The GRAS transcription factor CsTL regulates tendril formation in cucumber

Abstract

Cucumber (Cucumis sativus, Cs) tendrils are slender vegetative organs that typically require manual removal to ensure orderly growth during greenhouse cultivation. Here, we identified cucumber tendril-less (tl), a Tnt1 retrotransposon-induced insertion mutant lacking tendrils. Map-based cloning identified the mutated gene, CsaV3_3G003590, which we designated as CsTL, which is homologous to Arabidopsis thaliana LATERAL SUPPRESSOR (AtLAS). Knocking out CsTL repressed tendril formation but did not affect branch initiation, whereas overexpression (OE) of CsTL resulted in the formation of two or more tendrils in one leaf axil. Although expression of two cucumber genes regulating tendril formation, Tendril (CsTEN) and Unusual Floral Organs (CsUFO), was significantly decreased in CsTL knockout lines, these two genes were not direct downstream targets of CsTL. Instead, CsTL physically interacted with CsTEN, an interaction that further enhanced CsTEN-mediated expression of CsUFO. In Arabidopsis, the CsTL homolog AtLAS acts upstream of REVOLUTA (REV) to regulate branch initiation. Knocking out cucumber CsREV inhibited branch formation without affecting tendril initiation. Furthermore, genomic regions containing CsTL and AtLAS were not syntenic between the cucumber and Arabidopsis genomes, whereas REV orthologs were found on a shared syntenic block. Our results revealed not only that cucumber CsTL possesses a divergent function in promoting tendril formation but also that CsREV retains its conserved function in shoot branching.

Cucumber CsTL has a divergent function compared with its Arabidopsis homolog, promoting tendril formation by acting as a cofactor with CsTEN to directly induce CsUFO.

Introduction

Cucumber (Cucumis sativus L.), an important crop in the Cucurbitaceae family, is widely cultivated throughout the world, and its fruits are either consumed fresh or processed into pickles (Huang et al. 2009; Guo et al. 2020). Cucumber has an indeterminate growth type, in which leaves emerge from the sides of the shoot apical meristem (SAM), while axillary meristems (AMs) develop in the leaf axils, giving rise to branches, tendrils, and flowers (Leonard 1962; Weng et al. 2010; Qi et al. 2013). In greenhouse cultivation of cucumbers for the fresh market, the manual removal of branches and tendrils is typically required to ensure orderly growth (Shen et al. 2019; Liu et al. 2021). This process is both time-consuming and labor-intensive. Moreover, branches and tendrils, which compete with fruits for nutrient allocation, are both lateral organs situated in the leaf axil of the cucumber plant, potentially exerting a negative impact on overall yield. Hence, comprehending the regulatory mechanism underlying tendril and branch formation holds great significance for the breeding of cucumber plants with optimal architecture.

The tendril is a specialized organ with a curving, threadlike shape in cucurbits and other climbing plants, providing the ability to climb by winding around their neighbors or other adjacent objects. This allows them to obtain more living space, improving photosynthetic efficiency and promoting pollination (Darwin 1875). The ontogenetic origin of tendrils can be classified into two main types: shoot-derived tendrils and leaflet-derived tendrils (Sousa-Baena et al. 2018). Cucurbit plants develop classical shoot-derived tendrils, whereas leaflet-derived tendrils are prevalent in species in diverse angiosperm families, including the Asteraceae, Fabaceae, Bignoniaceae, and Papaveraceae (Gourlay et al. 2000; Sousa-Baena et al. 2014, 2018). In the garden pea (Pisum sativum L.), TENDRIL-LESS (HD-ZIP) and its upstream genes UNIFOLIATA/LEAFY and LATHYROIDES (WUSCHEL-related homeobox1/WOX1) are involved in the initiation of tendrils (Hofer et al. 2001, 2009; Zhuang et al. 2012).

A tendril-less mutant of melon (Cucumis melo), known as “Chiba Tendril-Less” (ctl), was identified due to its absence of tendrils, which was accompanied by the development of tendril-like branches in the leaf axils. Fine mapping showed that this tendril-less trait in the ctl mutant was caused by a defect in TCP1, a member of the TEOSINTE BRANCHED1, CYCLOIDEA, and PROLIFERATING CELL FACTORS1/2 (TCP) family (Mizuno et al. 2015). Genetic and mapping analysis of cucumber line CG9192 revealed a causal mutation in CsTEN, a gene homologous to melon TCP1, that was associated with a tendril-less phenotype (Wang et al. 2015). Subsequent investigations revealed not only that CsTEN functions through its C-terminal region, which binds to enhancers of its downstream target genes, but also that its N-terminal region acts as an atypical histone acetyltransferase (Yang et al. 2020). This tendril-less trait is also observed in the tendril-less1 mutants of cucumber and is the result of a nonsynonymous single-nucleotide polymorphism (SNP) in GENERAL CONTROL NONDEREPRESSIBLE 5 (GCN5), which encodes a histone acetyltransferase (Chen et al. 2017). Two mutants with defective tendril morphogenesis, tmd1 and uft, were identified in cucumber as mutants of Tendril Morphogenesis Defective and Unusual Flower and Tendril. Map-based cloning showed that the deletion of a fragment within the promoter and an SNP mutation (C to T) in the second exon of CsaV3_1G010110 [cucumber UNUSUAL FLORAL ORGANS (CsUFO)] gave rise to morphological variation of tendrils in tmd1 and uft, respectively (Chen et al. 2021; Hong et al. 2023). Biochemical assays confirmed that CsTEN directly binds to the promoter of CsUFO, positively regulating its expression (Hong et al. 2023).

In higher plants, the development of branches occurs in two stages: axillary bud initiation and outgrowth (Leyser 2003). In tomato (Solanum lycopersicum), lateral suppressor (Ls), which encodes a GRAS family transcription factor, plays a role in the initiation of axillary buds (Schumacher et al. 1999; Feng et al. 2021). Defects in Arabidopsis LATERAL SUPPRESSOR (LAS) lead to a significant reduction in the number of axillary buds compared with wild-type (WT) plants (Greb et al. 2003). Similarly, a functional analysis of rice (Oryza sativa) MONOCULM 1 (MOC1) revealed its essential role in the formation of AMs and the regulation of tiller number (Li et al. 2003). Amino acid sequence analysis indicated that tomato Ls, Arabidopsis LAS, and rice MOC1 encode homologous proteins (Pysh et al. 1999; Feng et al. 2021). In Arabidopsis and tomato, the boundary-specific expression pattern of LAS/Ls is conferred by a highly conserved enhancer/suppressor element (Raatz et al. 2011). The Arabidopsis class III HD-ZIP gene REVOLUTA (REV), a gene downstream from LAS, regulates the initiation of AMs that produce branches in the axils of rosette and cauline leaves (Talbert et al. 1995; Greb et al. 2003). REV up-regulates SHOOT MERISTEMLESS (STM) expression in leaf axil meristematic cells, leading to the establishment of AMs (Shi et al. 2016). In summary, prior research has demonstrated that orthologs of LAS and REV in various species perform a conserved function in branch formation.

In this study, we identified cucumber tendril-less (tl), a Tnt1 retrotransposon-induced insertion mutant that is characterized by the suppression of tendril initiation. Fine mapping revealed a Tnt1 retrotransposon insertion in CsaV3_3G003590, designated as CsTL, which encodes a GRAS family transcription factor that is homologous to Arabidopsis LAS. This insertion was identified as the cause of the tendril-less phenotype observed in tl. We further showed that knocking out CsTL resulted in the absence of tendrils without affecting shoot branching, and OE of CsTL in cucumber confirmed its function in regulating tendril formation. Moreover, we demonstrated that CsTL interacted with CsTEN, influencing tendril formation by increasing the expression of CsUFO. Additional gene knockout investigations involving CsREV revealed that CsTL governs tendril formation independently of CsREV. Collectively, these findings provide important insights that can form the basis of a strategy to develop labor-saving cucumber varieties lacking tendrils or branches, enabling more efficient and productive cultivation.

Results

Identification of the candidate mutated gene in the cucumber tendril-less (tl) mutant

In the protected cultivation of cucumber, tendrils usually need to be manually removed, leading to a substantial waste of biomass and considerable labor costs. In the quest for tendril-less cucumber resources, we screened a Tnt1 retrotransposon insertion mutant collection derived from the model cucumber cultivar 9930 (Zhang et al. 2018). We successfully identified a tendril-less (tl) mutant in the segregating generation of TN89, a cucumber line that was selected from the cucumber Tnt1 transposon mutant library. The tl mutant exhibited a complete absence of tendrils, with only branches and flowers differentiating at the leaf axils (Fig. 1, A to D). Genetic analysis indicated the mutation was due to a single Mendelian recessive gene (Supplementary Table S1). Subsequently, we employed bulked segregation analysis-seq (BSA-seq) to locate the candidate gene mutated in tl. After calculating statistical confidence intervals of P < 0.01 between the two extreme phenotypic bulks, a 3.73 Mb genomic region on Chromosome 3 was identified by combining the results of the two algorithms (Fig. 1, E and F). Considering that the tl mutation was caused by the insertion of the Tnt1 retrotransposon, we used the sequences of the four Tnt1 insertion sites within the candidate region to design markers for genotyping a total of 340 plants in the F2 population. In the relevant recombinants, we discovered that CsaV3_3G003590, which was identified by the TN3G29 marker, was highly likely to be the candidate gene, and thus named it CsTL (Fig. 1, G and H). By examining the genomic DNA (gDNA), we confirmed the insertion of a Tnt1 retrotransposon in the second exon of CsaV3_3G003590/CsTL (Supplementary Fig. S1). To better characterize CsTL, a phylogenetic tree was constructed using an alignment of sequences of genes encoding GRAS family transcription factors in cucumber, tomato, and Arabidopsis (Supplementary Files 1 and 2). In this tree, CsTL clustered with Arabidopsis LAS and tomato Ls (Supplementary Fig. S2).

Figure 1.

Fine mapping and expression analysis of CsTL, a novel gene controlling tendril formation in cucumber. Growth habit of WT plants (A) and tl mutant plants (B) 2 months after sowing. Boxplots depicting the distribution of tendril number (C) and branch number (D) in both 2.5-month-old WT and tl plants (n = 9). Significance analyses were performed with two-tailed Student's t-test (**P < 0.01). E) and F) Preliminary mapping of tl by using BSA-seq. E) Identification of the candidate interval for the tl gene was achieved through the SNP index association analysis method. The tl locus is situated within a 3.73-Mbp interval on Chromosome 3. The abscissa represents the chromosome name, while the ordinate denotes the SNP index value. F) Results of BSA-seq mapping of the tendril-regulating gene based on the Euclidean distance (ED) algorithm. The abscissa is the chromosome name, and the ordinate is the ED value. The colored dots represent the ED values of each SNP site. The black line represents the fitted ED value. The red line represents the correlation threshold. G) and H) Map-based cloning of genes in the tl locus. G) Four Tnt1 insertion sites within the candidate region used for designing markers for genotyping a total of 340 plants in the F₂ population. H) Annotated genes in the final interval. CsaV3_3G003590 is cucumber TL candidate. I) Expression analysis of CsTL in different cucumber organs. Error bars represent mean ±SD. Values are means ± SD (n = 3). Significance was determined using one-way ANOVA (P < 0.05). J) to L) In situ hybridization analysis of CsTL expression in SAM tissue. The sense probes of CsTL were hybridized as controls (CK) (J). The black arrows in K and L indicate the mRNA signals. Scale bars = 50 μm.

We next examined the expression levels of CsTL in various organs of the WT cucumber CU2 line by reverse transcription quantitative PCR (RT-qPCR). CsTL was expressed at much higher levels in the shoot apex, lateral buds, and young tendrils than in roots, stems, leaves and flower buds (Fig. 1I). To examine the cellular distribution of CsTL transcripts, we performed an RNA in situ hybridization experiment using CsTL and young shoot apexes. A CsTL-specific probe produced strong signals in the two side areas of the SAM (Fig. 1, J to L). Stronger CsTL signals were detected in the lateral tissues, which exhibited active differentiation of AMs and the formation of lateral organ primordia (Fig. 1, K and L). Moreover, the results of in situ hybridization assay in SAM cross sections indicated CsTL transcription on both sides of the central region of the SAM and in lateral meristematic tissues (Supplementary Fig. S3A). A CsSTM-specific probe produced signals both in the outer ring region of the SAM and in all lateral meristematic tissues (Supplementary Fig. S3, B and C). A sense probe representing CsSTM was used as an experimental control (Supplementary Fig. S3D). These findings suggested an overlap in the signal positions of CsTL and CsSTM, implying that CsTL might have a role in regulating organ initiation.

Knocking out CsTL inhibited tendril formation, whereas OE of CsTL led to the production of multiple tendrils in cucumber leaf axils

To validate the function of CsTL, its sequence was obtained from the WT cucumber CU2 line. Two single-guide RNAs (sgRNAs) targeting the first exon of CsTL near the deduced protein N-terminus were designed to knock out CsTL using the CRISPR/Cas9 (clustered regularly interspaced short palindromic repeats/CRISPR-associated system 9) gene editing system (Fig. 2A). Using this system, nine homozygous CsTL mutants (CsTL^CR), including CsTL^CR-1 (with a 4-bp deletion), CsTL^CR-3 (with a 51-bp deletion), and CsTL^CR-9 (with a 47-bp deletion), were obtained (Fig. 2A). In WT plants, phenotypic observations and schematic representations revealed that tendrils were present in the leaf axils from node 3 or 4 to node 20 and that branches formed in the leaf axils from node 2 to node 14 (Fig. 2, B and E). However, there were no tendrils in the leaf axils from nodes 1 to 20 in plants of three T₂-generation gene-edited mutant lines CsTL^CR-1, CsTL^CR-3, and CsTL^CR-9, and lateral branches were produced normally and distributed from nodes 2 to 15 (Fig. 2, C to E; Supplementary Fig. S4). These results indicated that the plant morphology of CsTL knockout lines was consistent with that of tl mutant.

Figure 2.

Functional analysis of CsTL by CRISPR/Cas9-engineering and OE system. A) sgRNA target sites in the first exon of CsTL and editing results in cucumber determined by Sanger sequencing. The two target sites of the sgRNAs and adjacent sequences in the first exon are shown for both the WT plant and the gene-edited mutants. Deleted nucleotides are denoted by dashes. The sgRNA sequences are highlighted in red, and the PAM sites are marked in blue. Genotypic analysis of CsTL knockout plants with the CU2 background revealed partial deletions in the DNA sequence of the first exon. B) to D) Comparison of phenotypic differences between WT, CsTL^CR-1, and CsTL^CR-3 knockout lines. LB, lateral branch; T, tendril. Scale bars = 5 cm. E) Diagram illustrating the positioning of tendrils and branches at the initial 20 nodes in the WT and CsTL mutant plants. Each layer represents a node in a cucumber plant. A green square signifies a node that produces a normal tendril but no branch, while a boxed green square indicates a node that produces both a normal tendril and a branch. A yellow square represents a node that produces neither a branch nor a tendril, and a boxed yellow square represents a node that produces a branch but no tendril. Comparison of the internal composition of shoot apexes in WT plant (F), CsTL knockout plant (G), and HW plant (H) at 30 days of age using 3D micro-CT imaging. TP, tendril primordium; FP, flower primordium; LBP, lateral branch primordium. Scale bars = 300 μm. Tendril phenotypes in WT (I), OE-2 (J), and OE-3 (K) plants. Tendrils are indicated by white arrows. T, tendril. Scale bars = 5 cm.

Subsequently, we observed the shoot apices of 2-month-old cucumber plants of the WT, the CsTL knockout line (CU2 background), and the wild cucumber variety Hardwickii (HW). In WT plants, tendrils, lateral branches, and flowers were produced normally in the leaf axils (Supplementary Fig. S5A). In the CsTL knockout plant, only lateral branches and flowers appeared in the leaf axils, whereas in HW plants, only tendrils and lateral branches were observed in the leaf axil (Supplementary Fig. S5, B and C). This suggests that the HW plants underwent only vegetative growth over an extended period of growth. In further observation using a microcomputed tomography (micro-CT) microscopic imaging system, we found that in 30-day-old WT plants, tendril primordia, lateral branch primordia, and floral primordia developed normally in the leaf axil (Fig. 2F). In CsTL^CR-1 line plants at 30 days of age, only lateral branch primordia and floral primordia were present in the leaf axil, suggesting that the CsTL mutant exhibited a defect in tendril initiation rather than tendril outgrowth (Fig. 2G). Meanwhile, in the wild HW plants at 30 days of age, only tendril primordia and lateral branch primordia were observed (Fig. 2H). These results indicated that the initiation of cucumber tendrils had already occurred in the early developmental stages in both the cultivated variety and the wild cucumber plants. Additionally, the micro-CT imaging results for WT plants, CsTL knockout plants, and HW plants indicated distinct initiation characteristics for tendrils, lateral branches, and flowers.

To further characterize the function of CsTL, a vector for the OE of CsTL was constructed for genetic transformation in cucumber. Four OE transgenic lines were obtained, namely OE-1, OE-2, OE-3, and OE-4. Positive PCR results targeting the T-DNA region of the vector indicated that the OE constructs were successfully integrated into the genomes of these four lines (Supplementary Fig. S5D). The expression levels of CsTL in OE-1, OE-2, OE-3, and OE-4 increased by 16-, 67-, 44-, and 15-fold, respectively, compared with the control plants (Supplementary Fig. S5E). In WT plants, only one tendril was generated per leaf axil, whereas the OE-3 line produced two tendrils in a single leaf axil at the early stage (Fig. 2, I and K). In addition, the internodes of the OE-2 line produced multiple whorls of leaves accompanied by multiple tendrils during the vegetative growth period (Fig. 2J). The extra-tendril phenotype was not observed in the OE-1 and OE-4 lines, possibly due to the expression level of CsTL in these two lines being not nearly as high as it was in the OE-2 and OE-3 lines.

Two tendril regulators, namely CsTEN and CsUFO, showed reduced transcription in CsTL knockout lines but were not direct downstream targets of CsTL

To identify the downstream targets and regulatory network of CsTL in cucumber, RNA sequencing (RNA-seq) analysis was carried out using shoot apexes from WT plants and the CsTL knockout lines. 124 upregulated and 382 downregulated genes were identified as differentially expressed genes (DEGs) (Supplementary Data Set 1). Heat map analysis showed that the expression levels of 19 transcription factor genes, including Tendril (CsTEN, CsaV3_5G039160) and Unusual Floral Organs (CsUFO, CsaV3_1G010110), were decreased in the CsTL knockout lines compared with WT plants (Fig. 3A). RT-qPCR analyses confirmed that CsTEN and CsUFO were significantly downregulated in the CsTL knockout lines (Fig. 3, B and C).

Figure 3.

Analyses of relationships between CsTL and CsTEN and CsUFO. A) Heat map of gene expression levels of DEGs in WT and CsTL^CR lines from three independent biological replicates (n = 3). A colored legend positioned to the left of the map indicates FPKM (fragments per kilobase of transcript per million mapped reads) values. The data for the WT serve as the control (CK). RT-qPCR analyses of tendril-related genes CsTEN(B) and CsUFO(C) expression in CsTL^CR lines. Significance analyses compared with WT were performed with two-tailed Student's t-test (**P < 0.01). The values represent the mean ± SD (n = 3). Phenotypic comparison of WT (D), CsTL^CR line (E), CsTEN^CR line (F), and Csufo mutant (G). H) Yeast one-hybrid assay showing that CsTL could not bind to the promoter of either CsTEN or CsUFO. I) Schematic diagrams of the effector and reporter constructs used in the LUC reporter transient expression system. J) Effects of CsTL on the CsTEN and CsUFO promoter activity. The LUC/REN ratio represents the LUC activity relative to the internal control REN. Values are mean ± SD (n = 6).

We then used the CRISPR/Cas9 system to knock out CsTEN and obtained seven CsTEN knockout lines, including CsTEN^CR-1, CsTEN^CR-2, and CsTEN^CR-5, homozygous lines with 7-bp, 8-bp, and 5-bp deletions, respectively (Supplementary Fig. S6A). Similar to the tendril-less phenotype of CsTL knockout plants, none of the three CsTEN knockout lines produced tendrils in the leaf axils compared with WT plants (Fig. 3, D to F; Supplementary Fig. S6, B to D). In addition, CsUFO-m, a cucumber mutant resulting from a 1,581-bp deletion in the promoter region of CsUFO (Supplementary Fig. S7), exhibited leaf-like structures with serrated edges on the elongated side of tendrils (Fig. 3G). Additionally, compared with the WT plant, the tendrils of the CsUFO-m plant lost the ability to coil around supporting structures (Fig. 3, D and G). In light of these findings, our initial hypothesis posited that CsTL exerts direct regulatory control over CsTEN and CsUFO by binding to their respective promoters. However, a yeast one-hybrid assay showed that CsTL did not directly bind to the promoters of either CsTEN or CsUFO (Fig. 3H). To further examine whether CsTL could directly regulate the expression of CsTEN and CsUFO, we performed a dual-luciferase (LUC) reporter (DLR) assay in Nicotiana benthamiana leaves. The relative intensity of LUC signals was not changed upon co-transformation of Pro35S:CsTL with ProCsTEN:LUC or ProCsUFO:LUC constructs, as compared to the empty vector control (Fig. 3, I and J), suggesting that CsTL could not directly regulate the expression of either CsTEN or CsUFO.

CsTL interacted with CsTEN to regulate transcription of CsUFO

The functional similarity between CsTEN and CsTL in tendril formation as well as their overlapping expression patterns in lateral tissues (Fig. 1L; Supplementary Fig. S8) suggested that CsTL might physically interact with CsTEN. Results of a yeast two-hybrid assay showed that CsTL directly interacted with CsTEN (Fig. 4A). In addition, in vitro pull-down assays showed that when Glutathione-S-Transferase (GST)-fused CsTL was incubated with CsTEN, visible bands corresponding to CsTL were seen on a resulting immunoblot, whereas the negative controls without addition GST-CsTL yielded no band (Fig. 4B). To verify the protein–protein interaction between CsTL and CsTEN in vivo, bimolecular fluorescence complementation (BiFC) assays and firefly LUC complementation imaging (LCI) assays were performed in N. benthamiana leaves. The combination of CsTL–CsTEN gave obvious fluorescence signals, whereas the negative controls gave no signals (Fig. 4, C and D). A co-immunoprecipitation (Co-IP) assay confirmed that CsTL could form a complex with CsTEN (Fig. 4E).

Figure 4.

CsTL directly interacts with CsTEN to enhance CsTEN-mediated CsUFO expression. A) Yeast two-hybrid assay showed that CsTL interacts with CsTEN in yeast cells grown on SD-Leu/-Trp/-His/-Ade/X-α-gal selective medium by observation of the growth of transformants. The combinations containing either empty pGADT7 or empty pGBKT7 vectors were used as negative controls. B) In vitro GST pull-down assay further confirming the physical interaction between CsTL and CsTEN. The combination of GST and His-CsTEN was used as negative control. C) BiFC assay validating the interaction between CsTL and CsTEN in N. benthamiana epidermal leaf cells. Scale bars = 20 μm. D) Firefly LUC complementation imaging assay further showed that CsTL interacts with CsTEN. CsTL-cLUC and nLUC-CsTEN were transiently co-expressed in N. benthamiana leaves, and the remaining combinations were used as negative controls. E) Co-IP analysis showed the interaction of CsTL with CsTEN in N. benthamiana leaves. Co-IP assay was performed using an agarose conjugated anti-GFP antibody. The expression levels of CsUFO(F) and CsTEN(G) were assessed in 35S:CsTL-GR transgenic plants following treatments with DEX, CHX, and DEX + CHX, respectively. Asterisks denote significant differences between samples treated with different chemicals (**P < 0.01, Student's t-test). The values are presented as mean ± SD (n = 3). H) Firefly LUC and Renilla reniformis LUC activity assay in N. benthamiana leaves by co-expression of Pro35S:CsTL and/or Pro35S:CsTEN with ProCsUFO:LUC. The LUC/REN ratio from the empty vector (62-SK) combined with ProCsUFO:LUC was used as a calibration. Values are mean ± SD (n = 6).

A previous study had shown that CsTEN directly bound to the CsUFO promoter and positively regulated its transcription in cucumber tendrils (Hong et al. 2023). Our RT-qPCR results indicated that the expression levels of CsTEN did not change significantly in CsTL OE lines, whereas CsUFO showed elevated expression in these lines (Supplementary Fig. S9, A and B). We transformed WT cucumber plants with a vector carrying glucocorticoid receptor (GR) sequences fused to CsTL (Pro35S:CsTL-GR) sequences. The presence of the appropriate PCR-produced bands, using primers specifically targeting the T-DNA region of the vector, confirmed the successful integration of the Pro35S:CsTL-GR constructs into the genomes of the transgenic plants (Supplementary Fig. S9C). CsTL-GR transgenic plants were treated with dexamethasone (DEX), the protein synthesis inhibitor cycloheximide (CHX), or a combination of DEX and CHX. RT-qPCR results revealed that, compared with mock treatment, CsTEN and CsUFO expression levels in transgenic plants remained unchanged after 4 h of CHX treatment (Fig. 4, F and G). After 4 h of DEX treatment, CsTEN expression showed no significant change, whereas CsUFO expression levels significantly increased (Fig. 4, F and G). Similarly, after 4 h of treatment with both DEX and CHX, CsTEN expression remained unchanged, but CsUFO expression levels also significantly increased, suggesting the upregulation of CsUFO expression by CsTL (Fig. 4, F and G). To further explore the relationship between CsTL, CsTEN, and CsUFO, a DLR assay was performed. The transcriptional activation of CsUFO was notably increased when co-infiltrated with both CsTL and CsTEN, in stark contrast to the comparatively weak activation observed in the presence of CsTL alone or CsTEN alone (Fig. 4H). Taken together, the data suggested that CsTL physically interacted with CsTEN to activate CsUFO transcription during tendril formation in cucumber.

Regulation of tendril initiation by CsTL was independent of CsREV

The Arabidopsis class III HD-ZIP gene REVOLUTA (REV), which acts downstream of LAS, regulates the initiation of lateral meristems in the axils of both rosette and cauline leaves (Talbert et al. 1995; Greb et al. 2003). In cucumber, we found that co-infiltration of CsTL and CsREV:LUC into N. benthamiana leaves led to an increased luciferase/renilla luciferase (LUC/REN) ratio (Supplementary Fig. S10). However, an in situ hybridization assay revealed no difference in the mRNA distribution pattern of CsREV (CsaV3_6G013930) between WT plants and the CsTL knockout lines, with signal enrichment in AMs (Fig. 5, A to C). RT-qPCR analysis also showed that the expression levels of CsREV did not change in either the CsTL knockout lines or the CsTL OE plants compared with WT plants (Fig. 5D; Supplementary Fig. S11A). To further clarify the function of CsREV, we created four CRISPR/Cas9-edited CsREV knockout lines, including CsREV^CR-1, CsREV^CR-2, and CsREV^CR-3, and these lines were found to have 2-bp (TA), 3-bp (ATA), and 3-bp (TAA) deletions, respectively (Fig. 5E). The axils of WT plants produced tendrils, branches, and flowers, whereas in CsREV^CR-1, CsREV^CR-2, and CsREV^CR-3 plants, the majority of axils gave rise to only tendrils and flowers and lacked branches (Fig. 5, F to K; Supplementary Fig. S12). A schematic representation also demonstrated a significant reduction in the number of branches in the leaf axils of CsREV^CR-1 knockout plants compared with the WT plants, whereas tendril initiation occurred normally (Fig. 5L). Similarly, there was no alteration in the xpression level of CsTL in CsREV knockout plants (Supplementary Fig. S11B). These data indicated that CsTL regulates tendril formation independently of CsREV in cucumber without affecting branch development, whereas CsREV possesses a conserved function in controlling shoot branching but not tendril formation.

Figure 5.

Role of CsREV in branch development in cucumber. A) to C) In situ hybridization analysis of CsREV expression in the shoot apex of WT (B) and CsTL mutant (C). The sense probe of CsREV was hybridized as negative control (CK) (A). AM, axillary meristem. Scale bars = 20 μm. D) Expression levels of CsREV in CsTL knockout lines. Values are mean ± SD (n = 3). E) Schematic diagram of the CsREV gene showing the two CRISPR/Cas9-targeting sites located in its first exon and the induced mutations at the first site in mutant lines 1, 2, and 3. Comparison of WT (F–H) and CsREV^CR-1 plants (I–K) from overall plant appearance (F and I) to individual leaf axils (G, H, J, and K). Scale bars = 5 cm. L) Diagram illustrating the positioning of tendrils and branches in the initial 20 nodes of both WT and CsREV^CR plants. Each layer corresponds to a node in a cucumber plant. Green squares denote nodes generating regular tendrils without branches, while boxed green squares signify nodes producing both normal tendrils and branches. Yellow squares represent nodes producing neither branches nor tendrils, with boxed yellow squares indicating nodes generating only branches without tendrils.

Divergence in syntenic relationships of TL/LAS-containing genomic regions and conservation of REV-containing regions between cucumber and Arabidopsis

To understand the functional divergence of TL/LAS and functional conservation of REV between cucumber and Arabidopsis from an evolutionary perspective, we examined the syntenic relationships between genomic regions containing these two genes in the genomes of cucumber and Arabidopsis. Interestingly, the Arabidopsis genomic region syntenic with the CsTL-containing region does not contain AtLAS (Fig. 6A). Similarly, we found that the two regions on cucumber Chromosomes 3 and 4 that are syntenic with the AtLAS-containing region do not contain CsTL (Fig. 6B). In contrast, consistent with the proposed conservation of REV function in these two species, REV orthologs are also located on highly syntenic regions (Fig. 6C). The absence of syntenic TL/LAS-containing regions in these two genomes is also consistent with the functional divergence we have observed in cucumber vs. Arabidopsis.

Figure 6.

Syntenic analyses of CsTL-containing region and CsREV-containing region between cucumber and Arabidopsis genome and within cucumber genome. A) Syntenic relationship between CsTL-containing region on Chromosome 3 of cucumber and its syntenic region on Chromosome 1 of Arabidopsis. The location of CsTL is highlighted in magenta. B) Syntenic analysis of cucumber Chromosomes 3 and 4 with the AtLAS-containing region in Arabidopsis. C) Syntenic relationship between CsREV-containing region on Chromosome 6 of cucumber and AtREV-containing region on Chromosome 5 of Arabidopsis. The location of CsREV is highlighted in green. D) Syntenic relationship between CsREV-containing region on Chromosome 6 of cucumber and another region on Chromosome 6. Gray lines indicate homologous gene pairs in the syntenic regions. The location of CsREV is highlighted in green. E) Syntenic relationship between CsTL-containing region on Chromosome 3 of cucumber and two regions on Chromosome 5 and 6. Gray lines indicate homologous gene pairs in the syntenic regions. The location of CsTL is highlighted in magenta.

Despite CsTL's functional divergence into controlling tendril initiation, CsTL appears to be a single-copy gene, with no duplicated copies carrying out its original function, even considering the presence of a whole-genome duplication (WGD) event shared by all Cucurbitaceae species (denoted CucWGD1 here) (Guo et al. 2020). This was confirmed by syntenic analysis showing that, in the cucumber genome, the genomic region containing CsTL on Chromosome 3 is flanked by two regions that are syntenic with Chromosomes 5 and 6, respectively, whereas the CsTL-containing region itself does not have any syntenic region in the cucumber genome (Fig. 6E). Similarly, only a single copy of CsREV was found in the cucumber genome, and similar syntenic analysis of the CsREV-containing region demonstrated that this region shared synteny with another region on Chromosome 6, but no REV homologous gene was found in this syntenic region (Fig. 6D). Taken together, these results indicate that, after undergoing the CucWGD1 event, the duplicated copies of TL and REV were likely lost. However, the remaining copy of TL likely gained a novel function in regulating tendril development, while the remaining copy of REV did not undergo functional divergence but retained its conserved function in regulating branches instead.

Discussion

CsTL and CsREV regulate the formation of tendrils and branches in cucumber, respectively

The Lateral suppressor gene(Ls/LAS/MOC1) encodes a GRAS family protein that has a conserved function in regulating branch or tiller formation in tomato, Arabidopsis, and rice (Schumacher et al. 1999; Greb et al. 2003; Li et al. 2003). The mRNAs of tomato Ls and Arabidopsis LAS are enriched in the band-shaped region on the adaxial side of the young leaf primordium, and loss of LAS/Ls function inhibits shoot branching during vegetative growth. In contrast, the LAS/Ls homolog in rice, MOC1, regulates tillering during both vegetative and reproductive growth (Schumacher et al. 1999; Greb et al. 2003; Li et al. 2003). Here, we identified CsTL, a gene homologous to LAS/Ls/MOC1, as the gene mutated in the cucumber tendril-less (tl) mutant (Fig. 1). Using CRISPR/Cas9 gene editing technology, we were able to knock out CsTL and found that tendril formation in the leaf axils was repressed in cucumber (Fig. 2). OE of CsTL gave rise to plants that produced two tendrils per leaf axil or generated multiple whorls of leaves and tendrils at individual internodes (Fig. 2). Interestingly, we observed no obvious differences in the numbers and phenotypes of branches in the tl mutants, CsTL knockout lines, or CsTL OE lines compared with WT plants (Figs. 1 and 2). Our findings demonstrate that CsTL plays a crucial role in cucumber tendril formation without affecting the development of branches, a result in stark contrast to the function of its homologous genes in Arabidopsis, tomato, and rice. The functional divergence of CsTL function in cucumber might be attributed to the following facts: (i) The plant architecture of cucumber differs significantly from that of the other model species previously examined; there are no tendrils in Arabidopsis, tomato, or rice. During the vegetative period of growth, Arabidopsis and rice plants produce only vegetative shoots (branches or tillers) in the leaf axils. Upon the transition to the reproductive phase, the apical meristems of these vegetative shoots undergo transformation into inflorescence meristems, subsequently giving rise to flowers (Fletcher 2002). Tomato plants also generate only lateral branches in leaf axils, with inflorescences emerging near the upper edge of the leaf insertion. In comparison, cucumber exhibits indeterminate growth, which is theoretically capable of continuously differentiating AMs to produce lateral branches, tendrils, and flowers in the leaf axils. (ii) Cucumber possesses a SAM that remains active, enabling continuous upward growth, while also continuously generating lateral branches, tendrils, and flowers in its leaf axils. Consequently, cucumber CsTL continues to function throughout the ongoing differentiation of tendrils. (iii) The protein sequence similarity between CsTL and tomato Ls, Arabidopsis LAS, and rice MOC1 is only 55%, 45%, and 35%, respectively (Supplementary Table S2), and this may be the reason for the functional divergence among these species.

In Arabidopsis, REVOLUTA (REV), together with the upstream gene LAS, regulates branch initiation (Otsuga et al. 2001; Shi et al. 2016) (Fig. 7A). Notably, the expression level and mRNA distribution of CsREV did not differ between CsTL knockout lines and WT plants (Fig. 5, A to D). Disrupting CsREV in cucumber resulted in the absence of branches in leaf axils, with no discernible difference in tendril initiation between CsREV knockout plants and WT plants (Fig. 5, E to K). Hence, CsTL autonomously regulates tendril formation in cucumber independently of CsREV. Additional experiments are required to discover the genes operating downstream of CsTL in the process of cucumber tendril formation.

Figure 7.

Model for engineering ideal plant architecture in cucumber. A) In model plant Arabidopsis, Lateral suppressor (LAS) together with its downstream gene REVOLUTA (REV) directly modulate shoot branching. B) However, our results indicate that the LAS homolog in cucumber, CsTL, regulates tendril formation by directly forming a complex with CsTEN to further enhance CsUFO expression, while CsREV conservatively functions in regulating branching in cucumber. C) Cucumber varieties cultivated in a protected environment traditionally produce fruits, tendrils, and branches in leaf axils, leading to competition for nutritional allocation among these three lateral organs. Through the editing of CsTL and/or CsREV in cucumber, it may be possible to develop cucumber varieties without tendrils, without branches, or without both tendrils and branches.

CsTL promotes tendril formation by forming a complex with CsTEN that directly activates CsUFO expression in cucumber

In this study, knocking out CsTEN in cucumber resulted in arrested tendril formation, and this phenotype was consistent with that of the CsTL knockout lines (Fig. 3, E and F). In situ hybridization assays showed that the distribution of CsTL mRNA overlapped with that of CsTEN in lateral tissues, suggesting that CsTL and CsTEN might function together in these tissues (Fig. 1, J to L; Supplementary Fig. S8). A series of detailed biochemical experiments revealed a direct interaction between CsTL and CsTEN, suggesting that these proteins may together regulate tendril formation by forming a protein complex (Fig. 4, A to E). A recent study identified a mutant defective in tendril morphogenesis, and the results of fine mapping implicated the candidate gene to be CsUFO (Hong et al. 2023). Analysis of transcript levels in this study found that the expression of CsUFO was significantly decreased in the CsTL knockout lines and increased in the CsTL OE lines (Fig. 3C; Supplementary Fig. S9B). Although biochemical analysis demonstrated that CsUFO was not the direct downstream target gene of CsTL (Fig. 3, H to J), we found that the presence of CsTL enhanced the transcriptional activation activity of CsTEN on CsUFO (Fig. 4H). All these data indicated that CsTL regulates tendril formation by forming a protein complex with CsTEN to induce CsUFO expression.

Distinct evolutionary trajectories of two tendril development genes, namely TL and TEN

Our evolutionary analyses indicated that, following the CucWGD1 event shared by all Cucurbitaceae species, one duplicated copy of CsTL was likely lost during (Fig. 6, A and B) and the other copy gained a specific function in tendril development that was different from its conserved function in regulating branch development in other plant species. In contrast, for TEN, which belongs to the CYC/TCP family, previous studies showed that, following CucWGD1, both copies of the ancestral CYC/TCP genes were retained. One copy, namely TEN, gained a function specifically in tendril development, and the function of the other duplicated CYC/TCP gene remains unknown (Guo et al. 2020; Zhang et al. 2022). Therefore, despite functioning in tendril development and by their direct interaction, genes encoding TEN and TL went through different evolutionary paths to achieve their current functions in tendril development.

Cucumber TL and its ClL homologs in watermelon exhibit differences in both expression patterns and functions

Through map-based cloning and gene editing in cucumber, we demonstrated that cucumber TL regulated the initiation of tendrils. In comparison, the loss of function of the CsTL homologous gene in watermelon, ClLs, led to the absence of all lateral organs, including branches, tendrils, and flowers (Jiang et al. 2023). In order to explore the underlying reason for the functional differences between CsTL and ClLs, we cloned the ∼2-kb promoter sequences for these two genes and used mVista to conduct a comparative sequence analysis by examining the upstream noncoding sequences of AtLAS and the cloned promoter sequences of ClLs and CsTL. Compared with the approximately 2-kb sequence upstream of AtLAS, the upstream region of ClLs demonstrates higher overall sequence similarity than that of CsTL (Supplementary Fig. S13A). We also identified more conserved noncoding sequences (CNSs) in the ClLs promoter than in CsTL promoter (Supplementary Fig. S13A). Upon comparing the upstream regions of AtLAS and CsTL with those of ClLs/LAS, we identified a region in the CsTL promoter (highlighted in the green box) that exhibited low sequence similarity with the ClLs promoter, whereas the corresponding region in ClLs showed much higher sequence similarity with the upstream region of AtLAS and contained a CNS region (Supplementary Fig. S13B).

Based on these results, we speculated that the disparities in the promoter regions between ClLs and CsTL could be the fundamental cause for the divergence in function and expression patterns observed between these two genes. To test this hypothesis, we constructed promoter–β-glucuronidase (GUS) reporter vectors using either the CsTL promoter (ProCsTL) or the ClTL promoter (ProClLs). Following genetic transformation in cucumber and watermelon using ProCsTL-GUS and ProClLs-GUS expression vectors, respectively, we observed a significantly higher promoter strength for watermelon Ls compared with cucumber TL. Before visible lateral organ formation, the GUS staining signal of watermelon ProClLs-GUS transgenic plants was concentrated in young leaves and leaf axils, whereas the cucumber ProCsTL-GUS transgenic plants showed only sporadic and weak GUS signals in the leaf axils (Supplementary Fig. S14, A, B, E, and F). After visible lateral organ formation, the GUS staining signal of watermelon ProClLs-GUS transgenic plants remained strong and was concentrated in the leaf axils and lateral organs, whereas the GUS staining signal of cucumber ProCsTL-GUS transgenic plants remained weak in the leaf axils and showed no signals in young tendrils and other young lateral organs (Supplementary Fig. S14, C, D, G, and H). These results may explain why knocking out ClLs in watermelon results in the absence of all lateral organs, whereas in cucumber, knocking out CsTL led only to the loss of tendrils.

The application of a strategy to customize cucumber varieties with ideal plant architecture for protected cultivation

In the protected cultivation of cucumbers, an abundance of branches and tendrils can compromise ventilation and reduce light transmission, necessitating significant time and labor costs for their removal (Liu et al. 2021). Moreover, the wounds resulting from the removal of branches and tendrils also elevate the risk of infection by harmful viruses and germs. For cucumbers cultivated for the fresh market in protected environments, in which fruits develop directly in leaf axils, plants without tendrils and branches would be ideal for time- and labor-saving purposes (Liu et al. 2021). Here, we demonstrated that CsTL is responsible for tendril formation by forming complexes with CsTEN to promote transcription of CsUFO, and also that CsREV conservatively regulates shoot branching in cucumber (Fig. 7B). Our results suggest a feasible strategy for breeding cucumber varieties lacking tendrils and branches by editing CsTL and CsREV, simultaneously (Fig. 7C).

Materials and methods

Plant materials and growth conditions

The cucumber (C. sativus L.) cultivar 9930 was used to develop the Tnt1 retrotransposon insertion mutant library (Zhang et al. 2018) and the resulting segregating F₂ population (340 individuals). The tendril-less mutant (tl) was found and investigated in the TN89 line during the screening of the cucumber Tnt1 transposon mutant library in the spring of 2021. The cucumber inbred line CU2 (South China type) was employed for genetic transformation in this study and represents the WT line. To initiate germination, plump seeds were soaked in water and then allowed to sprout in an incubator at 28°C for 36 to 48 h in darkness. The germinated cucumber seeds were sown into each plastic pot with general peat soil, and then, cucumber seedlings at the four-true-leaf stage were transplanted to a greenhouse at the Jingyang County Vegetable Industry Comprehensive Service Center where they were meticulously cultivated under strict water (watering schedule: during the seedling stage, water every 7 to 10 days; in the flowering phase, water every 5 to 7 days; and during the fruiting stage, water every 3 to 5 days), fertilizer management (using high-potassium fertilizer in quantities ranging from 10 to 15 kg per 667 square meters), and pest control (utilizing insect-resistant netting) regimes. The tmd1 mutant, which contains a mutation in CsUFO, was generously provided by Professor Zhujun Zhu (College of Horticulture Science, Zhejiang A&F University, Jinhua, Zhejiang, China). For analysis of transient gene expression, leaves from 4- to 6-week-old N. benthamiana plants (growing conditions: photon flux density, calibrated at 350 μmol m⁻²s⁻¹, was provided by RXZ-500D-LED lamps manufactured by Ningbo Jiangnan Instrument Factory, Ningbo, China; 16-h light/8-h dark, 24°C/18°C) were utilized.

Interspecific mapping population

Regional linkage analysis and fine mapping were conducted utilizing the variants identified from BSA-seq (Niu et al. 2018). Genome resequencing was undertaken to pinpoint sites of potential Tnt1 insertions, which were subsequently validated through PCR amplification. For genotyping in the F2 population, four Tnt1 insertion sites within the candidate interval were selected to design markers. Sequencing was carried out by Beijing Biomarker Technology Cooperation (Beijing, China).

Phylogenetic and syntenic analyses

The deduced amino acid sequences of genes encoding the GRAS family transcription factors of cucumber, tomato (S. lycopersicum), A. thaliana, and Amborella trichopoda were retrieved from the NCBI databases (https://www.ncbi.nlm.nih.gov/). These sequences were aligned using MAFFT, version 7 (Katoh et al. 2019), and the alignment was then used as the input for a maximum likelihood phylogenetic analysis using the IQ-TREE web server with 1,000 UltraFast bootstraps (http://iqtree.cibiv.univie.ac.at/) with the JTT + F + G4 model (Trifinopoulos et al. 2016) (Supplementary Files 1 and 2). Phylogenetic tree visualization was performed using iTOL v6 (https://itol.embl.de/). Syntenic analyses and visualization were performed using the MCScan pipeline in the JCVI python library (https://github.com/tanghaibao/jcvi).

Vector construction

To perform CRISPR/Cas9 gene editing of CsTL, CsTEN, and CsREV, we designed two specific sgRNA target sites for each gene using the website CRISPR-P (http://crispr.hzau.edu.cn/CRISPR2/). The Target1-(gRNA-Sc)-(U6–26t)-(U6–29p)-Target2 fragments for editing each gene were amplified using the pCBC-DT1T2 vector template and primers that contained the sgRNA sequences. Subsequently, these fragments were seamlessly integrated into the binary CRISPR/Cas9 vector pKSE402, using BsaI endonuclease and T4 DNA ligase (both from New England Biolabs, China). Vectors pKSE401 and pCBC-DT1T2 were kindly provided by Professor Qijun Cheng from China Agricultural University. The thermal incubation program for this procedure included steps at 37°C for 5 h, 50°C for 5 min, and 80°C for 10 min. This resulted in the generation of the pKSE402-CsTL, pKSE402-CsTEN, and pKSE402-CsREV gene editing vectors, as previously described (Xing et al. 2014). For the construction of the OE vector for CsTL, the full-length cDNA sequence of CsTL was amplified and cloned into pCAMBIA1305.4, yielding Pro35S:CsTL-NOS. The corresponding constructs were introduced into Agrobacterium tumefaciens strain GV3101 for Agrobacterium-mediated genetic transformation of cucumber as previously described (outlined below) (Hu et al. 2017). In this study, the gene encoding green fluorescent protein (GFP) was used as the reporter gene in the CRISPR/Cas9 vector and the OE vector. To generate the Pro35S:CsTL-GR construct, the sequence encoding the rat GR was PCR-amplified and incorporated into the existing OE vector Pro35S:CsTL. This procedure led to the creation of Pro35S:CsTL-GR-NOS.

Cucumber transformation

Cucumber seeds were immersed in warm water at 40 to 60°C for 1 h to facilitate the removal of the seed coat. Subsequently, peeled seeds were subjected to a sequence of treatments, including a 15-s rinse in 70% (v/v) ethanol and a 15-min immersion in a 6% (v/v) sodium hypochlorite solution. Following these steps, the seeds were washed with sterile distilled water and germinated on MS medium (Murashige and Skoog medium, containing 4.43 g/L of MS salts, 30 g/L sucrose, 2.6 g/L phytagel, pH 5.6 to 5.7) at 28°C in the dark for 48 h. Following germination, cotyledons were excised and exposed to a diluted A. tumefaciens GV3101 (AngYu Biotechnologies, China) culture. Subsequently, they were transferred to co-culturing medium (MS medium supplemented with 0.5 mg/L 6-benzylaminopurine and 1 mg/L ABA) for 3 days. Finally, the cucumber cotyledon explants were shifted to co-culturing medium supplemented with timentin (200 mg/L) for approximately 30 days to foster bud differentiation. Identification of shoots expressing GFP fluorescence was carried out through observation under a stereoscopic fluorescence microscope (MZ10F, Leica, Germany), and these shoots were then cultured on solid MS nutrient medium supplemented with 0.5 μM indole-3-acetic acid.

gDNA was extracted from transgenic plant leaves utilizing the cetyltriethylammnonium bromide method (Murray and Thompson 1980). Subsequently, specific PCR primers (Supplementary Table S3) were designed to amplify the sequences surrounding the target sites, which were validated in the inbred progeny plants. For comprehensive details on the primer names and their corresponding sequences, see Supplementary Table S3.

RNA in situ hybridization

In situ hybridization was performed as described previously (Zhang et al. 2013). The forward and reverse primers were designed based on sequences specific to CsTL, CsTEN, CsREV, and CsSTM, with the SP6 promoter included in the forward primer and the T7 promoter included in the reverse primer. In vitro transcription of RNA probes was performed with the DIG RNA Labeling Kit (Roche, Germany). Primer information is given in Supplementary Table S3. The terminal buds of CU2 seedlings at the four-leaf stage were fixed in 3.7% formalin–acetic acid–alcohol (FAA with 50% ethanol, 5% glacial acetic acid, and 3.7% formaldehyde), followed by sectioning (8 μm) and in situ hybridization.

X-ray micro-CT scanning

The terminal buds of WT plants, CsTL knockout plants, and HW plants at 30 days of age were fixed in a 10% (w/v) phosphotungstic acid solution in FAA for 7 days and stored at 4°C. After dehydration, the samples were subjected to CO₂ critical point drying (CPD) using a CPD 300 auto critical point dryer (Leica Microsystems, Germany) (Bellaire et al. 2014). The dried samples were then scanned using a micro-CT scanner (Bruker SkyScan 1172; Bruker Corp., Billerica, Massachusetts, USA) with a voltage of 50 kV and a current of 201 μA. Image stacks were generated using NR ECON v.1.6 for 3D reconstruction (Yao et al. 2019; Cheng et al. 2023).

RNA-seq

Samples consisting of 1-month-old shoot apexes were collected from both WT and CsTL knockout plants, with three biological replicates prepared for each sample. Total RNA extraction was carried out using the TRIzol reagent (Vazyme, China). RNA library construction and sequencing were conducted by Beijing Biomarker Technology Cooperation on the Illumina NovaSeq 6000 platform. Gene expression levels were estimated using fragments per kb per million reads values. The analysis of RNA-seq data was performed on the BMKCloud platform (http://www.biocloud.net). Utilizing transcriptome data, heat maps were generated with Origin software (version 2022) using normalized Z scores. DEGs were identified by the intersection of genes analyzed through DESeq2 (false discovery rate <0.05 and fold change ≥2) (Love et al. 2014). Detailed information on the sequencing data is provided in Supplementary Data Set 1.

RNA extraction and RT-qPCR

Total RNA was extracted from the shoot apexes of both transgenic plants and WT plants at the same growth stage using the RNA prep Pure Plant Kit (Tiangen, China) in accordance with the manufacturer's instructions. Subsequently, the RNA was reverse-transcribed into cDNA utilizing the PrimeScript RT Reagent Kit (Takara, China). RT-qPCRs were conducted using ChamQ SYBR qPCR Master Mix (Vazyme, China). Each sample in the RT-qPCR experiments involved three biological replicates and three technical replicates. Cucumber UBIQUITIN (UBI, Csa000874) served as the internal control for gene expression (Shen et al. 2019). Primer information is given in Supplementary Table S3.

Yeast one-hybrid assay

The full-length coding sequence of CsTL was amplified and subsequently cloned into pB42AD (Clontech, California, USA), resulting in the generation of effector constructs. Additionally, the promoter region of CsTEN (2,123-bp upstream of the ATG start codon) and the promoter region of CsUFO (1,980-bp upstream of the ATG start codon) were amplified using PCR and then cloned into the EcoRI and SalI sites of pLacZi2 (Clontech, California, USA) to create the reporter constructs. For subsequent analysis, specific combinations of effectors and reporters were co-transformed into the yeast strain EGY48, following previously established procedures (Li et al. 2010). The transformants were cultured on SD (minimal synthetic dropout medium)/-Trp/-Ura plates for 48 h. Subsequently, they were transferred to SD/-Trp/-Ura/Gal/Raf/X-Gal plates, which contained 5-bromo-4-chloro-3-indolyl-β-D-galactopyranoside at a concentration of 40 μg/mL, to induce the development of blue coloration. Detailed information regarding the primers is provided in Supplementary Table S3.

Yeast two-hybrid assays

The full-length coding sequence of CsTEN was cloned into pGBKT7 (Clontech, California, USA) to create the bait vector. Subsequently, it was co-transformed into the AH109 yeast strain (AngYu Biotechnologies, China) with the pGADT7 (Clontech, California, USA) activation domain vector, which carried the full-length coding sequence of CsTL. Concurrently, an empty pGADT7 vector was co-transformed with pGBKT7-CsTEN into the AH109 yeast strain to serve as a negative control. All resulting yeast strains were then plated on SD/-Leu/-Trp medium and cultured at 30°C. The colonies growing on SD/-Leu/-Trp medium were subsequently plated onto SD/-Ade/-His/-Leu/-Trp/X-alpha-Gal and incubated for 3 to 5 days at 30°C before observing the results. Primer information is given in Supplementary Table S3.

LCI assay

The complete coding sequence of CsTL (excluding the stop codon) was amplified and subsequently inserted into the modified plant expression vector pCAMBIA1300-cLUC (Chen et al. 2008). Simultaneously, the entire coding sequence of CsTEN (lacking the stop codon) was cloned into pCAMBIA1300-nLUC, positioned between the KpnI and SalI recognition sites (Chen et al. 2008; Zhang et al. 2017). Primer details are provided in Supplementary Table S3. Plasmid sequences were confirmed by Sanger sequencing before being introduced into A. tumefaciens strain GV3101. Subsequently, these strains were combined with a strain carrying the P19 silencing suppressor and co-infiltrated into the leaves of N. benthamiana (Zhou et al. 2018). Following 48-h incubation, the potassium salt of D-luciferin (AAT Bioquest, USA) was evenly applied to the posterior side of the infiltrated leaves in conditions shielded from light. Subsequently, LUC activity was assessed utilizing a plant-imaging system (CCD; Lumazone Pylon 2048B, USA).

BiFC assays

The coding sequence of CsTL was cloned into the BiFC vector pSPYNE (Waadt et al. 2008) and fused to the 3′ end of the gene encoding yellow fluorescence protein (YFP). Simultaneously, CsTEN was cloned into pSPYCE and fused to the 5′ end of sequences encoding YFP. These constructs were individually introduced into A. tumefaciens strain GV3101 and co-infiltrated into the abaxial side of N. benthamiana leaves to detect specific interactions, following a previously described methodology (Hu et al. 2017). Fluorescence signals were imaged under an excitation wavelength of 515 nm using a laser set at 15% power, with a gain of 750, and detected using an Olympus BX63 fluorescence microscope (Tokyo, Japan). The primer sequences utilized for the assays are listed in Supplementary Table S3.

Recombinant protein expression and purification from Escherichia coli

To isolate the recombinant GST-CsTL protein, the plasmid pGEX-GST-CsTL was introduced into E. coli (BL21) cells and introduced into E. coli Rosetta (DE3) cells (AngYu Biotechnologies, China), and protein expression was induced by adding 0.5 mM isopropyl-beta-D-thiogalactopyranoside, with shaking at 20°C for 18 h. GST-tagged proteins were isolated using affinity purification with glutathione-Sepharose 4B medium (Smart-Lifesciences, China), and the His-tagged proteins were purified using Ni-NTA agarose (Smart-Lifesciences, China). The pGEX-GST and pET-30a (His) vectors were provided by Qingmei Guan of Northwest A&F University, Shaanxi, China. Detailed primer information is provided in Supplementary Table S3.

In vitro pull-down assays

For GST protein pull-down assays, cell lysates in GST pull-down buffer (0.1% NP-40 in phosphate buffered saline (PBS) buffer: 140 mM NaCl, 2.7 mM KCl, 10 mM Na₂HPO₄, 1.8 mM KH₂PO₄, pH 7.3) were incubated with GST-fused proteins for 2 h at 4°C. Subsequently, glutathione beads were washed four times with PBS. Immunoprecipitated GST-CsTL was then incubated with His:CsTEN in the pull-down buffer at 4°C for 2 h, followed by six washes with PBS and separation by 12% SDS–PAGE. To test for protein–protein interactions, immunoblotting was carried out using antibodies against GST (TransGen, China, dilution 1:1000) and His (TransGen, China, dilution 1:1000). GST beads alone were used as a negative control. Primer information is provided in Supplementary Table S3.

Dual-LUC assay

The CsTL coding sequence was incorporated into pGreenII 62-SK (Sangon Biotech Co.) via the homologous recombination method (Hellens et al. 2005). Simultaneously, the 1,980-bp promoter of CsUFO, the 2,609-bp promoter of CsREV, and the 2,123-bp promoter of CsTEN were cloned into the pGreenII 0800-LUC vector (Sangon Biotech Co.) using a similar approach. Subsequently, both constructs were introduced into A. tumefaciens GV3101 (pSoup-p19; AngYu Biotechnologies, China) and co-infiltrated into the leaves of N. benthamiana plants. Sequences of the primers used for generating these vectors are found in Supplementary Table S3. After an incubation period of 48 h, total proteins were extracted from the leaves using the dual-luciferase reporter gene assay kit (Yeasen, China) following the manufacturer's instructions. The relative LUC activity (fLUC/rLUC) was measured using the Infinite M200pro full-wavelength microplate reader (Tecan, Switzerland).

Co-IP assay

The full-length coding sequence of CsTL (excluding the stop codon) was cloned into pBinGFP2 to create a construct encoding GFP-tagged CsTL. Simultaneously, the full-length coding sequence of CsTEN (excluding the stop codon) was cloned into the pSUPER1300-MYC vector to generate Myc-tagged CsTEN. The primer sequences are provided in Supplementary Table S3. These plasmids were individually introduced into A. tumefaciens strain GV3101 and subsequently co-infiltrated into 5-week-old N. benthamiana leaves along with an Agrobacterium strain containing P19 (Zhou et al. 2018). Following 48 h of incubation, the samples were ground into a fine powder in liquid nitrogen in an extraction buffer [0.05 M HEPES (pH 7.5), 0.15 M KCl, 1 mM EDTA, 0.3% (v/v) Triton X-100, 1 mM DTT, and 1× EDTA-free protease inhibitor cocktail (Roche, Basel, Switzerland)]. The immunoprecipitated samples were then separated on 12% SDS–PAGE gels and detected using an anti-GFP antibody (66002-1-lg; Proteintech, USA, dilution 1:20,000). The pBinGFP2 and pSUPER1300-MYC vectors were provided by Qingmei Guan of Northwest A&F University.

DEX and CHX treatments

The DEX (D4902, Sigma-Aldrich) and CHX (01810, Sigma-Aldrich) treatments were conducted following procedures outlined in previous research (Sun et al. 2019; Krizek et al. 2020). Terminal buds of Pro35S:CsTL-GR transgenic plants were immersed in a MS solution containing one of the following treatments: mock (0.1% DMSO + 0.015% Silwet), DEX (10 µM DEX/0.1% DMSO + 0.015% Silwet), CHX (10 µM in 0.1% DMSO + 0.015% Silwet), or DEX + CHX (10 µM DEX + 10 µM CHX in 0.1% DMSO + 0.015% Silwet).

Promoter cloning and comparative analysis

Sequences (∼2 kb) upstream of the ATG start codons of CsTL (from CU2 cucumber) and ClLs (from “TC” watermelon) were amplified with PCR. The upstream region for AtLAS was aligned with the cloned CsTL promoter sequence and the ClLs promoter sequence using Shuffle-LAGAN (Brudno et al. 2003) in the mVista webserver (https://genome.lbl.gov/vista/mvista/submit.shtml), and their sequence similarities were examined with a calculation window size of 100 bp. CNSs among these sequences were identified using the parameter of 20 bp size and 70% identity.

GUS staining

To assess GUS activity, terminal buds from ProCsTL-GUS and ProClLs-GUS transgenic plants were submerged in a GUS staining solution (Coolaber, China) according to the manufacturer's instructions. Images were captured using a stereoscopic fluorescence microscope (MZ10F, Leica, Germany).

Statistical analysis

Student's t-tests were performed using GraphPad Prism 8 software (GraphPad Software Inc.) to determine significance when comparing groups. Statistical significance was inferred at the specified P-values noted for each experiment. Additional details of the statistical analyses are provided in Supplementary Data Set 2.

Accession numbers

The raw sequencing data were deposited at the NCBI under the accession number PRJNA1086127. The accession numbers for genes in this study are CsTL (CsaV3_3G003590), CsTEN (CsaV3_5G039160), CsUFO (CsaV3_1G010110), and CsREV (CsaV3_6G013930).

Author contributions

Z.L. and J.S. designed and supervised this study; J.S., Y.J., J.P., L.S., Q.L., W.H., P.S., B.Z., H.Z., X.K., Y.G., and T.Y. performed experiments and analyzed the data; and Z.L., J.S., L.S., and Y.J. wrote and revised the paper.

Supplementary data

The following materials are available in the online version of this article.

Supplementary Figure S1. Genomic PCR confirmation of Tnt1 insertion in CsTL gene in tl mutant. (Supplementary Fig. 1).

Supplementary Figure S2. Phylogenetic analysis of GRAS family genes of cucumber, Arabidopsis, tomato, and Amborella. (Supplementary Fig. 1).

Supplementary Figure S3. In situ hybridization analysis of CsTL and CsSTM expression in cucumber shoot apical meristem (SAM) tissue. (Supplementary Fig. 1).

Supplementary Figure S4. Phenotypic characterization of CsTL knockout lines. (Supplementary Fig. 2).

Supplementary Figure S5. Observation of the distribution of lateral organs in different cucumber plants and molecular characterization of CsTL OE plants. (Supplementary Fig. 2).

Supplementary Figure S6. Phenotypic characterization of CsTEN knockout lines. (Supplementary Fig. 3).

Supplementary Figure S7. Schematic diagram illustrating the deletion of a 1,581-bp fragment in the promoter region of CsUFO in the tmd1 mutant (CsUFO-m). (Supplementary Fig. 3).

Supplementary Figure S8. Analysis of the expression pattern of CsTEN in shoot apices of the cucumber inbred line CU2 was conducted using RNA in situ hybridization. (Supplementary Fig. 4).

Supplementary Figure S9. Verification of the relationship between CsTL, CsTEN, and CsUFO. (Supplementary Fig. 4).

Supplementary Figure S10. CsTL can induce the activity of the CsREV promoter. (Supplementary Fig. 5).

Supplementary Figure S11. Expression analysis of CsREV and CsTL in transgenic plants. (Supplementary Fig. 5).

Supplementary Figure S12. Phenotypic characterization of CsREV knockout lines. (Supplementary Fig. 5).

Supplementary Figure S13. Comparison of upstream noncoding sequences of AtLAS, ClLs, and CsTL with mVista. (Supplementary Fig. 6).

Supplementary Figure S14. Utilizing CsTL and ClLs promoters for GUS expression in cucumber and watermelon, respectively. (Supplementary Fig. 6).

Supplementary Table S1. Genetic analysis of F2 population.

Supplementary Table S2. Amino acid sequence identity between cucumber TL, tomato Ls, Arabidopsis LAS, and rice MOC1 proteins.

Supplementary Table S3. Primers used in this study.

Supplementary Data Set 1. RNA-seq assay of differentially expressed genes between the shoot apexes of CsTL knockout lines and WT plants.

Supplementary Data Set 2. Results of Student’s t-tests of quantitative data.

Supplementary File 1. GRAS family genes from cucumber, Arabidopsis, tomato, and Amborella were utilized to construct the phylogenetic tree.

Supplementary File 2. Multiple protein sequences were used to construct the phylogenetic tree.

Funding

This work was supported by the National Key Research and Development Program of China (grant number 2022YFD1602000), the National Natural Science Foundation of China (32202519, U22A20498, U23A201593, and 32072596), the Key Research and Development Program of Shaanxi (2023-YBNY-011), and the Technology Innovation Team of Shaanxi (2021TD-32).

Data availability

All relevant data are included in the article and Supplementary materials.

Dive Curated Terms

The following phenotypic, genotypic, and functional terms are of significance to the work described in this paper:

• CPD Gramene: AT5G05690

• CPD Araport: AT5G05690

• AAT Gramene: AT2G22250

• AAT Araport: AT2G22250

• TCP1 Gramene: AT1G67260

• TCP1 Araport: AT1G67260