Promoter replacement by genome editing creates gain‐of‐function traits in Arabidopsis

Modifying target genes through genome editing technology has promising agricultural applications (Wang and Doudna, 2023). Typical genome editing employs site-directed nucleases (SDNs), such as clustered regularly interspaced palindromic repeats (CRISPR)-associated protein 9 (Cas9), to generate DNA double-strand breaks (DSBs) at specific sites in the genome, leading to DNA mutations caused by errors during repair. The mutations generated by genome editing using SDNs can be classified into three categories (Zhang et al., 2021). SDN-1 mutations are typically short deletions or insertions, arising through errors during DNA repair via non-homologous end-joining. SDN-2 and -3 mutations result from homology-directed repair, whereby foreign DNA sequences are introduced into the genome DNA. Even though genome-edited plants carrying SDN-1 type mutations were produced through transgenesis, they are not heavily regulated and are socially acceptable in many countries as long as their transgenes have been segregated out. Therefore, SDN-1 genome editing is useful for crop improvement, and many SDN-1 genome-edited strains have been developed. However, most SDN-1 genome-edited strains developed so far harbour loss-of-function mutations. Two DSBs on the same chromosome may cause a deletion or an inversion. Such inversions have been reported to be achieved in plants using CRISPR-Cas9 (Schmidt et al., 2019). Because such modifications do not involve foreign DNA sequences, they are considered to be SDN-1 type mutations. Most induced inversions cause loss-of-function mutations. However, an inversion that creates a gene fusion involving the coding region of the target gene and the promoter of another gene may affect the expression of the target gene, leading to phenotypic changes. To test this possibility, we aimed to exchange the promoters of two genes on the same chromosome via targeted inversion in the model plant Arabidopsis (Arabidopsis thaliana), using CRISPR-Cas9 to introduce DSBs near the transcription start sites. We chose FLOWERING LOCUS T (FT) on chromosome 1 as the target gene. FT encodes florigen and leads to very early flowering when overexpressed (Kardailsky et al., 1999). As a promoter donor, we used HTA3, encoding a histone H2A variant. HTA3 is expressed in various young tissues, and its loss-of-function mutants exhibit normal growth (Yi et al., 2006). HTA3 is also located on chromosome 1, about 3.6 Mb upstream of FT, with the two genes in divergent orientations. Thus, an inversion of the 3.6-Mb fragment between the two genes that contains their promoters would result in a promoter exchange, placing FT under the control of the HTA3 promoter. We therefore designed two single-guide RNAs (sgRNAs) to introduce DSBs near the transcription start sites of FT and HTA3 (Figure 1a, b). In this strategy, the HTA3 promoter should be joined to the FT coding region (H-F junction), and the FT promoter to the HTA3 coding region (F-H junction). We employed multiplex genotyping PCR that detects both the H-F fusion structure and the F-F wild-type structure (Figure 1a, c, d). PCR genotyping of bulked DNA samples from 170 independent T1 transgenic lines identified the H-F fusion in one bulked DNA sample containing six individuals (Figure 1d). One of the six individuals showed a positive signal for the H-F fusion (Figure 1e). We collected T2 seeds from this T1 plant, genotyped 14 T2 plants for the H-F fusion and examined their flowering time when grown under short-day conditions (Figure S1a). All individuals with at least one copy of the H-F fusion flowered early, whereas individuals that lacked the H-F fusion flowered normally. Thus, the H-F fusion was inherited by the progeny and is associated with early flowering. Sequencing of the PCR fragment derived from the H-F fusion confirmed that the HAT3 promoter was fused with the FT coding region, together with a 9-bp deletion (Figure 1b). For the F-H fusion, sequencing of the corresponding PCR amplicon detected a precise fusion between the FT promoter and HTA3 coding region at the intended breakpoint positions. These findings indicated that an inversion of the 3.6-Mb genomic fragment between FT and HTA3 occurred in this individual; we confirmed this result by whole-genome sequencing (Table S1). Thus, the inversion efficiency in this experiment was 0.65% (1/170 transformants). We generated a CRISPR-Cas9 transgene-free population segregating for the inversion by crossing a plant heterozygous for the inversion with Col-0. We observed a perfect one-to-one correspondence between the presence of an inversion and early flowering, although the flowering phenotype behaved as a semi-dominant trait (Figure 1g). The Col-0 plants transformed with a genomic fragment encompassing the inversion exhibited an early flowering phenotype, confirming that the inversion led to earlier flowering (Figure S1b). FT is not induced in the wild-type under short-day conditions, but was expressed at high levels in plants homozygous for the inversion under these conditions (Figure 1h). The misexpression of FT may therefore be responsible for the early-flowering phenotype of the plants harbouring the inversion. RT-PCR indicated that the FT transcripts produced in the inversion line comprise their entire respective coding sequences (Figure S2). When grown under short-day conditions, all transgene-free F3 plants homozygous for the inversion flowered as early as transformants overexpressing FT from the cauliflower mosaic virus (CaMV) 35S promoter (T5 individuals homozygous for a single-locus transgene) (Figure 1i). While the inversion line exhibited a stably inherited early-flowering phenotype, the flowering time of CaMV35S:FT plants was more variable, with some flowering as late as Col-0, suggesting possible silencing of the transgene. The growth and fertility of the transgene-free line with the inversion were comparable to those of the wild type (Figure 1j, Figure S3). These observations suggest that the genome-editing inversion system presented here can introduce gain-of-function mutations that confer stable agriculturally important traits without affecting other traits. In this study, we successfully modified the expression of a target gene using SDN-1 genome editing in Arabidopsis, a feat that has previously been achieved primarily through the creation of genetically modified organisms. This system potentially has interesting applications. For instance, a herbicide-resistant rice strain was developed through the overexpression of a target of herbicide by CRISPR-Cas9-mediated inversion (Lu et al., 2021). Beyond overexpression, modifying tissue specificity or enabling the inducible expression of a gene of interest could be achieved by selecting an appropriate promoter. The development of additional technologies, including the induction of translocations (Beying et al., 2020), would expand the applicability of this system by enhancing the availability of promoters. Furthermore, the recently reported Bridge RNA system, which achieved programmed recombination in Escherichia coli, could be incorporated into this system for innovative crop improvement (Durrant et al., 2024). M.K. and T.N. conceived the study and designed the experiments. T.N. and Y.N. performed the experiments. M.N. and M.K. analysed the data. We thank Tsuyoshi Nakagawa for providing the pGWB601 vector. The data that support the findings of this study are openly available in Data1 at https://figshare.com/account/items/28828577/edit, reference number 10.6084/m9.figshare.28828577. The genome sequence data of the inversion line are openly available in DDBJ. The associated BioProjec and Bio-Sample numbers are PRJDB20596 and SAMD00901493, respectively. Figure S1 Early flowering in plants harbouring a genomic fragment containing an HTA3-FT fusion. Figure S2 FT expression in the HTA3-FT inversion plants. Figure S3 Fertility analysis of plants homozygous or heterozygous for the HTA3-FT inversion. Table S1 Putative structural polymorphisms detected by LUMPY in chromosome 1. 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.