant calli emerged after four to five rounds of selection and were transferred to regeneration medium for shoot development (Figure 1A). DNA was extracted from T0 plants, and editing efficiencies for each gene were analyzed by next-generation sequencing and Sanger sequencing. We designed three pairs of primers, covering the left, middle, and right portions of the T-DNA region, to identify transgene-free plants that were negative for all three markers (Supplemental Figure 2A and 2B; Supplemental Table 2).

Figure 1.

One-step generation of prime-edited transgene-free T0 rice by Agrobacterium-mediated transient transformation.

(A) Tissue culture steps for the selection of transient co-editing events. Transfected rice calli were recovered for 7 days after 3 days of co-culture. After four rounds of pyroxsulam selection, resistant calli were transferred to regeneration medium for shooting, followed by a rooting step. Scale bar, 2 cm.

(B) Structure of the constructs used for co-PE-editing. 35S-C-U6p, 35S-CmYLCV-U6 promotor; HSPt, Arabidopsis HSP terminator; ZmUbi1p, maize ubiquitin1 gene promoter; Csy-P2A-PE, Csy4 gene fused with the PE linked by P2A, which will break during translation, resulting in two separate proteins; Hyg, hygromycin-detoxifying gene; Csy4RS, Csy4 recognition site; RB and LB, the right and left borders of the T-DNA.

(C) Co-PE-editing efficiencies for OsEPSPS (TAP–IVS; T173I, A174V, P177S), OsXa5 (V39E), and OsCold1 (K187T). Some pyroxsulam-resistant plants were identified as OsALS–W548L mutants, and they are not indicated in this figure. Ho, homozygous edits; He, heterozygous edits; Chi, chimeric edits; Re, DNA repair-derived byproducts with only some of the target bases edited when installing multiple-base substitution edits; InDel, insertion and deletion; Double, co-PE-edits in a homozygous W548M background. The presence of T-DNA was examined using three pairs of primers that amplified the left, middle, and right regions of the T-DNA, and triple-negative plants were considered transgene free. Next-generation sequencing (NGS) and Sanger sequencing results were used to determine specific co-PE-editing outcomes.

(D) Desired co-editing efficiency for the three target genes by three PE variants. All plants have either homozygous or heterozygous W548M.

(E) Illustration of Agrobacterium-mediated transient co-prime-editing processes. Agrobacterium delivers multiple copies of T-DNA into a single plant cell. In the nucleus, PE component genes on the T-DNA driven by strong promoters are transiently overexpressed, resulting in pyroxsulam resistance conferred by ALS–W548M. Upon pyroxsulam selection, both transgene-free and transgenic T0 plants emerge, as these T-DNAs do not necessarily integrate into the plant genome. The heterozygous mutations are not shown. Pyr, pyroxsulam.

Our results indicated that transgene-free T0 plants with W548M edits were readily identified upon transformation with either of these two PE vectors. However, no co-PE edits to OsEPSPS were observed (Supplemental Figure 1B and 1C; Supplemental Table 3). These results suggested that transient prime editing did occur, but the efficiency was too low to enable OsALS–W548M to serve as a selection marker for co-editing of other genes. Luckily, recent studies had reported further improvements in PE editing efficiency using different reverse transcriptases (RTs) in both animal and plant cells (Doman et al., 2023; Cao et al., 2024), and Csy4 had been reported to improve PE efficiency in wheat through better processing of pegRNAs (Ni et al., 2023). We therefore hypothesized that these improvements could enable the PE to efficiently perform transient editing, thereby producing transgene-free T0 plants. To this end, we incorporated Csy4 and two new RTs (Tf1 RT and M-MLV RT variant2) with the ePE3max backbone to produce C4PE3max (containing M-MLV RT variant1), C4PE6c (containing Tf1 RT), and C4PE6d (containing M-MLV RT variant2) (Doman et al., 2023) (Figure 1B; Supplemental Sequences 1–3). All three types of PE vector were constructed with the same RNA-expression cassette (Supplemental Sequence 4), spaced by the Csy4 recognition site, including one epegRNA–sgRNA pair to perform OsALS–W548M editing and another epegRNA–sgRNA pair to perform the desired editing of another gene of interest.

As expected, C4PE6c and C4PE6d displayed high efficiencies at the OsEPSPS–TAP, although C4PE3max still displayed low editing efficiency similar to that of ePE3max. C4PE3max, C4PE6c, and C4PE6d generated 4, 12, and 24 T0 plants after pyroxsulam selection, respectively. Next-generation sequencing showed that all these herbicide-resistant plants indeed harbored a mutated OsALS, either homozygous W548M or heterozygous W548M/L. It should be emphasized that C4PE6c and C4PE6d generated OsEPSPS TAP–IVS edits, producing 3/12 and 15/24 homozygously edited plants, respectively. More importantly, transgene-free co-PE-edited T0 plants were identified. C4PE6c and C4PE6d produced 1/12 and 11/24 plants, respectively, that were transgene free and harbored homozygous co-prime-edited alleles for both OsALS and OsEPSPS (Figure 1C; Supplemental Table 4). These results demonstrated that our PE optimization was successful for the transient co-prime-editing of two genes, largely validating our central hypothesis.

We next tested two additional targets, OsCold1 and OsXa5, reported to be effectively edited by PEs (Cao et al., 2024), for co-PE-editing with OsALS–W548M (Supplemental Table 1). The results indicated that, except for C4PE3max at the OsXa5 target, all the other PEs efficiently produced transgene-free prime-edited homozygous T0 plants (Figure 1C; Supplemental Figure 3). C4PE6c and C4PE6d showed higher efficiency than C4PE3max for co-editing of OsXa5, similar to the results obtained for OsEPSPS. Ultimately, 17/50 and 18/50 transgene-free homozygously edited plants were identified upon transformation with C4PE6c and C4PE6d, respectively (Figure 1C; Supplemental Table 4). For OsCold1, however, all three vectors produced highly efficient co-editing, resulting in a total of 25 transgene-free homozygously edited plants. C4PE3max performed editing at an efficiency comparable to that of C4PE6c and C4PE6d, likely because the specific mutation K187T was able to be efficiently edited by PE3max (Cao et al., 2024) (Figure 1C; Supplemental Table 4). Co-editing of genes of interest occurred not only in homozygous W548M plants but also in heterozygous W548M plants. Overall, co-editing of other genes in the W548M plant background was performed most effectively by C4PE6d (Figure 1D; Supplemental Tables 5–7). We selected 30 T0 plants harboring homozygous PE edits for both ALS and another gene of interest, i.e., 10 each for OsALS with OsEPSPS, OsCold1, and OsXa5, respectively, and examined the off-target potential at the three most likely positions for each target. Because these plants were homozygously edited, all were 100% on-target edited, and no off-target editing was observed (Supplemental Figure 4). We resequenced (25×) two T0 plants harboring homozygous PE edits for ALS and each of the other genes of interest, and all six examined plants were free of transgenes (Supplemental Figure 5). These results demonstrated again that transient co-PE-editing was significantly improved by Csy4 and RT variants, thus providing a new way to perform prime editing in plants.

We expected to obtain a similar number of pyroxsulam-resistant plants with the same type of duplex PE, such as CP4PE3max, because all harbored the same PE for ALS–W548M mutations and the same amount of callus was used for each transformation. However, we observed a large amount of variation, e.g., there were 4, 3, and 37 herbicide-resistant plants for the ALS–EPSPS, ALS–Xa5, and ALS–Cold1 duplex PEs based on C4PE3max, respectively (Figure 1C). We attribute this phenomenon to variations in rice transformation efficiency, as variations in transformation efficiency are very common in different rounds of experiments. For this reason, the number of herbicide-resistant plants cannot serve as a standard for prime editing efficiency. Because the number of edited plants harboring mutations of interest was dependent on the number of pyroxsulam-resistant plants, the co-editing efficiency at the targets of interest was not affected by rice transformation efficiency and can therefore be reliably calculated.

Agrobacterium can deliver multiple copies of T-DNA into a single plant cell. Once these copies reach the plant nucleus, the genes on the T-DNA driven by strong promoters will be transiently overexpressed, but these T-DNAs do not necessarily integrate into the plant genome. Therefore, upon pyroxsulam selection, we identified both transgene-free and transgenic T0 plants (Figure 1E). Transgene-free PE-edited plants were generated by transient expression of PE elements on T-DNAs that failed to integrate into the plant genome. Thus, the following factors contributed to the success of co-editing in our system. First, the prime-editing efficiency was greatly improved by the addition of Csy4 and new RT variants to the ePE3max backbone. Second, the Agrobacterium LBA4404/pVS1-VIR2 strain used in this study has an additional copy of the VIR gene cluster in the pVS1-VIR2 vector, enabling the delivery of more T-DNAs into infected plant cells (Zhang et al., 2019a). Third, homozygous OsALS–W548M was efficiently prime edited and served as an effective marker upon pyroxsulam selection of rice calli (Figure 1A). We observed that T0 plants with heterozygous W548L mutations occasionally emerged under pyroxsulam selection, with no simultaneous co-editing of genes of interest. However, the occurrence of W548L was eliminated by increasing the pyroxsulam selection pressure from 5 to 8 mg L⁻¹, enabling the better use of OsALS–W548M as a selection marker (Supplemental Figure 1D; Supplemental Table 3). It should be noted that ALS–W548M has already been commercialized for weed control in rice with no apparent fitness cost (Chen et al., 2021), making ALS–W548M an ideal selection marker for transient co-editing of genes of interest via PE.

The ALS–P197 mutation has also been used as a base-editing selection marker for the direct production of T0 plants. Zhang et al. reported that TaALS–P174 and TaACCase–A1992 were co-base-edited via CBE in wheat upon herbicide selection (Zhang et al., 2019b). Huang et al. selected transgene-free gene knockout by Cas12a as a result of co-base-editing of ALS-P197 by a CBE in tomato and citrus (Huang et al., 2023). Breeding requires versatile gene modifications that are often beyond the capability of co-base-editing; therefore, development of a PE-based co-editing strategy can more effectively benefit crop improvement. While our manuscript was under review, Zou et al. performed a similar co-PE-editing study using a Cas9-based PE to perform OsALS–S627I editing, which conferred resistance to bispyribac-sodium in rice, as well as co-PE-editing of other targets of interest. Indeed, one transgene-free co-PE-edited plant was identified (Zou et al., 2024). The generation of ample numbers of transgene-free co-PE-edited plants in our study also demonstrates that the acquisition of transgene-free PE edits in rice at the T0 generation is feasible.

In summary, we established a new strategy for PE-mediated transient co-editing in rice by comprehensively combining contributive factors including Agrobacterium, PE structure, tissue culturing, and pyroxsulam selection system. We presented the co-editing of three genes of interest in this study; however, this does not guarantee the successful co-editing of other genes. We strongly recommend screening out a highly effective epegRNA and nicking sgRNA pair for performance of co-editing together with ALS–W548M. For sexually reproducing plants, this system can accelerate the breeding process. We believe that this strategy could also be used with PEs for trait improvement in vegetatively propagated crops.

Funding

This study was supported by grants from the National Key Research and Development Program of China (2023YFD1202905) and the National Natural Science Foundation of China (32272629).

Acknowledgments

No conflict of interest is declared.

Author contributions

L.J. and Q.-J.C. conceived and designed the research. Y.L., T.N., D.Q., J.W., T.L., Z.C., W.S., and X.J. performed the experiments and analyzed the data. L.J., Q.-J.C., Y.L., and T.N. wrote the manuscript with input from all co-authors.

Published: December 20, 2024