Jingwen Li, Yu Wang, Joon Hyuk Suh*

Department of Food Science and Human Nutrition, Citrus Research and Education Center, University of Florida, FL 33850, USA

Keywords:

Multi-omics

Single-omics

Tea polyphenols

Tea breeding and processing

A B S T R A C T

Polyphenols are one of the most important metabolites in tea due to their unique biological activities and health benefits, arousing great attention of researchers to investigate biochemical mechanisms of polyphenols during tea plant growth, development and tea processing.Although omics has been used as a major analytical platform for tea polyphenol research with some proven merits, a single-omics strategy remains a considerable challenge due to the complexity of biological system and functional processes of tea in each stage of tea production.Recent advances in multi-omics approaches and data analysis have enabled mining and mapping of enormous number of datasets at different biological scales from genotypes to phenotypes of living organisms.These new technologies combining genomics, metagenomics, transcriptomics, proteomics and/or metabolomics can pave a new avenue to address fundamental questions regarding polyphenol formation and changes in tea plants and products.Here, we review recent progresses in single- and multi-omics approaches that have been used in the field of tea polyphenol studies.The perspectives on future research and applications for improvement of tea polyphenols as well as current challenges of multi-omics studies for tea polyphenols are also discussed.

1.Introduction

Tea, an infusion of variously processed leaves fromCamellia sinensis, is one of the most popular non-alcoholic beverages all over the world [1].There are hundred types of tea today with a continuous development of tea plants and processing techniques in many countries [2].Traditionally, tea is classified into 6 categories: white tea, yellow tea, green tea, oolong tea, black tea, and dark tea based on their manufacturing processes [3].This classification is widely recognized by the distinct nature of each tea (color, flavor, etc.) as well as the changes in their ingredients during tea processing [4].For example, the production of black tea has a common fermentation process for the complete oxidation of tea catechins; inversely, the production of green tea has a process of destroying enzymes, named enzyme inactivation, to prevent the oxidation of tea catechins; heaping for yellowing of yellow tea, piling of dark tea, fine manipulation of oolong tea, and withering of white tea make these teas have different flavors and distinct nutritional attributes from each other [4].Thus,together with tea plant breeding, tea processing is considered as an important factor to determine tea quality.

With cultural popularity of tea, a wide range of its biological effects make tea more popular [5-7].Many literatures reported that tea provides high nutritional values and health benefits, such as anti-obesity [8], anti-oxidation [5], anti-carcinogenic [9], antiinflammatory effects [10], and so on.These beneficial effects are attributed to bioactive chemicals in tea plants, mainly known as polyphenols [11,12].Hence, tea polyphenols have been regarded as quality indicators as well as functional compounds of tea products [13].The formation of major polyphenols in tea is present in Fig.1.

Fig.1 The formation of major polyphenols in tea plants.PAL, phenylalanine ammonia-lyase; C4H, cinnamate 4-hydroxylase; 4CL, 4-coumarate-CoA ligase;FGS, flavan-3-ol gallate synthase; ECGT, epicatechin 1-O-galloyl-β-D-glucose-O-galloyltransferase.

Due to their high bioactivities and health effects, multiple methodologies have been used to investigate biochemical mechanisms of polyphenols in tea plants [14-21].Among them, metabolomics has been most widely applied in tea polyphenol research.Because metabolomes are directly linked to physiological status of tea, influenced by various internal (e.g., genes) and external(environments, tea processing, etc.) factors, metabolites can be used as the “readout” of the phenotypic traits of tea plants, which could help evaluation of tea grade and quality [22].Thus, at metabolite level, polyphenols have been extensively studied for their types,formation and bioactivities [15,23,24].However, metabolomics alone cannot provide a complete view of biological network related to polyphenols due to complexity of biological system involving numerous players with different scales, including genomes,transcriptomes, proteomes and metabolomes.Therefore, in-depth technologies and strategies are required to integrate these features and uncover biochemical pathways of polyphenol formation and regulation during tea plant growth, development and processing.This effort may provide clues on how to improve beneficial polyphenols in tea plants and products.

Recently, “multi-omics” has received great attention in diverse fields of science [25].Multi-omics approach includes the combination of two or more omics techniques (genomics, transcriptomics,proteomics, and/or metabolomics) and integrated the interpretation of big data from biological processes at levels of genes, transcripts,proteins, and metabolites using various data analytical techniques such as merging, normalization, comparative analysis, and correlations analysis, etc.[26-28].Compared to single-omics, multi-omics approach can step-wisely confirm the associations and interactions of features from genotypes to phenotypes, comprehensively demonstrating the complex biological system in living organisms.Statistical methods such as machine learning algorithms for analysis of high-dimensional datasets have been developed with multiomics, accelerating application of multi-omics in various purposes of studies [29-32].Multi-omics technique is still infant, but it has been increasingly applied in plant science including tea research.Multiomics of tea research, particularly for understanding of the formation and alteration regarding tea polyphenols, would provide valuable information on how tea plants regulate polyphenols, and establish the profound knowledge and guidance for the future tea polyphenol research.All these information can give ideas of tea breeding and processing with desirable levels of polyphenols for health effects and tea quality.

This review focuses on the recent progress in single- and multiomics approaches, and their applications in tea polyphenol research,including theoretical aspects, technical strategies and data analysis.We will also discuss current challenges of omics techniques, and future perspective of multi-omics in the field of tea research.

2.Single-omics study for tea polyphenols

The development of high-throughput gene sequencing technology has accelerated the progress in genomics in life science [33].Along with the development, studying downstream targets of gene sequences termed as functional genomics (transcriptomics, proteomics and metabolomics) has emerged to identify gene functions and related biochemical/molecular mechanisms: transcriptomics for screening gene transcriptions and potential post-transcription [34]; proteomics for elucidating proteins translated from transcriptomes, including their sequences, modifications, structures, and interactions [35]; and metabolomics for monitoring metabolites derived from the activities of proteins, including their levels and dynamic changes [27].All these single-omics techniques have been employed to investigate genetic regulation and molecular and chemical mechanisms of tea polyphenols (Table 1) [36-39].Among them, metabolomics has been most widely used in tea polyphenol research due to the characteristics of metabolomes that directly represent phenotypic status [22].

Table 1Summary of literatures regarding tea polyphenol research using single-omics approach.

2.1 Genomics

Genome sequencing/genotyping data can provide information about genetic sequence and gene variations present in tea plants.The rapid development and innovation of next-generation sequencing(NGS), also known as high-throughput sequencing, have led to a significant progress in genomics for tea research, including deciphering novel genes, investigation of epigenetics associated with various biological processes, identification of tea quality-related genes, and exploration of genetic diversification of tea plants, etc.[40].The formation of tea polyphenols and related genetic regulatory networks have been studied using NGS technologies [41].A complementary DNA library was constructed from leaf tissue of the blister blight-resistant tea cultivar TRI2043.The key genesCsANR1andCsANR2were functionally characterized, which were found to encode the anthocyanidin reductase (ANR) with epimerase activity,converting cyanidin to (+)-epicatechin (EC) or (–)-catechin (C) [42].Single nucleotide polymorphism (SNP) site, a single-base variation at a DNA sequence among individuals, may be associated with the phenotypic trait of tea plants [40].The SNPs were investigated in the chalcone synthase (CHS) genes to identify their relationship with the total content of polyphenols inC.sinensis[43].Fifty-seven tea cultivars were screened to determine the SNPs of CsCHS using polymerase chain reaction (PCR) with genomic DNA(gDNA) and cDNA of tea leaves as templates.Two and four SNPs potentially related to the content of polyphenols were found to be positioned inCsCHS1andCsCHS2respectively, accounting for distinct mechanisms of polyphenol formation in the different tea plants.These genomics results give researchers ideas about the molecular design to breed new tea cultivars with improving tea polyphenols and nutritional values.However, NGS-based genomics has some technical limitations.For instance, the average assemblies forC.sinensisvar.sinensisandC.sinensisvar.assamicawith the scaffold N50 sizes are still limited by the short read length generated from NGS technology [40].Therefore, additional efforts will be needed to improve the quality and completeness of the genome assemblies of tea plants.Moreover, genome information can only represent one level of the complex biological system relating final products.Molecular interaction between transcripts, enzyme reaction,metabolites as well as phenotypic traits related to polyphenols remain to be elucidated [44].In addition, genomes could be co-regulated at the transcription/protein/metabolite level under external or internal stimuli.How genes are triggered and expressed, and how SNPs in genes affect the formation of phenols need to be further investigated.These complex features can be uncovered viacombination with other omics methodologies, such as transcriptomics, proteomics, and metabolomics.

2.2 Metagenomics

Metagenomics has been used to sequence genomes in microorganisms under a particular environment [45].Traditionally,the pure cultures have been used to culture microbiome in the laboratory, whereas these methods do not reflect the natural environmental conditions where microorganisms are grown in the real world.With sequencing uncultured microbes directly obtained from their original habitats, metagenomics has become a powerful tool to identify the relationship between microbes communities and habitats at the genomic level [46].The procedure of metagenomics approach includes the collection of environmental samples, extraction of metagenomic DNA, construction of metagenomic libraries,library screening/editing, and library screening for targeted or novel genes [47].16S rRNA gene sequencing and whole-genome shotgun sequencing have been used for the metagenomics approach [48-50].For the research on tea polyphenols, metagenomics has been mainly applied in tea fermentation process.The taxonomic and functional analysis of the microbial community in fermenting Pu-erh tea was conducted using pyrosequencing methodology [51].The taxonomic analysis identified three dominant bacterial phyla, Proteobacteria,Actinobacteria, and Firmicutes, and one dominant eukaryotic phylum,Ascomycota.Although direct relationship between metagenomics sequences and polyphenol biosynthesis was not concluded, 0.36% of functional genes presented in Pu-erh tea microbiome were categorized to be related to the secondary metabolite pathways which included polyphenol metabolism (e.g., flavonoids).Recent evidence also reveals microbiomes includingPenicllium, Rhizopus,Saccharomyces,Bacteriumand dominantAspergillusspp.(e.g.,A.niger,A.glaucus,A.terreus,A.candidus) have an important role in developing tea characteristics, such as taste, aroma and color, with modifying the type and level of metabolites (e.g., polyphenols), in Pu-erh tea [52].However, metagenomics alone is unable to completely reflect the interaction between metabolites, proteins, and microbiomes during the fermentation.The mechanism of the fermentation microbiome,especially dominantAspergillusspp., working on polyphenols metabolism requires additional explanation at protein or metabolic levels.For example, the changes in microbial composition, enzyme activities, polyphenol levels and their correlations in the pilefermentation process were investigated in primary dark teas (e.g.,Fu brick tea, Hei brick tea, Hua brick tea, and Qianliang tea) [37].During the pile-fermentations, along with the growth of microbial communities, such asAurantimonas,Klebsiella,Methylobacterium,andAspergillus, the activities of polyphenol oxidase (PPO) were increased, leading to the oxidation of catechins and a decrease in the total content of polyphenols.With determination of these extra features at enzyme (e.g., PPO) and metabolite (e.g., polyphenols)levels in fermented tea, the polyphenol metabolism could be finally understood.Hence, the combination of metagenomics with other omics, such as proteomics and metabolomics, would help investigate the detailed metabolic mechanisms of tea polyphenols and the interaction between microbes and metabolites in the environment.

2.3 Transcriptomics

Like the genomics, transcriptome sequencing (RNA-Seq) is based on NGS techniques to decode messenger RNA (mRNA), micro RNA (miRNA), and long non-coding RNA (lncRNA) in cells and/or tissues [34,53].The mRNA, miRNA, or lncRNA are reversely transcribed to establish a cDNA library, depending on which one is the target for RNA sequencing [34].RNA-Seq-based transcriptomics provides information on genetic pathways, function annotation, gene expression qualification, differentially expressed genes (DEGs), gene regulatory network reconstruction as well as gene module and motif analysis, etc.[54-56].Transcriptomics has been used to elucidate genetic regulatory mechanisms of polyphenols in tea plants [44,57]under different biotic [58]and abiotic stresses [20,59].With RNASeq analysis, the expression of genes linked to tea polyphenols under drought stress was demonstrated [38].DEGs related to secondary metabolism were found, and some of them were identified as key genes involved in the biosynthesis of catechins and other metabolites (e.g., caffeine, theanine).The transcriptional levels of CHS, ANR, dihydro flavonol 4-reductase (DFR), leucoanthocyanidin reductase (LAR), and anthocyanidin synthase (ANS) tended to decrease and subsequently increase in response to drought stress,which is consistent with the changes in epicatechin gallate (ECG),epigallocatechin gallate (EGCG), and gallocatechin gallate (GCG)contents.These results provided transcriptional evidence for genecontrolled polyphenol mechanisms in tea plants under abiotic stress.Using the transcriptomics, the polyphenol metabolisms in different tea cultivars has been studied as well [60-63].DEGs involved in catechins biosynthesis were identified in two different tea plants,C.sinensis(tea) andC.oleifera(oil tea) based on RNASeq analysis [64].The level of DEGs was confirmed by quantitative reverse transcription PCR (qPCR) analysis, revealing there is a considerable difference in the gene regulation underlying polyphenol formation between the two species (tea and oil tea).The expression levels of key regulatory genes of catechins, caffeine, and theanine biosynthesis pathways were consistent with their content in each tea.Transcriptomics has been also used to explore the expression of genes during tea processing [65].LncRNAs were found to play an important role in changes in secondary metabolites during the withering process of oolong tea leaves [66].Different withering methods (solarwithering and indoor-withering) led to variations in the expression of IncRNAs.The expression of 7 up-regulated (LTCONS_00056216,LTCONS_00044497,LTCONS_00031811,LTCONS_00001863,LTCONS_00090121,LTCONS_00030131, andLTCONS_00101116)and 3 down-regulated (LTCONS_00054003,LTCONS_00060939,andLTCONS_00000233) lncRNAs and their targets were proven to be linked to the low content of the total polyphenols, flavonoids and catechins (C, catechin gallate (CG), epigallocatechin (EGC),GCG, ECG, and EGCG) in the solar-withering process.The result indicated the withering process possibly modifies the type and level of polyphenols in tea plants through altered epigenetic regulatory mechanisms.Transcriptomics, especially based on RNA-Seq, has become a common functional genomics tool in plant science,with a merit of generating large-scale gene expression data at low cost.It can help understand molecular features between genes and proteins, which are connected to the metabolism of tea polyphenols.Nonetheless, RNA-Seq-based omics has intrinsic limitations existing in the NGS technology, such as difficulties to analyze large genetic structural changes in sequences [67].In addition, extra attention is needed for advanced library construction to minimize analytical biases (e.g., adapter removal, alignment)and processing errors from complex RNA-Seq data [68].Importantly, with information of the downstream targets (e.g.,proteins), transcriptomics could work better to elucidate regulation and formation of polyphenols in tea plants and products.

2.4 Proteomics

Proteomics generally deals with large-scale exploration of proteins including their levels, compositions, structures and activities,contributing to the understanding of gene and cellular functions [69].The application of proteomics based on mass spectrometry (MS) in tea science has increased as a result of the development of ionization techniques (e.g., matrix-assisted laser desorption ionization (MALDI),electrospray ionization (EI)) and quantification methods (e.g., isobaric tags for relative and absolute quantification (iTRAQ)).Protein profiles between winter and spring tender shoots of a tea cultivar were analyzed using MALDI-time of flight mass spectrometry (MALDITOF-MS) [70].The differently expressed proteins (DEPs) were determined to link metabolism of polyphenolic flavonoids in different seasons, demonstrating molecular mechanisms of seasonal changes in flavonoids.In another study, dynamic changes in proteomes of postharvest tea leaves during withering stages were investigated by using iTRAQ labeling coupled with MS analysis [39].There were 863 unique DEPs identified in different withering stages.Among them, 5 DEPs were proven to be involved in the biosynthetic pathway of catechins, indicating that the level of catechins can be altered in tea leaves by these enzymes during a withering process.Proteomics provides a high opportunity to discover alterations in polyphenols in tea plants and products under certain cellular conditions because the level and composition of proteins are more dynamic compared to genomes, providing more direct information on structural and functional framework of cellular life (polyphenolassociated metabolism, etc.) [71].However, there are some inherent challenges of proteomics for the study of tea polyphenols.Proteins have secondary and tertiary structure, which can be easily denatured by various factors, such as enzymes, heat, and light.Besides, proteins cannot be amplified like DNA, showing relatively low sensitivity,requiring specific and expensive instruments (e.g., nano-liquid chromatography) [71].Lastly, like other single-omics techniques,proteomics itself cannot represent the whole genetic and chemical mechanisms regarding tea polyphenols.For instance, Li et al.[72]identified polyphenol biosynthesis-related DEPs between young leaves and buds of tea plants using iTRAQ techniques, but the protein expressions were not linearly correlated with the levels of the corresponding transcripts, implying the translations from transcript to polyphenol-relevant proteins may be regulated by other molecular factors.Thus, the joint proteomics and other omics approach would be necessary for in-depth studying of biochemical mechanisms of polyphenols in tea.

2.5 Metabolomics

Metabolomics aims at qualitative and quantitative analysis of a wide range of metabolites in biological samples [73].The information of metabolites can be the bridge between genotypes and phenotypes to interpret physiological status in organisms.There are two instrumental methods majorly coupled with metabolomics: nuclear magnetic resonance (NMR) and MS.NMR is useful for structure elucidation and study of metabolite classes with non-invasive and unbiased properties [74-76].MS is feasible for quantification and tentative identification of metabolites, with high-throughput and highly sensitive features [75].Metabolomics has been extensively used in tea polyphenol research regarding tea plant growth under different environmental conditions, such as light [77], shade effects [16,78],zinc stress [79], and season changes [80].For instance, using UPLCQ-TOF MS, the metabolites in green tea were investigated under different shade periods (0, 15, 18, and 20 days) to see the effect of shade on the content of polyphenols [16].The increasing shade was found to enhance the levels of quercetin-galactosylrutinoside(QGalR), kaempferol-glucosylrutinoside (KGluR), ECG, and EGCG,while decreasing quercetin-glucosylrutinoside (QGluR), kaempferolglucoside (KG), gallocatechin (GC), and EGC levels, suggesting that the shade treatment could modify the level and type of polyphenols in tea plants.Metabolomics has been employed for studying polyphenols during tea processing [81-85], such as the effect of high-temperature roasting on polyphenol formation in yellow tea [15].In the study,using MS-based untargeted metabolomics approach,N-ethyl-2-pyrrolidinone-substituted flavan-3-ols (puerins) was determined as a marker significantly affected by the period and level of temperatures during the roasting procedure.Subsequently, targeted metabolomics for the precursors of puerins, flavan-3-ols (EGC, EC, ECG, GC,EGCG, and GCG), was performed: the levels of EGC, EC, ECG and EGCG dramatically decreased, whereas epimerized catechins GC and GCG were increased.The integrated results showed that the degradation and epimerization of catechins were dependent on temperature during roasting of yellow tea.Another study discussed puerins and flavan-3-ols in pile-fermentation of Pu-erh tea using metabolomics as well [86].The untargeted metabolomics analysis identified 49 markers related to the pile-fermentation, revealing flavan-3-ol class decreases after long-term pile-fermentation.The targeted metabolomics analysis further confirmed that puerins significantly increase in the first stage by bioconversion of flavan-3-ols during the fermentation.Metabolomics has been also used to discriminate tea cultivars by screening tea metabolites [87,88].Metabolomics was performed on thirteen Chinese tea cultivars to study their manufacturing suitability [89].Based on the metabolite data, the tea cultivars could be clustered into two groups, GT and G&BT.The G&BT cultivars possessing higher levels of specific polyphenols (e.g., catechins, dimeric catechins) were shown to be suitable for manufacturing of both green tea and black tea, whereas the GT cultivars having more flavonoid glycosides were only preferable for green tea production.The above examples indicate that metabolomics is useful to assess alterations in the type and content of polyphenols, induced by internal and external factors of tea plants,and some tea metabolites, such as quality-related polyphenols, may be used as indicators of breeding programs for tea plants.However,genetic features underlying these metabolic changes cannot be identified by metabolomics, and need to be further investigated at upper biological levels (e.g., gene level).Indeed, genetic mechanisms of major polyphenols in tea are largely unknown [22,44].On top of that, there are some technical limitations of metabolomics for the tea polyphenol research.For example, compared to the human database, plant metabolome database is relatively small, which needs to be improved with additional efforts in identification of plant metabolites and their functions.Therefore, other omics technologies may compensate these limitations, and help elucidate comprehensive metabolic mechanisms of tea polyphenols linked to complicated features at different biological levels.

3.Multi-omics approach for tea polyphenols

The progress in omics techniques has revealed molecules do not possess biological functions independently, but their functions are linked to complex and delicate regulatory networks [90].As emphatically mentioned above, integration of different omics technologies termed “multi-omics” would be a powerful tool to uncover these sophisticated biological systems, with identification of in-depth functions and mechanisms regarding molecules of interest at different scales (genomes, transcriptomes,proteomes and metabolomes).The multi-omics approach has been increasingly employed to discover genetic regulations and biosynthetic pathways of polyphenols in tea plants and products [91,92].In this review, multi-omics that have been used in tea polyphenol research are discussed, except some approaches(e.g., joint metabolomics and genomics approach) that have been seldom reported (Table 2).

Table 2Summary of literatures regarding tea polyphenol research using multi-omics approach.

3.1 Integration of metabolomics and proteomics

The integrated technique of metabolomics and proteomics has been majorly used to identify molecular functions and metabolic pathways of polyphenols in tea processing [93].Chen et al.[94]combined metabolomics with proteomics to explore changes in nonvolatile bioactive compounds during white tea processing, demonstrating the relationship between DEPs involved in biosynthesis of phenylpropanoids and flavonoids, and the oxidation of polyphenolic compounds in a withering process.It was proven that the down-regulation of protein expressions,such as flavonoid 3’,5’-hydroxylase (F3’5’H), ANR and flavonol synthase (FLS), by the withering could accelerate polyphenol oxidation and lower their biosynthetic activities, resulting in a decrease inflavonoids and related metabolites.Novel functions ofAspergillus, a dominant fungi in the post-fermentation of Pu-erh tea, were elucidated using metabolomics coupled with proteomics as well [95].In the fermentation process, specific enzymes (e.g.,laccase, vanillyl-alcohol oxidase and benzoquinone reductase)were produced byAspergillus, changing the composition of polyphenols that affect taste (decrease in astringent taste, increase in mellow taste, etc.) and bioactivity of Pu-erh tea.

3.2 Integration of metabolomics and transcriptomics

The combination of metabolomics and transcriptomics has been used to explore anthocyanin accumulation patterns in white and pink tea flowers during their development stages [96].The metabolomic analysis revealed a specific anthocyanin class (cyanidinO-syringic acid, petunidin 3-O-glucoside, and pelargonidin 3-O-β-D-glucoside)was accumulated only in the pink tea flowers in their development.The subsequent transcriptomic analysis confirmed the accumulation of anthocyanins was governed by specific DEGs, such as FLS gene and dihydroflavonol-4-reductase gene, between two tea flowers.The regulation of catechins in tea seedlings was studied using metabolomics coupled with transcriptomics at different growth stages [57].The catechins were found to be accumulated in leaves >stems > roots irrespective of growth stages, and three transcription factors homologous to ANL2, WRKY44, and AtMYB113 were speculated to play key roles in the regulation of catechins in different parts of plants.The combined metabolomics and transcriptomics approach was also applied to the investigation of effects ofin vitrosucrose on polyphenol biosynthesis in tea plants [97].A treatment of sucrose led to an increase in the polyphenol contents, with upregulation of eleven key structural genes, such as CHS, ANR,flavanone 3-hydroxylase (F3H), and flavonoid 3’-hydroxylase(F3’H) involved in polyphenol biosynthetic pathways.The result implied the role of sucrose in polyphenol formation during tea plant growth and development.For tea processing, one study was performed to identify the genetic and chemical control of polyphenols in leaves during black tea fermentation [98].The joint metabolomics and transcriptomics approach identified mechanisms on how theaflavins (TFs), the major polyphenol in black tea, are produced from parent catechin molecules during the fermentation.This reaction is catalyzed by enzymes including PPO and peroxidase (POD), and the expressions of related genes were monitored.Catechin levels were significantly reduced with the expression ofCsAPX1,CsPOD18andCsPODthat are thought to promote synthesizing TFs from catechins through the related enzyme activities (e.g., PPO and POD) during the fermentation of black tea.In another study, using metabolomics combined with transcriptomics, alterations in metabolites and regulatory gene expressions in green tea leaves were investigated under different conditions of a withering process [99].The withering at low temperature was shown to delay the oxidation of polyphenols by upregulation of dihydroflavanol-4-reductase genes, indicating that a withering process can modify the level of polyphenols in green tea.

3.3 Integration of metabolomics and metagenomics

Metagenomics is useful to analyze bacterial/fungal diversity,succession and association during fermentation procedure, revealing the role of the microbiome in tea processing [17,18].The integration of metabolomics and metagenomics has been used to study the relationship between microorganism activities and metabolite changes under diverse fermentation conditions.Fungal community succession and metabolite alterations during the manufacturing process of Fu brick tea were analyzed using metagenomics and metabolomics [100].The metabolite profiling showed that metabolites including sugars,flavonoids, organic acids, amino acids, caffeine, C, EGCG, ECG,GCG, EGC and EC were decreased during the process.The generaAspergillus,Cyberlindnera, andCandidawere predominant at the early stage of the manufacturing process, but onlyAspergilluswas dominated at the end of the process.The level ofAspergillusgenus was negatively correlated with catechin contents (e.g.,EGCG, GCG and ECG), indicating fungal communities and these polyphenols might be mutually influenced in the process of Fu brick tea.The influence of the microbial community on the metabolism of polyphenols during solid-state fermentation was investigated in Pu-erh tea using the combination of metagenomics and metabolomics [101].Aspergillusand some other genera (Bacillus,Rasamsonia,LichtheimiaandDebaryomyces) were found to affect the formation of flavor compounds such as theabrownins (TBs)derived from polyphenols at the different stages of the solidstate fermentation.These studies demonstrate the relationship between microbiomes and polyphenols, implying that polyphenol production can be adjusted by altered microbiome composition during tea processing.

3.4 Integration of proteomics and transcriptomics

Theoretically, proteomes and transcriptomes are closely related together, and their levels have a high correlation at the same tissue under similar conditions.However, in reality, many literatures have reported that the level of proteomes are not linearly matched with that of transcriptomes due to a variety of factors (e.g., posttranscriptional regulation) [102,103].Thus, the joint proteomics and transcriptomics approach can provide comprehensive information on biochemical mechanisms of polyphenol regulation and synthesis in tea, which may not be revealed by single-omics approach (proteomics or transcriptomics).Using proteomic and transcriptomic analyses,DEPs and DEGs were investigated between tender purple leaves (TPL)and mature green leaves (MGL) of tea plants, respectively, during leaf development [104].The DEPs related to anthocyanin formation,such as FLS, CHS and chalcone isomerase (CHI), were found to be regulated at the transcriptional level, and these enzymes were accumulated in TPL samples, indicating why TPL displays purple color in its development stage.Physiological differences between winter and spring tender tea shoots were studied using proteomics coupled with transcriptomics [105].The result showed expressions of shikimate dehydrogenase (SDH), dehydrogenation quinic acid (DHQ)and FLS were significantly decreased in the winter tender shoots,leading to the lower content of tea polyphenols.This work also discovered some expression discrepancies between DEPs and DEGs(e.g., eukaryoticgalactinol synthase, monodehydroascorbate reductase and fructokinase), demonstrating these proteins are differentially regulated at the transcriptional level (non-linear correlation).The accumulation pattern of phenolic compounds was studied in different development stages of tea leaves at proteome and transcriptome levels [106].During leaf development, nongalloylated catechins (EC and EGC) are increased by hydrolysis of galloylated catechins (ECG and EGCG).This phenomenon was proven by the higher activity of galloylated catechins hydrolase (GCH) and the lower activity of galloyl-1-O-β-D-glucosyltransfera (UGGT), epicatechins 1-O-galloylβ-D-glucoseO-galloyltransferase (EGGT).The enzyme activities were further confirmed with the expression levels of related genes(CsUGT75E2,CsUGT75E3,CsSCPL1,CsSCPL3, etc.), indicating the association between the enzymes and genes and their roles in the tea leaf development.Dynamic changes in catechins at postharvest leaf samples during tea processing (withering temperature) were investigated using the combination of proteomics and transcriptomics approach [107].DEPs (CHI, F3H and ANR) and DEGs (CsCHI,CsF3H, and CsANR) linked to catechin formation were observed under different withering temperatures, and were verified with catechin contents (e.g., GC, EGC and EC) in these conditions.Interestingly, this study again showed some DEGs not linearly correlated with the corresponding DEPs, indicating there are other factors affecting the expressions of proteins.

3.5 Integration of proteomics with metagenomics

Integration of proteomics and metagenomics has been used for studying fermentation processes in tea polyphenol research.With identification of microorganisms participating in tea fermentation,proteomics gives useful information on microbiome-derived enzyme reactions and alterations related to polyphenol formation.Microbial communities and proteins linked to polyphenols in solid-state fermentation of Pu-erh tea were studied by the joint metaproteomics and metagenomics approach [108].Integrated data indicated thatAspergillusis the primary fungus of the fermentation and is the major host of identified proteins including microbial extracellular enzymes, such as catalase (Q877A8), catalase-peroxidase (A2Q7T1)and peroxiredoxin (Q5ASN8).The proteins had the function of polyphenol oxidation with decreased amount of catechins (e.g.,EGC, EC, EGCG, GG and ECG).The relationship between polyphenol-associated proteins and microbiomes in fermentation of Pu-erh tea was investigated using metaproteomics combined with metabarcoding [109].Aspergilluswas dominant at all stages of fermentation, while the generaRasamsoniaandThermomyceswere only dominant at the intermediate stage.The microbiomederived enzymes (e.g., glycoside hydrolases, glycosyltransferases,tannase, catecholO-methyltransferase, phenol 2-monooxygenase,catechol 2,3-dioxygenases, catechol 1,2-dioxygenase and quercetin 2,3-dioxygenase) revealed roles in hydrolysis, oxidation, modification and degradation of polyphenols (C, EGC, ECG, EGCG, etc.).

3.6 Integration of transcriptomics and genomics

Using transcriptomics coupled with genomics, genes related to the biosynthetic pathway of flavonoids (CsF3’H, CHS, CHI and F3H) inC.sinensiswere cloned and functionally characterized [36].The gene expression levels were investigated by real-time quantitative PCR (RT-qPCR) analysis, indicating that the expression of all genes increases to a maximum at stage 3 among the four stages of tea seedling.The levels of flavan-3-ols (e.g., C, EGCG and GCG)reached a maximum at the same stage (stage 3), demonstrating the role of these genes in biosynthesis of flavan-3-ols.Genes (CsCHSs,CsF3′H,CsANRs,CsLARsandCsPALa) involved in phenylpropanoid and flavonoid pathways inC.sinensis(L.) O.Kuntze cv.‘Shuchazao’were also screened and validated using integrated genomics and transcriptomics [110].A correlation analysis between gene expressions and catechin contents showed that the expressions ofCsCHSs,CsF3’H,CsANRsandCsLARsare closely associated with the level of EC, and the expression of CsPALa is correlated with EGCG and proanthocyanidin (PA).

3.7 Data analysis for multi-omics study

Although a multi-omics approach is very useful for tea polyphenol research, data analysis is a challenge since it requires the combination of high-dimensional data from different biological layers, followed by identification of coherent signatures to achieve a complete view of biological mechanisms.However, among a huge number of observations, only a few are commonly revealed as meaningful features related to target subjects (e.g., polyphenol metabolism).Thus,statistical methods for narrowing down such features are necessary for the analysis of multi-omics data, having a high dimension and a relatively small sample size [29,111].

Various statistical methods have been used to integrate and interpret multi-omics datasets [32].They are mainly based on machine learning algorithms, which can automatically learn to recognize complicated patterns using empirical data to make predictions or decisions for unknown future events [111].Linear regression is one of the traditional machine learning models to predict relationship between independent and dependent variables by fitting a linear equation to observed data.It has been applied to some single-omics studies in different fields [112-114].However, the linear regression may not be suitable for the multi-omics approaches with a large number of variables, usually showing non-linear associations.Hence,multivariate analysis such as partial least squares discriminant analysis(PLS-DA) and partial least square regression (PLSR) has been used for multi-omics studies to integrate multi-layered features and reduce redundant information to construct linear models [115-117].Unsupervised methods including hierarchical clustering analysis(HCA) and principal component analysis (PCA) have also been employed to discover unknown patterns in multi-omics datasets without output variables and previous training data by analyzing correlation and covariance between quantitative variables [118].In the multi-omics study for tea polyphenols, metabolite data was analyzed by PCA and PLS-DA, demonstrating the metabolite pattern of fresh tea leaves was clearly different from that of withering tea leaves based on 132 differentially expressed metabolites having reliable variable importance in projection (VIP) values [99].With investigation of these metabolites with related gene expressions,the effect of withering temperature on polyphenol metabolism was identified in green tea.In another study [106], HCA was used for gene expression data to see the relationship among groups of different development stages in tea plants.The HCA successfully categorized the development stages based on DEGs.The result was further correlated with proteomes and metabolomes to confirm in-depth biochemical mechanisms underlying tea plant development.

Recently, advanced machine learning-based algorithms, such as support vector machines (SVM), random forests (RF), and artificial neural networks (ANN), have started being applied in multi-omics studies (clinical study, etc.) [31,119,120].These machine learning models may be more suitable for multi-omics study since they can deal with more sophisticated factors and variables at different biological scales, elucidating their linkage and significance with better performance and prediction accuracy.They also show good fit for biological data having non-linear relationship [119,120].However,despite their usefulness, these methodologies have not been applied to multi-omics data in tea polyphenol research yet.Zhu et al.[121]made an attempt to use the above methods (SVM, RF and multilayer perception as ANN) to establish an accurate prediction model for the fermentation degree of black tea, but the study did not involve multi-omics dataset.Therefore, the application of advanced machine learning algorithms in multi-omics studies related to tea polyphenols would help elucidate biological functions and mechanisms of polyphenols in tea accurately and clearly.

4.Conclusions and perspectives

In this review, we discussed the recent progress in single- and multi-omics approaches, and their applications in tea polyphenol studies.The given research examples demonstrate multi-omics has a superior capacity to uncover complex biological mechanisms of polyphenol formation and regulation in tea plants by dissecting and integrating of observed features at different biological scales (genes,transcripts, proteins and metabolites).The knowledge and information obtained from these studies can be utilized to adjust polyphenol levels during different stages of tea production, which might be beneficial for tea producers and industry (Fig.2).

Fig.2 Application of multi-omics technologies in tea polyphenol research from tea plant breeding to tea processing.

Although multi-omics is a powerful tool for tea polyphenol research, there are some challenges and limitations of this technique,necessary to address in the future.Multi-omics has not been studied well in plant science, especially in tea research.To date, only a few studies have been conducted for tea polyphenols using multiomics techniques.In addition, some important factors involved in tea polyphenol synthesis have been ignored and unreported.For instance,the association between soil microbiomes and tea polyphenols is currently unclear, although the soil microbiomes influence the type and level of nutrients by constant interaction with plants and the environment.Indeed beneficial polyphenols were found to be increased with shaping microbial composition by treatment of organic fertilizers [122].Similarly, the relationship of tea storage conditions and tea polyphenols remains to be elucidated as well [123].Analysis of these features (soil microbiomes, storage conditions,etc.) using multi-omics would add another layer of knowledge to better understand polyphenol formation and alteration in tea plants.Meanwhile, as discussed above, there are critical challenges in data analysis for multi-omics approaches.The advanced machine learning algorithms may improve this issue, building accurate statistical models to predict biochemical mechanisms of tea polyphenols.Data integration tools for microbiome data and other forms of data from newly emerging techniques (e.g., imaging technologies) also need to be developed [30,32,124].

Nonetheless, there have been continuous efforts to enhance multi-omics platforms, which can narrow the gap between current knowledge and as yet unknown, expanded knowledge for tea polyphenol research.Each omics technique has been improved to make up for its own drawback.The third-generation sequencing has been applied in solving the read-length and bias problems of NGS, and even the fourth-generation sequencing has been initiated to enable genome analysis directly (in situ) in the cell and tissue [125].MS technologies have been continuously developed with enhanced sensitivity and resolution, resulting in better performance of metabolome and proteome profiling [126-128].In addition, bioinformatics and statistical data interpretation methods have been updated to help annotate gene functions and identify metabolites [129-131].For example, a computational tool for systemic classification of metabolites, which can accurately predict compound class (e.g., polyphenol class) of unknown metabolites,has been introduced based on machine learning algorithms (deep neural network) [132].With further efforts in improving and adding these techniques and methodologies, multi-omics approaches offer comprehensive understanding of complex biological functions and mechanisms behind polyphenol regulation and formation in tea plants and products.Such knowledge and information may give clues on developing tea cultivars as well as modifying tea processing methods with desirable polyphenol profiles.They would be a milestone for sustainable tea industry and optimized production practices of tea with good quality and health benefits.

Conflicts of interests

The authors declare that they have no competing interests.