Targeted insertion of large DNA fragments through template-jumping prime editing in rice

Targeted insertion of large DNA fragments holds great promise in crop breeding but is extremely challenging in plants. Prime editing (PE) can efficiently install small genomic insertion through replacement but faces challenges in mediating insertion of >100 bp. To enable larger insertion, a paired PE strategy termed GRAND editing was developed in human cells. In plants, genomic insertions of up to 135 bp were achieved through GRAND editing (Xu et al., 2024), but no plants with larger targeted insertion were generated through PE-mediated replacement.

Here, we attempted to insert three truncated promoters, including 207-bp pGluB4, 204-bp p10kDa, and 206-bp p16kDa, to the 5′ UTRs of OsC1, OsB2, and OsB1, respectively, through PE. An improved prime editor termed ePPEplus (Ni et al., 2023) and epegRNA (pegRNA with a tevepreQ1 motif at the 3′ terminus) were used in the PE assays of this study. Expressions of the ePPEplus and epegRNAs were driven by ZmUbi and eCmYLCV promoters (eCmYLCV, 35S enhancer-CmYLCV promoter), respectively. All PE assays in this study were conducted in a japonica cultivar termed Heixiangnuo (HXN). Firstly, we conducted the insertion editing using GRAND editing but detected no targeted insertions in the GRAND editing transgenic plants (Table S1; see Supplemental Methods and Figures S1 and S2 for the design).

Recently, a PE technology termed template-jumping PE (TJ-PE) was developed for large insertion in human cells (see Figure S3 for the mechanism) (Zheng et al., 2023). TJ-PE pegRNA (TJ-pegRNA) contains one reverse transcriptase template (RTT) and two primer binding sites (PBSs), with one PBS matching the pegRNA target and another matching the nicking gRNA target (Figure S3). TJ-PE could mediate 200–800-bp insertion in replacement of the fragment between the two TJ-PE nicks in human cells. Thus, we also conducted the insertion editing using TJ-PE with the same target sites as the above GRAND editing (Figure 1a and Figure S4). In the TJ-PE assays, the TJ-epegRNAs were expressed with pre-tRNA and hepatitis delta virus ribozyme (HDV) processing systems to generate mature epegRNAs (Figure S5). Each TJ-epegRNA expression cassette and the corresponding nicking gRNA cassette were constructed in a binary vector with an ePPEplus cassette to generate one vector for each insertion editing (Figure S5). With each of the TJ-PE vectors, 97–137 transgenic plants were generated through Agrobacterium-mediated transformation. Among these TJ-PE transgenic plants, only two edited plants with truncated pGluB4 insertions at OsC1 (one with 149-bp pGluB4 insertion and another with 139-bp pGluB4 insertion) were identified, and no edits were detected in other transgenic plants (Table S2; Figure S6a–c). The 149- and 139-bp insertions occurred at OsC1 TJ-epegRNA nicking site with a precise junction at the 3′ site of the nick but with small genomic deletions at another junction (Figure S6c; Dataset S1).

Figure 1

Open in figure viewerPowerPoint

Targeted insertion mediated by TJ-PE at OsC1, OsB2, and OsB1. (a) TJ-PE targets for the insertion editing of OsC1, OsB2, and OsB1. (b) Schematic of the vector for Csy4PS-assisted TJ-PEs. Csy4RS: Csy4 recognition site; white box: promoter; grey box: terminator. (c) Efficiencies for Csy4PS-assisted TJ-PE insertions with RNase H⁻ and RNase H⁺ prime editors. (d) Comparison of HXN and the plant with homozygous precise 207-bp pGluB4 insertion.

Recently, the Csy4 pre-pegRNA processing system (Csy4PS)-mediated multiplex PE was reported to be more efficient than the pre-tRNA processing system-mediated multiplex PE (Ni et al., 2023). Thus, the above TJ-epegRNAs were also expressed with Csy4PS to generate mature epegRNAs (Figure 1b). These Csy4PS-assisted TJ-epegRNA expression cassettes and the corresponding nicking gRNA cassettes (same to the above nicking gRNA cassettes) were constructed in a binary vector with a Csy4-P2A-ePPEplus cassette, generating one vector for each insertion editing (Figure 1b). Agrobacterium-mediated transformation with these vectors generated 144, 142, and 126 transgenic plants for the TJ-PE insertions of 207-bp pGluB4, 204-bp p10kDa, and 206-bp p16kDa, respectively. Excitingly, we detected targeted insertions of 49-207-bp promoters (pGluB4, p10kDa, and p16kDa) in 112, 86, and 13 plants from the 144 (112/144, 77.8% efficiency), 142 (86/142, 60.6% efficiency), and 126 (13/126, 10.3% efficiency) plants, respectively (Figure 1c). Among these edits, targeted insertions of 207-bp pGluB4, 204-bp p10kDa, and 206-bp p16kDa were detected in 28 (28/144, 19.4% efficiency), 11 (11/142, 7.7% efficiency), and 5 (5/126, 4.0% efficiency) plants, respectively, and all the other edits (most of the insertion edits detected) were truncated insertions of the intended fragments (Figure 1c and Figures S7–S9). Among the above plants with insertions of the whole intended fragments, precise replacement-mediated insertions (insertion in replacement of the fragment between the two TJ-PE nicks, and termed precise insertion thereafter) of 207-bp pGluB4, 204-bp p10kDa, and 206-bp p16kDa were detected in 16 (16/144, 11.1% efficiency), 8 (8/142, 5.6% efficiency), and 4 (4/126, 3.2% efficiency) plants, respectively (Figure 1c, Figures S7a–S9a and S10a–c). Except for the precise insertions, all the other targeted insertions in the above plants occurred at TJ-epegRNA nicking sites with precise junctions at the 3′ sites of the nicks but with InDel (insertion and deletion) mutations at another junction (Figures S7b–S9b; Datasets S2–S4). For control assays, nicking gRNA cassettes were removed from the above Csy4PS-assisted TJ-PE vectors, generating one control vector for each editing. 87–139 transgenic seedlings were generated with each of the control vectors, but no insertion edits were detected in the transgenic plants, confirming the essential role of nicking gRNA in TJ-PE.

Next, we tried to insert 403- and 1002-bp pGluB4 to OsC1 through Csy4PS-assisted TJ-PE using the same targets as above (see Figure 1b for vector structure). We detected targeted insertions in 32 and 12 plants from the 59 and 95 transgenic plants generated for the 403-bp (32/59, 54.2% efficiency) and 1002-bp (12/95, 12.6% efficiency) insertions, respectively (Figure 1c). Two seedlings with 403-bp pGluB4 insertions and one seedling with 710-bp pGluB4 insertion were identified from the above 59 and 95 plants, respectively (Figures S10d,e, S11a, and S12a). All the other edits were targeted insertions of 39–321-bp pGluB4 (39–321 bp for the 403-bp insertion editing and 54–202 bp for the 1002-bp insertion editing) (Figures S11b,c and S12b). All the above 39–710-bp insertions occurred at the TJ-epegRNA nicking site with a precise junction at the 3′ site of the nick but with InDel mutations at another junction (Figures S11a and S12a; Datasets S5 and S6).

Removing RNase H domain from prime editor could improve the efficiency of regular PE in plants (Zong et al., 2022). Considering that RNase H domain may also have effects on TJ-PE efficiency, we constructed RNase H⁺ TJ-PE vectors (Figure S13) by incorporating an RNase H fragment to the end of the ePPEplus fragment (in-frame incorporation) in the above Csy4PS-assisted TJ-PE vectors. In the PE assays with these RNase H⁺ vectors, we detected targeted insertions (including truncated insertions and insertions of the whole intended fragments) in 27, 21, and 17 seedlings from the 80, 86, and 95 transgenic plants generated for the 207-bp pGluB4, 204-bp p10kDa, and 403-bp pGluB4 insertions, respectively (Figure 1c). Excitingly, most of the detected edits in these plants were targeted insertions of the whole intended fragments, and insertions of 207-bp pGluB4, 204-bp p10kDa, and 403-bp pGluB4 were detected in 23 (23/80, 28.8% efficiency), 18 (18/86, 20.9% efficiency), and 15 (15/95, 15.8% efficiency) plants, respectively (Figure 1c and Figures S14–S16). These efficiencies for insertions of the whole intended fragments (207-bp pGluB4, 204-bp p10kDa, and 403-bp pGluB4) are obviously higher than those achieved with ePPEplus TJ-PEs (28.8%, 20.9%, 15.8% vs 19.4%, 7.7%, and 3.4%, respectively) (Figure 1c), suggesting that RNase H domain promotes TJ-PE-mediated insertions of several hundred base pairs. Among the above edits, precise insertion of 403-bp pGluB4 was detected in 2 plants (2/95, 2.1% efficiency) (Figure 1c and Figure S16a). However, compared to the TJ-PEs with ePPEplus, the frequencies for precise insertions of 207-bp pGluB4 and 204-bp p10kDa were not obviously increased in RNase H⁺ TJ-PEs (Figure 1c). Further improvements should be required to increase the precision of TJ-PE-mediated insertion in rice. Except for precise insertions, all the other insertions generated by RNase H⁺ TJ-PE occurred at TJ-epegRNA nicking sites with precise junction at 3′ sites of the nicks but with InDel mutations at another junction (Figures S14b–S16b; Datasets S7–S9).

In T₁ generation, homozygous precise insertions were identified from all the above precise insertion lines. Additionally, in T₁ generation, homozygous 710- and 403-bp pGluB4 insertions were also identified from all the 710- and 403-bp pGluB4 insertion lines generated with ePPEplus. In contrast to the purple leaves of HXN, all the plants with homozygous 207-bp (precise insertion), 403-bp (precise insertion and the insertions with InDel mutations at one junction), and 710-bp pGluB4 insertions at OsC1 showed green leaves, consistent with the key role of OsC1 in anthocyanin biosynthesis (Figure 1d and Figure S17). The expressions of OsC1, OsB2, and OsB1 in seedling shoots could be detected in HXN but not in the corresponding edited plants with homologous precise insertion of 207-bp pGluB4, 204-bp p10kDa, or 206-bp p16kDa (Figure S18), suggesting that the target genes were severely disrupted by the precise insertions.

Collectively, this study established an effective method to install targeted insertion of several hundred base pairs in rice, expanding the editing scope of PE in plants and providing a valuable genome editing tool for crop breeding.