Introduction

Fruit color is an important economic trait and is determined by the concentration and composition of anthocyanins. Anthocyanins are a group within the flavonoid family of plant secondary metabolites that determine the color of flowers and plant organs. To date, more than 700 anthocyanins have been documented in diverse plant species (Zhao and Yuan 2021). The most common anthocyanins are the derivatives of six widespread anthocyanidins, namely cyanidin (Cy), delphinidin (Dp), pelargonidin (Pg), peonidin (Pn), petunidin (Pt), and malvidin (Mv) (Zhang and Zhu 2023; Zhao et al. 2022). In addition to their well-known physiological functions of serving as pollinator and seed dispersers, anthocyanins also play essential roles in protecting plants from being damaged by light, UV radiation, cold, drought, nutrient deficiency, and pathogen attack (Cappellini et al. 2021; Hichiri et al. 2011). Substantial studies have proved that anthocyanins have protective effects against age-related degenerative disease and other chronic disorders, such as cardiovascular and neurodegenerative diseases, cancer, and diabetes (Câmara et al. 2022; Salehi et al. 2020; Shu et al. 2023). Therefore, there has been much interests in the identification of regulatory genes controlling anthocyanin biosynthesis in major horticultural crops as tools in breeding programs for crop improvement.

Anthocyanin accumulation in plants depends on the coordinated expression of anthocyanin biosynthetic genes in the flavonoid branch of the phenylpropanoid pathway (Cappellini et al. 2021; Sunil and Shetty 2022; Zhang and Zhu 2023). Several core structural genes are well known to encode enzymes that catalyze the biosynthesis of anthocyanins, including phenylalanine ammonia lyase (PAL), chalcone synthase (CHS), chalcone isomerase (CHI), flavanone-3-hydroxylase (F3H), flavonoid 3’-hydroxylase (F3’H), flavonoid 3’5’-hydroxylase (F3’5’H), dihydroflavonol 4-reductase (DFR), anthocyanidin synthase (ANS), and UDP-glucose flavonoid 3-glucosyltransferase (UFGT) (Chaves-Silva et al. 2018; Sunil and Shetty 2022; Zhao et al. 2022). In addition, the competition of branch pathway genes such as those coding flavonol synthase (FLS), anthocyanidin reductase (ANR), leucoanthocyanidin reductase (LAR) also influence the accumulation of anthocyanins (Liu et al. 2021).

Anthocyanin biosynthesis is regulated at the transcriptional level by well-studied R2R2-MYB, basic helix-loop-helix (bHLH), and WD40 Repeat (WDR) class of transcription factors (TFs). These regulators form a ternary MBW complex that bind to the promoter of genes in the phenylpropanoid and anthocyanin biosynthetic pathway (Chaves-Silva et al. 2018; Hichiri et al. 2011; Liu et al. 2021; Zhang and Zhu 2023). The MYB TFs determine the specificity of the complex for different biosynthetic gene promoters (Lloyd et al. 2017; Xu et al. 2015). It has been characterized that R2R3-MYBs in subgroup (SG) 4, SG5, and SG6 are involved in the regulation of anthocyanin biosynthesis (Dubos et al. 2010; Hichiri et al. 2011; Stracke et al. 2001). Other TFs, including WRKY, bZIP, MADS, NAC are also reported to be associated with regulation of anthocyanins (Liu et al. 2021; Shen et al. 2022).

Pomegranate (Punica granatum L.) is one of the oldest and known fruits native to Iran and surrounding areas and today widely cultivated in tropical and subtropical regions (Stover and Mercure 2007). Fruits vary considerably in color from white to purple-red, along with varied differences in red color pigmentation patterns. Colored pomegranate has attracted considerable research interests due to its anthocyanin content with enhanced antioxidant capacities, leading to pomegranate fruits being described as “superfruit”. In pomegranate, it has been well characterized that cyanidin-3-O-glucoside (Cy3G), cyanidin-3,5-O-diglucoside (Cy3,5dG), delphinidin-3-O-glucoside (Dp3G), delphinidin-3,5-O-diglucoside (Dp3,5dG), pelargonidin-3-O-glucoside (Pg3G), and pelargonidin-3,5-O-diglucoside (Pg3,5dG) are dominant anthocyanins (Zhao and Yuan 2021).

Up to now, anthocyanin biosynthetic genes have been successfully isolated and their expression profiles in various pomegranate cultivars or tissues have also been monitored (Attanayake et al. 2019; Ben-Simhon et al. 2011, 2015; Harel-Beja et al. 2019; Kaur et al. 2019; Rouholamin et al. 2015; Zhao et al. 2015). However, only few of anthocyanin-related TFs have been functionally studied. For instance, PgWD40, dependent on PgAN2 (MYB) function, was responsible for specific control of cyanidin synthesis (Ben-Simhon et al. 2011). PgMYB5-like strongly interacted with PgbHLH, resulting in the production of intermediates of the flavonoids pathway, especially dihydroflavonols, while anthocyanins were not produced (Arlotta et al. 2020). In addition, PgbZIP16 and PgbZIP34 acted as a positive regulators in the biosynthesis of anthocyanins (Wang et al. 2022).

Bright red peel color for pomegranate is an essential exterior fruit quality attribute determining the commodity sale of fresh market. Furthermore, the anthocyanins rich in fruit peel/husk are valued for their promising biological activities (Fahmy and Farag 2021). Cultivated pomegranate fruits show substantial diversity of fruit peel color in red intensity, ranging from light to dark red. Purple-red pomegranate, also called as “black” cultivar, is characterized by an exceptionally high amounts of anthocyanins (Trainin et al. 2021; Zhao et al. 2013). It was shown that the dark-red trait resulted from a mutation of ANR gene, reducing competition over anthocyanidins formation and resulting in overproduction of anthocyanins (Trainin et al. 2021). Though the expression profiles for anthocyanin genes have been gained at the early stage when the color divergence occurred in fruits (Trainin et al. 2021), the differences in their transcriptional profiles as development progresses have not been systematically elucidated between red and purple-red pomegranate.

In this study, two cultivars with distinct color phenotypes, ‘Hongbaoshi’ (red-skinned, R) and ‘Moshiliu’ (purple-red-skinned, P), were used to get more insights into the color variation throughout fruit development. We analyzed temporal anthocyanin metabolomic and transcriptomic profiles of fruits with Ultra-Performance Liquid Chromatography coupled with Tandem Mass Spectrometry (UPLC-MS/MS) and RNA-seq technologies, trying to explore the key anthocyanins and genes responsible for exterior color mutation. The knowledge of anthocyanin-related fruit exterior color mutation may be used for supporting breeding efforts towards developing new pomegranate cultivars with tailored anthocyanin composition and concentration.

Materials and methods

Fruit sampling

From June to September, fruits at different development stages were harvested from ‘Hongbaoshi’ (a red-skinned cultivar) and ‘Moshiliu’ (a purple-red-skinned cultivar) pomegranate. The two cultivars were grown in the National Pomegranate Germplasm Nursery (Yicheng District, Zaozhuang City, Shandong Province, China, 117°49 N, 34°75 E). Four fruit developmental stages included 10 June (S1), 10 July (S2), 10 August (S3), and 10 September (S4), which were designated as R1~R4 and P1~P4, respectively. Samples from three trees per cultivar were collected and pooled (each tree is a biological repeat). Fruit skins from each replicate were carefully peeled, manually collected and combined to assure uniformity. All samples were immediately frozen in liquid nitrogen and stored at -80 °C for subsequent use.

Anthocyanins extraction and UPLC-MS/MS analysis

The peel samples were freeze-dried and then grounded to fine power using a mixer mill (MM 400, Retsch) with a zirconia bead for 1.5 min at 30 Hz. Equal amounts of peels (50 mg) were prepared, and anthocyanins were extracted overnight at 4 °C with 500 µL 50% aqueous methanol (containing 0.1% HCl). Ultrasound extraction was carried out for 5 min. The homogenates were centrifuged at 12,000 rpm/min for 3 min. Then the collected supernatants underwent repeated centrifugation, and the two supernatant fractions were combined together and filtrated by 0.22 μm filter before subsequent analysis.

The sample extracts were analyzed using UPLC (ExionLC™ AD, SCIEX) coupled with MS/MS (QTRAP 6500+, SCIEX) platform. The UPLC was equipped with a C18 column (Waters ACQUITY BEH, 1.7 μm, 2.1 mm*100 mm). The mobile solvent A was 0.1% methanoic acid in water, while solvent B was 0.1% methanoic acid in methanol. The gradient elution procedure was as follows: 5% B at 0 min, 50% B at 6.0 min, 95% B at 12.0 min, 5% B at 14.0 min, and the column was equilibrated for 2 min prior to the injection of next sample. The flow rate was 0.35 mL·min-1, the column temperature 40 °C, the injection volume 2 µL.

The mass spectrometry (MS) experiments were performed through an electrospray ionization (ESI) interface, operating in the positive ion modes. ESI source temperature was 500 °C, and curtain gas was 35 psi. Declustering potential (DP) and collision energy (CE) for individual ion pairs were done with further DP and CE optimization. The quality control samples were injected every ten experimental samples throughout the analytical run to provide a set of data from which repeatability could be assessed.

Qualitative and quantitative analysis of anthocyanins

Anthocyanin compounds were identified by MS cheminformatics based on the self-built database MWDB (Metware Biological Science and Technology Co., Ltd. Wuhan, China). The MultiQuant software 3.0.3 and Analyst 1.6.3 were used to process the MS data of different samples, to integrate and correct chromatographic peaks. Peak area integration was performed on the mass spectral peaks of all the substances, and the standard calibration curve was used for individual anthocyanins quantitation (Table S1). The data were obtained from three independent biological replicates.

Identification of differential anthocyanin metabolites (DAMs)

Principal component analysis (PCA) and orthogonal partial least-squares disciminant analysis (OPLS-DA) were carried out by SIMCA software (Umetrics, Umea, Sweden), with data from 24 samples (two cultivars × four developmental stages × three biological replicates). Differential anthocyanins screening in groups were performed by variable importance in projection (VIP) valules ≥ 1 and the|log2FoldChange| > 1.

RNA extraction, library construction and sequencing

Total RNA was extracted separately from peel samples using a TRIzol RNA extraction kit (Invitrogen, Carlsbad, CA, USA). The integrity of RNA was evaluated using agarose gel and an Agilent 2100 Bioanalyzer (Agilent Technologies, Inc. CA, United States). The purity of RNA was determined using NanoDrop spectrophotometer (Thermo Fisher Scientific, Inc. Waltham, MA, United States). The mRNA was enriched from total RNA by magnetic beads conjugated with oligo (dT) and then fragmented into short lengths of approximately 300 bp. The short RNA frangements were used to synthesize first-stranded cDNA with random primers. And then, the first-stranded cDNA was used as templates for second-strand cDNA synthesis. To generate cDNA sequencing libraries, the RNA fragments were subjected to 3’-end repair, adapter ligation, size selection, reverse transcription and PCR amplification. The constructed 24 cDNA libraries were sequenced using Illumina HiSeq 2000 platform to generate 150 bp paired-end reads.

Transcriptomic data analysis and differentially expressed genes (DEGs) identification

The image files which were got from the sequencing platform were transformed to the original data in FASTQ format (raw data). The obtained raw data were filtered by Cutadapt (v2.7) (Martin 2011) to get high-quality sequence (clean data) for further analysis. The clean reads were separately mapped to Punica granatum L. genome (ASM765513v2) using HISAT2 (2.1.0) software (Kim et al. 2015), allowing up to no more than two mismatches in each segment alignment. The read count values on each gene were compared by HTSeq (0.11.1) (Anders et al. 2015) as the original expression of the gene. Fragments Per Kilobase per Million mapped reads (FPKM) were determined to quantify gene expression levels for all genes in each sample. Genes with FPKM values <1.0 were defined to be minimally expressed and were removed from data set.

Correlation between all samples were determined via calculating Pearson correlation coefficient (PCC) and hierarchical clustering analysis (HCA). PCA were conducted by SIMCA software (Umetrics, Umea, Sweden). DEGs were identified between ‘Hongbaoshi’ and ‘Moshiliu’ at each of four developmental stages as well as between immediately adjacent two stages within each cultivar using DEseq R package (Robinson et al. 2010), with a cutoff threshold of|log2FoldChange| > 1 and flase discovery rate (FDR) < 0.05. The confirmed DEGs were subjected to Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analysis. Identification and classification of transcription factors were predicted by the online Plant Transcription Fctor Database (PlantTFDB) v5.0 (Jin et al. 2017) and Plant Transcription factor & Protein Kinase Identifier and Classifier (iTAK) (Zheng et al. 2016). The expresseion heatmap was generated by Tbtools software (Chen et al. 2020).

Identification of candidate structural genes and R2R3-MYB TFs involved in anthocyanin biosynthesis

According to the KEGG pathway enrichment results of DEGs in comparison groups, we searched for DEGs implicated in flavonoid and anthocyanin pathways. The different expressed MYBs in comparison groups were alignment together to identify R2R3-MYBs, and then those involved in the anthocyanin biosynthesis were screened by phylogenetic analysis. Protein sequences of R2R3-MYBs from Arabidopsis thaliana were retrieved from TAIR database. The sequences were aligned using MEGA 11, and the phylogenetic tree was subsequently constructed by neighbor-joining method. A diagram of the phylogenetic tree was generated using the ITOL webserver (https://itol.embl.de/).

Correlation analysis

In order to obtain the relationships of genes function and anthocyanin metabolites, PCC was used to assessed the correlations between MYBs and structural genes, and between structural genes and anthocyanin content using the OmicStudio tools at https://www.omicstudio.cn/tool/62. The absolute value of correlation coefficient > = 0.95 and P-value < 0.05 was considered significantly correlated. The associated correlation coefficients were displayed as correlation heatmaps.

Quantitative real-time PCR (qRT-PCR)

Total RNAs were extracted from ‘Hongbaoshi’ and ‘Moshiliu’ samples separately. After that the first-strand cDNAs were synthesized and were diluted to serve as the qRT-PCR templates. A total of 12 candidate strucutral genes including PAL, 4CL, CHS, CHI, F3’H, F3’5’H, DFR, ANS, UGT, FLS, LAR, ANR were selected to monitor their expression levels. The qRT-PCR reaction mixtures contained 10 µL SYBR Green (Thermo Fisher Scientific, Waltham, MA, USA), 7.8µL RNase-free water, 1 µL cDNA, and 0.6 µL of each forword and reverse primer. The reaction conditions were initial enzyme activation for 15 min at 95 ℃, followed by 40 cycles of denaturation for 10 s at 95℃, annealing for 20 s at 58 ℃, and extension for 30 s at 72℃. The data were normalized to Pgactin (GU376750 ) transcript levels and relative expression level of the target RNA was calculated according to 2-△△CT method (Livak and Schmittgen 2001). The primers used for qRT-PCR were shown in Table S2. All reactions were performed with three replicates.

Results and discussion

Component and concentration of anthocyanins in pomegranate

It has been well-known that anthcoyanins are major pigments in pomegranate plants. Here, two cultivars with distinct fruit skin color were harevested at four time points during fruit development (Fig. 1a) and the anthocyanin fingerprints were investigated using an targeted metabolomics approach based on UPLC-MS/MS. The fruit of ‘Hongbaoshi’ underwent a transition from orange-red colored young fruit (R1) towards the development of green-red fruits (R2). Later, the fruit regained red color and reached maximal levels of red intensity (R3 and R4). The fruit of ‘Moshiliu’ was characterized by purple-red with no obvious color variation during fruit development by visual inspection (Fig. 1a).

Fig. 1
figure 1

The anthocyanin profiles in two cultivars. a Two cultivars from four sampling stages. The upper row represents red ‘Hongbaoshi’ and the lower row represents purple-red ‘Moshiliu’. b The proportion changes of anthocyanin groups. c Concentration changes of the six major anthocyanins. d Change patterns of mono- and di-glucoside anthocyanins. R1~R4 and P1~P4 indicate ‘Hongbaoshi’ and ‘Moshiliu’ collected from four developmental stages, respectively. Cy, Cyanidin; Dp, delphinidin; Pg, pelargonidin; Pn, peonidin; Pt, petunidin; Mv, malvidin; Cy3G, cyanidin-3-O-glucoside; Cy3,5dG, cyanidin-3,5-O-diglucoside; Dp3G, delphinidin-3-O-glucoside; Dp3,5dG, delphinidin-3,5-O-diglucoside; Pg3G, pelargonidin-3-O-glucoside; Pg3,5dG, pelargonidin-3,5-O-diglucoside

Full size image

Consistent with the visual inspection, analysis of fruit anthocyanins demonstrated distinctive anthocyanin accumulation profiles between two cultivars. A total of 50 anthocyanins were identified in peel samples (Table S1, Fig. S1), including 12 cyanidins, 11 delphinidins, 10 peonidins, 8 pelargonidins, 7 petunidins, and 2 malvidins. A srupulous search of the literature revealed that 22 anthocyanins were firstly reported in our study, which contained 5 cyanindins (Cyanidin-3,5,3’-O-triglucoside, Cyanidin-3-O-(6-O-malonyl-glucoside), Cyanidin-3-O-5-O-(6-O-coumaryl)-diglucoside, Cyanidin-3-O-sambubioside, and Cyanidin-3-rutinoside-5-glucoside), 6 delphinidins (Delphinidin-3-O-(6-O-acetyl)-glucoside, Delphinidin-3-O-(6-O-malonyl-glucoside), Delphinidin-3-O-(coumaryl)-glucoside, Delphinidin-3-O-rhamnoside, Delphinidin-3-O-sophoroside, Delphinidin-3-rutinoside-5-glucoside), 2 pelargonidins (Pelargonidin-3,5,3’-O-triglucoside, and Pelargonidin-3-rutinoside-5-glucoside), 5 peonidins (Peonidin-3-O-(6-O-p-coumaryl)-glucoside, Peonidin-3-O-5-O-(6-O-coumaroyl)-diglucoside, Peonidin-3-O-sambubioside, Peonidin-3-O-sambubioside-5-O-glucoside, Peonidin-3-O-sophoroside), 4 petunidins (petunidin-3,5-O-diglucoside, petunidin-3-O-(coumaryl)-glucoside, petunidin-3-O-sambubioside-5-O-glucoside, and petunidin-3-O-sophoroside). Previously, more than 100 anthocyanin pigments were listed in pomegranate various cultivars, organs and tissues (Zhao and Yuan 2021), making the fruit one of the most popular functional foods.

Quantitatively, more abundant anthocyanins in ‘Moshiliu’ than in ‘Hongbaoshi’ fruit indicated a phenotypic color variation for both cultivars (Fig. S1a). Among several types of anthocyanin detected, Cy3G, Cy3,5dG, Dp3G, Dp3,5dG, Pg3G, and Pg3,5dG were predominant anthocyanins in all samples across developmental stages (Fig. S1b), which were consistent with previous results (Bar-Ya’akov et al. 2019; Fellah et al. 2018; Russo et al. 2018; Zhao and Yuan 2021; Zhao et al. 2013). In addtition, cyanidin-3-O-xyloside, which was also previously identified in leaves, flowers, fruit peels of ‘Tunisia’ cultivars (Fellah et al. 2018), was also considered as one of the major major anthocyanins found in ‘Moshiliu’ pomegranate.

The Cy, Dp, and Pg derivants accounted for more than 98% total anthocyanins for both cultivars (Fig. 1b). In red ‘Hongbaoshi’ pomegranate, a large amount of pelargonidins (orange color) presented in R1, with a rapid decline with time advancement. Similar results were also reported that a large amount of Pg3,5dG in flowers declined when the fruit began to develop (Arlotta et al. 2020; Ben-Simhon et al. 2011). The propotion of Cy and Dp anthocyanins were fluctuated in red ‘Hongbaoshi’, whereas in purple-red ‘Moshiliu’ Cy derivants were gradully decreased and Dp anthocyanins were increased over the time cousrse of fruit development. The fluctuation but not a linear changes for Cy and Dp in red ‘Hongbaoshi’ has also been mentioned by previous studies (Ben-Simhon et al. 2011; Harel-Beja et al. 2019; Zhao et al. 2013). In fact, at the onset of the coloration period, fruits exhibit transition from red coloring to red fading at the early stage, and then accumulating anthocyanins gradually in the later stages of fruit development. As a result, the dynamics of biosynthesis and degradation of anthocyanin together may influence changes in compositon and ratio of different anthocyanins.

We subsequently analyzed the six major anthocyanis in both cultivars and development stages (Fig. 1c). When comparing the two cultivars, a significantly higher content of anthocyanins in ‘Moshiliu’ was observed than in ‘Hongbaoshi’, especially Cy3G and Dp3G (Fig. 1c), in line with previsous observations (Zhao et al. 2013). It was also revealed that the dominant anthocyanins shifted from Pg3,5dG towards Cy3G in ‘Hongbaoshi’, from Cy3,5dG towards Dp3G in ‘Moshiliu’. Considering the exterior fruit color and anthocyanins profiles, the red ‘Hongbaoshi’ and dark-red ‘Moshiliu’ can be termed as Cy-type and Dp-type pomegranate, respectively.

Additional analysis showed that the amount of 3,5-diglucosides was higher than that of 3-glucosides in the early stages, while in later maturity stages mono-glucosides reached higher concentration than the di-glucosides (Fig. 1d). This trend was consistent with previous reports demonstraing similar shifts between mono-glucosides and di-glucosides with time advancement (Hernández et al. 1999; Singh et al. 2019).

Overall, both cultivars exibited similar anthocyanin-related trend during fruit development as previously reports (Zhao et al. 2013), with decreased and then increased accumulation of anthocyanins towards ripening. This developmental patten of pigmentation differed from other fruits such as blueberry (Lafferty et al. 2022), fig (Zhang et al. 2023), and Chinese bayberry (Li et al. 2023) that pigmentation intensity displayed a linear change with fruit maturation. This phenomenon may be explained as a dilution effect of anthocyanins accompanied by a rapidly expand in the size of the fruit towards ripening.

We can conclude that the anthocyanin profiles in pomegranate are developental regulated. The typical chacteristics were the high proportion of Pg in early stages in ‘Hongbaoshi’ but high Dp in late stages in ‘Moshiliu’. The considerably high content of Cy and Dp was the main reason behind the color variation between two cultivars. Unlike other horticultural fruit crops with various types of anthocyanins, such as grape (Balík et al. 2013; He et al. 2010; Huang et al. 2009), blueberry (Barnes et al. 2009), litchi (Rivera-López et al. 1999), the paramount anthocyanins in pomegranate were only 3-glucoside and 3,-5 diglucosides of Cy, Dp, and Pg, which covered three main branches of anthocyanin biosynthesis. Therefore, a well-defined anthocyanin composition makes pomegranate an excellent system for use to study the regulation mechanism of anthocyanin biosynthesis.

Differential anthocyanin metabolites (DAMs)

The PCA score plot results showed an obvious separation between sample groups, suggesing that there were differences in anthocyanins between two cultivars (Fig. S2a). Additionaly, a closer cluster relationship in ‘Hongbaoshi’ sampels than that in ‘Moshiliu’ suggested a homogeneous anthocyanin profile in ‘Hongbaoshi’. Three replicated samples in both cultivars grouped together indicated accurate repeatability and high reliability of samples and UPLC data.

In order to define the differentially accumulated anthocyanins resulting in color variation between ‘Hongbaoshi’ and ‘Moshiliu’ pomegranate, we looked for DAMs in samples (Fig. 2, Table S3). In ‘Hongbaoshi’ cultivar, the accumulated anthocyanins were significantly different, including 6 (3 up-regulated and 3 down-regulated), 6 (6 down-rgulated), 7 (7 up-regulated) anthocyanins in R2 vs.R1, R3 vs. R2, and R4 vs. R3, respectively. More differential accumulated anthocyanins were found in ‘Moshiliu’ than that in ‘Hongbaoshi’, with 13, 7, and 20 anthocyanin in three comparison groups. The results suggest a more complicated pitucure for purple-red trait of ‘Moshiliu’, since it dependes on the deposition of more different types of anthocyanins.

Fig. 2
figure 2

Venn diagrams showing the differential anthocyanins in comparison groups. a The pairwise comparisons in ‘Hongbaoshi’. b The pairwise comparisons in ‘Moshiliu’. c The pairwise comparisons in the same developmental stages between ‘Hongbaoshi’ and ‘Moshiliu’

Full size image

There were 3, 2, and 12 differential anthocyanins commonly shared in cultivar and developmental stages pairwise comparisons, suggesting that anthocyanin accumulation was characterized by homogeneity within cultivars but heterogeneity in developmental stages across cultivars. Thus, it can speculate that anthocyanin profiles are more sensitive to developmental signals in pomegranate.

Overview of the transcriptomic data

The transcriptomic analysis can provide crucial system-level framework for understanding the molecular basis of fruit color diversity. In this study, the triplicate sampling of four development stages in two cultivars yielded 24 RNA samples for RNA sequencing analysis. More than 38 million reads were generated per sample with the average Q30 value exceeded 89% (Table S4) which suggested that the transcriptome results were good for further analysis. After raw data filtering, a total of 1.05 billion high-quality reads (average 43 million reads from each sample) were generated for ‘Hongbaoshi’ and ‘Moshiliu’ cultivars from different development stages (Table S4). The obtained high-quality reads were mapped on pomegranate genome, showing that the proportion of total mapped reads were more than 94%. On average, more than 95% mapped reads were uniquely mapped to pomegranate genome.

FPKM of each transcript was determined to represent the normalized expression level by uniquely mapped reads processing for each sample. In general, genes with a mean FPKM>1 in at least one of the 8 triplicate samples were recognized as to be expressed. Overall, the number of expressed genes in samples varied from 59.2% (R4) to 67.1% (R1) in ‘Hongbaoshi’, and 62.4% (P4) to 66.1% (P2) in ‘Moshiliu’. The number of expressed genes centered on those with 1 ≤ FPKM ≤ 100 (Fig. S3a). The distribution of expression range count was similar in two cultivars, however, the least number of genes with 10 ≤ FPKM ≤ 100 was observed in the fourth stage in both cultivars (Fig. S3a).

Identification of DEGs involved in anthocyanin biosynthesis

In order to compare the transcriptomes representing four development stages in two cultivars, a Pearson’s distance correlation matrix was prepared (Fig. S3b). The matrix and the clustering dendrogram showed a clear distinction between the first two (S1 and S2) and last two developmental stages (S3 and S4) regardless of the cultivar, indicating that there were substantial transcriptomic differences between early and late developmental stages. PCA analysis demonstrated a high similarity between ‘Hongbaoshi’ and ‘Moshiliu’ samples at the same developmental stages (Fig. S2b), implying that the transcriptional regulation in pomegranate is more dependent on developmental signals other than cultivars. Also, three biological replicates of each sample were clustered closely, which confirmed the high reliability and repeatability of samples.

The differences in gene expression levels were analyzed by comparing each pair of consecutive developmental stages (Fig. 3, Table S5). Totally, 5795, 5319, and 4174 DEGs were identified between R2 vs. R1, R3 vs. R2, R4 vs. R3, respectively. Similarly, 5850, 5275, and 5109 DEGs were found in P2 vs. P1, P3 vs. P2, P4 vs. P3, respectively. The number of DEGs in each pairwise comparison showed a trend with a progressive decrease throughout developmental stages in both cultivars. The up-regulated DEGs were dominant in both cultivars when comparing S2 with S1. On the contrary, the number of down-regulated DEGs were dominant between S3 vs. S2. However, when comparisons were made between S4 vs. S3, a simimlar number of up-regulated and down-regulated DEGs were observed in both cultivars, impling that relatively stable levels of gene expresstion were manifested when fruits attained maturity. The Venn diagram in Fig. 3 showed that there were 994 and 1196 common shared DEGs in ‘Hongbaoshi’ and ‘Moshiliu’, respectively. These results demonstrated that fruit development involved the dynamic transcriptional modulation of a large number of genes.

Fig. 3
figure 3

The differential expressed genes in comparison groups. a The DEGs identified in comparison groups. b The DEGs identified in pairwise comparisons of ‘Hongbaoshi’. c The DEGs identified in pairwise comparisons of ‘Moshiliu’. d The DEGs identified in the same developmental stages between ‘Hongbaoshi’ and ‘Moshiliu’

Full size image

In order to analyze fruit color differences in ‘Hongbaoshi’ and ‘Moshiliu’, the DEGs of both cultivars in the same fruit development stage were compared. Comparisons of gene expression levels between both cultivars showed that there were 3029, 1686, 3666, and 3173 transcripts differentially expressed between P1 vs. R1, P2 vs. R2, P3 vs. R3, P4 vs. R4, respectively (Fig. 3), with 997, 498, 2016, and 1754 DEGs up-regulated, respectively. The number of stage-uique genes varied from 391 (S2) to 1531 (S3), and there were 267 DEGs common to all pairwise comparison groups (Fig. 3). In brief, it can be seen that a high similarity within two cultivars among different fruit development stages and a large variation across cultivars in the same fruit devlopment stages for the overall transcriptomic data.

To identify the metabolic pathways related to anthocyanin/flavonoid biosynthesis that were enriched, KEGG enrichment was conducted by comparing different stages in two cultivars. As illustrated in Fig. S4, DEGs in three comparison groups of ‘Hongbaoshi’ cultivar were significantly enriched in the anthocyanin biosynthesis related pathway, mainly ‘flavonoid biosynthesis’ (ko00941), ‘phenylpropanoid biosynthesis’ (ko00940), and ‘biosynthesis of secondary metabolites’ (ko01110), suggesting that these DEGs in three KEGG pathway participanted in fruit color regulation in ‘Hongbaoshi’. In ‘Moshiliu’ comparison groups, three anthocyanin biosynthesis related pathways were significantly enriched in P2 vs. P1, but there were only ‘phenylpropanoid biosynthesis’ and ‘biosynthesis of secondary metabolites’ seperately enriched in P3 vs. P2 and P4 vs. P3 groups, implying that P1 to P2 was a crucial stage in anthocyanin accumulation of ‘Moshiliu’. This result also demonstrated the transcriptional regulation varied from anthocyanins/flavonoids to other metabolites with fruits ripening. This can also be seen from Fig. S5, which shown the enrichment profiles of common DEGs in two cultivars. The common DEGs in ‘Hongbaoshi’ were firstly enriched in ‘flavonoid biosynthesis’ while this KEGG pathway was not enriched in common shared DEGs of ‘Moshiliu’. Different from ‘Moshiliu’, the DEGs in ‘Hongboshi’ enriched in three anthocyanin-related pathway with time advancement, demonstrating that anthocyanin biosynthesis in ‘Hongbaoshi’ fruit was one of the major metabolic activities.

KEGG enrichment analysis of DEGs in the same fruit development stage of both cultivars could be helpful to understand the crucial stages for color variation between red ‘Hongbaoshi’ and purple-red ‘Moshiliu’. Of the four stages (Fig. S5), DEGs in both S2 and S3 significantly enriched in ‘flavonoid biosynthesis’ and ‘anthocyanin biosynthesis’, indicating that S2 and S3 were critical time periods for color variation of both cultivars.

Transcriptional profiles of the genes related to anthocyanin biosynthesis

In pomegranate, most structural genes that encode enzymes involved in anthocyanin biosynthesis pathway have been obtained by PCR (Arlotta et al. 2020; Ben-Simhon et al. 2015; Zhao et al. 2015), transcriptomic (Ono et al. 2011), and genomic (Yuan et al. 2018) methods. In the meanwhile, expression profiles of anthocyanin genes were studied in various cultivars at transcriptional levels, trying to find the key genes controlling the anthocyanin accumulation (Ben-Simhon et al. 2011; Harel-Beja et al. 2019; Kaur et al. 2019). However, the data have not led to a definite conclusion, suggesting that the regulatory mechanisms underlying gene expression divergence in pomegranate might be different with cultivars or developmental stages.

In this study, according to the KEGG pathway enrichment analysis of DEGs in comparison groups, we searched for DEGs implicated in flavonoid pathways. Totally, 21 anthocyanin biosynthesis-related DEGs, including PAL (LOC116196349), C4H (LOC116199390), CHS1 (LOC116196157), CHS2 (LOC116196364), CHI1 (LOC116187009), CHI2 (LOC116194939), CHI3 (LOC116196656), F3H1 (LOC116201161), F3H2 (LOC116211316), F3’H (LOC116215135), F3’5’H (LOC116200281), DFR1 (LOC116214145), DFR2 (LOC116215495), ANS (LOC116213234), UGT1 (LOC116193155), UGT2 (LOC116204995), FLS1 (LOC116198870), FLS2 (LOC116203029), LAR1 (LOC116202057), LAR2 (LOC116210922), and ANR (LOC116208810) were identified. All individual gene expression patterns were displayed in Fig. 4. In ‘Hongbaoshi’ pomegranate, most of DEGs were up-regulated in R2 and R4, but down-regulate in R1 and R3. In ‘Moshiliu’ pomegranate, the transcriptional levels of four genes C4H, CHS1, F3H1, F3’H gradually increased during fruit development. Nine genes (CHS2, CHI2, CHI3, F3H2, F3’5’H, DFR1, DFR2, FLS2, LAR1, and LAR2) showed higher expression levels in ‘Moshiliu’ with fruit development than that in ‘Hongbaoshi’ fruit. Previously, Rouholamin et al. (2015) reported that DFR gene transcripts in purple-red genotype showed 3 and 5 times more than red and green skin genotypes. However, according to the opinion of Trainin et al., except for F3’5’H gene, other structural genes, including CHS, CHI, F3H, DFR, F3’H, LDOX, ANR and LAR, were expressed at a much higher level in the peel of red pomegranate as compared to “black” genotype in young fruits (Trainin et al. 2021). Thus, the expression of anthocyanin-related genes is related to multiple temporal and environmental factors.

Fig. 4
figure 4

The expression levels of genes involved in anthocyanin biosynthesis in ‘Hongbaoshi’ and ‘Moshiliu’ pomegranate during fruit development. The heatmap of each DEG was constructed based on log2 (FPKM) and the values were row scaled

Full size image

It is notable that the expression levels of ANR gene were significantly lower in ‘Moshiliu’ throughout fruit development, whose average expression differed nearly 25-fold compared with that of ‘Hongbaoshi’. Previous analysis strongly suggested that a single base deletion of ANR gene potentially inactivated or reduce the activity of the ANR protein, reducing competition over anthocyanidins formation and resulting in overproduction of anthocyanin (Trainin et al. 2021). These results demonstrated that the high expression levels of upstream genes and extremely low expression of ANR might play an important role in anthocyanin accumulation in ‘Moshiliu’ pomegranate. The color mutation arose from single branch-pathway gene was different from results reported in Camellia sasanqua (Fan et al. 2023) and Perilla frutescens (Liu et al. 2023), whose purple phenotypes arose from expression differences of several main anthocyanin pathway genes.

Our data showed that most of structural genes involved in anthocyanin pathway were expressed at a lower level in ‘Moshiliu’ as compared with ‘Honghaoshi’ in S1 and S2 stages, which were consistent with the results of Trainin et al. (2021). Instead, this pattern in expression levels were reversed when fruits development reached S3 and S4, including those genes involved in branch pathways, such as FLS2, LAR1, and LAR2. In can be inferred that the expression patterns of anthocyanin-related genes in red and purple-red pomegranate were associated with fruit development, which can be divided into two temporally distinct phases S1~S2 and S3~S4. This phenomenon can be explained that at early stages negative feedbacks in purple-red pomegranate were generated owing to high anthocyanin accumulation thereby resulted in reduced transcription of most of anthocyanin-related genes. From the middle to late developmental stages, the negative feedbacks were weak with the increased expression of ANR, thus caused higher expression levels for most genes in purple-red pomegranate.

Anthocyanin biosynthesis-related TFs and R2R3-MYBs

TFs in all DEG lists, belonging to 54 different families, exhibited differential regulation in at least one of the cultivars and developmental stages. The members of AP2/ERF, bHLH, bZIP, C2C2-Dof, C2H2, HB-HD-ZIP, MADS-MICK, MYB, NAC, WRKY TF families were highly represented in these samples (Table S6). Totally, there were 510, 460, and 388 different expressed TFs in ‘Hongbaoshi’ comparison groups, 489, 397, 482 in ‘Moshiliu’, and 257, 156, 312, 284 different expressed TFs in developmental stage comparison groups, respectively.

It is well established that the expression of anthocyanin biosynthesis-related structural genes is regulated by a conserved MYB-bHLH-WDR (MBW) regulatory complex (Miller et al. 2016; Xu et al. 2015; Zhang and Zhu 2023). The activity of R2R3-MYB genes has been suggested to be the primary cause of natural variation in anthocyanin pigmentation in plants. In pomegranate, though many studies monitored the expression profiles of MYBs involved in anthocyanin accumulation in various fruit cultivars or tissues (Arlotta et al. 2020; Ben-Simhon et al. 2011; Khaksar et al. 2015; Rouholamin et al. 2015; Zhao et al. 2015), the analysis of MYBs belonging to different subgroups (SGs) have not been described so far.

Hence, we specifically searched R2R3-MYB TFs in our comparative transcriptome datasets. In this study, R2R3-MYBs have been divided into 24 subgroups according to sequence similarity to Arabidopsis thaliana (Fig. 5a). It has been reported that R2R3-MYB TFs from SG4, SG5, and SG6 mediate in the regulation of anthocyanin biosynthetic pathway (Dubos et al. 2010; Hichiri et al. 2011). The MYB TFs from SG6 are involved in the regulation of anthocyanin accumulation, while those from SG5 are implicated in the biosynthesis of anthocyanin and (or) proanthocyanin (PA). In contrast, SG4 MYBs are regarded as repressors of the flavonoid/anthocyanin pathway (Allan and Espley 2018; Yan et al. 2021). Based on the phylogenetic clustering results, a total of 10 differential expressed R2R3-MYBs were identified, including 5 members from SG4, 3 from SG5, and 2 from SG6 (Fig. 5). Comparison of the transcript levels revealed that LOC116202447 and LOC116287201 showed a lower abandance in both cultivars (FPKM < 5). Therefore, we focused analysis on remainting 8 R2R3-MYBs.

Fig. 5
figure 5

Analysis of the MYB SG members in pomegranate. a Phylogenetic tree of MYBs constructed by neighbor-joining method. b Heatmap of expression levels of SG4, SG5, and SG6 MYBs involved in anthocyanin biosynthesis. R1~R4 and P1~P4 indicate ‘Hongbaoshi’ and ‘Moshiliu’ collected from four developmental stages, respectively

Full size image

SG6 R2R3-MYBs are well-characterized anthocyanin regulators. Most studies have indicated that the levels of MYB expression increase with the intensity of red color in fruit, such as kiwifruit MYB110 (Peng et al. 2019); apple MdMYB90-like (Sun et al. 2021), MdMYB10 (Espley et al. 2007), and MdMYBA (Ban et al. 2007); pear PyMYB10 (Feng et al. 2010); peach PpMYB10.1 (Tuan et al. 2015); grape VvMYBA1 and VvMYBA2 (Walker et al. 2007). The constantly elevated expression of SG6 LOC116202527 has also been observed in red ‘Hongbaoshi’, which its transcript level increased simultaneously with anthocyanins accumulation. However, the expression of LOC116202527 was dramatically decreased in purple-red ‘Moshiliu’ when compared with ‘Hongbaoshi’. The increased anthocyanin production with down-regulation of MYB genes has not been recorded in other plants, which suggests that the SG6 LOC116202527 may not a determinate factor of anthocyanin biosynthesis in ‘Moshiliu’. Similarly, it has been reported that a MYB gene AN2, whose sequence was identical to that of LOC116202527, was constantly expressed in all genotypes irrespective of fruit skin color (Rouholamin et al. 2015). These results indicate that it may be cultivar/genotype-dependent when LOC116202527 functions as a regulator of anthocyanin biosynthesis.

The transcription abundance of LOC116192926 and LCO116215847 from SG5 were steadily increased in ‘Moshiliu’ in comparison with ‘Hongbaoshi’, while LOC116193214 was only highly expressed in P3 stage of ‘Moshiliu’. Typically, SG5 MYBs regulate PA biosynthesis, recent studies have demonstrated that some MYBs also contribute to the regulation of both anthocyanin and PA biosynthesis, such as kiwifruit MYBC1 (Peng et al. 2020), grape VvMYB5b (Deluc et al. 2008), strawberry FhMYB5 (Li et al. 2019). This study revealed much higher transcription levels of two SG5 MYBs in ‘Moshiliu’ than ‘Hongbaoshi’, indicating their involvement in biosynthesis of anthocyanins. Given the fact that there was a sharp decrease in expression of LOC116202527 of SG6 in high anthocyanin-accumulating ‘Moshiliu’, it is reasonable to speculate that there might be a balance among the members of MYB genes in controlling anthocyanin biosynthesis in pomegranate. This have also been reported in red cabbage (Yuan et al. 2009) with a simultaneous up- and down-regulation of anthocyanin MYB transcription factors. These results suggest a possible coordinated regulatory role for MYBs in SG5 and SG6 to activate the anthocyanin pathway in pomegranate.

SG4 MYBs mainly act as repressors of anthocyanin biosynthesis by directly inhibiting or indirectly binding to other proteins to repress the expression of structural genes (Yan et al. 2021). In this study, the expression profiles of four SG4 MYBs could be divided into two types: LOC116203744 and LOC116212505 were up-regulated in ‘Hongbaoshi’, while LOC116205515 and LOC116212778 were up-regulated in ‘Moshiliu’. It has been proposed that there are two types of expression patterns considering the relationships between the expression of SG4 repressor genes and anthocyanin accumulation (Chen et al. 2019). One type is that the expression of genes is negatively associated with anthocyanin accumulation, such as Ginkgo biloba GbMYBF2 (Xu et al. 2014), and polar PtrMYB182 (Yoshida et al. 2015); the other type is that genes are highly expressed during active stages of anthocyanin accumulation (Chen et al. 2019), including Vitis vinifera C2 (Xie et al. 2020), and Platycodon grandiflorus PlgMYBR1 (Kim and Hyun 2023). The expression profile of SG4 MYB LOC116203744 and LOC116212505 belonged to the former style, while LOC116205515 and LOC116212778 exhibited the later pattern. These results thereby allow us to predict that SG4 MYB repressors play roles by either preventing the accumulation of anthocyanins or providing feedback repression to finely control and limit anthocyanin levels.

Overall, each R2R3-MYB transcription factor gene from SG4, SG5, and SG6 controlling anthocyanin biosynthesis seems to have a specific role under developmental signals. Transcriptional activators and repressors may cooperatively regulate anthocyanin biosynthesis in pomegranate. The expression of these MYB genes is important in determining genotypic differences in anthocyanin content. However, we cannot rule out the involvement of other TFs in regulation of anthocyanin biosynthesis. More studies are needed to fully understand the roles of different SG MYB members in activating or inhibiting anthocyanin biosynthesis, and the interaction with other TFs including bHLH and WDR.

The correlation of anthocyanin-related genes and anthocyanin content

Correlation analysis of anthocyanin structural genes and MYB-TFs were performed to identify genes related to anthocyanin biosynthesis. As shown in Fig. 6a of ‘Hongbaoshi’, most of MYBs (except LOC116202527) were correlated positively with nearly all structural genes. Especially, the transcriptional profiles of LOC116203744 and LOC11625847 were found to be significantly correlated with most early and late biosynthetic genes expression, respectively. In ‘Moshiliu’ (Fig. 6b), though the expression of most MYBs were positively correlated with that of structural genes, only several MYBs significantly correlated with several committed genes in the biosynthetic pathway, such as LOC116193214 and CHI1, LOC116212505 and UGT2. The correlation between LOC116202527 and structural genes was positive in ‘Moshiliu’ but negative in ‘Hongbaoshi’, suggesting the opposing transcriptional roles of LOC116202527 in both cultivars. Our data showed that four MYBs from SG4 were more or less positive with structural genes, indicating their involvement in the biosynthesis of anthocyanins. However, it has been found previously that the expression pattern of a SG4 member PgMYB4-like was not correlated with any of the selected structural genes, inferring that the gene linked to the production of flavonoids other than the anthocyanins (Arlotta et al. 2020).

Fig. 6
figure 6

The correlation analysis of anthocyanin-related genes and anthocyanin content. * and ** indicate significant correlation at P < 0.05 and highly significant correlation at P < 0.01, respectively. a The structural genes and TFs in ‘Hongbaoshi’. b The structural genes and TFs in ‘Moshiliu’. c The structural genes and anthocyanin content in ‘Hongbaoshi’. d The structural genes and anthocyanin content in ‘Moshiliu’

Full size image

Because major anthocyanin components in pomegranate were 3-glucosides and 3,5-diglucosides of Cy, Dp and Pg, the correlation analyses were carried out between anthocyanin-related genes and the six major anthocyanins. It can be seen from Fig. 6c that the expression of most structural genes except CHI1 were positively related with anthocyanins but less with Pg3,5dG (except F3’5’H) in ‘Hongbaoshi’ pomegranate. This result was in agreement with a previous investigation on Wonderful cultivar that the transcriptional profile of key genes like PgCHS, PgDFR, and PgUGT was highly significantly correlated with most of anthocyanins content (Arlotta et al. 2020). In contrast, the expression of anthocyanin genes in ‘Moshiliu’ pomegranate only positively correlated with Cy3G and Dp3G, while negative correlations were found between genes and other four kinds of anthocyanins (Fig. 6d). The negative relationship between expression levels of anthocyanin-related genes and anthocyanin content suggested that there may be other TFs involved in the biosynthesis of anthocyanins in ‘Moshiliu’. Also, these results indicate that the anthocyanin biosynthesis was differently regulated in red and purple-red cultivar as mentioned by Trainin et al. (2021).

Confirmation of gene expression using qRT-PCR

To further validate the expression of genes in ‘Hongbaoshi’ and ‘Moshiliu’, qRT-PCR during fruit development was performed for 12 representative genes (Fig. 7). Most genes exhibited similar expression profiles between qRT-PCR and RNA-seq data. Pearson correlation analysis showed high correlation coefficients (0.700~0.972), indicating that the FPKM values were highly correlated with the qRT-PCR, which confirmed the reliability of the RNA-seq data.

Fig. 7
figure 7

Comparison of expression profiles of representative genes by RNA-seq and qRT-PCR. Columns represent expression determined by qRT-PCR (left-axis), while lines represent expression by RNA-seq in FPKM values (right-axis). The x-axis in each chart represents the four developmental stages (S1-S4). Correlations between RNA-seq and qRT-PCR expressions were calculated and their associated P value are indicated

Full size image

Conclusion

Transcriptomic and metabolomic data were integrated to explore the anthocyanins, structural genes, and R2R3-MYB regulatory genes linked to red color mutation for red ‘Hongbaoshi’ and purple-red ‘Moshiliu’ pomegranate. Metabolites analysis demonstrated that the skin color differences were predominantly due to large differences in the content of Cy and Dp anthocyanins. Transcriptomic data implied distinct roles played by various anthocyanin-related genes in two cultivars. The extremely low expression ANR gene was responsible for the high accumulation of anthocyanins in purple-red ‘Moshiliu’. Moreover, the R2R3-MYBs from SG4, SG5, and SG6 exhibited expression patterns which were unique to different cultivars. Therefore, further studies are required to identify MYB genes serving as key regulators that control anthocyanin biosynthesis individually or coordinately in pomegranate.