Natural Variation in CYCLIC NUCLEOTIDE‐GATED ION CHANNEL 16 Controls Fruit Weight in Tomato

ABSTRACT

Fruit weight is a crucial agronomic trait influencing yield and quality in tomato domestication. Although numerous quantitative trait loci (QTLs) controlling fruit weight have been reported, the underlying molecular mechanisms remain largely unresolved. Here, we identify natural variation in the promoter region of CYCLIC NUCLEOTIDE‐GATED ION CHANNEL 16 (CNGC16) associated with both fruit weight and gene expression differences via genome‐wide association study (GWAS) in tomato accessions. Geographical and subpopulation analyses reveal that the large‐fruit CNGC16 haplotype was selectively favoured during domestication and may contribute to local environmental adaptation. Functional assays demonstrate that CNGC16 regulates fruit weight by promoting cell division and maintaining cytosolic Ca²⁺ homeostasis in fruit cells. Mechanistically, CNGC16 physically and genetically interacts with calmodulin 6 (CaM6) to mediate calcium signalling that controls fruit size. These findings uncover a novel calcium‐dependent pathway regulating tomato fruit weight and propose CNGC16 and CaM6 as promising molecular targets for crop improvement.

1. Introduction

Fruit weight and size are primary determinants of yield and quality in many cultivated vegetable crops (Rowland et al. 2020). Unravelling the genetic, molecular, and physiological bases of fruit weight regulation is essential for targeted crop improvement.

Tomato ( Solanum lycopersicum L.) is among the most widely cultivated vegetable crops worldwide and serves as both a rich source of vitamins and health‐promoting compounds and a model organism for genetic, developmental, and physiological studies (Gao et al. 2019; Li et al. 2023; Tomato Genome, C 2012; Zhang et al. 2024). Over the past decade, approximately 30 quantitative trait loci (QTLs) associated with fruit weight have been identified (Tanksley 2004). However, only a few genes corresponding to key QTLs—fw2.2, fw3.2, fw11.3, locule (lc), globe, and fascinated (fas)—have been cloned (Frary et al. 2000; Barrero and Tanksley 2004; Cong et al. 2008; Munos et al. 2011; Sierra‐Orozco et al. 2021; Chakrabarti et al. 2013). Tomato fruit development proceeds through four main stages: fruit set, cell division, cell expansion, and ripening (Gillaspy et al. 1993), with final fruit size predominantly determined by cell division and expansion (Mauxion et al. 2021). Specifically, fw2.2 and fw3.2 regulate cell number (Chakrabarti et al. 2013; Tran and Billakurthi 2024), while fw11.3 influences cell size (Mu et al. 2017). The fas and lc loci control locule number, and globe affects fruit shape (Munos et al. 2011; Barrero and Tanksley 2004; Cong et al. 2008; Sierra‐Orozco et al. 2021). Despite these advances, the molecular mechanisms by which these QTLs regulate fruit weight remain incompletely understood.

Recent reverse genetic studies have uncovered several regulatory networks influencing tomato fruit size. The conserved WUSCHEL‐CLAVATA (WUS‐CLV) feedback loop modulates floral meristem size, increasing locule number and hence fruit size (Xu et al. 2015). F‐box proteins STERILE APETALA1 (SAP1) and SAP2, components of an Skp1‐Cullin‐F‐box (SCF) complex, target KINASE‐INDUCIBLE DOMAIN INTERACTING (KIX8) and KIX9 for degradation, thereby regulating fruit size (Cook et al. 2025; Li et al. 2025). The jasmonic acid regulator pathway MYC2 activates the expression of lateral organ boundaries domain 40 (LBD40), which induces the cell expansion gene EXPANSION 5 (EXPA5) to promote cell enlargement (Liu et al. 2024). The Brassinazole Resistant 1.3 (BZR1.3) transcription factor regulates tomato fruit weight by recruiting the H3R2me2a demethylase Jumonji C‐domain‐containing protein 10 (JMJ10) to demethylate histones at target loci, indicating a pivotal role for histone arginine methylation in size control (Zeng et al. 2025). Furthermore, light signalling influences fruit size via the E3 ubiquitin ligase TLFP8 and the light signalling suppressor PIF1a, which mediates degradation of CDKB2, a key protein in fruit development (Zhang et al. 2025). These studies illustrate the integration of environmental and genetic signals in fruit size regulation.

Calcium (Ca²⁺) functions as a crucial secondary messenger in eukaryotic cells, playing vital roles in plant development and environmental responses. Cellular Ca²⁺ levels are tightly regulated by influx through plasma membrane channels and release from internal stores, followed by removal via exchangers and ATPase to maintain low resting cytosolic concentrations essential for signalling fidelity (Cook et al. 2025; Tan et al. 2020; Tang et al. 2020; Tian et al. 2019, 2020; Wang, Song, et al. 2024; Clapham 2007). Ca²⁺ signals are decoded by sensors, such as calmodulin (CaM), Ca²⁺‐dependent protein kinases (CDPKs), and calcineurin B‐like proteins (CBLs) with their interacting kinases (CIPKs) (Pan et al. 2019; Tang et al. 2020; Zhou et al. 2014). Cyclic nucleotide‐gated channels (CNGCs) mediate Ca²⁺ influx and are integral for plant development (Cui et al. 2020; Dietrich et al. 2020; Wang et al. 2017). Loss of CNGC function disrupts calcium oscillations critical for cellular responses (Cui et al. 2020; Tian et al. 2019). In Arabidopsis, CNGC18/8 interaction with CaM2 drives oscillatory channel activity essential for pollen tube growth (Pan et al. 2019), while CPK1 activates CNGC5/6/9 to promote root hair development (Zhu et al. 2025). The CNGC2/4 complex mediates pathogen‐associated molecular patterns (PAMPs)‐triggered calcium influx and defence activation (Tian et al. 2019). Natural variation in CNGC4 influences vegetative phase changes via Ca²⁺‐modulated miR156 expression in Arabidopsis (Wang, Song, et al. 2024). Nevertheless, the role of CNGCs in maintaining basal Ca²⁺ homeostasis and broader developmental processes is not fully elucidated.

In this study, we performed a genome‐wide association study (GWAS) on tomato fruit weight and identified significant genetic variation in the promoter region of CNGC16 correlated with fruit weight variation through differential gene expression among tomato accessions. Overexpression of CNGC16 increased fruit weight, while loss‐of‐function mutants showed smaller fruits and increased resting Ca²⁺ levels. We further demonstrate that CaM6 physically and genetically interacts with CNGC16 in regulating fruit weight. Our findings also indicate that CNGC16 underwent selection during tomato domestication and exhibits geographical distribution patterns. Collectively, this work uncovers a novel role for the Ca²⁺ signalling pathway in fruit weight regulation and highlights CNGC16 and CaM6 as promising targets for molecular breeding to improve tomato yield and quality.

2. Results

2.1. The CNGC16 Gene Is Associated With Fruit Weight

In a previous study, a genome‐wide association study (GWAS) was conducted on fruit weight using a germplasm collection of 163 tomato accessions representing the geographical, morphological, and genetic diversity of cultivated tomatoes and its closest wild relatives from their regions of origin (Mata‐Nicolas et al. 2020). This study identified a significant association peak near the fw2.2 locus and proposed two allelic variants as candidate genes with substantial effects, although their functions were not experimentally validated. To uncover novel genes involved in fruit weight regulation, we reanalyzed this germplasm collection by performing a GWAS with improved detection parameters. Besides the significant peak on chromosome 2, we detected a total of 352 significant polymorphisms distributed across 8 chromosomes, including two previously reported QTLs, fw7.2 and fw11.3, indicating improved detection efficiency (Figure S1; Table S1).

The leading SNP (Chr02:50688169) we identified on chromosome 2 co‐localizes with a previously reported significant association peak, which is 1.56 Mb away from the fw2.2 gene (Mata‐Nicolas et al. 2020). We identified a candidate gene, Solyc02g088560, positioned approximately 80 kb from the leading SNP (Figure 1a), encoding a CYCLIC NUCLEOTIDE GATED ION CHANNEL 16 (CNGC16). Notably, Arabidopsis mutants of the CNGC16 homologue exhibited shorter siliques in our prior study (Figure S2) (Wang, Song, et al. 2024). We detected 18 polymorphisms within CNGC16 (Figure 1b), of which 12—including 9 SNPs and 3 insertions/deletions—reside in the promoter region (Figure S3). Linkage disequilibrium analysis revealed a significant Linkage Disequilibrium (LD) block (R² > 0.7) among these polymorphisms (Figure 1b), allowing us to define two major haplotypes. Accessions carrying Hap1 showed significantly larger fruit weights compared to those with Hap2 (Figure 1c). These results suggest a robust association between natural variation in CNGC16 and fruit weight in tomato.

FIGURE 1.

Natural variation in CNGC16 correlates with fruit weight and its expression levels. (a) Manhattan plot depicting the significant genomic peak associated with fruit weight on chromosome 2. The red dashed lines delineate the most significant SNP of CNGC16. The black dashed line and red label indicate the distance from the leading SNP to the SNP (Chr02:50612226) in CNGC16. (b) Linkage disequilibrium (LD) plot for SNPs associated with fruit weight surrounding the peak in the CNGC16 region. LD analysis was performed using TASSEL 5.2, with the colour scale (white to red) representing the LD value (R²). The black dashed box highlights SNPs within the LD block. (c) Dot plots displaying fruit weight distribution between two haplotypes (Hap1 and Hap2). Statistical differences were assessed using Welch's t‐test. (d) Relative expression levels of CNGC16 in Hap1 and Hap2 accessions, represented as mean ± SD from three biological replicates. (e) Schematic of the reporter and control constructs employed in the dual‐luciferase assay. LUC denotes firefly luciferase; REN refers to Renilla luciferase. (f) Impact of promoter polymorphisms on CNGC16 activity. Constructs pCNGC16 ^Hap1 ::LUC and pCNGC16 ^Hap2 ::LUC were transiently expressed in tobacco leaves. Relative reporter activity (LUC/REN) was calculated. Points represent mean values from three technical replicates across fifteen independent biological replicates. ‘**’ indicate a significant difference between samples (Student's t‐test, p < 0.05).

2.2. Natural Variation in the Promoter Region Confers CNGC16 Expression Differences

To investigate how promoter variations influence CNGC16 expression, we analysed RNA‐seq data from developing tomato fruit samples of two contrasting genotypes, MM (Hap1) and PP (Hap2), collected in our laboratory. The MM (Hap1) accessions showed consistently higher CNGC16 expression levels at all developmental stages examined (Figure S4a). Furthermore, CNGC16 exhibited elevated expressions in flowers and fruits (Figure S4b). This was corroborated by RT‐qPCR analysis of 10 Hap1 accessions with large fruits and 10 Hap2 accessions with small fruits, confirming significantly higher CNGC16 expression in Hap1 accessions (Figure 1d).

To directly assess the impact of promoter variants on transcriptional activity, we cloned the promoter regions from Hap1 and Hap2 into a luciferase reporter system (Figure 1e). Each promoter was fused upstream of the firefly luciferase (LUC) gene, with Renilla luciferase (Ren) under the constitutive cauliflower mosaic virus (CaMV) 35S promoter serving as an internal control for normalization. The constructs were transiently expressed in tobacco leaves, followed by RT‐qPCR quantification. The Hap1 promoter drove a 2.4‐fold higher luciferase expression than the Hap2 promoter (Figure 1f), indicating elevated transcriptional activity associated with the Hap1 allele.

Collectively, these findings demonstrate that natural variation in the CNGC16 promoter contributes to differential expression levels across tomato accessions, which correlate with fruit weight variation.

2.3. Overexpression of CNGC16 Increases Fruit Size

To investigate whether CNGC16 expression influences fruit weight in tomato, we generated transgenic lines overexpressing CNGC16 under the constitutive 35S promoter in the Ailsa Craig (AC) background. Two independent lines, CNGC16‐OE#4 and CNGC16‐OE#7, showed significantly elevated CNGC16 transcript levels in 10‐day‐old fruits compared to wild‐type controls (Figure 2a). Consistent with our hypothesis, both overexpression lines exhibited marked increases in fruit weight, along with larger vertical and transverse fruit diameters relative to wild‐type plants (Figure 2b,e). These results demonstrate that the ectopic overexpression of CNGC16 enhances tomato fruit size and overall weight.

FIGURE 2.

CNGC16 controls fruit size in tomato. (a) RT‐qPCR analysis showing CNGC16 expression levels in two independent overexpression lines (OE4 and OE7). (b) Fruit size phenotypes of wild‐type (WT), OE4, and OE7 plants. (c–e) Quantification of single fruit weight, vertical diameter, and transverse diameter for WT, OE4, and OE7 plants (n = 15 fruits, mean ± SD). (f) Fruit size phenotypes of WT and CNGC16 mutants. (g–i) Quantification of single fruit weight, vertical diameter, and transverse diameter for WT and CNGC16 mutant plants (n = 15 fruits, mean ± SD). All experiments were conducted with three independent biological replicates. Values are presented as mean ± SD across all datasets. Statistical differences from WT were assessed using Student's t‐test; ‘**’ indicates significant differences from WT at p < 0.01.

2.4. Loss of CNGC16 Decreases Fruit Size

To further elucidate the function of CNGC16 in fruit weight regulation, we generated loss‐of‐function mutants in the AC background using CRISPR‐Cas9 genome editing. We isolated three independent mutant alleles: cngc16‐1, harbouring a 1 bp insertion at 181 bp downstream of the start codon, resulting in premature translation termination; and cngc16‐2 and cngc16‐3, each containing a 1‐bp insertion at 181 bp and 173 bp respectively, predicted to produce truncated proteins starting at amino acids 61 and 59 (Figure S5). Under field conditions, all three mutants exhibited significantly reduced fruit weight compared to the wild type, with average fruit weights of 32.89 g, 32.90 g and 32.79 g for cngc16‐1, cngc16‐2 and cngc16‐3, respectively (Figure 2f,g), versus 35.18 g for wild type. The mutants also showed marked decreases in both vertical and transverse fruit diameters (Figure 2h,i), confirming that CNGC16 plays an essential role in determining fruit size and weight.

Cross‐sections of fruits along the equatorial plane revealed that CNGC16 overexpression exhibited increased pericarp thickness, while loss‐of‐function mutants showed reduced pericarp compared to wild type (Figure S6a,b). To determine whether CNGC16 influences fruit size, cell division, or expansion, we quantified pericarp cell layers and individual cell sizes after fruit growth ceased. The average pericarp cell area was similar among CNGC16‐OE#4, wild type and cngc16‐1 (Figure S6a,d). However, the number of pericarp cell layers was significantly increased in CNGC16‐OE#4 and decreased in cngc16‐1 (Figure S6a,c), indicating that CNGC16 regulates fruit size primarily through modulation of cell division.

2.5. Natural Variation in CNGC16 Expression Confers Fruit Size Differences

To further validate the functional impact of CNGC16 promoter haplotypes, we amplified the 3 k promoter regions of CNGC16 from Hap1 (BGV006208) and Hap2 (BGV012640) lines and fused them with the CNGC16 coding sequence from AC. These constructs were transformed into the cngc16‐1 mutant background. Two independent CNGC16 ^Hap1 lines with elevated expression levels (about 4.2 and 3.9‐fold higher than wild type) produced significantly larger fruits than wild type (Figure 3a,b), with average fruit weights of 41.21 g, 40.95 for CNGC16 ^Hap1 (H1#1) and (H1#2), respectively, compared to 37.71 g for wild type (Figure 3c). These lines also exhibited a significant increase in vertical and transverse fruit diameters (Figure 3d,e). In contrast, two independent CNGC16 ^Hap2 lines with expression levels comparable to wild type produced fruits of similar size to wild type plants (Figure 3a–e). These results further confirm that natural variation in the CNGC16 promoter underlies differences in gene expression and fruit size among tomato accessions.

FIGURE 3.

Natural variation in the CNGC16 promoter contributes to fruit size variation. (a) RT‐qPCR analysis of CNGC16 expression in WT and transgenic lines expressing CNGC16 ^Hap1 and CNGC16 ^Hap2 in the cngc16‐1 mutant background (H1#1, H1#2, H2#1, H2#2). The Actin gene (Solyc03g078400) was used as the internal control. (b) Representative fruit phenotype of WT, cngc16, and independent transgenic lines (H1#1, H1#2, H2#1, and H2#2). (c–e) Quantification of fruit weight (c), vertical diameter (d), and transverse diameter (e) for the indicated genotypes (n = 15 fruits per genotype, mean ± SD). Statistical significance was assessed by one‐way ANOVA followed by Duncan's multiple range test (p < 0.05).

2.6. Geographical Distribution of the CNGC16 Alleles

Since CNGC16 was identified as the causal gene underlying a fruit weight QTL, we further analysed the distribution of its two major haplotypes—Hap1 and Hap2—across 163 sequenced tomato accessions to explore their roles in tomato domestication. Geographic distribution analysis revealed a non‐random pattern, with Hap1 and Hap2 exhibiting distinct spatial distributions (Figure 4a). Hap1 is predominantly present at higher latitudes (Figure 4b) and lower longitudes (Figure 4c) compared to Hap2, suggesting that local environmental factors or human selection may have maintained the Hap2 allele in specific regions.

FIGURE 4.

Geographic and population distribution of CNGC16 haplotypes. (a) Spatial distribution of the two major CNGC16 haplotypes across geographic locations. (b) Boxplots showing the latitude ranges associated with each haplotype. (c) Boxplots showing the longitude ranges associated with each haplotype. Boxplots represent median, interquartile range (first and third quartiles), and individual data points per accession. ‘**’ denotes significant differences at p < 0.01 (Welch's t‐test). (d) Pie charts representing the frequencies of CNGC16 haplotypes within different populations. (e) Frequency distribution of CNGC16 haplotypes across subpopulations. (f) Composition of the 154 analysed tomato accessions: 26  S. pimpinellifolium (SP), 113  S. lycopersicum var. cerasiforme (SLC), and 15  S. lycopersicum (SLL).

Examining accession origins, Hap2 is primarily found in accessions from Peru, Colombia and Ecuador (Figure 4d) – areas considered centres of tomato domestication. This geographic clustering implies that selective pressures, possibly environmental or cultural, have shaped the distribution of CNGC16 alleles.

To evaluate selection during domestication, we investigated allele frequencies in Solanum pimpinellifolium (SP), Solanum lycopersicum var. cerasiforme (SLC) and Solanum lycopersicum var. lyco persicum (SLL). As expected, all SLL accessions carried the large‐fruit haplotype Hap1, whereas Hap1 frequencies were 36% in SP and 91% in SLC, indicating strong selection for Hap1 during the transition from wild to cultivated tomato (Figure 4e,f). These findings suggest preferential selection of the CNGC16 Hap1 allele associated with increased fruit weight during domestication.

2.7. CNGC16 , FW11.3 and FW7.2 Function Synergistically to Regulate Fruit Size

To explore relationships among the FW11.3, FW7.2 and CNGC16 loci (Figure S7a) in regulating fruit size, we performed haplotype analysis, identifying large‐fruit alleles at these loci as T, T and C, respectively (Figure S7b–d). These loci formed four major genotypes (Figure S7e, excluding missing/heterozygous genotypes), with the large‐fruit haplotype combination producing the largest fruits, and the small‐fruit CCA haplotype associated with the smallest fruits (Figure S7f). These results indicate that the three loci are associated to some extent with fruit enlargement and represent promising markers for breeding programs aimed at fruit size improvement.

2.8. Loss of CNGC16 Increases Cytosol Ca²⁺ in Fruit Cells

CNGCs function as calcium‐permeable channels integral to plant stress responses (Cui et al. 2020; Finka et al. 2012). To determine whether CNGC16 regulates cytosolic calcium in tomato fruits, we generated transgenic lines expressing the Ca²⁺ sensor protein Yellow Cameleon 3.6 (YC3.6) for in vivo monitoring of cytosolic Ca²⁺ dynamics (Iwano et al. 2024; Miyawaki et al. 2013). The YC3.6 construct was introduced into cngc16‐1, and isogenic wild‐type and cngc16‐1 lines expressing YC3.6 were established.

Confocal imaging at 15 days post anthesis (DPA) revealed significantly elevated cytosolic Ca²⁺ levels in the cngc16‐1 mutant relative to wild type as indicated by increased ratiometric YC3.6 signals (Figure 5a,b). These results indicate that CNGC16 is critical for maintaining cytosolic Ca²⁺ homeostasis in fruit cells.

FIGURE 5.

Elevated resting cytosolic calcium concentration [Ca²⁺]cyt in the cngc16‐1 mutant. (a) Representative fluorescence images showing the FRET–YFP to CFP ratio from the YC3.6 calcium reporter in fruit cells of wild‐type (WT) and cngc16‐1 plants. (b) Quantification of FRET ratios in fruit cells from WT and cngc16‐1 plants. (c–d) RNA‐seq‐based transcript levels of CAX3 (Solyc06g006110) (c) and CNGC13 (Solyc05g050380) (d) in cngc16 mutants compared to WT. (e–f) RT–qPCR analysis showing relative expression of CAX3 (e) and CNGC13 (f) in cngc16 mutants compared to WT. Data represent mean ± SD from three independent biological replicates. Statistical significance compared to WT was assessed by Student's t‐test; ‘**’ indicates significant differences at p < 0.01.

Interestingly, despite the canonical roles of CNGCs in mediating calcium influx (Cui et al. 2020; Tian et al. 2019), the loss of CNGC16 paradoxically resulted in increased cytosolic Ca²⁺. Similar phenomena have been reported for cngc4 and vacuolar calcium transporters CAX1 and CAX3 mutants, in which impaired vacuolar sequestration elevates cytosolic Ca²⁺ (Conn et al. 2011; Poethig 2024; Wang, Song, et al. 2024; Punshon et al. 2012).

To elucidate the underlying mechanism, RNA‐sequencing analyses of wild‐type and cngc16‐1 indicated strong downregulation of CAX3 and significant upregulation of CNGC13 in cngc16‐1 (Figure 5c,d). These expression changes were validated by RT‐qPCR in 10 DPA fruits (Figure 5e,f). Collectively, our data suggest that loss of CNGC16 elevates cytosolic Ca²⁺ potentially through suppression of vacuolar calcium transporters like CAX3 and compensatory upregulation of plasma membrane Ca²⁺channels such as CNGC13, revealing a complex calcium homeostasis network regulated by CNGC16.

2.9. CNGC16 Physically Interacts With CaM6

A previous yeast two‐hybrid screening identified CNGC16 as a potential CaM6‐interacting protein (Lin et al. 2023). To validate this interaction, we employed multiple complementary approaches. First, CNGC16 and CaM6 were fused to the N‐terminal and C‐terminal fragments of yellow fluorescent protein (YFP), respectively, and transiently expressed in Nicotiana benthamiana leaf epidermal cells. YFP fluorescence was observed on the plasma membrane when both proteins were co‐expressed (Figure 6a), suggesting interaction at the plasma membrane consistent with their known subcellular localizations (Figure S8) (Cui et al. 2020; Lin et al. 2023).

FIGURE 6.

CaM6 physically interacts with CNGC16. (a) Bimolecular fluorescence complementation (BiFC) assay showing the interaction between CaM6 and CNGC16. (b) Yeast two‐hybrid assays confirming the interaction between CaM6 and CNGC16. SM–LW: Synthetic medium lacking Leu and Trp; SM–L–W–H–A: Synthetic medium lacking Leu, Trp, His, and Ade. (c) Co‐immunoprecipitation (Co‐IP) assay demonstrating interaction of CaM6 and CNGC16 in planta. Western blots show total protein extracts (Input) and immunoprecipitated (IP) fractions. Proteins were immunoprecipitated using GFP‐conjugated beads and blotted with ani‐Myc and anti‐GFP antibodies. (d) AlphaFold3‐predicted CaM6‐CNGC16 complex structure (CaM6 in Green, CNGC16 in Plum). Red dotted lines indicate hydrogen bonds. iPTM, interface predicted TM‐score; pTM, predicted TM‐score. HDOCK identified 14 interacting amino acid pairs (Table S3).

Next, a dual membrane yeast two‐hybrid (Y2H) assay further confirmed the interaction. CNGC16 and CaM6 coding sequences were cloned into the prey vector pPR3‐N and the bait vector PBT3‐SUC, respectively, and co‐transformed into the yeast strain NMY51. While all yeast cells harbouring both constructs grew on SD‐Leu‐Trp medium, only those containing CNGC16‐pPR3‐N and CaM6‐pBT3‐SUC grew on the selective medium SD‐Leu‐Trp‐His‐Ade (Figure 6b), indicating a specific protein–protein interaction. The interaction was further corroborated by co‐immunoprecipitation (Co‐IP) assays. CNGC16‐GFP and CaM6‐Myc proteins were transiently expressed in N. benthamiana. CaM6‐Myc was only detectable when co‐expressed with CNGC16‐GFP (Figure 6c), confirming their association in planta. To explore interaction interfaces, AlphaFold 3 and HDOCK predicted the CaM6‐CNGC16 complex structure, yielding high interaction confidence scores (iPTM = 0.67, pTM = 0.49) and identifying 14 residue pairs forming hydrogen bonds (Figure 6d and Table S3). Collectively, these results demonstrate direct physical interaction between CNGC16 and CaM6 both in vitro and in vivo.

2.10. CaM6 Increases Fruit Weight in Tomato

To assess CaM6's role in fruit weight regulation, we generated 15 independent CaM6 overexpression lines and selected two lines (CaM6‐OE#3 and CaM6‐OE#5) with markedly elevated expression for phenotypic analysis (Figure 7a). Both overexpression lines exhibited significantly increased fruit weight and size compared to WT (Figure 7b–e), with average fruit weight of 43.23 g and 44.02 g versus 39.52 g in WT. Vertical and transverse fruit diameters were also significantly larger in CAM6‐overexpressing plants (Figure 7d,e). Conversely, two CaM6 knockout mutants, cam6‐3 and cam6‐12, obtained from a previous study (Wang, Ge, et al. 2024), displayed significant reduction in fruit weight and size compared to WT (Figure 7f–i), with average fruit weights of 33.32 g and 34.12 g versus 37.21 g in WT (Figure 7f,g). Both mutants exhibited notable decrease in vertical (Figure 7h) and transverse (Figure 7i) fruit diameters. Histological analysis revealed that CaM6 overexpression increased pericarp thickness, whereas CaM6 loss reduced pericarp thickness relative to WT fruits (Figure S9a,b). Although average pericarp cell areas were comparable among CaM6‐OE#3, WT, and cam6‐3 (Figure S9a,d), the number of pericarp cell layers was significantly increased in CaM6‐OE#3 and reduced in cam6‐3 (Figure S9a,c).

FIGURE 7.

CaM6 positively regulates fruit size. (a) RT‐qPCR analysis of CaM6 expression in two independent overexpression lines (OE3 and OE5). (b) Representative fruit phenotypes of WT, OE3, and OE5 plants. (c–e) Quantification of fruit weight (c), vertical diameter (d), and transverse diameter (e) in WT, OE3, and OE5 plants (n = 15 fruits, mean ± SD). (f) Representative fruit phenotypes of WT and CaM6 knockout mutants. (g–i) Quantification of single fruit weight (g), vertical diameter (h), and transverse diameter (i) in WT and cam6 mutant plants (n = 15 fruits, mean ± SD). All experiments included three independent biological replicates. Values are presented as mean ± SD across all datasets. Statistical differences from WT were assessed using Student's t‐test; ‘**’ indicates a significant difference at p < 0.01.

Given that calmodulin binding can positively or negatively modulate CNGC channel activity (Pan et al. 2019; Tian et al. 2019), we examined the genetic interaction between CNGC16 and CaM6 in fruit weight control. By generating cngc16‐1cam6‐3 double mutants and a CNGC16 overexpression line in the cam6‐3 mutant background, we found that the loss of CaM6 abolished the fruit weight increase induced by CNGC16 overexpression (Figure 8a,b). Moreover, the cngc16‐1cam6‐3 double mutant exhibited fruit weights similar to either single mutant (Figure 8a,b). These findings indicate that CaM6 physically and genetically interacts with CNGC16 to regulate fruit weight, with CNGC16's effect being partially CaM6‐dependent.

FIGURE 8.

Genetic interaction between CaM6 and CNGC16 in regulating fruit weight in tomato. (a) Representative fruit phenotypes of CNGC16‐OE#4, WT, cngc16‐1, cam6‐3, cngc16‐1cam6‐3, and CNGC16‐OE#4cam6‐3 plants. (b) Quantification of fruit weight for CNGC16‐OE#4, WT, cngc16‐1, cam6‐3, cngc16‐1cam6‐3, and CNGC16‐OE#4cam6‐3 plants (n = 15). Data represent mean ± SD from three independent biological replicates. Statistical significance was determined using one‐way ANOVA followed by Duncan's multiple range test (p < 0.05). (c) Proposed working model illustrating the roles of CaM6 and CNGC16‐mediated calcium signalling in regulating fruit weight.

3. Discussion

Fruit weight is a key trait in the domestication history of tomato (Lin et al. 2014). Wild tomato species, such as SP, typically produce fruits weighing approximately 1 g, whereas the domesticated variety SLL can bear fruits exceeding 1 kg (Rodriguez et al. 2011). Although more than 30 QTLs associated with fruit weight or size have been identified over the past decade, the regulatory mechanisms underlying these QTLs remain largely unresolved (Beauchet et al. 2021). Our study identifies CNGC16 as a key gene underlying natural fruit weight variation through GWAS and establishes its critical role in maintaining cytosolic Ca²⁺ homeostasis and promoting cell division in tomato fruit development as proposed (Figure 8c). These findings uncover a novel role for calcium signalling in fruit weight regulation and provide valuable genetic targets and frameworks for improving tomato fruit weight.

Homologues of CNGC16 in Arabidopsis and rice mediate increases in cytosolic calcium([Ca²⁺]_cyt) in response to various stresses, including pathogen attack, cold, and heat stress (Cook et al. 2025; Cui et al. 2020; Tian et al. 2019; Wang et al. 2019, 2021). Our previous work showed that mutations in vacuolar calcium transporter CAX1 and CAX3 cause phenotypes similar to those of cngc4 mutants, and these mutants exhibit elevated [Ca²⁺]_cyt (Wang, Song, et al. 2024; Conn et al. 2011). Here, using transgenic tomato plants expressing the YC3.6 calcium reporter, we provide direct evidence that loss of CNGC16 significantly increases [Ca²⁺]_cyt in fruit cells (Figure 5). This observation is paradoxical, as CNGCs are classically plasma membrane channels mediating calcium influx from extracellular space, yet cngc16 mutants exhibited elevated cytosolic calcium. We hypothesize that this results from compensatory activation or repression of other calcium channels or transporters. Supporting this, we observed strong downregulation of CAX3, a vacuolar calcium transporter, and upregulation of CNGC13 in cngc16 mutants (Figure 5c–f), indicative of a feedback mechanism to maintain calcium homeostasis. Future studies focusing on generating knockout and overexpression lines of CAX3 and CNGC13 in tomato will be crucial to elucidate their roles in controlling [Ca²⁺]_cyt and fruit weight, and to clarify how CNGC16 modulates their expression, deepening our understanding of the calcium signalling regulatory network during fruit development in tomato.

Previous studies have hypothesized roles for calcium signalling in tomato fruit size regulation (Beauchet et al. 2021). For instance, the SUN locus encodes a protein with a calmodulin‐binding domain influencing elongated fruit shape, and FW2.2, a major fruit size QTL containing a PLAC8 motif, has been suggested to regulate calcium signalling (Beauchet et al. 2021; Clevenger et al. 2015). However, direct experimental evidence was lacking until now. Our study uncovers a novel function for CNGC16 in contributing to fruit weight variation among tomato accessions (Figures 1, 2, 3) and establishes its essential role in maintaining [Ca²⁺]_cyt in tomato fruit cells (Figure 5). We further demonstrate that CaM6 physically interacts with CNGC16 (Figure 6) and acts as a positive regulator of fruit weight (Figure 7). Collectively, these findings provide direct evidence that calcium signalling is a regulator of tomato fruit size.

Although our study did not fully elucidate the mechanistic relationship between CaM6 and CNGC16, we observed that loss of CaM6 largely abolishes the fruit size increase induced by CNGC16 overexpression, and the cngc16 cam6 double mutant displayed a similar phenotype to each single mutant (Figure 8a,b). This suggests that CNGC16's role in fruit weight regulation is partially dependent on CaM6. Moreover, fruit weight in CNGC16‐OE#4cam6‐3 lines intermediate between cam6‐3 and CNGC16‐OE#4, implying possible functional redundancy with other CaMs or CaM6‐independent regulatory mechanisms. Future investigations are warranted to dissect how CaM6‐dependent and ‐independent pathways converge on fruit weight regulation.

Calmodulins are well known for decoding calcium signals and shaping Ca²⁺ dynamics (Kan et al. 2022; Wang, Ge, et al. 2024). How CaM6 transduces calcium signals to control fruit size remains an open question. We aim to identify and characterize downstream CaM6 interactors that regulate fruit weight, providing deeper insight into calcium‐dependent molecular mechanisms governing fruit development.

Natural variation in gene expression responsive to developmental and environmental cues underlies local adaptation (Wang et al. 2022, 2025; Wang, Miao, et al. 2024; Wang, Song, et al. 2024; Zhang et al. 2017). Such variations enable organisms to fine‐tune developmental responses to specific environments for enhanced survival and reproduction (Wang, Miao, et al. 2024; Zhang et al. 2017). In this study, we demonstrated that natural variation in the CNGC16 promoter correlates with gene expression differences and fruit weight variation (Figures 1 and 3). Promoter activity assays show that the Hap1 promoter has increased activity compared to Hap2 (Figure 1e,f), supporting that natural promoter polymorphisms directly contribute to expression differences. Consistently, CNGC16 overexpression leads to increased fruit weight, linking expression variation of CNGC16 to natural fruit size diversity (Figure 2). Functionally, our data show that CNGC16 overexpression increases fruit weight, whereas loss‐of‐function decreases it (Figure 2). The frequency of the Hap1 allele progressively rises from small‐fruited wild SP, through semi‐domesticated species (SLC), to fully domesticated large‐fruited cultivars (SLL) (Figure 4e,f), implicating CNGC16 in domestication‐driven fruit size evolution. Given extensive genetic diversity in the CNGC16 promoter—multiple SNPs and InDels—future fine‐mapping is necessary to identify causal variants responsible for differential expression and fruit weight.

CNGC homologues respond to various abiotic stresses such as temperature fluctuations and pathogen attacks (Cook et al. 2025; Cui et al. 2020; Tian et al. 2019; Wang et al. 2021), implicating CNGC16 expression variation as a potential mediator of plant environmental adaptation. The distinct geographic distribution of CNGC16 haplotypes (Figure 4a–d) further supports this hypothesis, suggesting that natural variation in CNGC16 expression may enable tomatoes to better cope with local environments. Whether CNGC16 also directly contributes to stress response awaits future investigation.

4. Materials and Methods

4.1. Plant Materials and Growth Conditions

The natural tomato accessions used in this study were obtained from the United States Department of Agriculture (USDA) (Mata‐Nicolas et al. 2020). The wild‐type tomato employed was Ailsa Craig (AC), which belongs to the Hap1 group. Seeds were surface sterilized with 75% ethanol for 10 min, followed by three rinses with distilled water prior to planting. For phenotypic analysis, plants were cultivated in an experimental field located in Hangzhou, China. For RT‐qPCR experiments, plants were grown under greenhouse conditions with a short‐day photoperiod (10 h light/14 h dark), a light intensity of 130 μmol m⁻² s⁻¹and a constant temperature of 26°C.

4.2. Phenotypic Characterization

To minimize positional effects, plants of different genotypes were randomly arranged within the greenhouse. For each genotype, fruit weight, vertical diameter, and transverse diameter were measured on fully ripened fruits collected from at least 15 plants. Mean and variances were calculated from three independent experiments and used for statistical analysis.

4.3. Genome‐Wide Association Study

A dataset comprising 609 463 polymorphisms (including SNPs and Indels) was downloaded from the Sol Genomics Network (https://solgenomics. net). Fruit weight data for 163 accessions were previously reported (Mata‐Nicolas et al. 2020). Genome‐wide association analysis (GWAS) of fruit weight was conducted using the rMVP package in R, employing a mixed linear model (Yin et al. 2021). To reduce the influence of rare alleles on the GWAS results, a minor allele frequency (MAF) cutoff of 0.05 was applied. Manhattan plots were generated using the qqman R package.

4.4. Gene Expression Analysis

Total RNA was extracted from 10 days post‐anthesis (DPA) fruits using the RNA extraction kit (TransZol Plant, TRANSGEN BIOTECH, http://www.transgen.com.cn). First‐strand cDNA was synthesized with the Primer Script II 1st Strand cDNA Synthesis Kit (Wang, Song, et al. 2024). Quantitative real‐time PCR (RT‐qPCR) was performed using SYBR Green reagent (Takara) on a StepOne Plus Real‐Time PCR System (ABI). The ACTIN gene (Solyc03g078400) was used as an internal control for normalization. Each RT‐qPCR experiment included three independent biological replicates. Gene‐specific primers are provided in Table S2.

4.5. Promoter Activity Assay

Approximately 3 kb promoter fragments upstream of CNGC16 were amplified from BGC006327 (Hap1 group) and BGV013175 (Hap2 group), and cloned into the pGREENII 0800‐LUC vector to generate pCNGC16 ^hap1 :LUC and pCNGC16 ^hap2 :LUC constructs, respectively (Hellens et al. 2005). Constructs were transiently expressed in tobacco leaves, which were incubated at 25 C in the dark for 16 h post‐infiltration. Total RNA was extracted from the infiltrated leaves using TRIzol reagent (Invitrogen). Reverse transcription of mRNA was performed with the PrimerScript II 1st Strand cDNA Synthesis Kit (Takara) as previously described (Wang et al., 2018). Quantitative real‐time PCR assays were conducted using SYBR Green reagent (Takara) on a StepOne Plus Real‐Time PCR System (ABI). Primers used for qPCR are listed in Table S2. Three independent biological replicates were performed for the analysis.

4.6. Generation of Transgenic Plants

Two gRNA targeting CNGC16 were designed using the CRISPR‐PLANT website (https://www.genome.arizona.edu/crispr/CRISPRsearch.html), and cloned into the pICH47751 and pICH57761 vectors, then assembled into pICH4723. The coding sequences (CDS) of CNGC16 and CaM6 were PCR‐amplified from AC and cloned into pCAMBIA1300S‐GFP. The YC3.6 calcium reporter vector (35S pGPTVII) was obtained from the Wang lab (Zhu et al. 2025). Constructs were transformed into Agrobacterium tumefaciens strain GV3101 and used to transform AC plants.

Homozygous CRISPR mutants were identified by sequencing T0 or T1 plants. Overexpression lines were initially screened by PCR and qRT‐PCR, and homozygous plants were selected on hygromycin resistance from the T2 generation. T3 homozygous lines were used for subsequent phenotypic analyses. For complementation, the 3 kb promoters of CNGC16 ^Hap1 and CNGC16 ^Hap2 were amplified from BGV006208 and BGV012640, fused to the AC CNGC16 CDS, and transformed into the cngc16‐1 background. Transgenic lines were screened by PCR and RT‐qPCR in T1, and T2 progeny were analysed phenotypically.

YC3.6 transgenic lines showing strong fluorescence signals in fruit cells were crossed into the cngc16‐1 background. Progeny carrying both the mutant phenotype and robust YC3.6 signals were selected. The cam6 mutants, obtained from the Lin Lab (Wang, Ge, et al. 2024), were crossed with CNGC16 overexpression lines for genetic interaction studies.

4.7. Cytosolic Ca²⁺ Imaging Analysis in Fruit Cells

FRET/CFP‐based cytosolic Ca²⁺ imaging was conducted on the epidermis of 15 DPA tomato fruits as previously described (Zhu et al. 2025). The fruit epidermis was immersed in a working solution containing 5 mM KCl, 100 μM CaCl₂, 10 mM Mes‐Tris (pH 5.7) within a chamber for 1 h prior to imaging. FRET/CFP emission ratio of YC3.6 was measured using confocal microscopy, and images were collected for data analysis.

4.8. Bimolecular Fluorescence Complementation in Nicotiana benthamiana Leaves

The CDS of CNGC16 and CaM6 were amplified from AC and cloned into PUC‐SPYNE and PUC‐SPYCE vectors, respectively. Constructs were introduced into Agrobacterium tumefaciens strain GV3101. For transient expression, strains harbouring BiFC constructs were co‐infiltrated with the p19 silencing suppressor strain into leaves of 6‐week‐old Nicotiana benthamiana plants. YFP fluorescence was observed 2–3 days post‐infiltration via confocal laser scanning microscope.

4.9. DUAL Membrane System Y2H Assays

The DUAL membrane system (Dualsystems Biotech P01201–P01229, Schlieren, Switzerland) was employed for yeast two‐hybrid (Y2H) assays. The CDS of CaM6 and CNGC16 were cloned into the pBT3‐SUC and pPR3‐N vectors from AC cDNA. Constructs were co‐transformed pairwise into yeast strain NMY51. Transformants were grown on selective media lacking Leu and Trp or lacking Leu, Trp, His, and Ade at 30°C for 3 days. Yeasts strains co‐transformed with pNubG Fe65 and pTSU2‐APP served as positive controls, while yeasts transformed with pPR3‐N and pTSU2‐APP served as negative controls.

4.10. Coimmunoprecipitation (Co‐IP) Assays

CDS of CNGC16 and CaM6 were cloned into pCAMBIA1300S‐GFP and pCAMBIA1300‐Myc, respectively. Constructs were co‐expressed in Nicotiana benthamiana leaf epidermal cells by agroinfiltration. Total proteins were extracted using buffer containing 50 mM Tris HCl, pH = 7.5, 150 mM NaCl, 0.2% (v/v) NP‐40, 5 mM DTT, and a protease inhibitor cocktail. Lysates were incubated with GFP agarose beads (MBL, New Jersey, USA) at 4°C for 2 h, washed five times with washing buffer (50 mM Tris–HCl, pH = 7.5, 150 mM NaCl, 0.1% NP‐40), and immunoprecipitated proteins were separated by 10% SDS‐PAGE and analysed by immunoblotting using anti‐GFP (Sigma‐Aldrich, Saint Louis, Missouri, USA) and anti‐Myc (Cell Signalling Technology, Danvers, MA, USA) antibodies.

4.11. Structure Modelling of the CaM6‐ CNGC16 Interaction

The full‐length amino acid sequence of CaM6 in complex with CNGC16 was modelled using AlphaFold Server Beta (v3.0) with default parameters (Abramson et al. 2024). The resulting PDB files were subjected to protein–protein docking analysis via HDOCK (Yan et al. 2020). Structures were visualized and analyzed in PyMOL.

Author Contributions

G.W., X.W. conceived and supervised the project. X.W., Y.W., and H.M. conducted experiments. X.W. and G.W. wrote the manuscript. All authors read and approved the final manuscript.

Funding

This work was supported by the National Natural Science Foundation of China, 32502685. Zhejiang Provincial Natural Science Foundation of China, LQ24C150001. Zhejiang Normal University.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Figure S1: The Manhattan plot of GWAS for fruit weight. The red horizontal dashed line indicates the default 5% false discovery rate (Benjamin‐Hochberg) threshold for GWAS. The arrows indicate the significant peaks co‐localized with the two previously reported QTLs.

Figure S2: Fruit phenotype of the mutants of AtCNGC4.

Figure S3: Diagram of the DNA sequence variation in the promoter region of CNGC16 between Hap 1 and Hap 2.

Figure S4: (a) The expression levels of CNGC16 at different developmental stages in MM and PP, respectively; (b) The expression levels of CNGC16 in the indicated tissues in AC.

Figure S5: CNGC16 mutants generated using the CRISPR/Cas9 system. Diagrams show the genomic sequences of wild‐type (WT) and CNGC16 mutant alleles, with exons, introns, and untranslated regions represented by dark boxes and connecting black lines. The sgRNA target site used for CRISPR/Cas9 editing is indicated.

Figure S6: CNGC16 increases fruit pericarp thickness by promoting cell division in tomato. (a) Microscopic cross‐sections showing cell morphology and arrangement in fruits of CNGC16‐OE#4, WT and cngc16‐1 at the mature green stage. Bar = 1 mm. Quantification of pericarp thickness (b), pericarp cell layers (c) and Mean cell size (d). Plants were grown in an experimental field located in Hangzhou under 14:10‐h photoperiods with daytime and nighttime temperatures of 20°C–32°C and 17°C–25°C, respectively., China. Collected 30‐DPA fruits were restricted by keeping the first two inflorescences, each carrying a maximum of three fruits. Different letters indicate significant differences at the p < 0.05 level according to Tukey Kramer test (n ≥ 5 fruits, mean ± SD).

Figure S7: Combined effects of fruit weight–associated loci, CNGC16, FW11.3 and FW7.2. (a) Information of the leading SNPs corresponding to FW11.3, FW7.2 and CNGC16 loci, respectively based on the GWAS analysis of fruit weight. Association analysis between the three SNPs corresponding to FW11.3 (b), FW7.2 (c) and CNGC16 (d) loci and fruit weight in natural accessions. (e) Groups based on the leading SNPs corresponding to FW11.3, FW7.2 and CNGC16 loci. (f) Association analysis between the groups and fruit weight. N indicates the accession number of the genotypes. Different letters indicate significant differences at the p < 0.05 level according to Tukey Kramer test. ‘**’ indicates significant differences from WT at p < 0.01, Student's t‐test.

Figure S8: Subcellular localization of GFP and CNGC16‐GFP in tobacco epidermal cells.

Figure S9: CaM6 increases fruit pericarp thickness by promoting cell division in tomato. (a) Microscopic cross‐sections showing cell morphology and arrangement in fruits of CaM6‐OE#3, WT and cam6‐3 at the mature green stage. Bar = 1 mm. Quantification of pericarp thickness (b), pericarp cell layers (c) and Mean cell size (d). Plants were grown in an experimental field located in Hangzhou under 14:10‐h photoperiods with daytime and nighttime temperatures of 20°C–32°C and 17°C–25°C, respectively, China. Collected 30‐DPA fruits were restricted by keeping the first two inflorescences, each carrying a maximum of three fruits. Different letters indicate significant differences at the p < 0.05 level according to Tukey Kramer test (n ≥ 5 fruits, mean ± SD).

Table S1: The most significant SNPs.

Table S2: Primers used in this study.

Table S3: Predicted interaction sites between CaM6 and CNGC16.

Data Availability Statement

The data that support the findings of this study are available on request from the corresponding author. The data are not publicly available due to privacy or ethical restrictions.