Modification of <scp><i>TAWAWA1</i></scp>‐mediated panicle architecture by genome editing of a downstream conserved noncoding sequence in rice

Genome editing has significantly advanced in recent years, with numerous attempts to integrate it into crop breeding (Gao, 2021). Many useful agronomic traits result from subtle changes in gene expression patterns conferred by natural variations (Olsen and Wendel, 2013). Therefore, the modification of regulatory sequences through genome editing presents a potential strategy to develop practical breeding resources. Promoters and cis-regulatory elements (CREs) of several target genes have been extensively edited to alter their expression patterns in many crop species including tomato (Rodriguez-Leal et al., 2017) and rice (Zhou et al., 2023). However, such approaches require numerous genome edits across a wide range of promoter regions or rely on molecular genetic evidence for responsible CREs. Identifying optimal genome-editing target sites within large genome regions to improve desirable agronomic traits remains challenging. Here, we describe creation of quantitative trait variations in panicle branching by precise genome editing of a conserved noncoding sequence (CNS) (Freeling and Subramaniam, 2009) located downstream of the rice yield-related gene TAWAWA1 (TAW1) (Yoshida et al., 2013) and demonstrate the potential of CNSs as targets for genome editing to fine-tune agronomic traits. TAW1 is a member of the ALOG (Arabidopsis LSH1 and Oryza G1) gene family encoding putative transcriptional regulators. In grass species, ALOG family proteins are essential for specification of floral organ identity and the normal development of spikelet and inflorescence architecture (Jiang et al., 2024; Yoshida et al., 2013). In a screen of a transposon-mutagenized rice population, Yoshida et al. (2013) isolated two allelic mutants, taw1-D1 and taw1-D2 exhibiting elevated TAW1 expression and increased panicle branching. Both mutants carried nDart1-0 transposons inserted approximately 0.9 kb downstream from the stop codon of TAW1 (Figure 1a) (Yoshida et al., 2013). Given the high conservation of genes governing inflorescence architecture across grass species, we hypothesized that conserved regulatory sequences would be located near the taw1-D1/-D2 insertion sites in these species. We first identified TAW1 homologues in monocot species (Table S1; Figure S1), and then compared their genomic sequences (Figure 1a). We identified a CNS (hereafter, TAW1-CNS) in grass species, including the BEP clade, within 50 bp downstream of the transposon insertion sites in taw1-D1/-D2 (Figure 1a, b). To elucidate the function of TAW1-CNS, we produced genome-edited lines with deletions of TAW1 downstream regions using CRISPR–Cas9 system (Figure S2). While deletion of the genomic region including taw1-D1/-D2 insertion sites or regions further upstream resulted in a normal phenotype, deletion of the entire TAW1-CNS region significantly increased the number of secondary branches per primary branch (i.e. panicle branching) (Figure 1c). This increased branching phenotype was reminiscent of that of the taw1-D2 mutant (Figure 1c) (Yoshida et al., 2013), strongly suggesting that the phenotypes of taw1-D1/-D2 alleles are not caused by potential enhancer-like activity of the transposon, but by transposon insertion-mediated disturbance of transcriptional repression activity of the TAW1-CNS region. Phenotypic difference between taw1-D1 and taw1-D2 (Yoshida et al., 2013) suggests existence of yet-unknown complex molecular mechanism controlling panicle branching. Semi-dominant inheritance of the panicle branching phenotype in the TAW1-CNS deletion mutant (Figure S3), consistent with that of taw1-D1/-D2 traits (Yoshida et al., 2013), further supported the hypothesis of the repressive function of TAW1-CNS. We next investigated panicle branching of genome-edited lines with mutations around TAW1-CNS (Figure S4) and compared transcription patterns in genome-edited lines with moderate (CR19) and severe (CR3) phenotypes. Whereas TAW1 transcription levels were positively correlated with panicle branching phenotype strength (Figure 1d–f), transcription levels of the TAW1-proximal genes were not significantly altered in the genome-edited lines (Figure S5). This indicates that TAW1-CNS acts as a gene-specific silencer of TAW1. We identified two highly conserved regions (H1 and H2) in TAW1-CNS, with conservation scores ≥0.6 in the BEP clade (Figure S6). Genome-edited lines with large deletions of all or substantial parts of the TAW1-CNS region including H1 and H2 (CR3, 24 and 25) exhibited severe panicle branching phenotypes (Figure 1d). Deletion of all or large parts of the H1 region showed no significant effect on panicle branching (CR9, 10, 15 and 16; Figure 1d). Conversely, deletion of all or part of the H2 region yielded a moderate increase in panicle branching (CR18 to 23; Figure 1d). These results suggest that alteration of the H2 region is essential for phenotypic effects, whereas alteration of the H1 region enhances the effects of mutations in the H2 region. Notably, we found two independent CArG-like sequences (binding elements of MADS-box transcription factors (TFs)) (Tilly et al., 1998) embedded in H1 and H2 (Figure S4), suggesting MADS TF-mediated repression of TAW1 transcription. Natural variation or genome editing of a putative CArG element downstream of SlWUS in tomato causes a dominant mutation with increased locule number phenotypes (Rodriguez-Leal et al., 2017). Similarly, the putative CArG elements in TAW1-CNS likely function as silencer elements. Although taw1-D1/-D2 alleles showed substantial increases in spikelet number per panicle, they also decreased fertility (Yoshida et al., 2013), suggesting the need for a novel allele with moderately increased spikelet number per panicle as the ‘optimal’ allele. The genome-edited line with deletion of a part of the H2 region (CR19) displayed normal fertility with moderate increase in panicle branching while the line with deletion in entire TAW1-CNS region (CR3) showed decreased fertility with severe panicle branching (Figure 1e, g), implying that the CR19 is a candidate for the optimal allele. Our results suggest that different genome-editing patterns in, across or near the CNSs could produce different phenotypic outcomes. Both of multiple CREs and their inter-elemental sequences in the CNSs may play important roles for fine-tuning transcription of proximal genes. Further investigation into whether the TAW1-CNS region is involved in transcriptional repression via formation of intragenic chromatin loops, as previously reported (Gagliardi and Manavella, 2020), would be of interest. Indeed, using the CART database (Zhu et al., 2024), we found that the TAW1-CNS region is in an open-chromatin state, suggesting a TF binding-mediated mechanism. Furthermore, we detected potential chromatin looping around the TAW1-CNS region, further supporting our hypothesis (Figure S7). Identification and utilization of silencer modules would enable upregulation of proximal genes and broaden the range of genome-editing applications. Consequently, CNSs around agronomically important genes could serve as useful target sites for genome editing to produce allelic series with variously altered gene expression and phenotypes to obtain ideal breeding materials. We sincerely thank J. Kyozuka for valuable suggestions. We also thank M. Kuroda and M. Endo for providing vectors. We also thank staff and members of our lab for their generous support. This research was supported by a Research Program on Development of Innovative Technology grant (JPJ007097) from the Project of the Bio-oriented Technology Research Advancement Institution (BRAIN), JSPS KAKENHI Grant Number 22H00373, the Cross-ministerial Strategic Innovation Promotion Program (SIP) ‘Technologies for Smart Bio-industry and Agriculture’ (funding agency: BRAIN) and the Research Supporting Program of Research Center for Advanced Analysis, NARO. T.K., W.M.I, T.M. and H.Y. are co-inventors on patent PCT/JP2024/016294 filed by NARO and Tohoku University related to work described in this manuscript. The remaining authors declare no competing interests. T.K. and H.Y. designed the experiments and wrote the manuscript. T.K., F.L., S.C. and H.Y. performed the transgenic analyses. W.M.I. and T.M. performed analyses to detect CNSs. S.C. and Y.K performed RNA-seq analyses. A.Y. conceptualized the research. F.L. and S.C. edited the manuscript. All authors have reviewed and agreed to publication of the manuscript. The data that support the findings of this study are openly available in Expression profile in rice panicles at https://www.ncbi.nlm.nih.gov/sra/PRJNA1234087. Table S1 TAW1 and TAW1 homologues in plants. Table S2 gRNA sequences for production of genome-edited rice. Table S3 Primers and oligonucleotides used in this study. Data S1 Supplemental Materials and Methods. Figure S1 Phylogenetic tree based on predicted amino acid sequences of ALOG family genes. Figure S2 Positions of mutations, gRNA targets, and the CNS in the TAW1 downstream region. Figure S3 The panicle branching phenotype of TAW1-CNS deletion is inherited in a semidominant manner. Figure S4 Positions of mutations and gRNA targets around the TAW1-CNS region. Figure S5 Transcription of TAW1 and proximal genes in young panicles. Figure S6 Genomic sequence alignment of TAW1-CNS region in 9 grass species. Figure S7 Open chromatin regions around the TAW1 gene. 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.