Putative Synergetic Genes Involved in the Syntheses of Carotenoids and Flavonoids in 'Cara cara' Navel Orange as Revealed by Integrated Transcriptome and Metabolomics Analysis

Carotenoids and avonoids are important secondary metabolites in plants, which exert multiple bioactivities and benets to human health. Although the genes that encode carotenogenesis and avonoids enzymes are well characterized, the transcriptional regulatory mechanisms that are related to the pathway genes remain to be investigated. In this study, Cara cara navel orange fruit at four development stages were used to identify the key genes and TFs for carotenoids and avonoids accumulation. In this study, Cara cara navel orange (CNO) was used to investigate the proles of carotenoids and avonoids by a combination of metabolomic and transcriptomic analyses. The important stage for the accumulation of the major carotenoid, lycopene was found to be at 120 DAF (days after orescence). The transcripts of ve carotenogenesis genes were highly correlated with lycopene contents, and 16, 40, 48, 24 and 18 transcription factors (TFs) were predicted to potentially bind their promoters, respectively. Narirutin in the esh at the early 60 DAF was most important and 17, 22, 14, 25, 24 and 16 TFs could potentially bind their respectively. Furthermore, 15/15 candidate TFs might regulate at least three key genes and contribute to carotenoids/avonoids accumulation in CNO fruit. Finally, a hierarchical model for the regulatory network among the pathway genes and TFs was proposed.

lycopene in citrus fruit not only improves the fruit quality but is also bene cial to human health.
In higher plants, the isomeric C5 isopentenyl pyrophosphate (IPP) is synthesized in 2-C-methyl-Derythritol-4-phosphate (MEP) pathway, and is used as the precursor of geranylgeranyl diphosphate (GGPP) by GGPP synthase gene, GGPPS. GGPP is the direct precursor for various carotenoids catalyzed by phytoene synthase (PSY), phytoene desaturase (PDS), ζ-carotene desaturase (ZDS), lycopene βcyclase (LCYB), carotene isomerase (CRTISO), carotene hydroxylase (HYD), violaxanthin de-epoxidase (VDE) and zeaxanthin epoxidese (ZEP), etc. [7]. Previous studies have shown that 1-deoxy-D-xylulose-5phosphate synthase (DXS) was an important rate-limiting enzyme in the MEP pathway in tomato, Arabidopsis and Kiwifruit [12,13]. Besides, the deoxyxylulose 5-phosphate reductoisomerase (DXR) was shown to be rate-limiting enzyme in Arabidopsis, over-expression of DXR increased the content of carotenoids [14]. PSY is the key rate-limiting enzyme in the biosynthetic pathway of carotenoids in tomato and citrus [15,16]. The contents of carotenoids increased signi cantly when PSY was over-expressed in both tomato and citrus [15,16]. In tomato fruit, DXS, PSY, PDS, ZDS and CRTISO are up-regulated at the mature stage, resulting in lycopene's over-accumulation [17]. The expression of DXS was positively correlated with the content of carotenoids in citrus [18]. 'Hong Anliu' sweet orange is a red esh mutant from 'Anliu' sweet orange, with its fruit accumulating lycopene around 180 DAF [18]. It was found that some biosynthetic pathway genes and transcription factors (TFs) contributed to the accumulation of lycopene by combining transcriptome, proteomic and metabolomic analyses [15,18,19]. However, to our knowledge, the molecular basis of lycopene over-accumulation is still not clear.
Although the genes that encode carotenogenesis enzymes are well characterized, the transcriptional regulatory mechanisms that are related to the pathway genes remain to be investigated. TFs directly regulate the expression levels of their target genes, and take part in various biological processes in plants. Some TFs have been found to be associated with the accumulation of carotenoids, such as TAGL1, FUL1/2 and ERF6 [20][21][22]. Lycopene level was increased by over-expression of TAGL1 or PpPLENA in tomato fruit [20][21][22], while decreased in TAGL1 or FUL1/2 loss of function mutant [22]. Some TFs can directly bind to the biosynthetic pathway genes' promoters, then affect the accumulation of carotenoids. For example, PIF1 and SlMADS1 specially bound the PSY promoter and decreased its transcription and thus altered the carotenoids pro le [23,24], whereas HY5 and RIN up-regulated the expression of PSY in tomato [25]. In addition, the CrMYB68 decreased the expression levels of CrBCH2 and CrNCED5 in citrus [26], whist CsMADS6 increased those of LCYb1, PSY and PDS in 'Hong Anliu' [15].
Cara cara navel orange (CNO) is a bud mutant with red-esh derived from Washington navel orange, due to the over-accumulation of lycopene and the high content of β-carotene. Similar results were found with some red esh grapefruits and sweet oranges, but with varied ratios of lycopene to β-carotenoid [6], indicating the different patterns of carotenoids within different genotypes. It was found that the expression levels of DXS, HDS and HDR in CNO were higher than those in Washington navel orange at the mature stage [27]. However, the reason for the up-regulation of these genes remains unknown. The mature stage might not be the key stage for the biosynthesis of carotenoids in CNO fruit.
Flavonoids, one of the most important secondary metabolites in citrus fruit, have multiple functions, such as anti-cancer, anti-virus, anti-in ammatory and anti-bacterial activities [28,29]. Flavonoids mainly include avanone, avonoid, avonol, dihydro avonol and anthocyanin, and the avanone glycosides are the major avonoids in citrus fruit1. Our previous study found that the contents of avonoids in four sweet oranges tended to be stable or slightly decreased during 90-210 DAF, and that the lycopene accumulation may have direct or indirect effects on the biosynthesis of avonoids [8]. The avonoids are synthesized from phenylpropanoid pathway, mainly including PAL, C4H, 4CL, CHS and CHI genes [30]. It was found that the expression levels of these genes were at high levels at 60 DFA, but simultaneously dropped to extremely low levels at 90 210 DAF in four pomelos, the key pathway genes might be regulated by some TFs [30]. Numerous TFs have been reported to regulate lignin and anthocyanin biosynthesis in phenylpropanoid pathway [31]. For example, CsMYBF1 was found to bind to CHS-1 promoter and then affect avanone accumulation in citrus [32]; Ruby was found to bind to CHS, DFR and ANS promoters, then regulate anthocyanin accumulation in blood orange and Zipi pomelo [9,33]. Furthermore, many MYB TFs were reported to directly bind and regulate PAL and C4H promoter to affect lignin accumulation in Arabidopsis and sweet orange [34,35]. However, to our knowledge, the regulatory mechanisms of TFs in the biosynthesis of both avonoids and carotenoids remain unclear in citrus fruit.
In this study, HPLC-based carotenoids and avonoids pro ling combined with RNA-seq based gene expression data is utilized to mine the important pathway genes and TFs contributing to carotenoids and avonoids pro les in juice sacs during fruit development of CNO fruit. The results may provide new insights into the mechanism governing speci c carotenoid and avonoids accumulation in citrus fruit.

Results
Carotenoids and Flavonoids Pro les in Juice Sacs of CNO during Fruit Development Juice sacs at six different developmental stages of CNO were determined for variation in carotenoids and avonoids pro les using HPLC. The contents of both carotenoids and avonoids varied greatly at different developmental stages.
Seven major carotenoids were identi ed, including lycopene, phytoene, phyto uene, lutein, violaxanthin, βcryptoxanthin and β-carotene, with the level of lycopene being the highest (Additional le 1: Table S1; 1A). The content of lycopene was low at 60 DAF, and signi cantly increased from 90 to 120DAF, reached to a maximum at 150 DAF, then maintained or slightly decreased at the mature stage (Fig. 1A). The increases of lycopene levels were 2.65 µg/g, 23.23 µg/g and 14.19 µg/g from 60 to 90, 90 to 120 and 120 to 150 DAF, respectively, indicating that the stage from 90 to 120 DAF might be the most important for the lycopene biosynthesis (Additional le 1: Table S1; Fig. 1A). For the other carotenoids identi ed, phytoene and phyto uene were also signi cantly increased from 90 to 120 DAF, then maintained or slightly decreased at the mature stage (Fig. 1A). Lutein was found at 180 DAF and its level increased at 210 DAF; β-cryptoxanthin and β-carotene were detected at 150 DAF and tended to increase from 150 to 210 DAF; violaxanthin was detected at 120 DAF, then signi cantly increased from 150 to 180 DAF (Additional le 1: Table S1).
A total of 11 avonoids were identi ed in CNO fruit at four development stages, with narirutin as the dominant one at 60 and 90 DAF (Additional le 2: Table S2). The contents of total avonoids were the highest and lowest at 60 DAF and 210 DAF, respectively. The results suggested that the 60 DAF stage might be the important stage for the accumulation of avonoids (Fig. 1B).
Pearson correlation coe cient was − 0.797 between total carotenoids and total avonoids levels in CNO fruit at six development stages, while − 0.719 was at 60 to 150 DAF, suggesting a negative correlation between total contents of carotenoids and avonoids. Furthermore, before 150 DAF, the level of carotenoids constantly increased, whereas the content of avonoids sharply decreased at 90 DAF with a slight increase at 120 DAF due to the increase of hesperidin content. After 150 DAF, the level of lycopene signi cantly decreased, whilst the levels of the other main carotenoids and all avonoids decreased or kept at stable levels ( Fig. 1). These results showed that there were collaborative changes in carotenoids and avonoids biosynthesis, with150 DAF as the turning stage.  Table S3). Based on their expression levels, the average fragments per kilobase of transcript per million fragments mapped (FPKM) higher than 0.5 were selected for further analysis (Additional le 4: Table S4).

Important Pathway Genes for Lycopene Accumulation
The stage from 90 to 120 DAF was the most important stage for lycopene accumulation based on metabolic data. Thus, the DEGs were subjected to KEGG enrichment analysis, and the DEGs were found to be mainly enriched in metabolic pathways, biosynthesis of secondary metabolites and plant hormone signal transduction (Additional le 5: Figure S1 B). Among the DEGs, some genes were clustered into carotenoids biosynthetic and terpenoid backbone biosynthesis, including DXS1 (Cs1g20530), DXS2 (Cs9g05150), HDS (Cs8g16700), HDR (Cs8g07020) and ZDS4 (Cs3g11060), and their expression levels in the fruit at 120 DAF were signi cantly higher than thoseat90 DAF. Furthermore, KEGG enrichment analysis on the DEGs at 90-120 DAF found that some genes were clustered into plant hormone signal transduction, such as Cs9g08850 (ethylene signal transduction), Cs1g17210 (jasmonate signal transduction) and Cs9g18020 (abscisic acid signal transduction), etc: which might also play important roles in lycopene over-accumulation (Additional le 5: Figure S1 B).
To mine the most important genes for the over-accumulation of lycopene during fruit development, all the expressed genes were grouped using Mfuzz analysis based on their expression pattern changes, which yielded nine clusters ( Fig. 3; Additional le 6: Table S5). Notably, Cluster 2, 4 and 7 showed a good positive correlation with the content of lycopene, and 1,804, 1,604 and 1,546 genes were grouped in Cluster 2, 4 and 7, respectively ( Fig. 3). Further gene function annotation found that DXS1, PSY1 and ZDS2 (Cs3g11060) were in Cluster4, DXR and GGPPS2 were in Cluster2, and MDS (Cs5g03050) was in Cluster7, implying their importance as candidate genes accounting for lycopene over-accumulation. In addition, LCYB (orange1.1t00772) with consistently low expression levels was in Cluster 6 ( Fig. 3).
Next, correlation analysis between the genes within carotenoid biosynthetic pathway and the content of lycopene showed that DXS1, DXR, GGPPS2 and PSY1 were highly and positively related to lycopene levels, with correlation coe cients of 0.997, 0.893, 0.834 and 0.997, respectively, while the values of the other lycopene biosynthesis related genes were lower than 0.800 (Table 1; Fig. 4). Herein, DXS1, DXR, GGPPS2 and PSY1 might be the key biosynthetic pathway genes accounting for the high level of lycopene in juice sacs of CNO fruits.  Table S7).
Some TFs were highly positively or negatively correlated with DXS1, DXR, GGPPS2, LCYB and PSY1 genes that contribute to carotenoids accumulation. For example, DXS1 was positively correlated with 10 TFs, and negatively correlated with six TFs, and the coe cients of seven TFs were higher than 0.900, while four were lower than − 0.900 ( Fig. 5; Additional le 8: Table S7). These TFs might regulate the expression levels of ve important genes that contribute to lycopene accumulation in CNO fruit.

Important Pathway Genes and TFs for Flavonoids Biosynthesis
The contents of avonoids tended to decrease during fruit development (Fig. 1B). The expression levels of the most avonoids biosynthetic genes were found to be high at 60 DAF in CNO, while dropped at 120 and 150 DAF, including PAL-1, PAL-3, 4CL-2, 4CL-5, CHS-1 and CHI (Additional le 9: Figure S2). Results of Mfuzz analysis of Cluster 1 showed that their expression tended to decrease (Fig. 3), and KEGG annotation found that the genes in the cluster were mainly involved in metabolic pathways, biosynthesis of secondary metabolites, phenylpropanoid biosynthesis and so on (Additional le 10: Figure S3). Coincidently, PAL-1, PAL-3, 4CL-2, 4CL-5, CHS-1 and CHI grouped into Cluster 1 were identi ed via gene annotation.
To further identify the potential TFs that may regulate the above ve pathway genes and the accumulation of avonoids, TFs were predicted using PlantRegMap. The results showed that there were potentially 213, 229, 95, 262, 227 and 102 TF binding sites in PAL-1, PAL-3, 4CL-2, 4CL-5, CHS-1 and CHI promoters, respectively (Additional le 11: Table S8). The pathway genes with its potential 17, 22, 14, 25, 24 and 16 TFs were found to have high Pearson correlation coe cients (|r|>0.85), with 111 and seven as potential positive and negative regulators, respectively ( Fig. 6; Additional le 12: Table S9). Among the 118 potential regulators, there were 64 TFs, mainly including nine TFs from WRKY family, eight from MYB, six from bHLH and four from ERF (Additional le 12: Table S9).
Some TFs were highly positively or negatively correlated with PAL-1, PAL-3, 4CL-2, 4CL-5, CHS-1 and CHI genes that contribute to avonoids accumulation. For instance, PAL-1 was highly and positively correlated with 16 TFs, while negatively correlated with one TF, among them, the Pearson correlation coe cients of 12 TFs were higher than 0.900 ( Fig. 6; Additional le 12: Table S9). These TFs might regulate the expression levels of the six important pathway genes that contribute to avonoids accumulation in CNO fruit.

Postulated Hierarchical Network in Regulating Carotenoids and Flavonoids Pro les in CNO
Further analysis identi ed 24 TFs that could regulate important genes involved in both carotenoids and avonoids biosynthesis pathways. Twenty of these TFs might positively and negatively regulate avonoid biosynthesis and carotenogenesis genes, respectively, whereas two of them had the opposite effects and the other two were positively correlated with the important genes in both biosynthetic pathways (Additional le 13: Figure S4). Thus, these TFs might directly regulate the genes related to the biosynthesis pathways for both carotenoids and avonoids.
Fifteen and 15 TFs that could bind at least three carotenoids and avonoids biosynthesis genes, respectively, were subjected to the potential regulatory relationship analysis. Potential binding sites were predicted in 30 candidate TFs (Additional le 14: Table S10), and Pearson correlation analysis found that 21 of these TFs had strong potential regulatory relationships (|r| >0.85) (Additional le 15: Table S11). For example, orange1.1t02314, a carotenoids biosynthesis-related TF, may also potentially regulate avonoids biosynthesis-related TFs (orange1.1t00472, Cs7g04700, etc.) (Fig. 7A); Cs3g23190, a potential avonoids biosynthesis-related TF, was regulated by other potential carotenoids biosynthesis-related TF, such as Cs9g03820 (Fig. 7B). Besides, Cs3g19420, a carotenoids biosynthesis-related TF, may also potentially regulate other carotenoids-related TFs, such as Cs7g11810, Cs6g12530 and Cs9g03820 (Fig. 7C). The results might identify some TFs which affect the biosynthesis of carotenoids and avonoids indirectly.

Discussion
The Key Stage for Lycopene and Carotenoids Accumulation Carotenoids were the important primary and secondary metabolites in plants [36]. Seven main carotenoids were detected in CNO juice sacs using HPLC (Additional le 1: Table S1), maybe because the intermediate lycopene was over-accumulated, which may affect the biosynthesis of downstream carotenoids and other branch products. Similar to previous result, lycopene was found to be the most abundant carotenoid in CNO [6]. In the study, 90-120 DAF was identi ed as the key stage for lycopene and carotenoid biosynthesis (Fig. 1A). However, 180 DAF was found to be the key stage for lycopene accumulation in a sweet orange, 'Hong Anliu' [18]. Thus, this study has for the rst time revealed an early regulatory mechanism governing lycopene over-accumulation in CNO fruit.

Important Pathway Genes Contributing to Carotenoids Accumulation
In MEP pathway, GGPP is the direct precursor for the production of the rst carotenoid, the colorless phytoene, catalyzed by PSY [7,16]. However, DXS and DXR within MEP pathway play important roles in regulating the biosynthesis of carotenoid, abscisic acid, GAs and monoterpene [37]. To date, most of the carotenoid biosynthetic genes were cloned from citrus and other plants, such as GGPPS, PSY, PDS, ZDS, CRTISO, LCYB, LCYE, DHY, ZED and CCD [7,11], among them, PSY, PDS, ZDS and CRTISO were upstream to lycopene biosynthesis, while LCYB, LCYE, DHY, ZED and CCD were downstream. It was reported that the expression level of PSY was related to carotenoids accumulation in tomato and citrus, and PSY played an important role in lycopene biosynthesis [15,16]. LCYB and LCYE contributed to the production of βcarotene and α-carotene, respectively, using lycopene as a precursor [38]. The expression level of PSY in CNO was signi cantly higher than that in Washington navel orange fruit [39], and the higher expression levels of DXR and GGPPS were found in CNO fruit than in Washington navel orange fruit at the mature stage [27]. However, PSY and LCYB were found to be important for the accumulation of carotenoids in the sweet orange, 'Hong Anliu' [15].
In this study, the four genes, DXS1, DXR, GGPPS2 and PSY1 were found to signi cantly change during fruit development, based on metabolomic data, KEGG, Mfuzz cluster and Pearson correlation analysis (Additional le 5: Figure S1; Fig. 2; Table 1), indicating the important contribution of these genes to carotenoids accumulation in CNO fruit. The expression levels of these four genes were the highest at 120 DAF, coinciding with the highest increased rate of lycopene level from 90 to 120 DAF (Fig. 3). Although another signi cant increase of lycopene content happened from 120 to 150 DAF with a slightly lower rate compared to the previous period, the expression levels of the four genes at 150 DAF were lower than those at 90 and 120 DAF.
It is known that the accumulation of carotenoids is a net effect of several biological processes, including biosynthesis, transportation and degradation. However, the expression of the gene closely related to lycopene synthesis, LCYB, was kept at relatively constant low levels throughout fruit development (Fig. 3). Moreover, another lycopene synthesis-related gene LCYE, was found to have a high Pearson correlation coe cient with lycopene content (Table 1). However, the contents of its direct products, α-carotene and lutein, were retained at low levels. This may be due to the relative low enzymatic activity before 150 DAF, which to some extent may also result in the accumulation of lycopene together with the lower LCYB expression level and weak ABA production.

Potential TFs May Play Important Roles in Lycopene Accumulation
TFs play important roles in plant biological regulatory networks. It was reported that some TFs regulated the biosynthesis of carotenoids. For example, CsMADS6 could bind PSY and LCYB promoters and directly regulate their expression levels, and thus affect the accumulation of carotenoids in sweet orange fruit, 'Hong Anliu' [15], while PIFs, HY5, RAP2.2, RIN and SlMADS1 can directly bind the PSY promoter to alter the carotenoids pro les [23,24,40]. In this study, 16, 40, 48, 24 and 18 TFs showed high Pearson correlation coe cients (|r|≥0.85) with DXS1, DXR, GGPPS2, PSY1 and LCYB (Additional le 8: Table  S7).The TFs, including AP2, GRAS, MADS, MYB and WRKY, were found to possibly bind DXS1, DXR, GGPPS2 and PSY1 promoters (Additional le 8: Table S7). Eight and seven TFs may positively and negatively regulate at least three carotenoids biosynthetic pathway genes respectively, and thus may play important roles in the over-accumulation of lycopene in CNO. It was reported that over-expression of CubHLH1 decreased lycopene content in tomato, while PpPLENA had the opposite regulation effect [20,22]. And some TFs could affect plant photosynthesis and result in carotenoid accumulation rather than directly bind carotenoid biosynthetic genes, such as Gloden2-like in tomato [22]. Therefore, the TFs identi ed in the study might regulate other biological processes, such as hormone signal transduction and photosynthesis, etc. to affect the special carotenoids pro le in the CNO fruit. However, as reported in 'Hong Anliu', the Pearson correlation coe cient was only − 0.762 between CsMADS6 (Cs7g16970) and PSY1, while other MADS TFs, Cs7g11810 and Cs1g21080 showed higher and positive Pearson correlation coe cients (> 0.900) in CNO fruit [15]. It would be of interest to nd out if they are more important than CsMADS6 in lycopene accumulation.

Pathway Genes and TFs that May Play Important Roles in Flavonoids Accumulation
Flavonoids are important secondary metabolites in citrus fruit, and biosynthesized via phenylpropanoid biosynthetic pathway, with PAL, C4H, 4CL, CHS and CHI genes involved in the pathway [30]. The contents of various avonoids were stable or decreased during fruit development1, and the expression levels of PAL, C4H, 4CL, CHS-1 and CHI were high at 60 DPA and signi cantly decreased at 90, 120 and 210 DPA in Fenghuang pomelo, KaoPan pomelo, Huanong red pomelo and HB pomelo [30]. Similarly in this study, the expression levels of PAL-1, PAL-3, 4CL-2, 4CL-5, CHS-1 and CHI genes were found to decrease during fruit development (Additional le 9: Figure S2), indicating the important roles of these genes for avonoids accumulation at early stages.
Many TFs were reported to regulate phenylpropanoid biosynthetic pathway genes, and then affect the contents of anthocyanin, lignin and avonoids. For example, MYBF1 was found to bind CHS promoter and then regulate the synthesis of avonol in citrus and grapevine [32,35], while many other MYB TFs were reported to bind PAL, 4CL and CHS promoters and then regulate anthocyanin and lignin biosynthesis [40]. In the study, 64 potential TFs with high Pearson correlation coe cients were predicted based on PAL-1, PAL-3, 4CL-2, 4CL-5, CHS-1 and CHI promoters' analysis ( Fig. 6). Some of them could potentially bind promoters of multiple genes, and thus may be most important TFs regulating avonoids accumulation in CNO fruit.

Potential TFs that may Co-regulate Carotenoids and Flavonoids Accumulation
In the study, in CNO fruit, the content of carotenoids and expression levels of key genes involved in carotenoids synthesis tended to increase, while the level of avonoids and expression levels of the key genes tended to decrease ( Fig. 1; Fig. 3). Further analysis revealed that both pathways' key genes might be co-regulated by 24 potential TFs directly (Additional le 13: Figure S4). For instance, Cs1g21310 might positively and negatively regulate avonoids and carotenoids accumulation. Regulatory network analysis found that one carotenoid biosynthesis related TF could regulate avonoids-related TFs, and 11 avonoids-related TFs could also bind carotenoids-related TFs, which might regulate the accumulation of avonoids and carotenoids indirectly.
Finally, considering the potential TFs with direct and indirect functions in carotenoids and avonoids biosynthesis, 34 TFs might serve as important TFs for the syntheses of carotenoids, avonoids and other metabolites in CNO fruits to form a regulatory network (Fig. 7, Additional le 13: Figure S4 and Additional le 15: Table S11). These TFs may form an integrative fruit quality regulatory network with various metabolic pathways in fruits.

Conclusions
Carotenoids and avonoids are important metabolites in citrus fruit, and lycopene was found to be the major compound contributing to the red esh in CNO, while narirutin as the dominant avonoid at the early stage. Carotenoids and avonoids pro ling of CNO juice sacs at six developmental stages identi ed 90-120 DAF as the key stage for the over-accumulation of lycopene. Transcriptomic analysis indicated that DXS1, DXR, GGPPS2, PSY1 and LCYB genes were the most important genes for carotenoids accumulation, while PAL-1, PAL-4, 4CL-2, 4CL-5, CHS-1 and CHI for avonoids biosynthesis. Both of fteen candidate TFs were found to potentially bind the promoters of at least three carotenoids/ avonoids key pathway genes to regulate their expression levels, and thus help generate the special carotenoids and avonoids pro les in CNO fruits. In addition, a total of 34 TFs were postulated as co-regulators in both pathways directly or indirectly.

Materials
The

Determination of Flavonoids
Flavonoids were extracted from 1 g juice sacs for each sample using the method by Chen et al. [8] with minor modi cation. Flavonoids were analyzed using the above mentioned HPLC system with a C 18 Hypersil GOLD column (250 mm × 4.6 mm × 5 µm, Thermo Scienti c).

RNA Extraction, Transcriptomic Sequencing and Analysis
According to the method for total RNA extraction as described by Liu
Pearson correlation coe cients between the above pathway genes with potential binding TFs were calculated, based on transcriptome data by R or function with 'Pearson' method. The Pearson correlation coe cient value was ltered by higher than 0.850 or lower than − 0.850, and the results were shown using Cytoscapev3.7.2.

Data Analysis
The contents of carotenoids and avonoids were calculated using the standard curves. The R packages (nortest, stats and pgirmess) were used for analysis of variance with ANOVA (P < 0.05). Bar chart and dot charts were made based on DEGs and KEGG analysis results using R package, ggplot2. A heatmap of carotenoids biosynthesis-related genes was constructed based on transcriptome data using TBtools[42].  Figure 1 The    The relationships between DXS1 (Cs1g20530), DXR (Cs5g05440), GGPPS2 (Cs8g02140), PSY1

Figure 7
The relationships between the most important carotenoid and avonoid related TFs. A, orange1.1t02314 was a carotenoid related TF, which was regulated by avonoid related TFs; B, Cs3g23190 was a avonoid related TF, which was regulated by carotenoid related TF; C, Cs3g19420 was a carotenoid related TF, which was regulating other carotenoid related TFs.

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download.