Integrated Analysis of Metabolomics and Transcriptomics of the Differences in Flower Colors of Hybrid Cherry Blossoms

Abstract

Flower color, as an important trait of ornamental plants, has been a research hotspot in recent years. In this study, we selected Prunus campanulata (Maxim.) (ZH, red), P. dielsiana (Schneid.) (WH, white), and two cherry blossom varieties ‘Yanzhi Fei’ (FH, deep pink) and ‘Yanzhi Xue’ (XH, pinkish white) obtained by open-pollination hybridization as material. By means of bioinformatics methods such as metabolomics and transcriptomics, it is expected to deeply study the molecular mechanism of the gradient changes in flower color between the parents and offspring of cherry blossoms. Metabolomics analysis indicated that a total of 84 flavonoid related metabolites were identified, among which 31 were associated with the anthocyanin metabolic pathway, including three major types of anthocyanin substances: cyanidin, delphinidin, and malvidin. Transcriptome analysis showed that a total of 7712 differential genes were detected between P. campanulata and P. dielsiana; there were 3948 differential genes between P. campanulata and ‘Yanzhi Xue’, 2802 between P. campanulata and ‘Yanzhi Fei’, and 2511 between ‘Yanzhi Xue’ and ‘Yanzhi Fei’. After screening based on anthocyanin accumulation, nine key enzyme genes were obtained. Joint analysis showed that the relative expression trends of structural genes such as PAL, 4CL, CHI, DFR, and CYP75B in the samples were consistent with those of anthocyanins, and they had a high correlation with downstream metabolites. The results of this study lay a certain scientific foundation for the future directional improvement and breeding of cherry blossom colors.

1. Introduction

Flower color is one of the most important characteristics of ornamental plants, and the occurrence of flower color changes is jointly influenced by many internal and external factors. Therefore, understanding the mechanism of color development and its regulatory mechanism can provide an important theoretical basis and prerequisite for the cultivation and improvement of new ornamental plant varieties. The main factor affecting plant flower color is the accumulation of plant pigments in petals, which can be divided into three major categories: flavonoids, carotenoids, and alkaloids [1]. Flavonoid is a general term for a class of secondary metabolites widely present in plants. There are many types of flavonoid compounds, among which anthocyanins are the most important factor for color formation [2]. For example, plants such as Malus pumila [3], Armeniaca vulgaris [4], and Populus przewalskii [5] are all rich in a large amount of anthocyanin substances. Carotenoids represent isoprene compounds, which are often found in yellow-flowered or yellow-fruited plants such as Tagetes erecta and Prunus persica, and serve as the key factors that give their fruits or petals a yellow appearance [6,7]. Alkaloids are mainly dark purple berberine, yellow papaverine, and red betaine [8], and are found only in a few plants.

Flavonoid compounds are involved in regulating various developmental processes in the plant body [9], including the formation of flower color, fruit color and fragrance, and changes in plant stress resistance [10]. Among flavonoids, anthocyanins have the most significant regulatory effect on plant flower color. Anthocyanins can make plant petals appear in many different colors from yellow to purple [11]. Anthocyanins belong to the class of flavonoid compounds. As important secondary metabolites in plants, they exhibit strong antioxidant activity [12]. Anthocyanins are the largest family of secondary metabolites in plants. Most anthocyanins will form glycosides through glycosidic bonds with one or more monosaccharides, that is, anthocyanins [13]. More than 635 anthocyanins have been identified in nature [14], and most of the current research has focused on six of them: delphinidin, peonidin, pelargonidin, cyanidin, petunidin and malvidin [15]. The synthesis of anthocyanins begins with the phenylalanine pathway. Precursor substances are first catalyzed by upstream enzymes encoded by genes such as Phenylalanine ammonia-lyase (PAL), 4-Coumarate-CoA ligase (4CL), Chalcone synthase (CHS), and Chalcone isomerase (CHI), and then various anthocyanins are formed through the action of downstream enzymes encoded by genes including Dihydroflavonol 4-reductase (DFR) and Anthocyanidin synthase (ANS). This process ultimately influences the pigmentation of plants [16,17,18].

At present, many studies have jointly analyzed the mechanism of plant flower color synthesis through metabolomics and transcriptomics. For example, Zhang et al. [19] showed that the flower color difference of different Rosa hybrida varieties may be affected by the content difference of flavonoids and carotenoids and related genes; Fan et al. [20] discovered that naringenin and dihydrokaempferol are key metabolites in anthocyanin biosynthesis pathway of E. grandiflorum by searching for key genes and metabolites in the formation of Eustoma grandiflorum bicolor flower, and CHS is the key gene regulating its color formation; Qiu et al. [21] screened key metabolites and key genes related to flower color synthesis of Dendrobium nobile through metabolome and transcriptome, providing theoretical basis for flower color breeding of D. nobile.

Targeted breeding for flower color in ornamental plants is one of the key demands in the horticultural industry. Traditional breeding models mostly rely on phenotypic selection, often facing practical challenges such as long cycles and insufficient targeting [22]. Marker-assisted selection (MAS) conducts screening through key gene/metabolite markers, which theoretically holds promise for improving breeding efficiency. However, its effective application typically requires the identification of core regulatory targets and molecular markers underlying flower color variation [23]. Currently, integrated multi-omics analysis of gradient flower color variation in cherry blossom hybrids has not been fully carried out, which may to some extent leave cherry blossom flower color breeding lacking precise molecular target support. Therefore, establishing a systematic analysis method for “metabolite-gene” associations is of great practical significance, and is expected to provide potential technical references for the targeted breeding of cherry blossom flower color.

Prunus campanulata (red-flowered) and P. dielsiana (white-flowered) are important parental materials for cherry blossom breeding [24,25,26]. Their hybrid offspring, ‘Yanzhi Fei’ [27] (dark pink-flowered) and ‘Yanzhi Xue’ [28] (pink-white-flowered), form a natural flower color gradient, which provides an ideal material for analyzing the molecular mechanism of flower color variation (Figure 1). The cultivation process of the hybrid offspring is as follows: from April to May 2012, a large number of seeds were collected from the female parent P. dielsiana; from May 2012 to February 2013, the collected seeds were subjected to sand storage and low-temperature preservation; in March 2013, field sowing was carried out to obtain seedlings; in February 2016, variant individual plants with bicolor petals, pink-white petals and dense flowers were identified among the flowering seedlings; from 2016 to 2018, through continuous observation and selection, the traits of these variant individuals were confirmed to be stable; in the spring of 2018, the stable variant individual was used as the female parent, its branches as scions, and P. campanulata seedlings as rootstocks for grafting propagation, resulting in 30 grafted plants that flowered in the same year with consistent trait performance; in 2019, continuous observation and verification showed that the traits of the clonal grafted seedlings remained consistent and stable.

Figure 1.

Comparison pictures of whole and individual cherry trees of four types. Note: (a), P. campanulata; (b), ‘Yanzhi Fei’; (c), ‘Yanzhi Xue’; (d), P. dielsiana. The 1 cm scale bar in this figure provides a size reference for the individual flowers in the lower section.

This research aims to reveal the molecular mechanism underlying the gradient flower color variation in cherry blossoms, while attempting to screen key enzyme genes and characteristic metabolites of anthocyanin biosynthesis, thereby providing data support for deciphering the molecular associations of flower color variation. The integrated multi-omics analysis approach adopted in this study can offer potential references for the targeted improvement of cherry blossom flower color, as well as provide insights for flower color research and breeding of related ornamental plants.

2. Results

2.1. Determination of Cherry Blossom Petal Color

The color of each cherry blossom petal sample was measured by combining the Royal Horticultural Society Colour Chart (RHSCC) analysis and the CIELab color measurement method. The RHSCC results showed that ZH, FH, and XH all belong to the Red Group, while WH belongs to the White Group. The L*, a*, b*, h*, and C* values of the sample petals were similar to the RHS results, with significant differences between the red and white groups. The L* value represents the lightness of the sample: WH, a white-flowered variety, had the highest average L* (97.52); the L* values of the other samples gradually decreased with the deepening of flower color, and ZH had the lowest average L* (46.02) (Table 1). The a* value indicates the degree of red-green hue: it was the highest in ZH petals (41.92) and negative in WH petals (−1.23). The b* value represents the degree of yellow-blue hue: it was the highest in WH petals (5.29) and the lowest in FH petals (−2.97). The C* value denotes the sample saturation, calculated from a* and b*, and its variation trend among the four samples was opposite to that of L*. The h value is calculated via the mathematical formula h = arctan (b*/a*) based on the color parameters a* (red-green axis) and b* (yellow-blue axis), spanning a range of 0° to 360°. Distinct intervals within this range correspond to specific hue categories (e.g., red around 0°/360°, yellow around 90°, green around 180°, and blue around 270°). For the red-group samples (ZH, FH, XH): all h values fall within the red hue range near 0°/360°, and as flower color fades (ZH → FH → XH), the h values exhibit a slight increment (5.61° → 356.16° → 348.95°). For the white-group sample (WH): the h value is 102.09°, which lies close to the yellow hue range.

Table 1.

Petal color classification and parameters of cherry blossom petal samples in full bloom stage.

Name	RHSCC	Color Parameters
		L*	a*	b*	C*	h
ZH	N66A	46.02 ± 0.97 d	41.92 ± 0.96 d	4.12 ± 1.08 b	42.12 ± 0.90 a	5.61 ± 0.32 a
FH	N57D	62.76 ± 1.07 c	31.18 ± 1.81 c	−2.97 ± 0.99 d	31.32 ± 1.79 b	356.16 ± 1.12 d
XH	65B	74.89 ± 0.96 b	15.20 ± 1.67 b	−1.46 ± 0.43 c	15.27 ± 1.78 c	348.95 ± 0.89 c
WH	NN155D	97.52 ± 0.92 a	−1.23 ± 0.25 a	5.29 ± 1.10 a	5.43 ± 1.11 d	102.09 ± 0.81 b

Note: The color parameters in the table were determined following the CIELab color space methodology; L* denotes lightness (higher values correspond to greater brightness), a* represents the red-green axis (positive values indicate a reddish tint, while negative values indicate a greenish tint), b* denotes the yellow-blue axis (positive values indicate a yellowish tint, while negative values indicate a bluish tint), C* signifies chroma (higher values signify enhanced color vividness), and h denotes the hue angle (with values near 0°/360° corresponding to red and near 90° corresponding to yellow, thus characterizing the specific hue category). The lowercase letters (a, b, c, d) following the data in the table are results of the multiple comparison letter marking method, which are used to indicate statistically significant differences of the same index among different samples. Samples marked with the same letter indicate no statistically significant difference in this index (p > 0.05), while samples marked with different letters indicate a statistically significant difference in this index (p < 0.05). FH, ‘Yanzhi Fei’; WH, P. dielsiana; XH, ‘Yanzhi Xue’; ZH, P. campanulata.

2.2. Determination of Flavonoid and Anthocyanin Contents in Petals

The contents of flavonoids and anthocyanins in four cherry blossom petal samples were determined. The results showed that the flavonoid content was the highest in WH, followed by XH and FH, with the lowest in ZH; in contrast, the anthocyanin content was the highest in ZH and the lowest in WH (Figure 2). The content differences of the two measured substances showed an opposite trend, and further analysis is required to decipher the detailed regulatory mechanisms involved.

Figure 2.

Total flavonoids and total anthocyanin content of cherry blossom petal samples. FH, ‘Yanzhi Fei’; WH, P. dielsiana; XH, ‘Yanzhi Xue’; ZH, P. campanulata. Different lowercase letters above the bars indicate significant differences between groups (p < 0.05), while the same lowercase letters indicate no significant differences between groups.

2.3. Metabolite Data Analysis

Principal component analysis (PCA) was performed on each cherry blossom sample (Figure 3). The results demonstrated that the samples were well separated among different groups, with high reproducibility within the same group. Additionally, the quality control (QC) samples in the validation group were densely distributed in the score plot. These findings indicate that the data possess high reliability and are suitable for further metabolomic analyses.

Figure 3.

PCA of cherry blossom petal samples. FH, ‘Yanzhi Fei’; WH, P. dielsiana; XH, ‘Yanzhi Xue’; ZH, P. campanulata. QC in the figure refers to Quality Control sample, which is used to verify the reliability of experimental data.

2.4. Differential Metabolite Analysis

A total of 750 differentially expressed metabolites (DEMs) were identified in the four cherry blossom samples, of which 84 were annotated as flavonoid metabolites (Table S1). Among these flavonoid-derived compounds, anthocyanins—including cyanidin, delphinidin, and malvidin—are directly implicated in flower color determination. Notably, cyanidin was the predominant anthocyanin component, accounting for the highest proportion among all detected anthocyanin metabolites.

These differential metabolites were normalized and plotted into cluster heat maps (Figure 4). The metabolites in the figure can be roughly divided into four clusters: cluster 1 contains two types of proanthocyanidins, which are highly abundant in FH and XH; cluster 2 includes important precursor substances for anthocyanin synthesis such as quercitrin and coumaroyl-CoA, as well as delphinidin, with significantly higher abundance in FH and ZH compared to the other two groups; most of the metabolites in cluster 3 are cyanidin and kaempferol, etc., with the highest abundance in ZH; cluster 4 contains pelargonidin, which is highly abundant in WH and XH.

Figure 4.

Heat map of clustering of key differential metabolites in cherry blossom petal samples. FH, ‘Yanzhi Fei’; WH, P. dielsiana; XH, ‘Yanzhi Xue’; ZH, P. campanulata. Note: The intensity of color shades reflects metabolite abundance—the deeper the red, the higher the relative metabolite abundance; the deeper the blue, the lower the relative metabolite abundance. The values corresponding to the colors are normalized relative abundance (Z-score normalization). Negative values indicate that the relative metabolite abundance is lower than the average level of all samples, while positive values indicate it is higher than the average level, with a value range of −2 to 2. The larger the absolute value, the more significant the difference.

Combined with the degree of metabolite differences among various groups (Figure 5), the shared anthocyanin and flavonoid-related metabolites were screened out and sorted by significance as follows: quercetin 3-O-(6-O-malonyl-beta-D-glucoside), kaempferol-3-O-galactoside, cyanidin 3-galactoside, cyanidin 3-rutinoside, delphinidin 3-(6″-malonylglucoside) 5-glucoside, quercetin-3-O-sophoroside, cyanidin, and quercetin. In the comparison groups between ZH and the other three samples, cyanidin 3-(6″-p-coumarylsambubioside) was in the most significantly less abundant position, and the relative expression level of this metabolite in ZH was also much higher than that in the other three samples. In the comparison group between XH and FH, cyanidin 3-(6″-p-coumaroylsambubioside) was also the most significantly less abundant metabolite. However, in the comparison group between WH and XH, the significance of its more abundant was not high. From this, it can be inferred that this metabolite is very likely to be an important cause of the gradient change in petal color among the four cherry blossom samples.

Figure 5.

Volcano Plots of Differential Metabolites in Six Control Groups of Cherry Blossom Petal Samples. Note: The y-axis of this figure represents the log₂ fold change (log₂_FC) value, which not only reflects the upregulation or downregulation trends of metabolites but also indicates the magnitude of differences in their abundance. A larger absolute value of the y-axis corresponding to the target metabolite implies more significant differences in its abundance among groups. FH, ‘Yanzhi Fei’; WH, P. dielsiana; XH, ‘Yanzhi Xue’; ZH, P. campanulata.

Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analysis of differential metabolites showed that the differential metabolites in the four cherry blossom samples were significantly enriched in the following pathways: phenylalanine metabolism, flavone and flavonol biosynthesis, phenylpropanoid biosynthesis, flavonoid biosynthesis, isoflavonoid biosynthesis, anthocyanin biosynthesis, and phenylalanine, tyrosine and tryptophan biosynthesis (Table S2). In addition, the differential metabolites in the anthocyanin synthesis pathway were significantly annotated to the cyanidin synthesis pathway, further indicating that the flower color differences in cherry blossoms may be related to cyanidin.

2.5. Transcriptome Sequencing Data Quality Analysis

A total of 61.50 GB of clean data was obtained from the 12 samples. Three plants each were selected for P. campanulata, ‘Yanzhi Fei’ and ’Yanzhi Xue’; only one plant was available for P. dielsiana, from which three branches were chosen for sampling. Three biological replicates were collected for each (selected unit), resulting in a total of 12 samples. The sequence data for each sample exceeded 6.01 GB, with Q30 base percentages all greater than 94.08% and CG content ranging from 45.13% to 45.62% (Table 2). The total alignment rate ranged from 84.69% to 90.25%, and the unique alignment rate ranged from 80.43% to 85.38% (Table 3). These data indicate that the sequencing reads have a high degree of matching with the reference genome, which can be used for subsequent analyses.

Table 2.

Sample Transcriptome Sequencing Quality Overview.

Sample Name	RawData (Gb)	CleanData (Gb)	Q20 (%)	Q30 (%)	N (%)	GC (%)
FH1	6.27	6.21	98.05%	94.16%	87,571%	45.14%
FH2	7.10	7.04	97.89%	93.88%	126,291%	45.13%
FH3	7.39	7.32	97.91%	93.89%	131,696%	45.14%
WH1	6.63	6.57	98.18%	94.51%	101,017%	45.47%
WH2	6.42	6.36	98.05%	94.21%	99,145%	45.62%
WH3	7.83	7.75	97.55%	93.06%	138,203%	45.47%
XH1	6.27	6.21	98.24%	94.68%	93,609%	45.51%
XH2	6.34	6.28	97.72%	93.41%	121,295%	45.46%
XH3	6.15	6.11	97.84%	93.66%	119,651%	45.60%
ZH1	6.26	6.20	98.24%	94.67%	94,055%	45.53%
ZH2	6.72	6.65	98.25%	94.68%	99,163%	45.47%
ZH3	6.23	6.16	98.03%	94.15%	90,956%	45.54%

Note: FH, ‘Yanzhi Fei’; WH, P. dielsiana; XH, ‘Yanzhi Xue’; ZH, P. campanulata.

Table 3.

Overview of Transcriptome Sequencing Reads Mapping Results for Samples.

Sample Name	Total	Unmapped (%)	Unique Mapped (%)	Multiple Mapped (%)	Total Mapped (%)
FH1	41,383,874	6,273,986 (15.16%)	33,322,101 (80.52%)	1,787,787 (4.32%)	35,109,888 (84.84%)
FH2	46,957,272	7,029,886 (14.97%)	37,914,406 (80.74%)	2,012,980 (4.29%)	39,927,386 (85.03%)
FH3	48,825,116	7,474,102 (15.31%)	39,267,883 (80.43%)	2,083,131 (4.27%)	41,351,014 (84.69%)
WH1	43,853,954	5,171,471 (11.79%)	36,989,431 (84.35%)	1,693,052 (3.86%)	38,682,483 (88.21%)
WH2	42,522,368	5,605,793 (13.18%)	35,310,673 (83.04%)	1,605,902 (3.78%)	36,916,575 (86.82%)
WH3	51,744,860	6,483,957 (12.53%)	43,272,262 (83.63%)	1,988,641 (3.84%)	45,260,903 (87.47%)
XH1	41,423,706	4,536,431 (10.95%)	35,021,450 (84.54%)	1,865,825 (4.50%)	36,887,275 (89.05%)
XH2	41,882,480	4,721,298 (11.27%)	35,303,980 (84.29%)	1,857,202 (4.43%)	37,161,182 (88.73%)
XH3	40,657,808	4,463,109 (10.98%)	34,382,933 (84.57%)	1,811,766 (4.46%)	36,194,699 (89.02%)
ZH1	41,341,232	4,030,394 (9.75%)	35,298,231 (85.38%)	2,012,607 (4.87%)	37,310,838 (90.25%)
ZH2	41,383,874	6,273,986 (15.16%)	33,322,101 (80.52%)	1,787,787 (4.32%)	35,109,888 (84.84%)
ZH3	46,957,272	7,029,886 (14.97%)	37,914,406 (80.74%)	2,012,980 (4.29%)	39,927,386 (85.03%)

Note: FH, ‘Yanzhi Fei’; WH, P. dielsiana; XH, ‘Yanzhi Xue’; ZH, P. campanulata.

Combined with the PCA plot (Figure 6a) and the clustering plot (Figure 6b), it is found that three biological replicates within each group show an obvious clustering trend, indicating good correlation among them (Tables S3 and S4). Additionally, as can be seen from the violin plot (Figure 6c), the expression trends of each sample in the experiment are consistent, further verifying the reliability of the transcriptome data.

Figure 6.

Reliability analysis of sample transcriptome results. Note: (a), PCA plot of cherry blossom petal samples; (b), Cluster analysis plot of cherry blossom petal samples; (c), Violin plot of cherry blossom petal samples. FH, ‘Yanzhi Fei’; WH, P. dielsiana; XH, ‘Yanzhi Xue’; ZH, P. campanulata.

2.6. Analysis of Differentially Expressed Genes

With |log₂Fold Change| ≥ 1 and FDR < 0.05 set as the screening thresholds, a total of 11,299 differentially expressed genes (DEGs) were screened from the four samples. The number of DEGs between groups was as follows: ZH vs. WH (7712 DEGs), ZH vs. XH (3948 DEGs), ZH vs. FH (2802 DEGs), and XH vs. FH (2511 DEGs) (Figure 7). Weighted Gene Co-expression Network Analysis (WGCNA) classified these DEGs into 17 modules. The genes included in each module are shown in the figure (Figure 8). After excluding modules with excessively large intra-group differences among samples, 15 modules were obtained, namely “magenta”, “blue”, “salmon”, “darkgreen”, “greenyellow”, “red”, “darkgrey”, “black”, “grey60”, “purple”, “cyan”, “darkorange”, “midnightblue”, “pink”, and “turquoise”. Combined with trait analysis, modules whose co-expression trends were associated with the flower color variation of the samples were further screened out. The heatmap of relative expression levels in each sample showed that the gene expression of the “magenta”, “blue”, and “salmon” modules was consistent with the flower color gradient (high expression in red samples) (Figure 9a).

Figure 7.

Statistical chart of differential genes among groups of cherry blossom samples. Note. FH, ‘Yanzhi Fei’; WH, P. dielsiana; XH, ‘Yanzhi Xue’; ZH, P. campanulata.

Figure 8.

WGCNA module division and differential gene distribution within modules of cherry blossom samples.

Figure 9.

Analysis of Module Correlations and Transcription Factors of Differential Genes in Cherry Blossom Samples. Note: (a), Relative expression levels of module genes in cherry blossom samples; (b), Statistical chart of the number of transcription factors in cherry blossom samples. FH, ‘Yanzhi Fei’; WH, P. dielsiana; XH, ‘Yanzhi Xue’; ZH, P. campanulata.

In addition, transcription factor analysis was performed on all differential genes in the 17 modules. The Plant Transcription Factor Database (https://planttfdb.gao-lab.org/, accessed on 1 November 2025) was used to annotate these differential genes, and the number of differential genes in each transcription factor (TF) family was summarized. The figure shows the top 15 transcription factor families ranked by the number of genes. Among them, the bHLH family contains the largest number of differential genes, accounting for 7.69% of the total; followed by the MYB family, accounting for 7.47% (Figure 9b). Studies have shown that transcription factors from the MYB and bHLH families are present in various plant species and play extensive roles. Specifically, research on Vaccinium uliginosum [29], Diospyros kaki [30], Ficus macrocarpa [31], and P. avium [32] has demonstrated that various transcription factors from the MYB and bHLH families exert a significant promoting effect on anthocyanin synthesis in plants.

To gain deeper insights into the transcriptome data, the identified DEGs were initially subjected to annotation against the Gene Ontology (GO) database, resulting in the successful assignment of 13,389 genes to three functional categories (biological process, BP; cellular component, CC; molecular function, MF). Taking the annotation results of the XH-FH group as an example (Figure 10), 5571 genes were annotated to BP, 3008 to CC, and 1279 to MF. In the BP category, the two most highly annotated terms were “cellular process” (1327 genes) and “metabolic process” (1173 genes); in the CC category, the top two were “cellular anatomical entity” (972 genes) and “protein-containing complex” (279 genes); and in the MF category, the most abundant terms were “binding” (1194 genes) and “catalytic activity” (1178 genes). KEGG enrichment analysis showed that the DEGs were mainly enriched in metabolic pathways. Among these, pathways closely associated with flower color regulation included phenylpropanoid biosynthesis (ko00940), flavonoid biosynthesis (ko00941), phenylalanine metabolism (ko00360), and anthocyanin biosynthesis (ko00942) (Figure 11).

Figure 10.

Map of differential gene GO enrichment pathway in cherry blossom petal samples. The GO enrichment results are classified into three major functional categories: Biological Process, Cellular Component, and Molecular Function. In the figure, “up” represents upregulated DEGs, and “down” represents downregulated DEGs. Note: FH, ‘Yanzhi Fei’; WH, P. dielsiana; XH, ‘Yanzhi Xue’; ZH, P. campanulata.

Figure 11.

Differential gene KEGG enrichment circle of cherry blossom petal samples. Note: This figure adopts a four-layer circular structure, systematically integrating information on the KEGG pathway classification, significance, expression trends, and enrichment intensity of differentially expressed genes (DEGs). The first layer (outermost ring) labels three core pieces of information: ① KEGG functional categories (green for Metabolism, orange for Cellular Processes, and pink for Environmental Information Processing); ② standardized KO numbers of pathways. Pathways are clustered by function, among which the core pathways directly related to cherry blossom flower color regulation are concentrated in the “Metabolism” category, including Phenylpropanoid biosynthesis (ko00940), Flavonoid biosynthesis (ko00941), Anthocyanin biosynthesis (ko00942), etc. The second layer is a circular color gradient band, mainly presenting two key data: enrichment significance (Q value) and the number of background genes. Darker colors indicate smaller Q values and more significant enrichment (Q ≤ 0.05 is considered significantly enriched). The third layer is a two-color segmented circle, clearly showing the expression trends of DEGs in each pathway: the purple segments represent the proportion of upregulated DEGs, and the blue segments represent the proportion of downregulated DEGs. The bars of varying heights in the innermost layer represent the magnitude of the enrichment factor, with taller bars indicating larger enrichment factors.

2.7. Screening of Key Differential Genes Regulating Flower Color

In this experiment, based on transcriptomics technology, key enzyme genes involved in the anthocyanin synthesis pathway were first screened out from all detected differentially expressed genes. Subsequently, the measured genes were re-screened using p-values, FDR values, and Log₂FC values, and finally, differential candidate genes that met the criteria for inter-group differences were obtained. Expression pattern analysis was conducted on the aforementioned screened differential genes, and the focus was ultimately placed on genes whose expression trends were consistent with the changes in anthocyanin accumulation levels across the four samples. A total of 9 enzyme genes related to anthocyanin metabolism were obtained: 1 PAL gene, 4 4CL genes, 1 CHI gene, 1 DFR gene, 1 CYP gene, and 1 ANS gene. The relative expression levels of these genes in the four samples decreased successively in the order of ZH, FH, XH and WH. In addition, the expression patterns of some other genes are opposite to the accumulation pattern of anthocyanins. For example, the expression level of the Flavonol synthase (FLS) gene is the highest in WH, while it is relatively lower in ZH, FH and XH. The FLS enzyme may compete for the substrate dihydroflavonol, thereby reducing the production of cyanidin precursor substances [33]. As a downstream gene in the anthocyanin metabolic pathway, the FAOMT gene can specifically methylate unstable flavonol-glycosides and anthocyanin-glycosides. The expression level of this gene is the highest in WH and the lowest in ZH. Therefore, it is inferred that the Flavonoid O-methyltransferase (FAOMT) gene can reduce the content of cyanidin in petals by methylating cyanidin-glycosides, thereby affecting flower color variation [34]. Consequently, the primary reason that the colors of ZH and FH lean toward red and dark pink is most likely the regulation of the anthocyanin metabolic pathway by key enzyme genes, including PAL (EVM0025591), 4CL (EVM0002909, EVM0006685, EVM17537, EVM0026083), CHI (EVM0007449), CYP (EVM0002884), DFR (EVM0021188), and ANS(EVM0003770). The main reason why the colors of XH and WH tend to be light pink and white is likely the significant upregulation of the above-mentioned key structural genes such as FLS and FAMOT.

2.8. Correlation Analysis of Transcriptome and Metabolome

DEGs from the transcriptome and DEMs from the metabolome were enriched in 138 and 94 KEGG pathways, respectively. Among these, eight pathways were commonly responsive to both omics (Tables S7–S12). The core associated pathways focused on those related to flower color regulation, including phenylpropanoid biosynthesis (ko00940), flavonoid biosynthesis (ko00941), and anthocyanin biosynthesis (ko00942). Specifically, 55 DEGs and 27 DEMs were enriched in the ko00940 pathway; 60 DEGs and 10 DEMs in the ko00941 pathway; and 1 DEG and 3 DEMs in the ko00942 pathway (Figure 12).

Figure 12.

Distribution map of differential genes and differential metabolites in the anthocyanin biosynthesis-related KEGG pathways of cherry blossom petal samples. Note: FH, ‘Yanzhi Fei’; WH, P. dielsiana; XH, ‘Yanzhi Xue’; ZH, P. campanulata. PAL (Phenylalanine ammonia-lyase gene), 4CL (4-Coumarate-CoA ligase gene), CHI (Chalcone isomerase gene), F3H (Flavanone 3-hydroxylase gene), DFR (Dihydroflavonol 4-reductase gene), ANS (Anthocyanidin synthase gene), CYP75B2 (Cytochrome P450 75B2 gene), FLS (Flavonol synthase gene).

Pearson correlation coefficient calculation was performed on the metabolome and transcriptome data, and based on the results, a gene-metabolite co-expression network diagram and a correlation heatmap were drawn (Table S13). The co-expression grid map showed that 81 differential genes had significant correlations with 13 differential metabolites (Figure 13). Among them, the key structural genes involved in phenylalanine biosynthesis and flavonoid biosynthesis are closely correlated with metabolites. For example: PAL shows a significantly positive correlation with R1 (rutin) and P1 (p-Coumaroyl-CoA), while exhibiting a significantly negative correlation with P2 (Pelargonidin-3-O-glucoside); 4CL1, CHI3, and CYP75B2 display significantly positive correlations with R1, P1, N1 (Naringenin), M1 (Malvidin 3-rutinoside), C1 (Cyanidin 3-rutinoside), K1 (Kaempferol-3-O-galactoside), and C3 [Cyanidin 3-(6”-p-coumarylsambubioside)], and a significantly negative correlation with P2. The key structural genes involved in anthocyanin biosynthesis also show high correlations with the key metabolites in the figure. ANR, ANS, DFR, and UFGT are positively correlated with C1, C3, K1, and M1, and significantly negatively correlated with P2; FAOMT exhibits a significantly negative correlation with several anthocyanin and anthocyanidin substances. The correlation heat map shows that the three metabolites, quercetin-3-O-glucosyl-rutinoside, rutin and quercetin, are positively correlated with various genes and have a high degree of correlation; the three 4CL genes had a relatively high degree of correlation with each metabolite, except for a poor correlation with proanthocyanidin metabolites.; the FAOMT gene had a low or even negative correlation with most metabolites (Figure 14).

Figure 13.

Co-expression network diagram of anthocyanin-related differential genes and anthocyanin differential metabolites in cherry blossoms. Note: The differential metabolites in the figure are abbreviated and their full names are as follows: D1-Dihydromyricetin, E1-Epicatechin, C1-Cyanidin 3-rutinoside, Q1-Quercetin, N1-Naringenin, C2-Cyanidin, P1-p-Coumaroyl-CoA, D2-Delphinidin 3-6″-malonylglucoside 5-glucoside, M1-Malvidin 3-rutinoside, R1-Rutin, K1-Kaempferol-3-O-galactoside, C3-Cyanidin 3-6″-p-coumarylsambubioside, P2-Pelargonidin-3-O-glucoside.

Figure 14.

Heatmap of correlation between anthocyanin-related differential genes and anthocyanin-differential metabolites. Note: The red color series represents a positive correlation, while the blue color series represents a negative correlation. The darker the color, the stronger the correlation. The screening threshold for significantly correlated pairs is a Pearson correlation coefficient |r| ≥ 0.7.

The above results indicate that the key structural genes in the anthocyanin metabolic pathway and its upstream pathways of cherry blossom samples have good correlations with various anthocyanins and flavonoids in cherry blossom samples. For example: the 4CL enzyme gene shows a significant positive correlation with its downstream metabolite P1; the DFR enzyme gene, which functions to reduce dihydroflavonol substances, also has a good positive correlation with downstream malvidin and cyanidin anthocyanin substances. In addition, while most enzyme genes are positively correlated with various metabolites, they are uniquely negatively correlated with P2. This is consistent with the previous speculation in metabolomics, suggesting that there is a certain competitive mechanism between the metabolic pathway of pelargonidin and the metabolic pathways of cyanidin and malvidin.

2.9. qRT-PCR Verification

To ensure the accuracy of transcriptome sequencing, this experiment selected 9 key enzyme genes and transcription factor genes related to anthocyanin synthesis in the samples for verification by quantitative real-time PCR (qRT-PCR). The results showed that the qRT-PCR results of each gene were consistent with the trends of the transcriptome analysis results (Figure 15). Therefore, it is judged that the transcriptome sequencing results have high credibility and can be used for various analyses.

Figure 15.

qRT-PCR verification results of partial structural genes in cherry blossom petal samples. Note: FH, ‘Yanzhi Fei’; WH, P. dielsiana; XH, ‘Yanzhi Xue’; ZH, P. campanulata. The left y-axis represents the gene FPKM (Fragments Per Kilobase of transcript per Million mapped reads) values obtained from transcriptome sequencing, with the black bars corresponding to this data; the right y-axis represents the gene relative expression levels from qRT-PCR validation, with the gray bars corresponding to this data. Different lowercase letters above the bars indicate significant differences between groups (p < 0.05), while the same lowercase letters indicate no significant differences between groups.

3. Discussion

3.1. Screening of Differential Metabolites Regulating Flower Color Changes in Samples

Among the differential metabolites, a total of 84 flavonoid metabolites were identified, among which anthocyanin metabolites such as cyanidin, delphinidin, and malvidin are directly related to flower color. Among all the detected anthocyanins, cyanidin accounts for the largest proportion. Analyses of anthocyanins in plants such as Rosa rugosa [35], Styphnolobium japonicum [36], and Corydalis edulis [37] have shown that the accumulation of a large amount of cyanidin leads to a reddish color change in the accumulating parts. The relative content of various cyanidin metabolites is the highest in ZH and the lowest in WH. Delphinidin and malvidin also show the same trend. Among the screened differential metabolites, there are five cyanidin metabolites, one delphinidin metabolite, one malvidin metabolite, one rutin metabolite, and one proanthocyanidin. Among them, substances such as cyanidin 3-(6″-p-coumaroylsambubioside), cyanidin 3-rutinoside, and cyanidin 3-galactoside are all located downstream of the cyanidin metabolic pathway. They are structurally stable anthocyanin substances formed by the combination of cyanidin and glycosides. Studies on the red peel of Malus pumila have found that cyanidin 3-rutinoside is the most important anthocyanin in apples, and its content in individuals with dark red peels is significantly higher than that in those with light red peels [38]. Similarly, anthocyanins in Pyrus pyrifolia are also dominated by cyanidin 3-rutinoside, and its content increases with the expansion of the colored area and the enhancement of color intensity of the peel [39]. These findings indicate that cyanidin is an important red pigment and also confirm the dominant role of cyanidin metabolites in regulating the flower color of cherry blossom samples. The results of KEGG enrichment indicated that the regulation of anthocyanins in cherry blossoms mainly focuses on the flavonoid biosynthesis pathway, flavone and flavonol biosynthesis pathway, phenylalanine biosynthesis pathway, as well as a small number of anthocyanin biosynthesis pathways. Among them, the flavonoid biosynthesis pathway has the highest number of enriched genes and the smallest Qvalue, which indicates that the regulatory effect of this pathway on anthocyanin synthesis in cherry blossoms is more significant compared with other pathways. Among the four samples, the content of naringenin in P. campanulata was the highest, followed by that in ‘Yanzhi Fei’, ‘Yanzhi Xue’ and P. dielsiana. This indicates that the accumulation of abundant naringenin plays a very important role in the subsequent accumulation of anthocyanins.

3.2. Comprehensive Analysis of Transcriptome and Metabolome

In this study, genes involved in the phenylalanine metabolic pathway and flavonoid metabolic pathway, including PAL, CHI, 4CL, and CYP75B, were identified in four cherry blossom samples, totaling seven genes. The expression levels of these genes in red-series samples were all higher than those in white-series samples. In the study on the Prunus salicina cultivar ‘Cuihongli’, it was found that compared with the ordinary cultivar, the early-maturing mutant cultivar had a higher expression level of the PAL enzyme gene and a higher anthocyanin content [40]. The CHI gene in Prunus virginiana [41] and the 4CL gene in Malus pumila [42] can both exert certain effects on anthocyanin synthesis.

After being catalyzed by the above-mentioned key enzyme genes, metabolites enter the anthocyanin metabolic pathway. The first metabolite in the anthocyanin metabolic pathway is dihydroflavonol, and the production of dihydroflavonol must be catalyzed by F3’H. Li et al. identified the CYP75 gene family and its expression patterns in Cymbidium goeringii and found that the CYP75B gene can promote the transformation of plant flower color to red by regulating the synthesis of F3’H [43]. Then, dihydroflavonol is reduced and synthesized by DFR and ANS to form various anthocyanins (pelargonidin, cyanidin and delphinidin).

In this study, a total of two genes, namely DFR and ANS, were identified in the anthocyanin metabolic pathway, and their relative expression levels were higher in red-series samples. DFR is an important rate-limiting enzyme in the anthocyanin metabolic process. Generally, DFR can evenly reduce various dihydroflavonols (DHQ, DHM, DHK). However, in some plants, there may be a situation where it selectively reduces different dihydroflavonol substrates, and this situation often leads to the specific expression of flower color. As one of the cherry blossom varieties with the deepest red color, the DFR enzyme in ZH is highly likely to specifically reduce dihydroquercetin rather than dihydrokaempferol. In yellow-fruited varieties of Rubus idaeus [44], the ANS gene undergoes nonsense mutation-mediated mRNA decay. This phenomenon directly impairs the expression of its ANS gene, resulting in the absence of anthocyanin accumulation in yellow fruits. These findings indicate that the nine key structural genes in the two pathways interact to ultimately influence the changes in flower color of cherry blossoms.

Omics data were simultaneously enriched in the KEGG database, with a total of 95 differential genes and 40 differential metabolites enriched in pathways related to anthocyanin synthesis. The results of the grid plot showed that relevant structural genes in the anthocyanin synthesis process, such as PAL, 4CL, CHI, CYP75B2, ANR, ANS, DFR, and UFGT, were positively correlated with various anthocyanins or anthocyanin precursor substances, with high correlation coefficients. Among them, PAL, 4CL, CHI, and ANR, as key enzyme genes in the upstream of anthocyanin metabolism, were positively correlated with intermediate products such as p-coumaroyl-CoA, naringenin, and various subsequent anthocyanin substances. This indicates that the accumulation of these genes can provide precursor substances for subsequent anthocyanin synthesis and can positively regulate this process, which is consistent with the research results on upstream genes in Acer truncatum [45]. Subsequently, entering the anthocyanin metabolic pathway, the enzyme genes ANS, DFR, and UFGT are directly related to the synthesis of anthocyanin substances such as malvidin 3-rutinoside, cyanidin 3-rutinoside, kaempferol-3-O-galactoside, and cyanidin 3-(6”-p-coumarylsambubioside). In strawberries, the ANS gene is directly involved in the synthesis of anthocyanins and proanthocyanidins [46]; the DFR gene in pears directly reduces dihydroflavonols to leucoanthocyanins [47]; and the UFGT gene in Citrus sinensis can specifically glycosylate unstable anthocyanins into anthocyanins [48]. Therefore, it is inferred that the above three structural genes play the same roles in cherry blossom samples. There is another type of FAOMT (methyltransferase) that shows a significant negative correlation with several anthocyanin substances. This type of gene can methylate flavonols and anthocyanins, thereby making them lose their original functions. In addition, correlation analysis further confirmed that there is a certain competitive effect among the downstream branches of the anthocyanin metabolic pathway in cherry blossoms, namely the cyanidin metabolic pathway, malvidin metabolic pathway, and pelargonidin metabolic pathway. The above results indicate that the key structural genes in the anthocyanin metabolic pathway and its upstream pathways in cherry blossom samples have a good correlation with various anthocyanins and flavonoids in cherry blossom samples. In addition, while most enzyme genes are positively correlated with various metabolites, they are uniquely negatively correlated with P2. This is consistent with the previous speculation from the metabolomic analysis, indicating that there is a certain competitive mechanism between the metabolic pathway of pelargonidin and those of cyanidin and malvidin.

Through integrated transcriptomic and metabolomic analyses, this study delineated the regulatory roles of cyanidin derivatives and key structural genes in the gradient variation of cherry blossom flower color. It not only unraveled the molecular mechanism underlying flower color divergence between parental lines and their hybrid offspring but also furnished critical molecular targets and practical underpinnings for the goal of targeted flower color improvement and breeding of cherry blossoms as proposed in the Introduction.

4. Materials and Methods

4.1. Plant Materials

The plant materials selected in this experiment were P. campanulata (ZH, red), P. ‘Yanzhi Fei’ (FH, deep pink), P. ‘Yanzhi Xue’ (XH, pinkish white), and P. dielsiana (WH, white) (Figure 1). All samples of this experiment were taken from Fuzhou Cherry Garden. In order to avoid the interference of environmental factors differences on the sample, unified sampling was selected at 2 pm on 6 March 2022. Ten samples with consistent growth and no pests were selected, of which three were selected for ZH, FH and XH, and only one WH mother plant, so three branches were selected for collection, each of which were collected for three biological replicates, and a total of twelve samples were used for metabolomics and transcriptomics sequencing analysis. After sampling, first wrap the petals in tin foil and immerse them in liquid nitrogen for 5 min; then, transfer them to a dry ice insulated container for storage and transport them back to the laboratory; finally, store them in a −80 °C ultra-low temperature freezer [49].

4.2. Determination of Flower Color

Petals collected at various stages (4 samples in total) were compared for color using the Royal Horticultural Society Colour Chart (RHSCC) under identical environmental conditions, and the color code of each sample was recorded separately, with three replicate comparisons performed for each group of samples. For more accurate quantitative analysis of flower color, a colorimeter was used to simultaneously determine color values. The D65 standard light source was selected, with an observation diameter of 8 mm, and the surface of petals 0.5 cm away from the base was measured uniformly, with 15 technical replicates. This yielded the lightness value (L*), red-green value (a*), yellow-blue hue value (b*), saturation value (C*), and hue angle (h). The saturation value (C*) was calculated using the formula: C* = (a² + b²) 1/2.

4.3. Determination of Total Flavonoid and Anthocyanin Contents

The determination of total flavonoid content was performed with slight modifications according to the method described by Feng et al. [50]. Briefly, 10 mg of sample was weighed and mixed with 1 mL of methanol, followed by extraction at 60 °C for 4 h. Then, 0.5 mL of the crude extract was taken and diluted to 5 mL with methanol. Subsequently, 0.3 mL of NaNO₂ solution was added, and the mixture was allowed to stand at room temperature for 5 min. After that, 0.3 mL of AlCl₃ solution was added (excluding the control group), and the mixture was incubated for another 6 min. Next, 4 mL of NaOH was added, and the total volume was adjusted to 10 mL with water. After standing for 15 min, the absorbance was measured at 510 nm. Rutin was used as the standard reference.

The flavonoid content (mg/g) was calculated using the formula:

Flavonoid content (mg/g) = C × Total volume of extract/Actual sample weight

C = 2.5469 × A₅₁₀ − 0.116

Anthocyanin extraction was carried out with minor modifications based on the method reported by Wang et al. [51]. Precisely, 10 mg of sample was weighed and homogenized with 1 mL of 5% formic acid solution. The mixture was centrifuged at 12,000 rpm at 4 °C for 10 min, and the supernatant was collected. The precipitate was re-extracted with 1 mL of 5% formic acid solution repeatedly until no additional supernatant could be obtained. The absorbance of the combined supernatants was measured at 530 nm.

The anthocyanin content (μg/g) was calculated using the formula:

Anthocyanin content (μg/g) = C × Total volume of extract/Actual sample weight

C = 49.16 × A₅₃₀ + 0.0491

4.4. Metabolite Extraction

An amount of 100 mg of frozen petal tissue was weighed, ground thoroughly in liquid nitrogen, and then mixed with 0.5 mL of pre-chilled 80% methanol (LC-MS grade, CNW Technologies, Düsseldorf, Germany) followed by vortexing for resuspension. After incubation on ice for 5 min, the mixture was centrifuged at 15,000× g and 4 °C for 20 min using a centrifuge (Heraeus Fresco17, Thermo Fisher Scientific, Waltham, MA, USA). The supernatant was collected and diluted to a final concentration of 53% methanol, transferred to a new centrifuge tube, and centrifuged again under the same conditions for 20 min. Subsequently, the supernatant was filtered through a 0.22 μm filter membrane and injected into an ultra-high performance liquid chromatograph (Vanquish, Q Exactive HFX, Thermo Fisher Scientific) for UHPLC-MS/MS analysis. The sample was loaded onto a Hypesil Gold column (100 × 2.1 mm, 1.9 μm) with a flow rate of 0.2 mL/min and a column temperature maintained at 35 °C. Linear gradient elution was carried out over 17 min using 2–100% methanol (LC-MS grade, CNW Technologies) as the mobile phase. For the positive ion mode, the eluents were Eluent A (0.1% formic acid aqueous solution) and Eluent B (methanol); for the negative ion mode, Eluent A was 5 mM ammonium acetate solution (pH 9.0, LC-MS grade, SIGMA-ALDRICH, St. Louis, MO, USA). A Q Exactive^TM HF-X mass spectrometer (Thermo Fisher Scientific) was employed in both positive and negative ion modes, with the following operating parameters: spray voltage of 3.2 kV, capillary temperature of 320 °C, sheath gas flow rate of 40 arb, and auxiliary gas flow rate of 10 arb.

4.5. Metabolome Analysis

After filtering the raw data by retention time and mass-to-charge ratio (m/z), peak alignment and peak extraction were performed. Molecular formulas were then predicted based on molecular ions and fragment ions, and the predicted results were submitted to the online databases mzCloud, mzVault, and Masslist for comparison to eliminate background ions. The quantitative results of peak areas were normalized (Tables S14 and S15). During the data quality control phase, metabolite extraction and instrumental analysis were conducted on 12 samples from 4 cherry blossom varieties as well as QC samples. PCA was implemented using the gmodels package (v2.18.1) in R, while Pearson correlation coefficients were calculated and correlation heatmaps were generated using the pheatmap package (v1.0.12) to verify sample reliability. Orthogonal Partial Least Squares-Discriminant Analysis (OPLS-DA) was performed with the ropls package in R, and the stability of the model was validated through permutation tests. DEMs were screened using the thresholds of VIP ≥ 1, |log₂Fold Change| ≥ 1, and p < 0.05 [52].

4.6. Transcriptome Sequencing and Analysis

Total RNA was extracted using the RNA prep Pure Polysaccharide- and Polyphenol-rich Plant Total RNA Extraction Kit (Tiangen, Beijing, China). After verifying RNA quality by RNase-free agarose gel electrophoresis, cDNA was synthesized via reverse transcription using the Vazyme#R232 Reverse Transcription Kit (Nanjing, China). Libraries were constructed by Genodio Biotechnology Co., Ltd. (Hefei, China) and sequenced on the Illumina Novaseq 6000 platform (San Diego, CA, USA). Raw sequencing data were processed with fastp software (v0.18.0, Beijing, China) to remove adapter sequences, low-quality reads containing more than 10% unknown nucleotides (N), and those with over 50% low-quality bases (Q value ≤ 20). Using the whole-genome sequence of Prunus serrulata (https://doi.org/10.6084/m9.figshare.12431846.v3) as the reference genome, clean reads were aligned to the reference genome with HISAT2 software (v2.2.1, Baltimore, MD, USA). Sample correlation was validated by PCA using R (v3.0.2, Aarhus, Denmark) and the gmodels package (v2.18.1). Differential expression analysis was performed with DESeq2 software (v1.38.3, Heidelberg, Germany). The WGCNA package in R software (v4.2.0, Los Angeles, CA, USA) was used to analyze the DEGs. After filtering low-expression genes, the soft threshold was determined based on the approximate scale-free topology (R² > 0.85). Modules were identified via the dynamic tree cutting method, and their correlations with flower color traits were analyzed to screen flower color-related modules.

4.7. Transcriptome and Metabolome Association Analysis

For the transcriptome-metabolome correlation analysis: pathway enrichment of both DEGs and DEMs was performed via the KEGG database. A model was constructed using the bidirectional orthogonal partial least squares (O-PLS) method, and the Pearson correlation coefficient (with a threshold of |r| ≥ 0.7) was calculated using R (v3.0.2). Finally, a correlation network plot was visualized using Cytoscape software (v3.10.0).

4.8. Differential Gene qRT-PCR Verification

Nine candidate genes related to anthocyanin synthesis were selected from the DEGs of the transcriptome for quantitative real-time PCR (qPCR) validation. Primers were designed using Primer3 (v2.6.1) software (Table 4) and screened via the Primer BLAST tool (https://www.ncbi.nlm.nih.gov/tools/primer-blast, accessed on 11 July 2025). The qPCR reactions were performed using the SYBR Green Premix Pro Taq HS qPCR Kit (Table 5), quantitative analysis was conducted on the ABI StepOnePlus system, and the relative gene expression levels were calculated using the 2^−∆∆Ct method.

Table 4.

Anthocyanin-related candidate gene primers.

Gene Name	Gene ID	Forward Primer	Reverse Primer
bHLH28	5756	ATTACGCCAACTGGTTCCTG	GCGAGTAGTTGGGAGAGTCG
bHLH120	24,979	AAGCTTGGTGAACGGATTGT	GGAGCGCTCAACACCTTAAC
MYB123	1552	AAGCAGTCCTCTGGTGCAGT	GTTGGTCGACGGTTGAGTTT
MYB13	21,262	ACAGAGAGAGGCCACAAGGA	CTCTTGCCCTTGCTCAAGTC
WRKY69	27,405	AAGGGTTGTTCTGCCAAAAA	GGTTGCAATTGGGTCAAGTT
CYP73A13	1120	ACGGAAGAAACTTTCGAGCA	CGTTGATCTCTCCCTTCTGC
CYP75B2	2884	CAAGCTCACAGACACCGAGA	TGGCGAAGGAGTTCTGCTAT
MYB101	2122	AGGGCTCGTTATCAGAAGCA	GGTGGAAGGGATGAAACTGA
PAL1	25,591	AGGCCTAATTCCAAGGCTGT	AGAGCCAGGCCTTCTTTAGG
Actin		AGCAACTGGGATGACATGGA	CAGGGGTGCCTCAGTAAGAA

Table 5.

qRT-PCR Reaction System and Procedure.

The Reaction System	Volume	The Reaction Process	Temperature	Time	Circular Number
2× SYBR Green Pro Taq HS Premix	10 μL	Pre-denaturation	95 °C	30 s	1
Forward primer	0.4 μL	Denaturation	95 °C	5 s	40
Reverse primer	0.4 μL	Annealing & Extension	60 °C	30 s	40
ROX	0.4 μL
RNase-free water	6.8 μL
cDNA	2 μL

5. Conclusions

This study systematically analyzed the variations in flower color-related gene expression and metabolite accumulation between parental and offspring cherry blossom varieties via integrated metabolomic and transcriptomic analyses. Cyanidin was identified as the predominant anthocyanin, and structural genes including PAL, 4CL, CHI, CYP75B, and DFR were demonstrated to collectively promote anthocyanin biosynthesis in cherry blossoms. This leads to variations in anthocyanin accumulation across samples and thereby influences the petal color traits of cherry blossoms.

This study establishes a robust foundation for subsequent investigations into the molecular mechanisms underlying cherry blossom flower color, offers critical theoretical underpinnings for the breeding of new red-series cherry blossom cultivars, and facilitates the transition of the cherry blossom industry from traditional random hybridization to precise targeted breeding.

Notably, this study has several limitations: Owing to objective constraints during the research phase, sampling was restricted to petals at the full bloom stage, and longitudinal sampling across distinct growth and developmental stages of cherry blossoms was not conducted. This hinders the comprehensive elucidation of the temporal dynamics of anthocyanin accumulation and the time-dependent features of their regulatory networks. In future studies, longitudinal samples covering key developmental stages (e.g., bud stage, early bloom stage, full bloom stage, and senescence stage) could be incorporated to perform sequential analyses, aimed at deciphering the temporal dynamics underlying anthocyanin biosynthesis and regulation. Additionally, functional validation assays (e.g., gene silencing, overexpression, and yeast two-hybrid assays) are needed to further delineate the specific functional mechanisms of key genes and transcription factors, refine the molecular regulatory network governing cherry blossom flower color, and provide a more comprehensive theoretical reference and technical support for the targeted genetic improvement of flower color in woody ornamental plants.

Supplementary Materials

The following supporting information can be downloaded at: https://zenodo.org/records/18067174, Table S1. Statistical Table of Differentially Expressed Metabolite Counts. Table S2. Metabolite KEGG Annotation Result Table. Table S3. Total Gene Expression Table for All Samples. Table S4. PCA Plotting Data for All Samples. Table S5. GO Enrichment Background File for Samples. Table S6. KEGG Pathway Enrichment Background File for Samples. Table S7. Pathway Annotation Table for DEGs and DEMs in the WH-vs.-FH Comparison Group. Table S8. Pathway Annotation Table for DEGs and DEMs in the WH-vs.-XH Comparison Group. Table S9. Pathway Annotation Table for DEGs and DEMs in the XH-vs.-FH Comparison Group. Table S10. Pathway Annotation Table for DEGs and DEMs in the ZH-vs.-FH Comparison Group. Table S11. Pathway Annotation Table for DEGs and DEMs in the ZH-vs.-WH Comparison Group. Table S12. Pathway Annotation Table for DEGs and DEMs in the ZH-vs.-XH Comparison Group. Table S13. Correlation Result Table for DEGs and DEMs. Table S14 Statistical Table of Qualitative and Quantitative Results for Metabolites in Positive Ion Mode. Table S15. Statistical Table of Qualitative and Quantitative Results for Metabolites in Negative Ion Mode.

Author Contributions

Conceptualization and conception, X.Y. and X.W.; methodology, T.W. and Y.Y.; data curation, X.Z., S.Q. and W.F.; software, T.W. and Y.Y.; writing—review and editing, Y.Y. All authors have read and agreed to the published version of the manuscript.

Data Availability Statement

The data presented in this study are available on request from the corresponding author; The relevant original experimental data included in the Supplementary Materials has been uploaded to https://zenodo.org/records/18067174.

Conflicts of Interest

The authors declare no conflicts of interest.

Funding Statement

This study was funded by the Modern Agriculture Key Project in Jiangsu Province, China (BE2020343); Forestry Science Technology Innovation and Popularization Project in Jiangsu Province, China [LYKJ (2021) 30].