Spatial control of anthocyanin and proanthocyanidin production in tomato fruits by competitive MBW complexes to promote seed dispersal Authors: Zilin Qian, Hexu Cai, Mengyi Yu, Xiaolin Geng, Huabing Lin, Yaoyao Sun, Xinyi Zhang, Qiuyu Tao, Yu Jiang, Luyao Yang, Yunhyeong Lee, Jili Li, Huolin Shen, Minmin Du, Wencai Yang Published: 2026 ABSTRACT AbstractAnthocyanins and proanthocyanidins (PAs) are flavonoid compounds with important roles in plant physiology and ecology, as well as beneficial effects on human health. Although they share a common biosynthetic pathway, these two types of flavonoids exhibit different tissue-specific accumulation patterns in many fruits and plant organs. In tomato, anthocyanins accumulate primarily in the fruit peel of some wild relatives, whereas PAs accumulate mainly in seeds. However, the molecular mechanisms that underlie these patterns remain to be characterized. Here, we demonstrate that the tissue-specific accumulation of anthocyanins and PAs in tomato fruits is governed by different MYB–basic helix-loop-helix (bHLH)–WD40 (MBW) transcriptional modules. The bHLH transcription factor SlAN1 and the WD40 regulator SlAN11, critical components of the MBW complex, are required for both anthocyanin and PA biosynthesis. By contrast, two MYB components of the MBW complex, SlAN2-like and SlMYB54, specifically regulate anthocyanin biosynthesis in the peel and PA biosynthesis in seeds, respectively. The two MYB components interact with SlAN1 to form two competitive MBW transcriptional modules that control the tissue-specific accumulation of anthocyanins and PAs in tomato fruits. Finally, we provide evidence that these tissue-specific accumulation patterns are likely orchestrated by a dual-layered protective mechanism to ensure successful seed dispersal. Our findings reveal that efficient biosynthesis of anthocyanins and PAs arises from strict tissue-specific expression of both regulatory and functional genes, suggesting strategies for the enhancement of anthocyanin and PA levels in tomato fruits.Key words: anthocyanin, proanthocyanidin, tomato, MBW complex, tissue-specific expression Introduction Flavonoids are a class of secondary metabolites that impart a wide spectrum of colors to various plant organs, thus attracting pollinators and seed dispersers; they also make important contributions to plant defense against biotic and abiotic stresses (Winkel-Shirley, 2001; Landi et al., 2015; Sharma et al., 2019). Anthocyanins and proanthocyanidins (PAs, also known as condensed tannins) are two of the major flavonoid subgroups in the plant kingdom. In addition to their ecological functions, anthocyanins and PAs offer substantial human health benefits due to their potent antioxidant properties (Kandil et al., 2002; Tena et al., 2020) and anticancer activities (Lin et al., 2017; Rauf et al., 2019). Anthocyanins and PAs are the terminal products of the flavonoid branch of the phenylpropanoid pathway. Anthocyanin synthesis is orchestrated by a suite of enzymes localized on the cytoplasmic face of the endoplasmic reticulum, including chalcone synthase (CHS), chalcone isomerase (CHI), flavanone 3-hydroxylase (F3H), flavonoid 3′,5′-hydroxylase (F3′5′H) (although it is notably absent in certain species such as Arabidopsis, apple, and peach), dihydroflavonol 4-reductase (DFR), leucoanthocyanidin dioxygenase (LDOX), and UDP-glucose flavonoid 3-O-glucosyltransferase (UFGT). PAs are polymers of flavan-3-ol units, specifically catechin and epicatechin, which are derived from the immediate precursors of anthocyanins (leucoanthocyanidin and anthocyanidin) through the catalytic activity of leucoanthocyanidin reductase (LAR) and anthocyanidin reductase (ANR), respectively. Although anthocyanins and PAs share a common origin and intermediates in the phenylpropanoid pathway, their biosynthesis diverges at the level of branch-specific enzymes and regulatory mechanisms (Constabel, 2018). The biosynthesis of anthocyanins and PAs is controlled by two gene classes: structural genes, which encode enzymes, and regulatory genes, which control transcription. Structural genes, such as those encoding phenylalanine ammonia-lyase (PAL), cinnamate-4-hydroxylase (C4H), 4-coumaroyl:CoA-ligase (4CL), CHS, CHI, F3H, F3′5′H, DFR, and anthocyanidin synthase (ANS), are responsible for generating intermediates (Tanaka et al., 2008; Gonzalez et al., 2009; Gonzali et al., 2009; Zhang et al., 2014; Liu et al., 2018). Branch-specific structural genes then refine these intermediates. For anthocyanins, 3-O-glycosyltransferase (3-GT), anthocyanin acyltransferase (AAT), and the transport protein glutathione S-transferase (GST) convert anthocyanidins into anthocyanins. For PAs, LAR transforms leucoanthocyanidins into catechin, and ANR/BANYULS (BAN) converts anthocyanidins into epicatechin, together with the multidrug and toxic compound extrusion (MATE) transporter TRANSPARENT TESTA12 (TT12) (Debeaujon et al., 2001) and the oxidizing enzyme TT10 (Pourcel et al., 2005) to facilitate PA polymerization (Yu et al., 2023). The transcription of structural genes is predominantly regulated by transcription factors (TFs), including members of the R2R3-MYB, basic helix–loop–helix (bHLH), WD repeat protein (WDR), and WRKY families. Notably, R2R3-MYB, bHLH, and WDR TFs assemble into the evolutionarily conserved MBW (MYB–bHLH–WDR) complex, which exerts transcriptional control over the biosynthetic networks of both anthocyanins and PAs (Quattrocchio et al., 1998; Baudry et al., 2004; Xu et al., 2014; Lloyd et al., 2017). WRKY TFs can also modulate anthocyanin and PA biosynthesis (Johnson et al., 2002; Verweij et al., 2016; Wang et al., 2024). Anthocyanins and PAs are ubiquitous secondary metabolites found in diverse plant species, including fruit crops, vegetables, flowers, and pigmented grains. However, their distribution varies markedly among tissues and organs. Anthocyanins have been detected at high levels in the floral tissues of petunia (Verweij et al., 2016), lily (Bi et al., 2023), chrysanthemum (Zhou et al., 2024), and cauliflower (Chiu et al., 2010), as well as in the tubers of potato (Du et al., 2025). They are also present in the fruit peel of eggplant (Yong et al., 2019), grapevine (Sun et al., 2024), apple (Ji et al., 2025), and pear (Bai et al., 2019); in the fruit flesh of kiwi (Wang et al., 2019), apple (Yang et al., 2021), and orange (Butelli et al., 2012); in the leafy structures of cabbage (Pourjavaher et al., 2017); and in the grains of maize (Cone et al., 1986), rice (Zhu et al., 2017), and peanut (Zhao et al., 2025). By contrast, PAs are found primarily in seeds, as evidenced by their presence in the seeds of grapevine (Qi et al., 2023) and wheat (Himi et al., 2023). Limited accumulation of PAs has also been observed in poplar leaves (Ullah et al., 2019), freesia flowers (Shan et al., 2025), and cotton fibers (Wang et al., 2022). Studies in Arabidopsis have revealed distinct tissue-specific localization patterns for these metabolites: anthocyanins are predominantly found in the cotyledons and hypocotyls, whereas PAs are confined to the seed coats (Nesi et al., 2001; Johnson et al., 2002). Although the biosynthetic pathways of anthocyanins and PAs have been studied extensively, the mechanisms that control their tissue-specific accumulation patterns remain poorly understood. Tomato (Solanum lycopersicum) is among the most widely cultivated and consumed vegetables worldwide, valued for its nutritional benefits and health-promoting properties. To date, most studies have focused on anthocyanin accumulation in the tomato fruit peel. Although fruits of cultivated tomato are notably deficient in anthocyanins, the transcriptional network that governs anthocyanin biosynthesis in the peel has been characterized, primarily through the introgression of genes from wild relatives (Zhang and Zhang, 2020). Anthocyanin biosynthesis in the tomato fruit peel is regulated by an MBW complex comprising one MYB TF, SlAN2-like (Sun et al., 2020; Yan et al., 2020); two bHLH TFs, SlJAF13 (Chen et al., 2022) and ANTHOCYANIN1 (SlAN1) (Sun et al., 2020); and one WDR protein, SlAN11. The SlAN2-like gene, which originated from S. chilense (Anthocyanin fruit allele, Aft) and S. lycopersicoides (Aubergine allele, Abg), is crucial for promoting anthocyanin accumulation in the fruit peel (Sun et al., 2020; Yan et al., 2020; Menconi et al., 2023); its positive effect is further enhanced by the presence of the recessive allele of atroviolacium (atv) from S. cheesmaniae (Cao et al., 2017; Colanero et al., 2018; Sun et al., 2020). In addition to its role in the fruit peel, SlAN1 has also been implicated in anthocyanin biosynthesis in the hypocotyl (Qiu et al., 2016). Nonetheless, although the primary anthocyanin biosynthetic network in the fruit peel has been characterized, the potential involvement of additional regulators in this pathway remains to be clarified, as does its potential role in PA biosynthesis in seeds. Here, we observed differences in the tissue-specific accumulation patterns of anthocyanins and PAs in tomato fruits: anthocyanins were found predominantly in the peel and PAs mainly in the seeds. We then demonstrated that two tissue-specific MYB TFs interact with the bHLH TF SlAN1, forming two distinct and competitive MBW transcriptional complexes to orchestrate the tissue-specific biosynthesis of anthocyanins and PAs in tomato fruits. Results Spatial and temporal patterns of anthocyanin and proanthocyanidin accumulation in tomato fruits Anthocyanins and PAs are among the primary end products of the flavonoid biosynthetic pathway and share a series of intermediates during their synthesis (Figure 1A). Here, we aimed to investigate patterns of anthocyanin and PA accumulation in tomato fruits. Because cultivated tomato fruits are typically deficient in anthocyanins, we used the elite inbred line Jingzi 8035 (wild type, WT), which carries the Aft (a functional allele of SlAN2-like) and atv mutations, enabling it to accumulate anthocyanins in the fruit peel. Although the precursors of both anthocyanins and PAs have been detected in fruit peels and seeds (Toor and Savage, 2005; Wang et al., 2020a; Kumar et al., 2021), we found that anthocyanins accumulated primarily in the peel and PAs mainly in the seeds (Supplemental Figure 1). The anthocyanin content of the peel increased over the course of fruit development (Figure 1B). By contrast, the soluble PA content in seeds increased significantly only after 30 days post-anthesis (DPA), peaking at the mature green (MG) stage and then declining (Figure 1C). We next performed RNA sequencing (RNA-seq) to examine the expression patterns of structural genes involved in the flavonoid biosynthetic pathway. Robust transcription of structural genes (SlPAL, SlC4H, Sl4CL, SlCHS1, SlCHS2, SlCHI, SlF3H, SlDFR, and SlANS) shared by anthocyanin and PA biosynthesis was detected in both peels and seeds (Figure 1D; Supplemental Table 1). By contrast, anthocyanin-specific structural genes (e.g., Sl3-GT and SlGST) were expressed exclusively in the fruit peel, and PA-specific structural genes (e.g., SlANR and SlTT10) were expressed only in the seeds (Figure 1D; Supplemental Table 1). These findings indicate that the production of anthocyanins and PAs is spatially restricted in tomato fruits, likely owing to the tissue-specific expression patterns of structural genes. Spatial and temporal patterns of anthocyanin and proanthocyanidin (PA) biosynthesis in tomato (A) The anthocyanin and PA biosynthetic pathway. Gray, purple, and brown arrows and lines represent the shared pathway, the anthocyanin branch, and the PA branch, respectively. (B) Images (top panel) of wild-type (WT) fruits at eight developmental stages: 10 days post-anthesis (DPA), 20 DPA, 25 DPA, 30 DPA, 35 DPA, mature green (MG), breaker (BR), and BR+5. Scale bars, 1 cm. Anthocyanin contents (middle panel) and fresh extracts (bottom panel) of light-exposed fruit peel of the WT at eight developmental stages. (C) Images (top panel) of 4-(Dimethylamino) cinnamaldehyde (DMACA)-stained seeds at eight developmental stages. Scale bars, 1 cm. Soluble PA content (middle panel) and fresh extracts (bottom panel) of seeds from the WT at eight developmental stages. (D) Expression levels of representative structural genes for anthocyanin and PA biosynthesis in fruit peel and seeds. The log2 (FPKM + 1) value of each gene from three biological replicates is shown. FPKM, fragments per kilobase of exon per million fragments mapped. Data in (B) and (C) are means ± SD (n = 3). The MBW components SlAN1 and SlAN11 are required for anthocyanin and PA production Various stages of flavonoid biosynthesis are regulated by MBW complexes. We investigated the roles of different MBW components, including the bHLH component SlAN1 and the WDR component SlAN11, in the regulation of anthocyanin and PA biosynthesis. RNA-seq and real-time quantitative reverse transcription PCR (RT–qPCR) analyses showed that SlAN1 and SlAN11 were highly expressed in both the fruit peel and seeds (Figure 2A–2C; Supplemental Figure 2A and 2B; Supplemental Table 2). To investigate the function of these MBW components, we generated Slan1 and Slan11 mutants using the CRISPR-Cas9 genome-editing system (Supplemental Figure 2C and 2D). Compared with the WT, both Slan1 and Slan11 exhibited marked reductions in anthocyanin pigmentation and expression of anthocyanin structural genes in the fruit peel (Figure 2D–2F; Supplemental Figure 2E). SlAN1 and SlAN11 participate in the biosynthesis of both anthocyanins and proanthocyanidins (PAs). (A and B) Temporal expression patterns of SlAN1(A) and SlAN11(B) in fruit peel and seeds. (C) Expression levels of SlAN1 and SlAN11 in fruit peel and seeds. The log2 (FPKM + 1) value of each gene from three biological replicates is shown. FPKM, fragments per kilobase of exon per million fragments mapped. (D) Images of WT, Slan1, and Slan11 fruit harvested at the green stage. Scale bars, 1 cm. (E) Anthocyanin contents (top panel) and fresh extracts (bottom panel) of light-exposed fruit peel from WT, Slan1, and Slan11 at 30 DPA. (F) Relative expression of shared and anthocyanin-specific structural genes in the fruit peel of WT, Slan1, and Slan11 at 30 DPA. (G) Representative images of WT, Slan1, and Slan11 seeds. The top four pictures show dry seeds harvested at the mature stage, and the bottom four pictures show seeds collected at the BR stage after DMACA staining. Scale bars, 1 cm. (H) Soluble PA content (top panel) and fresh extracts (bottom panel) of seeds from WT, Slan1, and Slan11 at the mature stage. (I) Relative expression of shared and PA-specific structural genes in WT, Slan1, and Slan11 seeds at the BR stage. For (A), (B), (F), and (I), SlUBI3 (Solyc01g056840) was used as the reference transcript. All data are presented as means ± SD (n = 3). Statistical significance was determined by two-tailed Student’s t test (∗∗p < 0.01). Dry seeds of Slan1 and Slan11 had a yellowish hue, in contrast to the brown coloration of WT seeds (Figure 2G). After DMACA–HCl staining of PAs in fresh seeds at the breaker (BR) stage, WT seeds showed uniform black staining, whereas Slan1 and Slan11 seeds showed a reddish-brown coloration (Figure 2G). Quantitative analysis confirmed that soluble PA contents were significantly lower in fresh seeds of the mutants than in those of the WT (Figure 2H). Consistent with these findings, transcript levels of PA-specific structural genes were markedly downregulated in seeds of Slan1 and Slan11 (Figure 2I; Supplemental Figure 2F). Collectively, these results indicate that SlAN1 and SlAN11 positively regulate the accumulation of both anthocyanins and PAs in tomato fruits. To further examine the role of SlAN1 in anthocyanin and PA pigmentation, we overexpressed SlAN1 in the tomato inbred line LA1996, which shows only scattered purple spots on the exocarp. Two independent transgenic lines in which SlAN1 expression was upregulated 14.7-fold (SlAN1GFP-OE#1) and 6.7-fold (SlAN1flag-OE#2) were used for subsequent experiments (Supplemental Figure 2G and 2H). The SlAN1-OE lines exhibited enhanced anthocyanin pigmentation in the fruit peel and a noticeably darker brown hue in the mature dry seeds (Supplemental Figure 2G). Accordingly, the anthocyanin content of the fruit peel (Supplemental Figure 2I) and the soluble PA content of the seeds (Supplemental Figure 2J) were significantly higher in SlAN1-OE plants than in LA1996. These data confirmed that SlAN1 acts as a positive regulator of both anthocyanin and PA biosynthesis in tomato fruits. SlAN2-like and SlMYB54 drive tissue-specific accumulation of anthocyanins and PAs in fruit peel and seeds Because the bHLH and WDR components of the MBW complex did not appear to have tissue-specific regulatory roles in anthocyanin and PA biosynthesis, we shifted our focus to the MYB regulators. The R2R3-MYB TF SlAN2-like has been reported to participate in anthocyanin biosynthesis in the fruit peel of tomato (Sun et al., 2020; Yan et al., 2020). Phylogenetic analysis revealed that SlAN2-like was closely related to AtPAP1 and AtPAP2 (Supplemental Figure 3; Supplemental Table 3), which are involved in anthocyanin biosynthesis in Arabidopsis (Borevitz et al., 2000; Stracke et al., 2001). We found that SlAN2-like was strongly expressed in the fruit exocarp (peel) but barely detectable in seeds, as revealed by both RT–qPCR and RNA-seq data (Figure 3A; Supplemental Figure 4A; Supplemental Table 4). To investigate the regulatory role of SlAN2-like in anthocyanin and PA biosynthesis, we generated two Slan2-like mutants using CRISPR-Cas9 genome editing (Supplemental Figure 4B). Compared with the WT, the Slan2-like mutants had significantly reduced anthocyanin pigmentation in the fruit peel, but PA accumulation in the seeds remained unchanged (Figure 3C, 3D, 3F, and 3G). Consistent with these observations, expression of anthocyanin-specific structural genes was markedly downregulated in the peel of Slan2-like, whereas PA-specific gene expression was unchanged (Figure 3E and 3H). We also observed that the seed coloration of varieties carrying the Aft gene (the functional allele of SlAN2-like) was similar to that of varieties with aft (a loss-of-function allele of SlAN2-like) after DMACA staining (Supplemental Figure 4C). These findings suggest that SlAN2-like functions as a peel-specific MYB component of the MBW complex, regulating anthocyanin biosynthesis without affecting PA production in seeds. SlAN2-like and SlMYB54 differentially regulate anthocyanin and proanthocyanidin (PA) biosynthesis in tomato. (A and B) Relative expression of SlAN2-like(A) and SlMYB54(B) in seven tissues as measured by RT–qPCR. (C) Images of Slan2-like, WT, and Slmyb54 fruit harvested at the green stage. Scale bars, 1 cm. (D) Anthocyanin contents (top panel) and fresh extracts (bottom panel) of light-exposed fruit peel from Slan2-like, WT, and Slmyb54 at 30 DPA. (E) Relative expression of shared and anthocyanin-specific structural genes in the fruit peel of Slan2-like, WT, and Slmyb54 at 30 DPA. (F) Representative images of Slan2-like, WT, and Slmyb54 seeds. The top pictures show dry seeds harvested at the mature stage, and the bottom pictures show seeds collected at the BR stage after DMACA staining. Scale bars, 1 cm. (G) Soluble PA content (top panel) and fresh extracts (bottom panel) of seeds from Slan2-like, WT, and Slmyb54 at the mature stage. (H) Relative expression of shared and PA-specific structural genes in the seeds of Slan2-like mutants, WT, and Slmyb54 at the BR stage. For (A), (B), (E), and (H), SlUBI3 (Solyc01g056840) was used as the reference transcript. All data are presented as means ± SD (n = 3). Statistical significance was determined by two-tailed Student’s t test (∗p < 0.05 and ∗∗p < 0.01). We next aimed to identify the seed-specific MYB components that regulate PA biosynthesis in tomato seeds. The tomato genome contains 121 R2R3-MYB TFs (Zhao et al., 2014), and we examined the expression levels of all these R2R3-MYB genes using transcriptome data from tomato fruit peel and seeds. Nine candidate MYB TFs with high seed-specific expression were identified through cluster analysis (Supplemental Figure 5; Supplemental Table 5). To clarify their functions, we performed a phylogenetic analysis that included these nine TFs and other R2R3-MYBs previously reported to participate in PA, flavonoid, and anthocyanin biosynthesis. This analysis showed that SlMYB54 was closely related to AtTT2 and that SlMYB5 was closely related to AtMYB5, both of which are known to be involved in PA biosynthesis (Nesi et al., 2001; Li et al., 2009) (Supplemental Figure 3; Supplemental Table 3). Two R2R3-MYB candidates, SlMYB54 and SlMYB5, were selected on the basis of their exclusive expression in seeds, as revealed by RT–qPCR and RNA-seq analysis (Figure 3B; Supplemental Figure 4A; Supplemental Table 4). Both proteins exhibited nuclear localization (Supplemental Figure 4D), consistent with their proposed roles as TFs. To investigate the potential roles of SlMYB54 and SlMYB5 in seed PA biosynthesis, we generated CRISPR-Cas9 knockout mutants for both genes (Supplemental Figure 4E and 4F). As anticipated, Slmyb54 single mutants exhibited significantly reduced PA accumulation in seeds, whereas anthocyanin biosynthesis in fruit peels was unaffected (Figure 3C, 3D, 3F, and 3G). Consistent with these observations, expression of PA-specific structural genes was markedly downregulated in the peel of Slmyb54, whereas anthocyanin-specific gene expression remained unchanged (Figure 3E and 3H). Although the Slmyb5 single mutant did not show an obvious reduction in PA accumulation, the Slmyb54Slmyb5 double mutant displayed a more severe phenotype than the Slmyb54 mutant (Supplemental Figure 4G and 4H), as evidenced by both PA levels and the expression of structural genes for PA biosynthesis (Supplemental Figure 4I and 4J). To further characterize the role of SlMYB54 in PA production, we generated six SlMYB54-Flag transgenic lines (SlMYB54-OE). Two representative lines, SlMYB54-OE#1 and SlMYB54-OE#2, were chosen for detailed analyses (Supplemental Figure 6A). The expression of SlMYB54 was 10-fold and 6-fold higher in SlMYB54-OE#1 and SlMYB54-OE#2 , respectively, than in WT plants (Supplemental Figure 6B). Accordingly, substantial accumulation of SlMYB54-Flag protein was detected in both SlMYB54-OE lines (Supplemental Figure 6B). Compared with WT seeds, SlMYB54-OE seeds had a darker brown hue (Supplemental Figure 6C) and significantly higher PA levels (Supplemental Figure 6D). Expression of the PA-specific structural gene SlANR was much higher in SlMYB54-OE seeds than in WT seeds (Supplemental Figure 6E). These results suggest that SlMYB54 and SlMYB5 function as seed-specific MYB regulators, specifically controlling PA biosynthesis in tomato seeds. SlMYB54 and SlMYB5 interact with SlAN1 as MYB components of the MBW complex Previous studies have identified SlAN2-like as an MYB component of the MBW complex, given its ability to interact with the bHLH component SlAN1 (Sun et al., 2020). Because SlMYB54 and SlAN1 expression overlapped in tomato seeds (Figure 4A and 4B), we next investigated whether SlMYB54 and SlMYB5 physically interacted with SlAN1. Yeast two-hybrid assays showed that the R3 domain of SlMYB54 (SlMYB54R3) and SlMYB5 (SlMYB5R3) interacted with SlAN1 in yeast cells (Figure 4C and 4D). Bimolecular fluorescence complementation (BiFC) assays revealed strong interaction signals in cells co-expressing SlAN1-nYFP with SlMYB54-cYFP or SlMYB5-cYFP (Figure 4E). The protein interaction between SlAN1 and SlMYB54 was also verified by co-immunoprecipitation (Co-IP) (Figure 4F) and pull-down assays (Figure 4G). These results demonstrate that SlMYB54 and SlMYB5 physically interact with SlAN1 in vivo and in vitro, supporting their roles as MYB regulators within the MBW complex that mediates PA biosynthesis in tomato seeds. SlMYB54 and SlMYB5 interact with SlAN1 (A) Relative expression of SlAN1 and SlMYB54 at eight stages as measured by RT–qPCR. Data are presented as means ± SD (n = 3). SlUBI3 (Solyc01g056840) was used as the reference transcript. (B) Expression patterns of SlAN1 and SlMYB54 in seeds examined using the proSlAN1:GUS and proSlMYB54:GUS reporter systems, respectively. Scale bars, 1 cm. (C and D) Interaction of SlMYB54 (C) or SlMYB5 (D) with SlAN1 as revealed by Y2H assays. R3-domain-containing fragments of SlMYB54 and SlMYB5 were separately fused with the DNA-binding domain in pGBKT7, and the full-length coding sequence of SlAN1 was fused with the activation domain in pGADT7. Transformed yeast cells were grown on SD/−4 medium to examine protein interactions. (E) BiFC assays showing the interaction of SlMYB54 or SlMYB5 with SlAN1 in Nicotiana benthamiana leaves. Images were obtained using a confocal microscope. YFP, YFP signal; mCherry, nuclear marker; bright, bright field; merged, combination of the YFP, bright field, and mCherry signals. Scale bars, 10 μm. (F) Co-IP assays showing the interaction of SlMYB54 with SlAN1 in N. benthamiana leaves. SlAN1-Flag and SlMYB54-YFP-HA were immunoprecipitated using anti-Flag antibodies and detected with anti-Flag or anti-HA antibodies. (G)In vitro pull-down assay of the interaction between SlMYB54 and SlAN1. GST or GST-SlMYB54 fusion proteins bound to anti-GST magnetic beads were used to pull down His-SlAN1. SlAN2-like and SlMYB54 differ in their capacity to bind to the promoters of anthocyanin- and PA-specific structural genes Because SlAN2-like and SlMYB54 are MYB components of the MBW complexes that specifically govern anthocyanin and PA biosynthesis, respectively, we postulated that they might selectively bind to the promoters of corresponding structural genes. The anthocyanin-specific structural gene Sl3-GT and the PA-specific structural gene SlANR were selected to test this hypothesis. Chromatin immunoprecipitation followed by quantitative PCR (ChIP–qPCR) revealed enrichment of SlMYB54-Flag, but not SlAN2-like-GFP, on the P1 and P5 fragments of the SlANR promoter, which contain an MYB-binding motif (AACCTAA) (Figure 5A). Electrophoretic mobility shift assays (EMSAs) demonstrated that GST-SlMYB54 recombinant protein bound directly to the P5 DNA probe, which harbored the MYB-binding motif, whereas SlAN2-like-MBP failed to do so (Figure 5B). SlMYB54, but not SlAN2-like, specifically regulates PA-specific structure gene SlANR (A) Schematic representation of SlANR showing the primers and probe used for ChIP–qPCR and EMSA. Gray segments represent PCR primers, the blue segment represents the EMSA probe, and stars indicate MYB-binding motifs. ChIP–qPCR analysis of SlAN2-like-GFP and SlMYB54-Flag enrichment at the SlANR promoter is shown. (B) EMSA showing that the GST-SlMYB54 fusion protein binds directly to the SlANR promoter in vitro but that the SlAN2-like-His-MBP fusion protein fails to do so. Biotin-labeled probes were incubated with GST-SlMYB54 protein, and shifted and free DNA (arrows) were separated on an acrylamide gel. The GST and His-MBP proteins were incubated with the labeled probe to serve as negative controls. Unlabeled probes were used as competitors. Mu-competitor, mutated probes in which the MYB-binding motif 5′-AACCTAA-3′ was replaced with 5′-TTTTTTT-3′. CBB staining shows the amount of recombinant protein applied to the gel. The asterisks indicate the target bands for the recombinant proteins. (C) Schematic representation of Sl3-GT showing the primers used for ChIP–qPCR. Gray segments represent PCR primers, and stars indicate MYB binding motifs. ChIP–qPCR analysis of SlAN2-like-GFP and SlMYB54-Flag enrichment at the Sl3-GT promoter is shown. (D) EMSA showing that the SlAN2-like-His-MBP and GST-SlMYB54 fusion proteins bind directly to the Sl3-GT promoter in vitro. Biotin-labeled probes were incubated with SlAN2-like-His-MBP and GST-SlMYB54 protein, and shifted and free DNA (arrows) was separated on an acrylamide gel. The GST and His-MBP proteins were incubated with the labeled probe to serve as a negative control. Unlabeled probes were used as competitors. Mu-competitor, mutated probes in which the MYB binding motif 5′-TAACCA-3′ was replaced with 5′-TTTTTT-3′. CBB staining shows the amount of recombinant protein applied to the gel. The asterisks indicate the target bands for the recombinant proteins. (E) Schematic representation of SlMYB54 showing the primers used for ChIP–qPCR. Gray segments represent PCR primers, and stars indicate MYB binding motifs. (F) Relative expression of endogenous SlMYB54 in seeds of WT, SlMYB54-OE#1, and SlMYB54-OE#2 plants at 30 DPA. (G) ChIP–qPCR analysis of SlMYB54-Flag enrichment at the SlMYB54 promoter. (H) Schematic representation of SlAN2-like showing the primers used for ChIP–qPCR. Gray segments represent PCR primers, and stars indicate MYB binding motifs. (I) Relative expression of endogenous SlAN2-like in the leaves of WT and SlAN2-like-OE#1 plants. (J) ChIP–qPCR analysis of SlAN2-like-GFP enrichment at the SlAN2-like promoter. For (A), (C), (G), and (J), chromatin from SlAN2-like-OE#1 and SlMYB54-OE#1 plants was immunoprecipitated using anti-GFP and anti-Flag antibodies, respectively. qPCR analysis of target-gene promoter enrichment in immunoprecipitated DNA is presented as a percentage of input. ACTIN (Solyc03g078400) was used as a nonspecific target. For (F) and (I), SlUBI3 (Solyc01g056840) was used as the reference transcript. All data represent means ± SD (n = 3). Statistical significance was determined by two-tailed Student’s t test (∗∗p < 0.01). By contrast, ChIP–qPCR demonstrated that both SlAN2-like-GFP and SlMYB54-Flag were enriched at the P7 fragment of the Sl3-GT promoter (Figure 5C), and EMSAs confirmed these results (Figure 5D). Dual-LUC assays showed that SlAN2-like activated Sl3-GT expression, whereas SlMYB54 did not (Supplemental Figure 7). These results indicate that only SlAN2-like transcriptionally activates Sl3-GT, despite the fact that both MYBs bind to the Sl3-GT promoter. ChIP–qPCR also revealed that SlMYB54-Flag was enriched at its own promoter (Figure 5E–5G), suggesting that an auto-activation mechanism governs its transcription. Consistent with this notion, the transcript level of endogenous SlMYB54 was significantly higher in SlMYB54-OE plants than in the WT (Figure 5F). Similarly, gene expression and ChIP–qPCR data also showed that SlAN2-like could stimulate its own transcription (Figure 5H–5J). SlMYB54 and SlAN2-like compete for interaction with SlAN1 Surprisingly, the fruit peel of the two SlMYB54-OE lines had a lighter purple color (Figure 6A) and reduced anthocyanin content compared with that of the WT (Figure 6B). This phenotype was also evident in the SlMYB5-OE lines (Supplemental Figure 8A–8C). We proposed two possible explanations for this observation. First, SlMYB54 overexpression might induce the expression of PA-specific structural genes in the peel, causing PA biosynthesis to consume the precursors necessary for anthocyanin biosynthesis. This hypothesis was supported by the finding that SlANR and SlTT12 expression were significantly higher in the fruit peel of the two SlMYB54-OE lines (Figure 6C). However, we found no evidence of PA production in the peels of these lines, a finding that could be attributed to either reduced expression of shared structural genes or a lack of other crucial regulators necessary for PA biosynthesis in the peel (Figure 6C). Second, ectopic expression of SlMYB54 in the peel might interfere with the interaction between SlAN1 and SlAN2-like, potentially suppressing SlAN2-like-mediated transcription of anthocyanin-specific genes. The transcription of anthocyanin-specific genes, including SlAN2-like itself, was reduced in the peel of SlMYB54-OE lines compared with the WT (Figure 6C). We therefore designed experiments to determine whether SlMYB54 interfered with the interaction between SlAN1 and SlAN2-like. Yeast three-hybrid assays revealed that the induction of SlMYB54 expression on SD/−5 medium significantly inhibited the interaction between SlAN2-likeR3 and SlAN1 (Figure 6D). Similarly, the induction of SlAN2-like expression on SD/−5 medium impaired the interaction between SlMYB54R3 and SlAN1 (Figure 6D). Furthermore, pull-down experiments revealed that the ability of SlAN2-like-His-MBP to bind and pull down SlAN1-His decreased with increasing levels of GST-SlMYB54 when the concentrations of SlAN2-like-His-MBP and SlAN1-His were held constant (Figure 6E). Similarly, the ability of GST-SlMYB54 to pull down SlAN1-His decreased with increasing levels of SlAN2-like-His-MBP but was unaffected by increasing amounts of the control, His-MBP (Figure 6F). These results indicate that SlMYB54 competes with SlAN2-like for interaction with SlAN1 and may thus affect SlAN2-like-mediated transcription of anthocyanin-related genes in fruit peel. SlMYB54 disrupts the SlAN2-like–SlAN1 interaction to inhibit anthocyanin biosynthesis (A) Images of WT and SlMYB54-OE fruits collected at the green stage. Scale bars, 1 cm. (B) Anthocyanin contents and fresh extracts of light-exposed fruit peel from WT and SlMYB54-OE plants at 30 DPA. FW, fresh weight. Values are means ± SD (n = 6). (C) Relative expression of SlAN2-like, shared structural genes, anthocyanin-specific structural genes, and PA-specific structural genes in fruit peel from WT and SlMYB54-OE plants at 30 DPA. SlUBI3 (Solyc01g056840) was used as the reference transcript. Data are presented as means ± SD (n = 3). (D) Competition of SlMYB54 and SlAN2-like for interaction with SlAN1 as revealed by yeast three-hybrid assays. Yeast cells (top two groups) co-transformed with pGADT7-SlAN1 and pBridge-SlAN2-likeR3-SlMYB54 were grown on SD/−2, SD/−4, and SD/−5 media to assess whether SlMYB54 interferes with the interaction between SlAN1 and SlAN2-like. In parallel, yeast cells (bottom two groups) co-transformed with pGADT7-SlAN1 and pBridge-SlMYB54R3-SlAN2-like were grown on SD/−2, SD/−4, and SD/−5 media to assess whether SlAN2-like interferes with the interaction between SlAN1 and SlMYB54. (E) MBP pull-down assay showing that SlMYB54 interferes with the SlAN2-like–SlAN1 interaction. (F) GST pull-down assay showing that SlAN2-like interferes with the SlMYB54–SlAN1 interaction. CBB staining in (E) and (F) shows 10% of the total recombinant protein applied to the gel. Asterisks indicate the target bands of the recombinant proteins. For (B) and (C), statistical significance was determined by two-tailed Student’s t test (∗∗p < 0.01). SlJAF13 acts upstream of SlAN1 to regulate anthocyanin biosynthesis in fruit peel In addition to SlAN1, SlJAF13 has also been identified as a bHLH subunit in the MBW complex, where it plays a crucial role in regulating anthocyanin biosynthesis in fruit peel (Yan et al., 2020; Chen et al., 2022). We used CRISPR-Cas9 technology to generate Sljaf13 mutants and observed that their PA production was comparable to that of the WT (Figure 7A and 7B; Supplemental Figure 9A). This finding suggests that the two bHLH components, SlAN1 and SlJAF13, have distinct functions. SlAN1 regulates both anthocyanin and PA biosynthesis, whereas SlJAF13 specifically affects anthocyanin biosynthesis, with no discernible effect on PA production (Figure 7A and 7B). Yeast two-hybrid (Y2H) assays demonstrated that SlJAF13 could interact with both SlAN2-like and SlMYB54 (Supplemental Figure 9D). This observation implies that the specific function of SlJAF13 in regulating anthocyanin biosynthesis is not dependent on its exclusive interaction with the anthocyanin-specific MYB component. Conversely, although SlJAF13 was highly expressed in the peel, its expression in seed coats was negligible, as evidenced by published RNA-seq data (Supplemental Figure 9E), and this expression pattern could account for its specific role in anthocyanin production in the peel. SlJAF13 acts upstream of SlAN1 and SlAN2-like to specifically regulate anthocyanin biosynthesis, and mutation of PA biosynthesis genes results in earlier seed germination and a softer seed coat (A and B) Phenotypes of fruits (A) and seeds (B) from WT, Sljaf13#1, Slan1#6, Slan2-like#1, Sljaf13 Slan1#2, and Sljaf13 Slan2-like#1-1. Fruits were collected at the green stage, and seeds were harvested at the BR stage and stained with DMACA. Scale bars, 1 cm. (C) Anthocyanin contents of light-exposed fruit peel from WT, Sljaf13#1, Slan1#6, Slan2-like#1, Sljaf13 Slan1#2, and Sljaf13 Slan2-like#1-1 at 30 DPA. FW, fresh weight. (D) Relative expression of the regulatory genes SlAN2-like, SlAN1, and SlAN11 in fruit peel of WT and Sljaf13 plants at 30 DPA. (E) Relative expression of SlJAF13 in fruit peel of WT, Slan1, and Slan2-like at 30 DPA. (F) Images of WT, Sljaf13#1, and Sljaf13 SlAN1GFP-OE#1 fruit at the green stage. Scale bars, 1 cm. (G) Germination rates at three time points and images of germinated seeds from WT, Slan2-like#1, Sljaf13#1, Slan1#6, Slan11#1, Slmyb54#11, and Slmyb54 Slmyb5#1 at 36 h, 48 h, and 60 h. (H) Images of dry seeds before grinding (top panel) and after grinding (lower panel) from WT, Slan2-like#1, Sljaf13#1, Slan1#6, Slan11#1, Slmyb54#11, and Slmyb54 Slmyb5#1 fruit. Scale bars, 1 cm. (I) Proposed model for the biosynthesis of anthocyanins and PAs in tomato. The bHLH transcription factor SlAN1 and the WD40 regulator SlAN11, essential components of the MBW complex, are required for the biosynthesis of both anthocyanins and PAs. By contrast, the MYB components SlAN2-like and SlMYB54 specifically regulate anthocyanin biosynthesis in the fruit peel and PA biosynthesis in the seeds, respectively. In the fruit peel, SlAN2-like, SlAN1, and SlAN11 assemble into an MBW complex that activates the expression of anthocyanin biosynthetic genes. Specifically, SlAN2-like binds to the promoter of the anthocyanin-specific gene Sl3-GT. In addition, SlJAF13 acts upstream of SlAN1, SlAN11, and SlAN2-like to fine-tune anthocyanin biosynthesis (Montefiori et al., 2015; Sun et al., 2020; Chen et al., 2022). In seeds, SlMYB54, SlAN1, and SlAN11 form a distinct MBW complex that regulates the expression of PA biosynthetic genes. SlMYB54 specifically targets the promoter of the PA-specific gene SlANR. Arrows and lines represent activation. Different letters in (C) indicate significant differences based on one-way ANOVA, and data in (D), (E), and (G) were analyzed by two-tailed Student’s t test (∗∗p < 0.01; ∗p < 0.05; and ns, not significant). For (D) and (E), SlUBI3 (Solyc01g056840) was used as the reference transcript. All data are presented as means ± SD (n = 3). The second notable difference between SlAN1 and SlJAF13 is that the Sljaf13#1 mutant completely lacked anthocyanins in the fruit peel, in marked contrast to Slan1#6 (Figure 7A and 7C). The expression of all genes associated with anthocyanin biosynthesis (Supplemental Figure 9F), including SlAN2-like and SlAN1 (Figure 7D), was significantly reduced in Sljaf13 compared with the WT. Conversely, the transcript level of SlJAF13 was unaltered in all Slan1 and Slan2-like lines (Figure 7E). These findings indicate that SlJAF13 regulates the expression of SlAN1, whereas SlAN1 does not exert regulatory control over the expression of SlJAF13. To confirm this finding, we created the Sljaf13Slan1#2 and Sljaf13Slan2-like#1-1 double mutants (Supplemental Figure 9A–9C); both completely lacked anthocyanins in the fruit peel, a phenotype virtually identical to that of Sljaf13 (Figure 7A and 7C). Furthermore, overexpression of SlAN1 was able to partially, if not completely, reverse the anthocyanin-deficient phenotype of Sljaf13. (Figure 7F). These findings, together with prior research demonstrating that SlJAF13 binds directly to the SlAN1 promoter, provide compelling evidence that SlJAF13 functions upstream of SlAN1 in the regulation of anthocyanin biosynthesis. Fruit-derived anthocyanins and PAs orchestrate a dual-layered protective mechanisms for successful seed dispersal Our study revealed a tissue-specific accumulation pattern of anthocyanins and PAs in tomato fruits. Anthocyanins predominantly localized to the fruit peel, whereas PAs were primarily accumulated in the seeds. This spatial segregation of anthocyanin and PA biosynthesis prompted us to investigate its physiological significance. Notably, anthocyanins have been shown to extend tomato fruit shelf-life by delaying over-ripening and enhancing resistance to necrotrophic pathogens such as Botrytis cinerea (Chen et al., 2018). We propose that this external anthocyanin deposition constitutes an evolutionarily optimized protective interface, which not only maintains the structural integrity of the dispersal unit (fruit) during biotic/abiotic transit stress, but also serves as a visual signal to modulate frugivore-mediated seed dispersal. Because PAs have been reported to inhibit seed germination in Arabidopsis (Jia et al., 2012), we examined the seed germination characteristics of the PA-deficient mutants. PA-deficient mutants, such as Slan1#6, Slan11#1, Slmyb54#11, and Slmyb54 Slmyb5#1, showed accelerated germination compared with the WT. Conversely, seed germination was largely unaffected in the PA-unaltered mutants Slan2-like#1 and Sljaf13#1 (Figure 7G). Subsequent steel-bead grinding experiments showed that the seed coats of the PA-deficient mutants Slan1#6, Slan11#1, Slmyb54#11, and Slmyb54 Slmyb5#11 were more susceptible to damage than those of the WT and the PA-unaltered mutants Slan2-like#1 and Sljaf13#1 (Figure 7H; Supplemental Figure 10). These results suggest that seed-specific PA accumulation may serve to prevent premature germination of seeds while they are still inside the fruit during dispersal. Discussion Tissue-specific biosynthesis of anthocyanins and PAs requires different MYB TFs Cultivated tomato (S. lycopersicum) was domesticated from Solanum pimpinellifolium (Lin et al., 2014). Because S. pimpinellifolium does not typically accumulate anthocyanins in its fruit, the fruit peel of common cultivated tomato also lacks the capacity for anthocyanin biosynthesis. However, introgression of genetic loci (including Aft, atv, and Abg) from wild relatives has enabled some tomato cultivars to accumulate anthocyanins in the peel (Gonzali et al., 2009). By contrast, PA accumulation can be detected in seeds of tomato accessions from different domestication stages, including S. pimpinellifolium, S. lycopersicum var. cerasiforme, and S. lycopersicum (Supplemental Figure 11), suggesting that the capacity for PA production may have been retained during tomato domestication, serving as a protective mechanism to maintain seed coat hardness and prevent premature germination. In the present study, we observed that anthocyanins and PAs show tissue-specific patterns of accumulation, with anthocyanins found predominantly in the peel and PAs mainly in the seeds. Although both anthocyanins and PAs are derived from the phenylpropanoid pathway and share some intermediates, their biosynthetic processes diverge through the activity of branch-specific structural genes (Constabel, 2018). We found that anthocyanin-specific structural genes (e.g., Sl3-GT and SlGST) were exclusively expressed in the fruit peel (Figure 1D), whereas PA-specific structural genes (e.g., SlANR and SlTT12) were expressed only in the seeds (Figure 1D). These findings indicate that the tissue-specific production of anthocyanins and PAs reflects the stringently regulated expression of these branch-specific structural genes. Flavonoid biosynthesis is governed by the well-characterized MBW complex, and our data indicate that the MYB components play an important role in orchestrating the tissue specificity of anthocyanin and PA biosynthesis. First, unlike the bHLH component SlAN1 and the WDR component SlAN11, which were expressed in both peel and seeds, the MYB TFs SlAN2-like and SlMYB54 were expressed exclusively in the peel and seeds, respectively. Second, in contrast to SlAN1 and SlAN11, which were required for both anthocyanin and PA biosynthesis, SlAN2-like specifically regulated anthocyanin production in the peel, whereas SlMYB54 specifically regulated PA biosynthesis in the seeds. Finally, SlAN2-like and SlMYB54 differed in their abilities to bind to the promoters of anthocyanin- and PA-specific structural genes. These findings are consistent with previous studies in which distinct types of MYB TFs were required for the biosynthesis of anthocyanins and PAs in several plant species and organs. For instance, AtMYB75 (PAP1) and AtMYB90 (PAP2) are key activators of anthocyanin biosynthesis in Arabidopsis (Borevitz et al., 2000; Stracke et al., 2001), whereas AtMYB123 (TT2), AtMYB4, and AtMYB5 regulate PA biosynthesis in the same species (Nesi et al., 2001; Li et al., 2009; Wang et al., 2020b). In grapevine, MYB TFs from the anthocyanin-associated clade, such as VvMYBA1 and VvMYBA2, and from the PA-associated clade, including VvMYBPA1, VvMYBPA2, and VvMYB5a, specifically govern the biosynthesis of these two flavonoid types (Bogs et al., 2007; Walker et al., 2007). In tomatoes, SlAN2-like (AN2-like) is responsible for anthocyanin biosynthesis in the fruit peel (Sun et al., 2020; Yan et al., 2020), whereas SlMYB75 (AN2) facilitates anthocyanin biosynthesis in the hypocotyl (Zhi et al., 2020). In contrast to SlAN1, which participates in the biosynthesis of both anthocyanins and PAs in tomato fruits, the other bHLH component of the MBW complex, SlJAF13, specifically governs anthocyanin biosynthesis in the fruit peel and has no discernible effect on PA production in the seeds. Thus, SlAN1 and SlJAF13 appear to have distinct regulatory roles. Indeed, SlJAF13 operates upstream of SlAN1 in regulating anthocyanin biosynthesis. First, the Sljaf13 mutants displayed a complete absence of anthocyanins in the fruit peel, in marked contrast to the Slan1 mutants (Figure 7A and 7B). Second, the expression of all genes involved in anthocyanin biosynthesis, including SlAN1, was markedly reduced in the Sljaf13 mutants, whereas SlJAF13 transcription was unaffected by the Slan1 mutation. Moreover, overexpression of SlAN1 could partially, if not fully, rescue the anthocyanin-deficient phenotype of the Sljaf13 mutants. Finally, a previous study demonstrated that SlJAF13 directly binds to the SlAN1 promoter (Chen et al., 2022). Given that a bHLH factor functions upstream of SlAN1 to regulate anthocyanin biosynthesis, it is plausible that a comparable regulatory framework may also govern PA biosynthesis, potentially through a separate bHLH factor that acts upstream of SlAN1. Four bHLH proteins were identified as potential candidates for such a regulatory role based on their seed-specific expression patterns (Supplemental Figure 12A–12D). However, Y2H assays indicated that neither SlMYB54 nor SlAN2-like interacted with these four proteins (Supplemental Figure 12E). Therefore, this hypothesis requires further exploration. Our study and previous reports have demonstrated that the bHLH TFs SlJAF13 and SlAN1 can each form an MBW complex with SlAN2-like and SlAN11 to activate the anthocyanin biosynthetic pathway (Sun et al., 2020; Chen et al., 2022). The MBW complex containing SlJAF13 acts as an upstream regulatory module. SlJAF13 promotes the transcription of SlAN1, thereby facilitating the assembly of a secondary MBW complex containing SlAN1, which mediates further downstream regulation. Such a hierarchical regulatory mechanism not only allows the primary MBW complex to respond to light signals and initiate an appropriate response but also enables the downstream complex to enhance the regulation of anthocyanin biosynthetic genes through a cascade amplification effect. Similar hierarchical regulatory mechanisms are widely present in the transcriptional networks that govern responses to diverse external signals (Du et al., 2017; Liu et al., 2019; Zhang et al., 2025). SlAN2-like and SlMYB54 interact with SlAN1 to form two competitive MBW complexes The observation that typical tomato fruit peel does not naturally produce PAs motivated us to engineer a novel tomato genotype capable of high PA production in the peel or pericarp through overexpression of SlMYB54. We obtained several SlMYB54-OE lines, but none of them showed detectable PA production in the peel. Surprisingly, SlMYB54-OE lines exhibited a lighter purple color (Figure 6A) and reduced anthocyanin content compared with the WT (Figure 6B). We propose two possible explanations for this observation. First, overexpression of SlMYB54 may have triggered the upregulation of PA-specific structural genes in the fruit peel, depleting the common precursors (anthocyanidins) essential for both anthocyanin and PA biosynthesis. This hypothesis gains support from the observation that SlANR and SlTT12 expression were significantly increased in the fruit peel of the two SlMYB54-OE lines (Figure 6C). In addition, ectopic expression of BAN (ANR) in tobacco flower petals and Arabidopsis leaves has been shown to reduce anthocyanin levels (Xie et al., 2003). Conversely, mutation of FvANR has been shown to activate the anthocyanin biosynthetic pathway in strawberry (Xu et al., 2024). However, we did not observe any PA production in the fruit peel of the SlMYB54-OE lines, which could be attributed to reduced expression of shared structural genes or a lack of other crucial regulators necessary for PA biosynthesis. Second, because both SlAN2-like and SlMYB54 interact with the bHLH TF SlAN1, it is possible that ectopically expressed SlMYB54 might disrupt the interaction between SlAN1 and SlAN2-like, potentially inhibiting SlAN2-like-mediated transcription of anthocyanin-specific structural genes in the peel. Consistent with this hypothesis, the transcript levels of anthocyanin-specific genes, including SlAN2-like itself, in the peel were lower in SlMYB54-OE lines than in the WT. In addition, we provided further evidence that SlMYB54 indeed interferes with the interaction between SlAN1 and SlAN2-like, which may then affect the SlAN2-like-regulated anthocyanin biosynthesis pathway in the peel. It should be noted that the proposed competitive interaction model is based primarily on ectopic expression experiments. Given the strong, non-overlapping tissue-specific expression patterns of SlMYB54 and SlAN2-like in planta, direct competition between these two TFs is unlikely to occur under natural physiological conditions. However, this does not exclude the possibility that such a competitive interaction mechanism may contribute to the tissue-specific expression patterns of SlMYB54 and SlAN2-like in fruits. Both SlAN2-like and SlMYB54 can activate their own transcription. This positive feedback regulation, acting in concert with the competitive interaction mechanism, might further enhance the tissue-specific expression patterns of SlMYB54 and SlAN2-like. This possibility warrants further investigation. On the basis of our findings, we propose that SlAN2-like and SlMYB54 interact with SlAN1 to establish two distinct, yet competitive, MBW complexes that regulate anthocyanin and PA biosynthesis, respectively, in different tissues of tomato fruits. In the fruit peel, light serves as a key inducer, activating the expression of SlAN2-like (Sun et al., 2020). This TF then interacts with the bHLH component SlAN1 to form an anthocyanin-specific MBW module, thereby promoting anthocyanin biosynthesis. By contrast, a signal associated with seed development may trigger the expression of SlMYB54, which binds to SlAN1 to form a PA-specific MBW module. Notably, these anthocyanin- and PA-specific MBW complexes not only enhance the expression of branch-specific structural genes but also stimulate the transcription of MYB genes, creating a positive feedback loop to maintain and amplify the tissue-specific expression of the MYB genes. The competition between SlAN2-like and SlMYB54 for binding to SlAN1 thus constitutes a finely tuned regulatory mechanism. This ensures that the expression of branch-specific structural genes and MYB genes, as well as the biosynthesis of the two flavonoids, are precisely localized to their respective tissues. The spatially restricted accumulation of anthocyanins in the fruit peel and PAs in the seeds likely confers significant adaptive advantages. Anthocyanins have been shown to extend tomato fruit shelf life by delaying over-ripening and enhancing resistance to necrotrophic pathogens such as Botrytis cinerea (Chen et al., 2018). We propose that the anthocyanin-containing peel constitutes an evolutionarily refined protective interface that maintains the structural integrity of the dispersal unit, the fruit, during biotic and abiotic stresses during transit and also provides a visual signal to promote frugivore-mediated seed dispersal. In contrast to the external protection mediated by anthocyanins, the endogenous accumulation of PAs in seeds likely serves to prevent premature germination within the fruit and to strengthen the mechanical resistance of the seed coat, thus safeguarding embryonic tissues. This compartmentalized defense strategy—extrinsic protection for the carrier (fruit) versus intrinsic fortification for the payload (seed)—may constitute an elegant evolutionary solution for optimizing both dispersal success and post-dispersal establishment. In addition to their important roles in seed dispersal, anthocyanins and PAs are well recognized for their strong antioxidant activity, which enables them to scavenge reactive oxygen species, alleviate oxidative stress, and promote human health (Tena et al., 2020; Chen et al., 2025; Fu et al., 2025). Increased accumulation of these flavonoid metabolites in edible tissues not only improves the nutritional value and functional properties of fruits but also enhances biotic and abiotic stress tolerance (Yu et al., 2023; Liu et al., 2025; Sun et al., 2025). Therefore, our findings highlight the potential for manipulating tissue-specific MYB-dependent pathways to breed nutrient-rich and stress-resistant tomato cultivars. In summary, this study reveals tissue-specific regulatory mechanisms that govern anthocyanin and PA biosynthesis in tomato fruits, providing a foundation for engineering fruits with enhanced levels of these bioactive compounds to promote human nutrition and plant resilience (Figure 7I). Methods Plant materials and growth conditions The purple-fruited tomato line Jingzi 8035 (wild type, WT) carrying both the Aft and atv genes was developed by progeny selection from a cross between LA1996 (Aft) and LA3736 (atv). Jingzi 8035 can accumulate large amounts of anthocyanins in the fruit peel, whereas LA1996 only shows purple spots on the exocarp. Jingzi 8035 was used as the genetic background to create single and double mutants of SlAN1, SlAN11, SlAN2-like, SlMYB54, SlMYB5, and SlJAF13 and overexpression lines of SlAN1 and SlMYB54. LA1996 was used to overexpress SlAN1. All tomato lines were grown in greenhouses under the same conditions for phenotypic data collection and transcriptome and metabolite analyses. N. benthamiana plants were grown in a growth chamber at 25°C under long-day conditions (16 h light/8 h dark) for 5 weeks and used for biochemical experiments. Seed collection Fruits were harvested at the red ripe stage for seed collection. The seeds were surface-sterilized by soaking in 3 N HCl and 10% (w/v) trisodium phosphate for 30 min, washed with ddH2O, and air-dried. Images of seeds were captured with a scanner (Epson, Perfection V850 Pro, Japan) after 5 min of treatment with 1 N sodium hypochlorite. Seed germination measurement At least 100 seeds from the WT and the Slan2-like#1, Sljaf13#1, Slan1#6, Slan11#1, Slmyb54#11, and Slmyb54 Slmyb5#1 mutant lines were divided into three groups and placed in Petri dishes on filter paper moistened with distilled water (Gao et al., 2025). The dishes were incubated at 25°C in the dark. Radicle emergence was observed and recorded at 12-h intervals over a 96-h period. Seedling emergence was also monitored every 12 h, and the corresponding percentage emergence was calculated. Seed grinding experiment Dry seeds from the WT and the Slan2-like#1, Sljaf13#1, Slan1#6, Slan11#1, Slmyb54#11, and Slmyb54 Slmyb5#1 mutant lines were placed into 2.0-mL centrifuge tubes with 10 seeds per tube. Two stainless steel balls (3 mm in diameter) were added to each tube, and the seeds were ground using a high-throughput grinder (Bionoon-96, Shanghai, China) at 30 Hz (900 rpm) for 15 min. After grinding, the seeds were washed twice with ddH2O and imaged using a scanner (Epson Perfection V850 Pro, Japan). Seed area was measured using ImageJ. The seed area ratio was calculated as the individual area after grinding divided by the average area before grinding. Gene editing and genetic transformation Gene-specific sgRNA target sites (20 nucleotides in length) were selected using the CRISPR-P 2.0 website (http://cbi.hzau.edu.cn/cgi-bin/CRISPR). CRISPR-Cas9 constructs were generated using previously described methods. The completed constructs were introduced into Agrobacterium strain GV3101 and transformed into WT plants. The gene-edited tomato lines were verified by PCR using gene-specific primers by sequencing the target regions. All T0 plants derived from CRISPR-Cas9 gene editing were retained and used to develop T1, T2, and further progenies. Target sites and primer information are listed in Supplemental Table 6. To obtain SlAN1 overexpression lines, the full-length coding sequence (CDS) of SlAN1 was PCR amplified and inserted into the pSuper1300-Flag and pSuper1300-GFP vectors. The constructed plasmids were transformed into the tomato line LA1996. To produce SlMYB54 overexpression lines, the CDS of SlMYB54 was amplified and inserted into the pSuper1300-Flag vector. A 2072-bp sequence upstream of the SlAN1 start codon (proSlAN1) was amplified from tomato genomic DNA and cloned into the pCAMBIA2300-GUS vector to generate the proSlAN1:GUS construct. Similarly, a 2410-bp promoter fragment of SlMYB54 (proSlMYB54) was amplified and cloned into the pCAMBIA2300-GUS vector. All constructs were introduced into Agrobacterium strain GV3101 for transformation into WT or LA1996 tomato plants. Primer information is provided in Supplemental Table 6. RNA extraction and gene expression analysis The fruit outer pericarp and seeds at 30 DPA were collected for gene expression analysis. Seven tissues (roots, stems, leaves, flowers, exocarp, mesocarp, and seeds at 30 DPA) were harvested for analysis of SlAN2-like and SlMYB54 expression patterns. Three independent biological replicates and three technical replicates were performed for each gene. Each biological replicate contained samples collected from at least three fruits. Total RNA was isolated using the Eastep Super Total RNA Extraction Kit (Promega, Beijing, China, cat. #LS1040) and reverse transcribed into cDNA with the PerfectStart Uni RT&qPCR Kit (TransGen Biotech, Beijing, China, cat. #AUQ-01) following the manufacturer’s instructions. RT–qPCR was performed with gene-specific primers (Supplemental Table 6) on a Real-Time PCR Detection System using the PerfectStart Uni RT&qPCR Kit. SlUBI3 (Solyc01g056840) was used as the reference transcript. Transcriptome sequencing RNA-seq was performed using RNA isolated from seeds, with pectin removed, and the peel of WT fruits at 30 DPA. Three biologically independent replicates were included for each tissue. RNA library construction and sequencing on the NovaSeq 6000 platform (Illumina) were performed by Novogene (Beijing). Clean reads were mapped to the reference genome (SL4.0) of Heinz1706. Differentially expressed genes were identified using the following criteria: FPKM > 1 in at least one tissue, log2FoldChange > 1, and p-value < 0.05. Data are provided in Supplemental Tables 1, 2, 3, 4, and 5. Subcellular localization The p35S:SlMYB54-YFP-HA and p35S:SlMYB5-YFP-HA vectors were constructed using the CDSs of SlMYB54 and SlMYB5, respectively. NF-YA4-mCherry was used as a nuclear marker. The recombinant plasmids were transformed into Agrobacterium strain GV3101 for infiltration into the leaves of 5-week-old N. benthamiana plants, with co-infiltration of Agrobacterium harboring P19. Leaves were collected after 24 h in the dark and 24 h in the light, and YFP fluorescent signals were observed using a laser-scanning confocal microscope (Zeiss, LSM 710, Oberkochen, Germany). The primer sequences are listed in Supplemental Table 6. BiFC analysis The open reading frames of SlMYB54 and SlMYB5 were individually cloned into the pSY-CE vector, and that of SlAN1 was cloned into the pSY-NE vector (Walter et al., 2004). These constructs were transiently transformed into N. benthamiana leaves using a previously established method. In brief, Agrobacterium strains harboring the pSY-NE construct were co-infiltrated with those carrying the pSY-CE constructs. YFP fluorescence was detected using a laser-scanning confocal microscope (Zeiss, LSM 710). The primer sequences are listed in Supplemental Table 6. Anthocyanin and PA measurement Anthocyanin content was measured as described previously (Deikman and Hammer, 1995; Li et al., 2024; Qin et al., 2025) with modifications. A pericarp sample at 30 DPA was collected with a paring knife and ground into powder in liquid nitrogen. Approximately 0.1 g of powder was placed into 1 mL of extraction buffer (18% [v/v] 1-propanol and 1% [v/v] HCl in water) and incubated overnight at 4°C. The absorbance values (A535 and A650) of the supernatant were measured using a spectrophotometer (AS ONE, Japan) after centrifugation at 12,000 rpm for 20 min. The anthocyanin content was calculated as (A535−A650)/g fresh weight. PA ([+]-catechin) content was measured as described previously (Li et al., 1996; Himi et al., 2023) with modifications. A seed sample at the BR stage was stripped of pectin and ground into powder in liquid nitrogen. Approximately 0.1 g of powder was placed into 1 mL of extraction buffer (0.58% [w/v] 4-dimethylaminocinnamaldehyde and 0.58% [v/v] 6 N hydrochloric acid in methanol) and incubated for 30 min on ice. The supernatant was collected after centrifugation at 12,000 rpm and 4°C for 10 min, and absorbance was measured at 640 nm using a spectrophotometer. PA content was calculated as A640/g fresh weight. All data were collected from three independent biological replicates and at least two technical replicates. DMACA staining DMACA–HCl staining was used to visualize the presence of soluble PAs in the seed coat, as they form a black complex. Fresh seeds were collected from fruits at the BR stage and stained with DMACA–HCl staining buffer (1% [w/v] 4-dimethylaminocinnamaldehyde and 1% [v/v] 6N hydrochloric acid in absolute ethanol) as described previously (Bogs et al., 2007). After incubation at 4°C overnight, the seeds were washed three times with 70% ethanol and photographed using a digital camera. GUS staining Seeds from ProSlAN1:GUS and ProSlMYB54:GUS transgenic plants were vacuum infiltrated with staining buffer (50 mM PBS [pH 7.0], 2 mM X-Gluc, 0.1% [v/v] Triton X-100, 1 mM potassium ferrocyanide, 1 mM potassium ferricyanide, and 20% [v/v] methanol) for 30 min and then incubated overnight at 37°C in the dark. The seeds were washed in 70% ethanol until the tissues became clear. Images were obtained using a scanner (EPSON, Perfection V850 Pro). Co-IP assay The full-length CDSs of SlAN1 and SlMYB54 without stop codons were ligated into the pCAMBIA1300-GFP vector and the pCAMBIA2300-YFP-HA vector, respectively. Transient transformation of tobacco was performed using previously described methods. Samples were collected and ground into powder in liquid nitrogen, then fully mixed with IP buffer (10 mM HEPES [pH 7.0], 100 mM NaCl, 1 mM EDTA, 10% glycerol [v/v], 1% Triton X-100 [v/v], 1 mM PMSF, and Roche protease inhibitor cocktail). Immunoprecipitation was performed with anti-Flag magnetic beads (Beyotime, Cat. #P2115) at 4°C for 2 h. The beads were collected with a magnetic rack and washed five times with PBS buffer, then denatured at 100°C for 10 min. Proteins retained on the beads were separated by SDS–PAGE and detected by immunoblotting with anti-HA (Sigma Aldrich, Merck KGaA, Germany, Cat. #H9658) or anti-FLAG (Sigma-Aldrich, Cat. #F7425) antibodies. Primer information is listed in Supplemental Table 6. In vitro protein expression To produce the GST-SlAN1 and GST-SlMYB54 fusion proteins, the full-length CDSs of SlAN1 and SlMYB54 were amplified and cloned into the pGEX-4T-1 vector. Proteins were expressed in Escherichia coli BL21 (DE3) and purified using GST Bind Resin (GE Healthcare, General Electric Company, USA, cat#17-0756-01). To produce the SlAN1-His fusion protein, the full-length CDS of SlAN1 was amplified and cloned into pET-28a. To produce the SlAN2-like-His-MBP fusion protein, the full-length CDS of SlAN2-like was amplified and cloned into pET-28a-MBP (modified by our lab). The pET-28a-MBP vector was constructed by cloning the full-length MBP CDS and the Factor Xa cleavage site into pET-28a via homologous recombination, using BamHI and EcoRI as the double restriction sites. His-tagged proteins were purified using Ni-NTA Agarose. Primers used for plasmid construction are listed in Supplemental Table 6. In vitro pull-down assays To detect the SlAN1–SlMYB54 interaction using in vitro pull-down assays, 1 μg of in vitro translated GST or GST-SlMYB54 protein was first bound to magnetic agarose beads (Beyotime, Cat#P2258) in PBS buffer (1 mM DTT and 0.1% NP-40) at 4°C for 2 h; 1 μg of in vitro translated SlAN1-His protein was then added and incubated at 4°C for 2 h. For each reaction, 20 μL of magnetic agarose beads for GST-tagged protein was used. The beads were then collected and washed five times with PBS buffer. Proteins retained on the beads were analyzed by immunoblotting using anti-GST antibody (Beijing Protein Innovation, Beijing, China, Cat#AbM59001-2H5-PU) and anti-His antibody (Beijing Protein Innovation, Cat#AbM59012-18-PU). For competitive pull-down assays, 1 μg of SlAN1-His with 1, 2, or 5 μg of SlAN2-like-His-MBP was incubated with immobilized GST-SlMYB54 or GST at 4°C in PBS buffer (1 mM DTT and 0.1% NP-40) for 2 h. Proteins retained on the beads were analyzed by immunoblotting with anti-His and anti-GST antibodies; 10% of the volume of each experimental group of each fusion protein was separated by SDS–PAGE as input, and CBB-staining of polyacrylamide gels was used as a loading control. Electrophoretic mobility shift assay The 5′-biotin-labeled DNA probes were synthesized by Tsingke (Beijing Tsingke Biotech Co., Ltd., Beijing, China) and annealed into double strands. DNA gel shift assays were performed using a Chemiluminescent EMSA Kit (Beyotime, Cat. #GS009). Each EMSA reaction (10 μL) contained 0.5 μg of purified recombinant protein, 2 μL of biotin-labeled DNA probe (100 nM), and 2 μL of binding buffer. Transferred DNA and purified proteins were crosslinked using a UV lamp on the benchtop for 15 min. Biotin-labeled DNA was detected using the Chemiluminescent EMSA Kit. Primers used for plasmid construction are listed in Supplemental Table 6. Y2H assay The Y2H assays were performed using the Matchmaker GAL4 Two-Hybrid System (Clontech, USA). To investigate interactions of SlAN1 with SlMYB54 and SlMYB5, the full-length CDS of SlAN1 was cloned into the pGADT7 prey vector, and R3-domain-containing fragments of SlMYB54 and SlMYB5 were independently inserted into the pGBKT7 bait vector. All recombinant constructs were co-transformed into Y2HGold yeast cells, with the empty pGADT7 vector serving as a negative control. The yeast cells were grown on plates with medium lacking Leu and Trp (SD/−2) as a transformation control or lacking Ade, His, Leu, and Trp (SD/−4) to detect protein interactions. The interactions were examined after 3 days of incubation at 28°C. Primers used for vector construction are listed in Supplemental Table 6. Yeast three-hybrid assay The yeast three-hybrid assays were performed using the Matchmaker GAL4 Two-Hybrid System (Clontech). To construct the pBridge-SlMYB54R3-SlAN2-like recombinant vector, an R3-domain-containing SlMYB54 fragment was cloned into MCS I of the pBridge vector (Clontech) through fusion with the GAL4 DNA-binding domain (BD), and the full-length CDS of SlAN2-like was inserted into MCS II. Expression of the SlAN2-like “bridge” protein was methionine dependent, being activated only under methionine-deficient conditions. All recombinant constructs were co-transformed into Y2HGold yeast cells. Transgene integration was verified by growth on SD/−Trp−Leu (SD/−2) minimal medium. To analyze protein interactions, the SlAN1–SlMYB54R3 interaction was assessed on SD/−Trp−Leu−His−Ade (SD/−4) plates under conditions of SlAN2-like suppression. SlAN2-like-mediated regulatory effects were evaluated on SD/−Trp−Leu−His−Ade−Met (SD/−5) plates through methionine-deprivation-induced expression. Interactions were observed after 3 d of incubation at 28°C. In parallel, the pBridge-SlAN2-likeR3-SlMYB54 construct was generated by inserting an R3-domain-containing SlAN2-like fragment into MCS I and the full-length SlMYB54 CDS into MCS II. This design enabled conditional expression of the SlMYB54 bridge protein to examine its effects on the formation of the SlAN1–SlMYB54R3 complex. The experimental procedures were the same as those described above. Primers used for vector construction are listed in Supplemental Table 6. ChIP–qPCR ChIP–qPCR was performed as described previously (Sun et al., 2024). ChIP was performed using 1 g of leaf or flower tissues from SlAN2-likeGFP-OE#1 (SlAN2-like-GFP) and SlMYB54-OE#1 (SlMYB54-Flag) transgenic plants. The chromatin complexes were analyzed by RT–qPCR using the PerfectStart Uni RT&qPCR Kit on a real-time PCR instrument. A DNA fragment of the ACTIN (Solyc03g078400) 3′ intergenic region was used as an internal control. Primers used for vector construction and PCR are listed in Supplemental Table 6. Statistical analysis Statistical significance was determined by one-way analysis of variance (ANOVA) and two-sample Student’s t tests performed in GraphPad Prism 10. Accession numbers The gene sequences used in this study are available at the SOL Genomics Network (SGN) under the following accession numbers: SlMYB54 (Solyc12g049300), SlMYB5 (Solyc01g005660), SlAN2-like (SlMYB114) (Solyc10g086290), SlAN1 (Solyc09g065100), SlJAF13 (Solyc08g081140), SlAN11 (Solyc03g097340), SlPAL (Solyc09g007900), SlC4H (Solyc05g047530), Sl4CL (Solyc06g068650), SlCHS1 (Solyc09g091510), SlCHS2 (Solyc05g053550), SlCHI (Solyc02g067870), SlF3H (Solyc02g083860), SlF3′5′H (Solyc11g066580), SlDFR (Solyc02g085020), SlANS (Solyc08g080040), Sl3-GT (Solyc10g083440), SlGST (Solyc02g081340), SlTT12 (Solyc12g006360), SlANR (Solyc03g031470), SlMYC2 (Solyc08g076930), SlbHLH-like (Solyc03g120530), SlbHLH4 (Solyc01g081100), and SlbHLH137 (Solyc06g066580). Data and code availability The data that support the findings of this study are available in the main text and supplemental information of the article. Acknowledgments We thank the Tomato Genetic Resources Center at the University of California, Davis, for providing seeds of tomato lines LA1996 and LA3736, Dr. Ming Zhou at the Beijing Academy of Agriculture and Forestry Sciences, and Associate Professor Chuanlong Sun at Shandong Agricultural University for providing the SlAN2-like-overexpressing transgenic line. Author contributions Z.Q., M.D., and W.Y. conceived and designed the experiments. Z.Q. and H.C. performed most of the experiments. Y.S., Y.J., Y.L., L.Y., X.Z., M.Y., J.L., and Q.T. conducted some experiments. Z.Q., X.G., and H.L. analyzed the data. H.S. provided intellectual and experimental support. Z.Q., M.D., and W.Y. drafted and thoroughly revised the paper. All authors read and approved the final version of the manuscript. Published: May 6, 2026 Footnotes Supplemental information is available at Plant Communications Online. Contributor Information Minmin Du, Email: duminmin@cau.edu.cn. Wencai Yang, Email: yangwencai@cau.edu.cn. Supplemental information Document S1. Supplemental Figures 1–12 Document S2. Supplemental Tables 1–7 Document S3. Article plus supplemental information References Associated Data Supplementary Materials Document S1. Supplemental Figures 1–12 Document S2. Supplemental Tables 1–7 Document S3. Article plus supplemental information Data Availability Statement The data that support the findings of this study are available in the main text and supplemental information of the article.