Search for a command to run...
Base editing (BE) is an essential tool for plant precision breeding due to its high efficiency. Although BE has been applied in cucurbits—such as creating herbicide-resistant watermelon with CBE (Tian et al. 2018) and non-flowering mutants via ABE (Wang et al. 2024)—its broader application remains limited by low transformation efficiency and few practical targets for traits like fruit quality. Here, we developed a visual fluorescence-based BE system to modify a key residue in ClPSY1, aiming to enhance β-carotene in watermelon. To overcome low transformation efficiency, a visual screening platform was established. The system was constructed by incorporating tdTomato—a bright and photostable red fluorescent protein—into editing vector pBSE401 (Xing et al. 2014). This backbone was used to engineer two editors: pABE001, which utilises TadA8e-V106W for precise A·T-to-G·C conversion (Fan et al. 2024), and pCBE001, which employs hA3A-Y130F for specific C·G-to-T·A conversion (Ren et al. 2021) (Figure 1a). Agrobacterium-mediated transformation was performed using a two-step selection strategy: initial selection on phosphinothricin followed by tdTomato fluorescence screening. The system allowed real-time identification of positive T0 plants (Figure 1b and Figure S1) and transgene-free T1 progeny via seed fluorescence (Figure 1c). Prior to trait engineering, editing of the endogenous ClTFL gene yielded precise base substitutions in tdTomato-positive T0 plants (Figure S2), confirming effective ABE and CBE function in watermelon. To validate and apply this editing platform in practical breeding, we selected a key quality trait in watermelon flesh—β-carotene accumulation. Watermelon flesh colour is determined by distinct carotenoid profiles, predominantly lycopene, neoxanthin/violaxanthin, and β-carotene. β-carotene, enriched in orange-fleshed varieties, is a valuable vitamin A precursor (Zhang et al. 2020). ClPSY1, encoding the rate-limiting enzyme, is central to orange flesh formation (Liu et al. 2022; Song et al. 2023). However, whether the phenotype is controlled by promoter SNPs (Liu et al. 2022) or an exon variant (A445G, K149E) (Song et al. 2023) remains unclear, leaving the causal SNP(s) unresolved. To identify the causative variant for orange flesh, we crossed yellow-fleshed ‘JLM’ and orange-fleshed ‘HBJ’. All F1 plants were orange, indicating dominant inheritance (Figure S3). Bulked segregant analysis mapped a major orange-flesh locus to a 0.8-Mb region on chromosome 1 containing ClPSY1, where a single A/G polymorphism (A445G, K149E) distinguished the parents within a 2-kb flanking region. Re-analysis of ClPSY1 sequences (intronic and exonic regions) from 414 natural accessions (Guo et al. 2019) identified 46 SNPs (Figure S4), among which only A445G was fixed: all 19 orange-fleshed accessions were G/G, while all 24 yellow-fleshed ones were A/A (Figure S5, Table S1). Structural modelling placed K149 in a conserved hydrophobic flap domain (Cao et al. 2019), where it alters residue charge and may affect substrate binding (Figure 1d, Figure S6). Together, these genetic and structural data robustly support K149E as the key functional mutation driving high β-carotene in orange-fleshed watermelon. Guided by these findings and the optimal positioning of the target “A” within the ABE editing window (Figure 1e), we introduced the K149E allele into yellow-fleshed ‘JLM’ via ABE-mediated transformation. Among 15 independent T0 lines, 6 (40%) carried the intended edit (5 heterozygous, 1 homozygous; Figure 1f). All edited T0 plants (including heterozygous and homozygous) produced orange flesh, confirming the mutation's dominant effect. Perfect co-segregation of the orange phenotype with the K149E allele was observed in T1 progeny from heterozygous T0 plants (Table S2). Transgene-free, homozygous ClPSY1ᴷ149ᴱ lines were isolated from the progeny of a homozygous T0 plant based on absence of tdTomato fluorescence. These lines stably exhibited orange flesh with no apparent growth penalty (Figure 1g,h, Figure S7). qRT-PCR (primers in Table S3) indicated unchanged ClPSY1 expression (Figure 1i), suggesting a protein-level mechanism. Transcriptional profiling of downstream carotenoid genes revealed specific up-regulation of ClLCYB (Figure S8), which redirects lycopene toward β-carotene synthesis and has been linked to orange flesh (Zhang et al. 2020). HPLC confirmed markedly elevated β-carotene in ClPSY1ᴷ149ᴱ lines (Figure 1j), reaching levels ten-fold higher than in red-fleshed varieties and comparable to natural orange-fleshed accessions (Figure 1k). Thus, the K149E substitution enhances β-carotene accumulation by altering ClPSY1 function and increasing metabolic flux through the carotenoid pathway. In summary, we established a visual base editing system in watermelon using tdTomato for real-time tracking and early selection of transgene-free lines. By introducing the novel ClPSY1ᴷ149ᴱ allele via ABE, we significantly increased β-carotene content without affecting growth. This work provides an efficient platform for precision quality breeding in cucurbits and demonstrates the potential of base editing to accelerate trait improvement in horticultural crops. S.T., J.Z. and Y.X. designed the research; S.T. and X.Z. performed the experiments; other authors analysed and interpreted the data; S.T., J.Z. and Y.X. drafted the manuscript; all approved. This work was supported by grants from the Beijing Academy of Agricultural and Forestry Sciences (KJCX20251008, JKZX202401 and JKPY2026003); National Natural Science Foundation of China 32172592; the Ministry of Agriculture and Rural Affairs of China (CARS-25). This work was supported by Beijing Academy of Agricultural and Forestry Sciences, KJCX20251008, JKZX202401, JKPY2026003, National Natural Science Foundation of China, 32172592, Ministry of Agriculture and Rural Affairs of the People's Republic of China, CARS-25. The data that support the findings of this study is available in Supporting Information. Figures S1–S8: Tables S1–S3: 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.