Precise fine-turning of GhTFL1 by base editing tools defines ideal cotton plant architecture

Abstract

Background: CRISPR/Cas-derived base editor enables precise editing of target sites and has been widely used for basic research and crop genetic improvement. However, the editing efficiency of base editors at different targets varies greatly.

Results: Here, we develop a set of highly efficient base editors in cotton plants. GhABE8e, which is fused to conventional nCas9, exhibits 99.9% editing efficiency, compared to GhABE7.10 with 64.9%, and no off-target editing is detected. We further replace nCas9 with dCpf1, which recognizes TTTV PAM sequences, to broaden the range of the target site. To explore the functional divergence of TERMINAL FLOWER 1 (TFL1), we edit the non-coding and coding regions of GhTFL1 with 26 targets to generate a comprehensive allelic population including 300 independent lines in cotton. This allows hidden pleiotropic roles for GhTFL1 to be revealed and allows us to rapidly achieve directed domestication of cotton and create ideotype germplasm with moderate height, shortened fruiting branches, compact plant, and early-flowering. Further, by exploring the molecular mechanism of the GhTFL1^L86P and GhTFL1^K53G+S78G mutations, we find that the GhTFL1^L86P mutation weakens the binding strength of the GhTFL1 to other proteins but does not lead to a complete loss of GhTFL1 function.

Conclusions: This strategy provides an important technical platform and genetic information for the study and creation of ideal plant architecture.

Keywords: Base editor; Cotton; Directed evolution; Genome editing; Plant ideotype.

Fig. 1

Comparison of GhABE7.10 and GhABE8e for A-to-G base editing efficiency in cotton. a Sequence comparison of wtTadA, TadA7.10, and TadA8e (V106W). Black boxes and black asterisks mark amino acid sites with differences between wtTadA and TadA7.10; red asterisks mark amino acid sites with differences between TadA7.10 and TadA8e (V106W). b Schematic diagram of the vector element of GhABE7.10 and GhABE8e. GhU6-7, Upland cotton U6 promoter; gRNA, tRNA-20bp target-gRNA; pOsUbi, rice ubiquitin 1 promoter; wtTadA, wild-type E. coli TadA gene; TadA7.10, TadA8e, engineered TadA genes; NLS, nuclear localization signal; NOS, nopaline synthase terminator. c Comparison of adenine editing efficiencies of all A-to-G conversion between GhABE7.10 and GhABE8e within sgRNA1 and sgRNA2 target region by amplicon deep sequencing. Each dot represents the editing efficiency of an independent sample. The dashed line represents the average of the editing efficiency of all samples. d Schematic depiction of the target element of GhABE8e for sgRNA3 to sgRNA11. e Base-editing efficiencies in multiple sites of GhABE8e at sgRNA3-sgRNA7. The base editing efficiency is the percentage of reads with target A•T to G•C substitution in total reads. The edited A site is indicated in red. f Editing windows of GhABE8e at sgRNA8-sgRNA11 in cotton. The middle line of the box represents the median and the bottom and top lines of the box represent the upper and lower quadrilles of the data, respectively. Tail extends to the minimum and maximum of data. g Editing efficiency of GhABE8e + sgRNA8 (left) and GhABE8e + sgRNA11 (right) in T0 and corresponding T1 seedlings

Fig. 2

Genome-wide analysis of DNA and RNA off-target effect for the GhABE8e system by whole genome sequencing and whole transcriptome sequencing. a Sequence alignment of sgRNA13 on cotton genome by IGV browser views. The A-to-G mutations edited by GhABE8e were detected at the A5, A6, and A8 of target region. The target sgRNA sequences are highlighted in different colors. b Numbers of total SNVs identified in the GhABE8e, GhABE8e without sgRNA, WT, and negative plants. c Distribution characteristics of DNA and RNA off-target SNVs (A-to-G/T-to-C) on cotton chromosomes in GhABE8e without sgRNA and GhABE8e edited plants. d UpSet diagram of off-target SNVs (A-to-G/T-to-C) at DNA and RNA levels in GhABE8e without sgRNA and GhABE8e edited plants. e Venn diagrams of DNA-SNVs and Cas-OFFinder predicted off-target sites for edited lines. f Venn diagrams of RNA-SNVs and Cas-OFFinder predicted off-target sites for edited lines. g The number of different locations of SNVs for GhABE8e without sgRNA and GhABE8e-edited T0 plants at DNA and RNA levels

Fig. 3

Procedure for artificially directed evolution of plant functional proteins through base editing. a Schematic of the procedure for artificial evolution of the CDS and promoter of GhTFL1 via GhABE8e. Created with BioRender.com. b Statistics of editing efficiency for artificial evolution of the CDS and promoter of GhTFL1 via GhABE8e

Fig. 4

The artificial evolution of GhTFL1 mediated by base editing generates novel plant architecture cotton mutants. a Mutants of the GhABE8e-induced L86P amino acid substitution in the CDS of the GhTFL1 gene exhibit an early flowering, compact inflorescence phenotype. Scale bar, 5 cm. GhTFL1^L86P mutants develop rare twin flowers. b The T-to-C base editing led to amino acid substitution from leucine to proline at amino acids position 86 of GhTFL1 and the editing efficiency is analyzed by CRISPResso2 for quantification. Three plants tested carried two types of mutations and were chimeric. c The T0 seedlings carrying the K53G and S78G mutations had a double ball phenotype. Scale bar, 5 cm. d The A-to-G base editing led to amino acid substitution from Lysine to Glutamicacid/Glycine and Serine to Glycine at amino acids position 53 and 78 of GhTFL1 and the editing efficiency is analyzed by CRISPResso2 for quantification. This plant tested carried two types of mutations and was chimeric. e The plants with base editing of GhTFL1 promoter at -783 bp showed that leaves become more numerous and smaller, with darker leaves and more nutritional growth, and the editing efficiency is analyzed by CRISPResso2 for quantification. Scale bar, 5 cm. Two plants tested carried two types of mutations and were chimeric. f The plants with base editing of GhTFL1 promoter at -485-504 bp showed excessive nutritional growth and the editing efficiency is analyzed by CRISPResso2 for quantification. Scale bar, 5 cm. This plant tested carried two types of mutations and was chimeric

Fig. 5

Phenotypic characterization of WT and T1 progenies from GhTFL1^L86P and GhTFL1^K53G+S78G mutants. a Representative images of WT and T1 progenies of GhTFL1^L86P at different growth stages. GhTFL1^L86P plants show the cluster terminal flowers at the top and all axillary meristems subsequently produce floral buds directly on the main stem. b Representative images of WT and T1 progenies of GhTFL1^K53G+S78G. The plants show semi-dwarf height, terminating clusters of flowers forming at the tips of the main stems and fruiting branches as monotypic fruiting branches, terminating sympodial growth. c, d Plant heights (c) and days to bud stage (d) of WT and the T1 progenies of GhTFL1^L86P and GhTFL1^K53G+S78G. All data are presented as mean ± sd. *P < 0.05, **P < 0.01, ***P < 0.001 and ****P < 0.0001 by two-tailed Studentʼs t-test. e The genotype analysis of the T1 progenies of GhTFL1^L86P and GhTFL1^K53G+S78G. The ordinate represents the target site information, including the 20 bp sgRNA sequence and the adjacent 3 bp PAM sequence, for the T1 generation individual plants of GhTFL1^L86P and GhTFL1^K53G+S78G. Each dot represents the editing efficiency of an independent sample

Fig. 6

Molecular mechanism of plant architecture in the cotton GhTFL1^L86P and GhTFL1^K53G+S78G mutants. a Comparative amino acid sequences and structural domain analysis of GhTFL1 from different plants. Three amino acid sites, K53, S78, and L86, are marked with green pentagrams; two structural domains are marked with blue boxes. b Analysis of conserved regions, contact interfaces, and mutation-sensitive regions of GhTFL1 protein. c Transcriptome analysis of stems and leaves of Ghtfl1, GhTFL1^L86P, and Jin668. Volcano plots of differential genes under comparison of the three lines in stems. d A firefly LCI assay confirms that GhTFL1 and GhTFL1^L86P interact with GhAP1 and Gh14-3-3, respectively. A quantitative comparison of luciferase signals showed that GhTFL1 interacted more strongly with GhAP1 and Gh14-3-3, respectively, than GhTFL1^L86P interacted with GhAP1 and Gh14-3–3. e Bimolecular fluorescence complementation (BiFC) assay showing that GhTFL1 and GhTFL1^L86P interact with GhAP1 in the nucleus and GhTFL1 and GhTFL1^L86P interact with Gh14-3-3 in the nucleus and membrane. The N-terminus of yellow fluorescent protein (YFP) was fused to GhTFL1 and GhTFL1^L86P, while the C-terminus of YFP was fused to GhAP1 and Gh14-3-3. f Model for antagonistic roles of TFL1 and FT in promoting branch or floral fate. Schematic model illustrating the role of GhTFL1. During the nutritional growth phase, GhTFL1 binds to Gh14-3-3, GhFD to repress the downstream flowering gene GhAP1, thereby maintaining the nutritional growth of the plant. When the GhTFL1 gene is completely knocked out, nutritional growth is prematurely terminated, resulting in early flowering, reduced plant height, and other phenotypes. When a single base mutation occurs at some of the GhTFL1 loci, the function of GhTFL1 to repress downstream flowering genes is diminished, resulting in mutants with intermediate phenotypes. Created with BioRender.com