Integration of metabolome and transcriptome reveals flavonoid accumulation in the intergeneric hybrid between Brassica rapa and Raphanus sativus

Brassica rapa and Raphanus sativus are two important edible vegetables that contain numerous nutritional ingredients. However, the agronomic traits and nutritional components of the intergeneric hybrid of B. rapa and R. sativus remain poorly understood. In this study, we used a stably inherited intergeneric hybrid of B. rapa and R. sativus as a model to study its metabolome and transcriptome profiles. Morphological and cytological analysis showed the intergeneric hybrid had the expected chromosome number and normal meiosis behavior. Moreover, the metabolome analysis showed multiple important secondary metabolites, including flavonoids and glucosinolates, were significantly upregulated in the hybrid. Furthermore, transcriptome data revealed that the expression level of the important genes involved in phenylpropanoid and flavonoid pathways was significantly upregulated in the hybrid. Ultimately, our data indicate the intergeneric hybrid will be a valuable bioengineering resource and promise to become a new-type hybrid vegetable with great medicinal value in future.

of the expression levels of multiple metabolites in plants [19][20][21][22][23][24] . High-throughput sequencing is widely used for the transcriptome analysis of different varieties, mutants and tissues of the Brassicaceae family. Transcriptome analysis has become an important tool to discover new signaling pathways and regulatory mechanisms of gene expression during hybridization. Recently, transcriptome sequencing has been combined with metabolomics analysis to increase the power of bioinformatics analysis [25][26][27] . Metabolomics analysis is a quantitative detection of all metabolites and their biochemical states in specific organisms or tissues, which provides a powerful way to study the metabolic phenotype of organisms in the study of functional genomics. Therefore, the combination of metabolomics with transcriptomic analysis has been used to explore the important genes participating in the regulatory pathways of specific metabolites in plants.
In this study, we used the intergeneric hybrid of B. rapa and R. sativus as a model to study its metabolic and transcriptome profiles. Morphological and cytological analysis first showed that the intergeneric hybrid displayed the expected chromosome number and normal meiosis behavior. We then investigated the overall metabolite landscape of B. rapa, R. sativus and their hybrid by metabolomics analysis. GC-MS and LC-MS identified a total of 623 metabolites of which 135 differentially expressed metabolites were found among B. rapa, R. sativus and the hybrid. Further analysis indicated that various phenylpropanoids, flavonoids and anthocyanins were upregulated in the hybrid compared with those in its parents. Moreover, transcriptome analysis found that the expression level of the genes that encode important enzymes involved in the phenylpropanoid and flavonoid/anthocyanin pathways was significantly upregulated in the hybrid in comparison with that in its parents. Taken together, the results demonstrated that the intergeneric hybrid of B. rapa and R. sativus was stably inherited and had more health-promoting metabolites than those of its parents, which indicate that the intergeneric hybrid has potential to become a new hybrid vegetable in the future.

Results and Discussion
Morphological and cytological analyses of the intergeneric hybrid of B. rapa and R. sativus.
The morphological analysis showed that the intergeneric hybrid of B. rapa and R. sativus exhibited intermediate characteristics between B. rapa and R. sativus with some exceptions including the color of leaves, the flower and the silique (Fig. 1A-F). Chromosome identification of the intergeneric hybrids was performed, and the plants with 38 chromosomes were selected for further studies ( Fig. 1G-I). Moreover, the chromosomes in somatic cells of the hybrid were hybridized in situ with the genomic DNA of R. sativus, and 18 of which exhibited strong signals in comparison with those of the other chromosomes ( Fig. 1H-I). This result demonstrated that the hybrid was derived from the intergeneric hybridization from R. sativus and B. rapa. In general, distant hybridization usually displays hybrid vigor, disease resistance and some other traits in the future generations 28 . However, several difficulties result from obtaining a hybrid from distant hybridization including cross incompatibility, hybrid death or sterility and progeny separation of the hybrid. Generally, except for the recombination of the nuclear genome, somatic hybridization by very distant species will lead to the hybridization of cytoplasmic organelles by protoplast fusion, which creates new genetic diversity and variations in these organelle genomes. Incorporation of whole genomes from two distant species through somatic fusion will introduce many exotic genetic materials and imbalances that induce somatic incompatibility, which might induce abnormal growth and development of the hybrid or hybrid regeneration 29 . For instance, somatic hybrids reported between the Brassica species in different tribes were generated by symmetric and asymmetric protoplast fusions 15 . In this study, the intergeneric hybrids with expected chromosome number and normal meiosis behavior were then sampled for further metabolomics and transcriptome analysis (Fig. 1G,J). Moreover, the intergeneric hybrid of B. rapa and R. sativus has been inherited for multiple generations and has high fertility. Therefore, the intergeneric hybrid of B. rapa and R. sativus has the potential to become a new type of vegetable and to provide valuable information to depict interactive networks between functional genes and their metabolites.
Metabolic profiling of B. rapa, R. sativus and their hybrid. To define a comprehensive landscape of metabolite reprogramming between B. rapa, R. sativus and their hybrid, we performed metabolite profiling of B. rapa, R. sativus and their hybrid by using LC-MS technologies. Three biological replicates of leaf tissues of R. sativus, B. rapa and the intergeneric hybrid were used for metabolic profile analysis. Figure 2A shows that the correlation coefficients of sample replicates were very high, which indicated that the metabolome data were reliable. A total of 623 metabolites were identified in B. rapa, R. sativus and their hybrid (Fig. 2B). Among the metabolites, 135 differentially expressed metabolites were detected between B. rapa and the hybrid, including 34 down-regulated and 101 up-regulated metabolites, respectively ( Fig. 2C and Supplementary dataset 1). For example, many flavonoids and anthocyanins were upregulated in the hybrid. We also observed that some metabolites, including quinate derivatives and amino acid derivatives, were upregulated in the hybrid. Between R. sativus and the hybrid, 155 differentially accumulated metabolites were detected, including 76 downregulated and 79 upregulated metabolites, respectively ( Fig. 2C and Supplementary dataset 1). The upregulated metabolites included flavonoids, amino acid derivatives, organic acid derivatives and so on. We also observed that 165 differentially accumulated metabolites were detected between B. rapa and R. sativus, including 68 downregulated and 97 upregulated metabolites, respectively ( Fig. 2C and Supplementary dataset 1). The main differentially expressed metabolites included flavonoids, quinate derivatives, anthocyanins, coumarin derivatives, organic acid derivatives and so on. We further used the KEGG database to annotate the differential metabolites and analyze their metabolic pathways. Figure 2D lists the top 20 KEGG pathways. Among the pathways, the top 3 pathways (ranked by p-value) included "Flavone and flavonol biosynthesis", "Phenylpropanoid biosynthesis" and "Isoflavonoid biosynthesis". Further analysis showed that multiple differentially expressed flavonoids were detected between B. rapa, R. sativus and the hybrid. As shown in Supplementary dataset 2, 46 differentially expressed flavonoids were identified between B. rapa and the hybrid, including 9 downregulated and 37 upregulated flavonoids, respectively. In addition, 70 differentially expressed flavonoids were identified between R. sativus and the hybrid, including 20 downregulated and 50 upregulated flavonoids, respectively. For example, luteolin is an important flavonoid and has many potential therapeutic activities such as antioxidant, anti-inflammatory, antimicrobial and anticancer activities. The results showed that the level of luteolin of the hybrid was much higher than that of R. sativus. In addition, we found that most derivatives of luteolin were significantly accumulated in the hybrid compared with those of its parents (Supplementary dataset 2). As secondary metabolites, flavonoids are widely distributed in plants. Flavonoids are the most important plant pigments, chemical messengers, physiological regulators and cell cycle inhibitors 30 . In addition, flavonoids play an important role in plant defense 31 . Moreover, flavonoids are often used as antioxidants, as a cardioprotective, as an antihypertensive, or for blood glucose regulation 32 . Brassica rapa and R. sativus are two edible vegetables that contain multiple nutrients and phytochemicals including flavonoids, phenolic compounds, and anthocyanins. In this study, we explored the overall metabolite reprogramming among B. rapa, R. sativus and their hybrid and found the level of most flavonoids was significantly upregulated in the hybrid compared with that in its parents. The incorporation of whole genomes from B. rapa and R. sativus introduced numerous exotic genetic materials in the hybrid, which led to the rearrangement of the genome and induced abnormal expression of key genes involved in many metabolic pathways. Moreover, we surprisingly found that sinigrin (an important glucosinolate) was enormously upregulated in the hybrid in comparison with that in its parents (Supplementary dataset 1). Various pharmacological activities of sinigrin reveal anticancer, antibacterial, antifungal and anti-inflammatory effects 33 . For example, sinigrin inhibited bladder cancer growth and blocked muscle invasion in an orthotopic rat bladder cancer model 34 . Collectively, the abnormal expression of flavonoids and sinigrin may represent new agronomic traits in the hybrid of B. rapa and R. sativus, which indicated that the hybrid of B. rapa and R. sativus has potential to become a novel hybrid vegetable with great medicinal value in the future. rapa and the intergeneric hybrid were sequenced using the Illumina Hiseq. 2000 platform. After removing the unknown reads, low-quality reads and adaptor sequences, a total of 18,873,973 and 92,858,774 and 44,155,651 clean reads were obtained for R. sativus, B. rapa and the hybrid, respectively. We conducted the de novo assembly of the clean reads of the hybrid from B. rapa and R. sativus by Trinity program, which generated a total of 23,671unigenes with an average length of 619 bp and an N 50 of 668 bp (Fig S1). Furthermore, the obtained unigenes were mapped to B. rapa and R.sativus genome, respectively. The results showed that 9817 unigenes come from B. rapa and 10,631 unigenes come from R. sativus. The transcriptome differences were further examined by differential expression analysis of RNA-seq data from B. rapa, R. sativus and the hybrid using the R package edgeR. When the B. rapa genome was used as the reference, 12,643 differentially expressed genes (6155 upregulated and 6488 downregulated) were identified among B. rapa, R. sativus and the hybrid (Fig. 3A, Supplementary  dataset 3). Quantitative real-time polymerase chain reaction (qRT-PCR) experiments validated the RNA-seq results using eight randomly selected differentially expressed genes (Fig. 3B).
Gene Ontology (GO) enrichment analysis was performed for the differentially expressed genes using the Cytoscape plug-in BiNGO (http://www.cytoscape.org/). Significantly enriched GO terms were clustered and visualized using the Cytoscape plug-in EnrichmentMap (EM) (http://www.cytoscape.org/). As shown in Fig. 3C, significantly enriched GO categories were clustered into nine separated modules. The largest module was the GO category of "metabolic process", which consisted of approximately 60 daughter categories, including "small molecule metabolic process", "glycoside metabolic process", "purine nucleotide process", "flavonoid metabolic process" and "metabolic process". The other main modules contained the GO categories of "response to stimulus", "protein modification", "chromatin organization", "negative regulation of gene expression", "reproductive process", "homeostasis", "cell death" and "immune response". Genome doubling reportedly influences metabolic profiles in autopolyploids. For example, Xing et al. 35 found that the tetraploid Catharanthus roseus produced www.nature.com/scientificreports www.nature.com/scientificreports/ more secondary metabolites, including vindoline, catharanthine and vinblastine, than those of its diploid parents. Importantly, our gene ontology analysis showed that most differentially expressed genes were enriched in "metabolic process". For example, we observed that several key genes involved in the flavonoid pathway, including CHI, F3H, FLS and DFR, were upregulated in the hybrid of B. rapa and R. sativus compared with those in its parents, which is consistent with the metabolome results that many flavonoids were upregulated in the hybrid in comparison with those in its parents. Taken together, Gene Ontology enrichment analysis suggested that the hybridization of B. rapa and R. sativus greatly influenced the metabolic process and activity by altering the expression level of some key genes involved in the metabolic pathways.

Identification of differentially expressed genes involved in phenylpropanoid and flavonoid
pathways. In the process of plant growth and development, the expression level of secondary metabolites is related to special gene expression, protein modification and environmental changes. Therefore, we identified the differentially expressed genes responsible for the synthesis of secondary metabolites by the transcriptome analysis of B. rapa, R. sativus and their hybrid. The results showed that many differentially expressed flavonoids of B. rapa, R. sativus and their hybrid were detected in phenylpropanoid and flavonoid pathways (Fig. 4A,B). Therefore, we further investigated the expression level of the genes that encoded important enzymes involved in phenylpropanoid and flavonoid pathways. As shown in Fig. 4A,C, transcriptome analysis identified 10 genes involved in phenylpropanoid and flavonoid pathways. For example, some key genes in phenylpropanoid and flavonoid pathways, including PAL, C4H, 4CL, CHS, CHI, F3H, FLS and DFR, were significantly upregulated. As an upstream enzyme, C4H forms 4-coumarate to yield coenzyme A (CoA)-thioester by 4CL. 4-Coumaroyl-CoA is the precursor of flavonoid and many other phenylpropanoid compounds 36 . Previous study showed that Eucalyptus Class II C4H is involved in the pathway of stress responses and wood lignin biosynthesis, while Class I C4H is constitutively expressed in any tissue and participates in the synthesis of phenylpropanoid metabolites 37 . In addition, F3H is a 2-oxoglutarate-dependent dioxygenase and forms multi-enzyme complexes with other enzymes involved in flavonoid biosynthesis 38 . For example, the expression level of F3H is regulated by salicylic acid, jasmonic acid www.nature.com/scientificreports www.nature.com/scientificreports/ and abscisic acid. Moreover, the increased expression of F3H leads to the tolerance to abiotic and biotic stress including drought, cold, UV radiation, saline conditions and so on 38 . Particularly, the expression level of F3H was significantly increased over 50-fold in the hybrid compared with that in its parents. As key genes in the process of anthocyanin synthesis, the expression levels of LDOX and UFGT changed slightly among B. rapa, R. sativus and their hybrid.
Interestingly, we also observed some differentially expressed genes involved in photosynthesis in the hybrid in comparison with those of its parents. For example, the quantitative RT-PCR experiment showed that RCA was upregulated in the hybrid (Fig S2A). RCA is usually required for the light activation of the Rubisco enzyme. This result was also supported by the measurement of chlorophyll relative content by SPAD-502. As shown in Fig  S2B, the average chlorophyll relative content of the hybrid, B. rapa and R. sativus was 43.6 ± 2.1, 34.4 ± 3.4 and 45.2 ± 1.5, respectively. Although the hybrid contained almost equal chlorophyll relative content to that of R. sativus, the chlorophyll content in the hybrid was significantly higher than that of B. rapa. In summary, both metabolome and transcriptome analyses showed that the intergeneric hybrid of B. rapa and R. sativus contained more health-promoting metabolites than those of its parents, suggesting the hybrid will be a valuable bioengineering resource with new agronomic traits for the improvement of vegetable varieties in the future.

Materials and Methods
plant materials. Brassica rapa (AA, 2n = 20, Chinese cabbage, pak choi and turnip) and R. sativus (RR, 2n = 18, big root radish) were used as female parent to synthesize the intergeneric amphidiploid (AARR, 2n = 38). Chromosome doubling techniques were performed on the hybrids, and the putative amphidiploid plants with relatively normal fertility were selected for further studies. Professor Jiangsheng Wu of Huazhong Agricultural University (Professor Wu obtained AARR materials through exchange with Soo-Seong Lee of Chung-Ang University) provided the materials. cytological analysis of the intergeneric hybrid. Styles of putative amphidiploid plants were used to determine the chromosome number of the intergeneric hybrids. The materials were treated with 2 mM 8-hydroxyquinoline for 4 hours at 22 °C, fixed in Carnoy for 4 hours and then stored in 70% ethanol at 4 °C before chromosome observation. The anthers were dissected out, cut in half and stained with 10%-modified carbol fuchsin for meiotic analysis 28 . DNA samples of R. sativus (RR, 2n = 18) were extracted for GISH (genomic in situ hybridization) analysis, which were labeled with biotin-11-dUTP (Fermentas, Lithuania) by nick translation. GISH for chromosome number and meiosis analysis were performed according to the methods described previously 39 . Metabolite profiling. Three biological sample sets were extracted as previously described before analysis using an LC-ESIMS/MS system 40 . The freeze-dried leaves of B. rapa, R. sativus and the hybrid were crushed using a mixer mill (MM 400, Retsch) with a zirconia bead for 1.5 min at 30 Hz. One hundred milligrams of the powder was then weighed and extracted with 1.0 ml of 70% aqueous methanol overnight at 4 °C. Following centrifugation at 10,000 g for 10 min, the extracts were absorbed and filtrated for further LC-MS analysis. The treated extracts were analyzed using an LC-ESI-MS/MS system (HPLC, Shim-pack UFLC SHIMADZU CBM30A system, www. shimadzu.com.cn/; MS, Applied Biosystems 6500 Q TRAP, www.appliedbiosystems.com.cn/). Metabolite quantification was carried out using a multiple reaction monitoring (MRM) method as previously described 41 .
RNA-seq library construction and transcriptome sequencing. Total RNA was extracted from the same leaf tissues (in maturity stage) of B. rapa, R. sativus and the hybrid used for metabolite profiling analysis using TRIzol. mRNA was isolated from total RNA using Dynabeads oligo (dT) (Invitrogen). First-and second-strand cDNA were synthesized using Superscript II reverse transcriptase (Invitrogen) and random hexamer primers. Double-stranded cDNA was fragmented by nebulization and used to generate RNA-seq libraries as