De novo creation of solid-placenta, few-seeded melon via targeted mutation of CmSLM/CmAGL11

Seeds, the primary reproductive organs of most angiosperms, are essential to the plant life cycle because they store genetic material and supply nutrients for early seedling growth (He et al. 2026). However, seedlessness markedly improves fruit quality for fresh consumption and processing by increasing the edible proportion and eliminating hard seeds, making it one of the most valuable horticultural traits (Wang et al. 2021). Seedless fruits may occur naturally or be induced through hormone application, ploidy manipulation, or genetic engineering (Pang et al. 2025). Previous studies have demonstrated that parthenocarpy and stenospermocarpy are two distinct mechanisms underlying seedless fruit formation (Royo et al. 2018). In parthenocarpy, fruits develop from the ovary without pollination or fertilization, as observed in cucumber. In watermelon, seedlessness can be achieved either through conventional breeding of triploid plants or through the application of hormone analogs such as N-(2-chloro-4-pyridyl)-N’-phenylurea (CPPU). In contrast, stenospermocarpy involves successful pollination and fertilization followed by abortion of seed development due to defects in integument or endosperm formation. This process results in normal fruit enlargement with only rudimentary seed remnants, as seen in seedless grapes (Royo et al. 2018). With recent advances in genetic engineering, genome editing has become a feasible approach for generating seedless fruits, such as seedless watermelon (Jiang et al. 2024).

MADS-box genes encode a major family of transcription factors that regulate diverse physiological and developmental processes in plants (Favaro et al. 2025). Within the ABCDE model of floral organ development, the C-class gene AG, which governs stamen and carpel formation, and the D-class genes STK (previously AGL11), SHP1, and SHP2, which regulate ovule development, all belong to the AG subfamily of type II MADS-box genes in Arabidopsis (Pinyopich et al. 2003; Mizzotti et al. 2012). In addition, functional studies of the tomato D-class gene SlMBP3/AFF have shown that its mutation results in all-flesh fruit with non-liquefied locules and severely deformed seeds lacking a seed coat (Zhang et al. 2019; Huang et al. 2021). In grape, a single amino acid substitution (R197L) in VviAGL11 confers the stenospermocarpic seedless phenotype (Royo et al. 2018). Despite these findings, it remains unclear whether the functions of these genes are conserved in cucurbit crops.

Melon is an important horticultural crop in the Cucurbitaceae family and is closely related to watermelon, although their edible tissues differ anatomically. In watermelon, the flesh primarily originates from the fleshy placenta, whereas in melon, the edible portion consists mainly of the mesocarp and endocarp, which expand during fruit maturation as the placenta liquefies. Genetic resources for parthenocarpy in melon are limited and remain underutilized in commercial breeding. Moreover, seed coat formation often persists even after CPPU treatment. Therefore, developing seedless or low-seed melon varieties is an urgent breeding objective.

In this study, we performed a genome-wide identification of MADS-box genes in melon. A total of 46 MADS-box genes were identified, comprising 13 type I and 33 type II members (Table S1). The type II genes were further classified into 12 subfamilies (Fig. S1; Table S1). Phylogenetic analysis of the AG subfamily identified four melon genes belonging to this group (Fig. 1A). Among them, MELO3C033521 showed the closest phylogenetic relationship with AFF, STK, and VviAGL11, with protein sequence similarities of 56.9%, 60.6%, and 88.3%, respectively (Fig. 1A). Based on this relationship and to facilitate functional analysis, this gene was designated CmSLM (Cucumis melo solid-placenta and less-seed melon). To reflect both its molecular homology and functional characteristics, it is referred to as CmSLM/CmAGL11 throughout this study.

Fig. 1.

Creation of solid-placenta and less-seed melon via CRISPR-Cas9-mediated gene editing. A Phylogenetic analysis of AG subfamily MADS-box genes. B Subcellular localization of the CmSLM/CmAGL11-GFP fusion protein. Scale bar = 20 μm. C In situ hybridization analysis of CmSLM/CmAGL11 mRNA in fruits at 5 days after pollination. D Gene structure and the two designed target sites of CmSLM/CmAGL11. E Editing outcomes of selected homozygous mutants. PAM sequences are shown in bold, and sgRNA target sequences are underlined. F Longitudinal section and G cross section of mature fruits. Scale bar = 5 cm. H Placental firmness, normal seed number per fruit (I), and seed production (J) in WT and three edited lines. Scale bar = 5 cm. K Scanning electron micrographs of mature dry seeds. Scale bar = 100 μm. L Paraffin section of seed coat cross-sections. The sub‑epidermal layer is outlined with a red rectangle. Scale bar = 50 μm. M Thickness of the seed coat sub‑epidermal layer. Values are means ± SD, n = 10. **** represents P < 0.0001 as determined by Student's t-test

Subcellular localization analysis showed that the CmSLM/CmAGL11-GFP fusion protein co-localized with the nuclear marker RFP-tagged HTB2 (Fig. 1B). Tissue-specific expression analysis revealed predominant expression in hermaphroditic flowers and fruits, with particularly high levels in the placenta (Fig. S2). Consistently, in situ hybridization of fruits demonstrated that CmSLM/CmAGL11 transcripts were mainly localized in the placenta, with strong signals detected in developing seeds (Fig. 1C), suggesting a role in seed development. To further elucidate the function of CmSLM/CmAGL11, a CRISPR/Cas9 construct containing two sgRNAs targeting the third and fourth exons of CmSLM/CmAGL11 was generated and introduced into the thin-skinned melon inbred line IVF105 via Agrobacterium-mediated transformation (Fig. 1D, S3). Five positive T₀ transgenic lines were obtained, and mutations at both target sites were confirmed by Sanger sequencing. Three independent T₂ homozygous mutant lines (cr1, cr2, and cr3) were selected for further analysis. The edited lines cr1 and cr2 carried frameshift mutations caused by base insertion or deletion at the first and second target sites, respectively, resulting in truncated proteins lacking the K-box domain (Fig. 1E, S4, S5). In contrast, cr3 exhibited a large-fragment deletion that eliminated both conserved domains (Fig. 1E, S5, S6). Furthermore, CmSLM/CmAGL11 transcript levels were significantly reduced in all three mutants (Fig. S6). Agronomic trait evaluation of the T₃ generation showed no significant differences between mutant lines and WT in plant height or fruit development period (Fig. S7A, B). During fruit ripening, fruit weight and soluble solids content were comparable between mutant and WT fruits, whereas pericarp firmness was significantly higher in the mutant lines (Fig. S7C-E).

Internal examination of mature fruits revealed pronounced differences between the mutant lines and WT (Fig. 1F, G). The placental tissue of WT fruits was fully liquefied, whereas all mutants retained non-liquefied, solid placentas, resulting in a 15- to 30-fold increase in placental firmness (Fig. 1H). Mature WT fruits typically contained ~400 seeds, while seed numbers were significantly reduced across all three mutant lines (Fig. 1I, G, S8A). In particular, the cr3 mutant produced only 0–3 normal seeds, accompanied by 13.5 ± 3.2 aborted seeds (Fig. 1I, G, S8A, S9). Both heterozygous edited plants and F₁ progeny from crosses between the cr3 mutant and WT displayed normal fruit phenotypes, indicating that the mutation is recessive. Compared to WT, mutant seeds exhibited thinner seed coats that were prone to cracking and displayed numerous longitudinal surface striations, together with a significant reduction in thousand-seed weight (Fig. S8B, S10). Scanning electron microscopy of mature dry seeds further showed that mutant seed coats possessed prominent cell wall protrusions and abundant band-like structures of variable length and width, in contrast to the smooth and regularly arranged epidermal cells observed in WT (Fig. 1K). Semi-thin sectioning revealed a substantial reduction in both size and thickness of the sub-epidermal layer in mutant seed coats, providing a structural basis for their increased brittleness (Fig. 1L, M). Collectively, these findings indicate that the melon CmSLM/CmAGL11 gene is involved in regulating placental liquefaction and supporting normal seed development.

In conclusion, loss of CmSLM/CmAGL11 function led to non-liquefied placentas and severely impaired seed development. In particular, large-fragment deletions produced predominantly solid placentas and near-seedless fruits with a phenotype resembling that of watermelon. Unlike CPPU-induced seedless melons, which retain empty seed coats, the solid-placenta, few-seeded melons generated in this study were produced through disruption of a MADS-box gene. Looking ahead, optimization of CmSLM/CmAGL11 target sites, removal of exogenous vector sequences, and combination with CPPU treatment may enable the development of elite, truly seedless melon cultivars.

Author’ contributions

JM, HC and CL designed and supervised the research. JM, WH, and ML performed the experiments. JM, HZ and GY wrote the manuscript. All authors read and approved the final version of the manuscript.

Funding

This study was supported by the Project for Scientific and Technological Innovation Capacity Building of BAAFS (KJCX20261007 and KJCX20261416), the Youth Research Fund of BAAFS (QNJJ202426), the Innovation and Development Program of Beijing Vegetable Research Center (KYCX202601), the Construction of Cucurbits Collaboration and Innovation Center (XTCX202301), and the National Natural Science Foundation of China (U21A20229).

Data availability

All data are available in the manuscript. The datasets and materials of this study are available from the corresponding author upon reasonable request.

Declarations

Ethics approval and consent to participate

Not applicable.

Consent for publication

Not applicable.

Competing interests

The authors declare that they have no conflict of interest.