Abstract

   Insights into flavor formation during fruit ripening can guide the
   development of breeding strategies that balance consumer and producer
   needs. Cherry tomatoes possess a distinctive taste, yet research on
   quality formation is limited. Here, metabolomic and transcriptomic
   analyses were conducted on different ripening stages. The results
   revealed differentially accumulated metabolites during fruit ripening,
   providing candidate metabolites related to flavor. Interestingly,
   several key flavor-related metabolites already reached a steady level
   at the mature green stage. Transcriptomic analysis revealed that the
   expression levels of the majority of genes tended to stabilize after
   the pink stage. Enrichment analysis demonstrated that changes in
   metabolic and biosynthetic pathways were evident throughout the entire
   process of fruit ripening. Compared to disease resistance and fruit
   color genes, genes related to flavor and firmness may have a broader
   impact on the accumulation of metabolites. Furthermore, we discovered
   the interconversion patterns between glutamic acid and glutamine, as
   well as the biosynthesis patterns of flavonoids. These findings
   contribute to our understanding of fruit quality formation mechanisms
   and support breeding programs aimed at improving fruit quality traits.

   Keywords: omics, cherry tomato, flavor, breeding

1. Introduction

   Improving fruit flavor while maintaining high yields, disease
   resistance, and long postharvest shelf life poses a significant
   challenge for breeders [[32]1,[33]2]. To achieve this goal, it is
   essential to gain a comprehensive understanding of how flavor compounds
   are formed throughout the fruit ripening process. Tomato (Solanum
   lycopersicum) serves as a primary model system for studying climacteric
   fleshy fruit development and ripening [[34]3]. Cherry tomatoes (Solanum
   lycopersicum var. cerasiforme), as one of the tomato subspecies, are
   widely planted in China and are favored by consumers due to their
   intense flavor. Along with apples, grapes, and bananas, cherry tomatoes
   are among the four priority fruits supported by the Food and
   Agriculture Organization of the United Nations. Compared to the
   “big-fruit” tomato, cherry tomatoes exhibit higher levels of sugar
   (~44% higher), organic acids (~5% higher), and volatile organic
   compounds [[35]4,[36]5,[37]6]. This makes them an excellent model for
   analyzing the formation of flavors.

   The ripening process of red-fruited-clade tomato fruit can be divided
   into five stages: the mature green stage, the breaker stage, the pink
   stage, the light-red stage, and the red ripe stage. These stages are
   characterized by physiological and biochemical changes that ultimately
   lead to alterations in the appearance, texture, flavor, and aroma of
   the fruit [[38]3,[39]7]. Key manifestations of this process include
   changes in ethylene levels, gradual alterations in fruit color, fruit
   softening, and modifications in the metabolism of sugars, acids, and
   other compounds, as well as the production of volatile substances
   [[40]8]. Specifically, the ripening process involves the accumulation
   and spread of autocatalytic ethylene, which leads to a series of
   phenotypic changes [[41]9]. The accumulation of carotenoids, such as
   lycopene and β-carotene, contributes significantly to changes in color
   [[42]3]. Fruit softening and increased juiciness are primarily caused
   by the partial disassembly of the cell wall [[43]10]. Additionally, the
   formation of sugars, organic acids, and volatiles, and even some bitter
   compounds, such as steroidal alkaloids and their glycosylated forms, as
   well as flavanone derivatives, can affect the fruit’s flavor and aroma
   [[44]11,[45]12].

   To date, numerous key genes related to fruit ripening in tomatoes have
   been identified, including ACC, ACS, and ACO genes involved in direct
   ethylene synthesis [[46]13,[47]14,[48]15,[49]16,[50]17], CNR [[51]18],
   PSY1 [[52]19,[53]20], and OG [[54]21] influencing fruit color
   transformation, PG [[55]22,[56]23,[57]24], PL [[58]25,[59]26], and PME
   [[60]27] affecting fruit softening and juice viscosity, as well as E8
   [[61]28], LIN5 [[62]29], PAR1, and PAR2 [[63]30] influencing fruit
   flavor. Furthermore, various metabolites that impact fruit flavor
   formation have been documented, encompassing sugars, acids, amino
   acids, fatty acids, terpenoids, bitter metabolites, and other
   substances [[64]12]. However, despite these advancements, connecting
   multiple ripening-related genes with the signaling and coordination
   mechanisms that induce ripening remains a significant challenge
   [[65]31]. The complex gene expression networks associated with
   metabolite changes during fruit ripening largely remain unknown.

   Omics approaches, such as metabolomics and transcriptomics, can
   contribute to our understanding of the genetic, hormonal, and metabolic
   networks that regulate tomato fruit development and ripening
   [[66]32,[67]33]. Integrative analysis of metabolite and transcript
   levels is a well-established approach that has been applied to various
   fruit crops including tomato [[68]34], ponkan [[69]35], kiwifruit
   [[70]36], and litchi [[71]37]. However, cherry tomatoes, being a
   popular commodity in the fresh market, have been relatively
   under-studied in this context. Hence, in this study, we conducted a
   combined metabolomic and transcriptomic analysis of cherry tomatoes to
   investigate the mechanisms underlying flavor formation during ripening.
   We selected the Yu-nu cultivar as the material due to its widespread
   cultivation in southeast China, as well as its reputation for high
   yield, medium firmness, and intense flavor. Considering the growth
   pattern and flavor changes, fruit samples at the mature green (MG),
   pink (Pk), and red ripe (RR) stages were collected for analysis
   ([72]Figure 1a). Through the identification of differentially
   accumulated metabolites (DAMs) and differentially expressed genes
   (DEGs), we explored the changes in metabolite accumulation and gene
   expression during fruit ripening. The integrated analysis provided a
   systematic perspective on the biosynthesis of flavor-related
   metabolites. These findings contribute to our understanding of tomato
   breeding strategies aimed at striking a balance between consumer
   sensory experience and expected production qualities.

Figure 1.

   [73]Figure 1
   [74]Open in a new tab

   Metabolome analysis of fruit at different stages: mature green (MG),
   pink (Pk), and red ripe (RR): (a) Representative picture of Yu-nu fruit
   at three sampling stages. (b) Venn diagram showing differentially
   accumulated metabolites (DAMs) between MG-Pk and Pk-RR stages. (c)
   Accumulation profiles of DAMs from MG to Pk. The heatmap on the left
   illustrates changes in DAM accumulation, and the bar charts on the
   right depict the enrichment of metabolite sets for increased (UP) and
   decreased (DOWN) DAMs. (d) Accumulation profiles of DAMs from Pk to RR.
   The heatmap on the left illustrates changes in DAM accumulation, and
   the bar charts on the right depict the enrichment of metabolite sets
   for increased (UP) and decreased (DOWN) DAMs.

2. Results

2.1. Metabolic Changes during Fruit Ripening

   In order to investigate the metabolic changes that occur during fruit
   ripening, we conducted a quantification of metabolites at three
   different stages of ripening: mature green (MG), pink (Pk), and red
   ripe (RR). A representative picture of the fruits is shown in
   [75]Figure 1a. Principal component analysis (PCA) based on raw mass
   spectrometry data revealed similarities among replications and
   noticeable differences between the three ripening stages ([76]Figure
   S1a and [77]Figure 1b). Overall, we identified a total of 420
   metabolites, categorized into various groups including 67 benzenoids,
   66 organic acids, 62 fatty acyls, 27 polyketides, 25 nucleotides and
   nucleotide derivatives, and 16 carbohydrates and other metabolites
   ([78]Tables S1 and S2, [79]Figure S1c).

   During the MG-Pk stage, a total of 71 differentially accumulated
   metabolites (DAMs) were detected ([80]Figure 1b), with 50 metabolites
   showing an increase and 21 metabolites showing a decrease ([81]Figure
   1c). The increased metabolites were found to be significantly enriched
   in aromatic metabolite sets, such as benzamides, purines, pyrimidines,
   cinnamic acids, and benzenediols, as well as sugar-related metabolite
   sets like disaccharides and glycosyl compounds. On the other hand, the
   decreased metabolites were significantly enriched in
   unpleasant-smelling pyridines, monosaccharides, and TCA acids. These
   changes in metabolite levels may have an impact on the initial
   formation of flavor and aroma.

   Similarly, during the Pk-RR stage, a total of 79 DAMs were detected
   ([82]Figure 1b), with 39 metabolites showing an increase and 40
   metabolites showing a decrease ([83]Figure 1d). The increased
   metabolites were enriched in aromatic metabolite sets such as
   benzamides, purines, phenylacetic acids, and phenols, as well as in the
   odor-related metabolite set of aldehydes. Likewise, the decreased
   metabolites were also enriched in aromatic metabolite sets including
   benzamides, indoles, cinnamic acids, and benzenediols. These
   metabolites may serve as intermediates in the process of flavor and
   aroma formation.

   Overall, across all three stages, a total of 24 DAMs were detected
   ([84]Figure 1b). Among these, 12 metabolites (2-hydroxyglutarate,
   2-isopropylmalic acid, adenosine, coumarin, galactaric acid, L-aspartic
   acid, L-glutamic acid, L-proline, p-coumaroyl quinic acid,
   phosphorylcholine, psilocybin, and salidroside) showed continuous
   increments, 4 metabolites (5-methyltetrahydrofolic acid, D-ribose,
   L-threonine, and orotic acid) showed continuous decrements, and 10
   metabolites (12-hydroxydihydrochelirubine,
   2-dehydro-3-deoxy-L-rhamnonate, D-xylitol, glycerophosphocholine,
   L-homophenylalanine, m-coumaric acid, putrescine, pyroglutamic acid,
   quercetin 3-2G-xylosylrutinoside, and trehalose) exhibited an initial
   increase followed by a decrease ([85]Figure 1b, [86]Tables S3 and S4).
   These DAMs might provide candidate metabolites related to flavor.

2.2. Gene Expression Changes during Fruit Ripening

   To investigate the expression patterns of genes during fruit ripening,
   we conducted whole-transcriptome sequencing at three distinct ripening
   stages. A total of 47.85 Gb of raw sequences with a read length of 150
   bp was generated. From this dataset, we retained 68,677,640
   high-quality reads, achieving an average alignment rate of 94.8%
   ([87]Table S5). Utilizing the RNA-seq analysis pipeline, we detected
   16,978 genes expressed during fruit ripening, with an average read
   count per million (CPM) equal to or greater than 1. Notably, 13,514
   genes were found to be expressed across all three stages.

   A total of 2169 differentially expressed genes (DEGs) were identified
   in fruits between the MG stage and Pk stage, including 680 upregulated
   genes and 1489 downregulated genes ([88]Table S6, [89]Figure 2a). Gene
   Ontology (GO) annotations of the DEGs revealed enrichment in biological
   processes such as organ development and morphogenesis, cellular
   components including the membrane and cell wall, and molecular
   functions like catalytic activity and transporter activity ([90]Table
   S7, [91]Figure 2b). Furthermore, Kyoto Encyclopedia of Genes and
   Genomes (KEGG) annotations of the DEGs indicated enrichment in
   metabolism, flavonoid biosynthesis, isoflavonoid biosynthesis,
   phenylpropanoid biosynthesis, as well as in the biosynthesis of other
   secondary metabolites ([92]Table S8, [93]Figure 2c).

Figure 2.

   [94]Figure 2
   [95]Open in a new tab

   Transcriptome analysis of fruit at the mature green (MG), pink (Pk),
   and red ripe (RR) stages: (a) Volcano plot showing the differentially
   expressed genes (DEGs) between MG and Pk. Red dots represent
   upregulated genes, while blue dots represent downregulated genes. The
   same applies to (d). (b) GO enrichment analysis of DEGs between MG and
   Pk. (c) KEGG pathway enrichment analysis of DEGs between MG and Pk. (d)
   Volcano plot showing the DEGs between Pk and RR. (e) GO enrichment
   analysis of DEGs between Pk and RR. (f) KEGG pathway enrichment
   analysis of DEGs between Pk and RR. (g) Expression profiles of genes
   involved in color, texture, and flavor from MG to RR. The color
   transition from blue to red corresponds to increasing expression
   levels.

   Subsequently, in fruits transitioning from the pink (Pk) stage to the
   ripe red (RR) stage, we identified a total of 1165 DEGs. Among these
   DEGs, 165 genes were upregulated, while 1000 genes were downregulated
   ([96]Table S9, [97]Figure 2d). The GO annotations of the DEGs revealed
   enrichment in the biological processes of metabolism and
   photosynthesis, the cellular components of the extracellular region,
   cell wall, and plastid, and the molecular functions of catalytic
   activity and lyase activity ([98]Table S10, [99]Figure 2e). Moreover,
   the KEGG annotations of the DEGs indicated enrichment in metabolism,
   photosynthesis, and biosynthesis of other secondary metabolites
   ([100]Table S11, [101]Figure 2f).

   The comparison between the MG-Pk and Pk-RR stages revealed only 356
   common DEGs, indicating distinct physiological and biochemical
   responses before and after the pink stage of fruit ripening. We focused
   on analyzing well-characterized genes involved in color, texture, and
   flavor from the DEGs ([102]Figure 2g). Interestingly, most of the genes
   related to texture and flavor showed significant upregulation or
   downregulation at the Pk stage, and their expression levels remained
   stable during the RR stage. For instance, AFF, known for promoting
   locule gel formation and improving firmness and solids content at low
   expression levels [[103]38], showed lower expression at the Pk stage
   and insignificant changes during the RR stage. Similarly, PAR2, which
   enhances the accumulation of 2-phenylethanol and reduces the
   accumulation of 2-phenylacetaldehyde, thereby influencing aroma and
   flavor [[104]30], displayed a similar expression pattern to AFF at the
   Pk-RR stage. However, the expression of several genes related to flavor
   showed continuous changes from the Pk stage to the RR stage. For
   example, ADH2, which encodes alcohol dehydrogenase and affects the
   aroma derived from the lipoxygenase pathway [[105]39], was continuously
   upregulated from the MG stage to the RR stage. This suggests that these
   genes might enhance flavor during the whole ripening stage.

   There were 80 KEGG terms commonly enriched in both the MG-Pk and Pk-RR
   stages, mainly involving metabolite biosynthesis and various metabolic
   pathways. These findings reflect the continuous variations in abundant
   metabolites during the fruit ripening process ([106]Table S8,
   [107]Table S11). Furthermore, 30 KEGG terms were exclusively enriched
   in the MG-Pk stage, encompassing several pathways involved in signaling
   and transport. Additionally, 13 KEGG terms were exclusively enriched in
   the Pk-RR stage, incorporating pathways related to energy and
   photosynthesis, which may be associated with the physiological
   transition from chloroplasts to chromoplasts and seed development
   [[108]40,[109]41,[110]42]. These enrichment analyses of DEGs reveal
   significant differences in metabolic pathways among the different
   ripening stages.

2.3. Correlation Analysis between DAMs and DEGs

   To understand patterns linking the transcriptome and metabolome,
   correlations were calculated between the abundance of DEGs and DAMs. A
   total of 5332 pairwise correlations were identified, involving 1289
   genes and 97 metabolites. Among these correlations, the top ten
   metabolites displayed the highest number of correlations with 934 genes
   ([111]Figure 3a, [112]Table S12), including seven metabolites
   (adenosine, amygdalin, cytosine, naringenin, naringenin chalcone,
   ubiquinone-1, and umbelliferone) that were found to be negatively
   correlated with the expression of a majority of DEGs and three
   metabolites (spermine, carvone, and beta-alanyl-L-lysin) that were
   positively correlated, indicating their involvement in a complex
   regulatory network during fruit ripening.

Figure 3.

   [113]Figure 3
   [114]Open in a new tab

   Correlation analysis between DAMs and DEGs: (a) Visualization of the
   correlation between DAMs (top ten) and DEGs in a network graph. Nodes
   represent DAMs and DEGs, and node size reflects the abundance of
   correlation. Correlated nodes are connected by edges, with pink
   indicating positive correlation and blue indicating negative
   correlation. (b) Box plot showing the abundance of correlated DAMs for
   well-characterized genes. Dots represent the well-characterized genes
   (color: GF, SlWRKY35, UNIFORM, WF, SlMYB12, and ZISO; disease
   resistance: Cf-2, Cf-10, Ph-2, Sm, Rx4, and four quantitative trait
   loci; flavor: AADC2, ADH1, ADH2, AGPL1, AGPL3, E8, GORKY, INVINH1,
   LOXA, PAL5, PAR2, SGT2, and SlCGT; and texture: AFF, FUL1, PG2a, qFIS1,
   SlERF.D7, Solyc11g011300, TBG4 and TBG6).

   Furthermore, to understand the correlation between gene expression and
   metabolite accumulation, we examined the DAMs that showed significant
   correlations with well-characterized genes related to color, texture,
   disease resistance, and flavor ([115]Table S13, [116]Figure 3b).
   Interestingly, our analysis revealed that the genes involved in flavor
   exhibited more pronounced changes in metabolite abundance compared to
   those correlated with color (T-test p-value = 8.83 × 10^−3). This
   observation suggests that alterations in fruit flavor are accompanied
   by a wide range of changes in metabolite composition, while fruit color
   may not exert a dominant influence on flavor. Additionally, none of the
   DAMs were significantly correlated with disease resistance genes, while
   two of eight firmness-related genes (ERF.D7 and TBG6) were found to be
   correlated with five or more metabolites ([117]Table S12), suggesting
   that firmness-related genes might led to a greater change in
   metabolites compared to disease resistance genes.

2.4. Integrative Analysis of KEGG Pathways

   To investigate the dynamic changes in fruit flavor during ripening, we
   performed an integrative analysis of the transcriptome and metabolome.
   Previous studies have demonstrated a significant association between
   glutamic acid and overall fruit flavor intensity and consumer
   preference [[118]2]. Meanwhile, more DAMs and DEGs were identified in
   the glutamate metabolism pathway, facilitating us to gain insights into
   the metabolic transformations in this pathway ([119]Table S14). As
   depicted in [120]Figure 4, the concentration of glutamic acid increased
   continuously throughout fruit ripening, concomitant with a gradual
   decrease in oxoglutaric acid levels, which is consistent with the
   flavor development process. Additionally, we observed the upregulation
   of Solyc04g014510 (gst1, encoding glutamine synthetase cytosolic
   isozyme 1-1) and Solyc11g011380 (GS1, encoding glutamine synthetase),
   along with the downregulation of Solyc03g083440 (LOC101254281, encoding
   glutamate synthase 1), indicating a reciprocal conversion between
   glutamic acid and glutamine during fruit ripening. Specifically, during
   the MG-Pk stage, glutamic acid is converted to glutamine, while during
   the Pk-RR stage, glutamine is converted back to glutamic acid.
   Furthermore, the conversion from glutamic acid to gamma-aminobutyric
   acid and then to succinic acid semialdehyde was found to occur during
   the Pk-RR stage, and the corresponding genes, Solyc11g011920 (GAD2,
   encoding glutamate decarboxylase isoform 2), Solyc01g005000 (GAD3,
   encoding glutamate decarboxylase isoform 3), and Solyc12g006470
   (GAME12, encoding gamma aminobutyrate transaminase 2), showed a lower
   expression level at this stage.

Figure 4.

   [121]Figure 4
   [122]Open in a new tab

   Glutamate metabolism pathway from KEGG, depicting the expression and
   accumulation profiles of DEGs and DAMs. The color gradient indicates
   the level of gene expression, ranging from blue (low expression) to red
   (high expression). Similarly, the color gradient represents the level
   of metabolite accumulation, transitioning from green (low accumulation)
   to pink (high accumulation).

   In the context of the flavonoid biosynthesis pathway, we identified
   nine DEGs and nine DAMs, providing a comprehensive view of the changes
   in flavonoid metabolism ([123]Figure 5, [124]Table S14). Specifically,
   the expression of Solyc01g096670 (LOC101246092, encoding p-coumaroyl
   quinate) showed a negative correlation with 5-O-caffeoylshikimic acid
   accumulation and a positive correlation with cholorogenic acid.
   Interestingly, both metabolites interacted with caffeoyl-CoA and
   displayed opposite accumulation profiles, suggesting a potential
   competitive relationship involving Solyc01g096670 in their
   biosynthesis. Moreover, when comparing the expression profiles of the
   DEGs Solyc02g083860 (F3H, encoding flavanone 3-dioxygenase),
   Solyc11g013110 (LOC101249699, encoding flavonol synthase), and
   Solyc03g115220 (F3′H, encoding flavonoid 3′-monooxygenase) with the
   accumulation profile of the dihydrokaempferol-kaempferol-quercetin
   pathway, we observed a closer resemblance, indicating that this pathway
   predominantly operates during fruit ripening, rather than the
   dihydrokaempferol-taxifolin-quercetin pathway. These findings
   contribute to our understanding of the mechanisms underlying the
   formation of fruit quality.

Figure 5.

   [125]Figure 5
   [126]Open in a new tab

   Flavonoid biosynthesis pathway from KEGG, illustrating the expression
   and accumulation profiles of DEGs and DAMs. The color gradient
   indicates the level of gene expression, ranging from blue (low
   expression) to red (high expression). Similarly, the color gradient
   represents the level of metabolite accumulation, transitioning from
   green (low accumulation) to pink (high accumulation).

3. Discussion

   The formation of fruit flavor is a complex and multifactorial process
   that is not yet fully understood. In this study, we aimed to gain a
   comprehensive understanding of fruit ripening in cherry tomatoes by
   integrating transcriptome and metabolome data. By identifying DAMs and
   DEGs, we were able to unravel the patterns of metabolite accumulation
   and gene expression during fruit ripening. And the integrative analysis
   provided valuable insights into the development of improved fruit
   flavor and tomato breeding.

   Fruit flavor development is typically associated with the ripening
   process [[127]43]. Previous research has extensively focused on
   identifying metabolites related to tomato fruit flavor
   [[128]30,[129]33,[130]44,[131]45]. In our study, we detected a total of
   124 DAMs in cherry tomatoes (cv. Yu-nu), some of which have been
   previously reported to be associated with flavor, such as glutamic acid
   and phenol [[132]2,[133]46]. Interestingly, certain key flavor
   compounds, including glucose, fructose, citric acid, and
   2-phenylethanol, which have been reported to contribute to flavor
   preferences [[134]1,[135]30], did not show significant changes from the