Differential metabolites and their transcriptional regulation in seven major tea cultivars (Camellia sinensis) in China

GAO Ting, HOU Bing-hao, SHAO Shu-xian, XU Meng-ting, ZHENG Yu-cheng, JlN Shan,WANG Peng-jie, YE Nai-xing#

1 College of Horticulture, Fujian Agriculture and Forestry University/Key Laboratory of Tea Science in Universities of Fujian Province, Fuzhou 350002, P.R.China

2 College of Horticulture, Northwest A&F University, Shaanxi 712100, P.R.China

Abstract Various genetic and biochemical characteristics exist in tea plant cultivars, and they largely determine production suitability and tea quality.Here, we performed transcriptomic and metabolomic analyses of young shoots of seven tea cultivars and identified major regulatory transcription factors (TFs) for the characteristic metabolites in different cultivars based on weighted gene co-expression network analysis (WGCNA).Phenotypically, we found that ‘Tieguanyin’ (TGY)and ‘Fujian Shuixian’ (FJSX), which are suitable for oolong tea, had higher catechin contents.The metabolites of ‘Jinxuan’(JX) were more prominent, especially the contents of phenolic acids, flavonoids, terpenes, and tannins, which were higher than those of the other six cultivars.Moreover, ‘Fudingdabai’ (FDDB), which is suitable for white tea, was rich in amino acids, linolenic acid, and saccharides.At the molecular level, hydroxycinnamoyl CoA quinate hydroxycinnamoyl transferase (HCT) (CsTGY12G0001876, and CsTGY06G0003042) led to the accumulation of chlorogenic acid in TGY.The main reason for the higher l-ascorbic acid content in FJSX was the high expression levels of L-galactono-1,4-lactone hydrogenase (GalLDH) (CsTGY13G0000389) and Myo-inositol oxygenase (MIOX) (CsTGY14G0001769, and CsTGY14G0001770), which were regulated by WRKY (CsTGY11G0001197).Furthermore, FDDB, ‘Longjing 43’ (LJ43),‘Shuchazao’ (SCZ) and ‘Baihaozao’ (BHZ) had higher free fatty acid contents, among which MYB (CsTGY14G0002344)may be a hub gene for the regulation of palmitoleic acid accumulation.More importantly, we found that the shoots of TGY were green with purple, mainly due to the accumulation of anthocyanins and the downregulation of the Mgprotoporphyrin IX nonomethyl ester cyclase (MPEC) (CsTGY10G0001989) gene that affects chlorophyll synthesis.These results will provide a theoretical reference for tea cultivar breeding and suitability.

Keywords: Camellia sinensis, transcriptomics, metabolomics, WGCNA

1.lntroduction

Tea is a popular drink worldwide due to its classic flavors and known health benefits that is usually made from the young shoots of tea plants (Camelliasinensis(L.) O.Kuntze) (Wang P Jet al.2021b).The chemical composition of tea is highly dependent on the cultivars,environmental factors and manufacturing practices(Fanget al.2021).Tea plant cultivars have different genetic and biochemical characteristics, and the cultivar characteristics determine their suitability as tea species and quality to a large extent.Various metabolites,including amino acids, flavonoids, lipids, and terpenoids,vary significantly in the fresh leaves of different tea cultivars (Maritimet al.2021).Differential abundances of metabolites such as amino acids, phenylpropanes,flavonoids, and terpenoids are significantly different between the cultivars that are suitable for green, white,oolong, and black tea (Chenet al.2022).Nearly 4 000 bioactive compounds have been identified in tea (Linet al.2012), of which catechins, purine alkaloids, and free amino acids are the most important compounds.Free amino acids play a key role in tea aroma and freshness.The astringency mainly originates from flavonoids, with caffeine as the main source of bitterness.Fatty acids are well-known aroma precursors in tea and vary among cultivars, seasons, and growth regions (Ramaswamy and Ramaswamy 2000; Govindasamyet al.2011).

Green tea is a non-fermented tea dominated by amino acids and soluble sugars, and there is a higher content of flavonol glycosides in green tea cultivars compared to black tea (Li P Let al.2018; Wang H Jet al.2021).For instance,the representative green tea cultivars Longjing 43 (LJ43,no.GS13007-1987), Baihaozao (BHZ, no.GS13017-1994),and Shuchazao (SCZ, no.GS2002008), which come from Zhejiang, Hunan, and Anhui provinces, respectively, were identified as Chinese national improved cultivars.White tea is a lightly fermented tea popular for its sweetness and outstanding health benefits (Li Xet al.2018).For example,Fudingdabai (FDDB, no.GS13001-1985) is a major cultivar suitable for making white tea, as well as green tea and black tea.It has played an important role in the history of Chinese tea breeding, and it has been used as the paternal or maternal plant in the cultivation of more than 20 modern tea cultivars released at the national or provincial levels(Zhang W Yet al.2021).Oolong tea is a semi-fermented tea known for its charming floral and fruity aromas, and it is mainly produced in Fujian, Guangdong, Taiwan of China,and other provinces (Linet al.2020).Among the cultivars,Tieguanyin (TGY, no.GS13007-1985) tea is a typical cultivar of Chinese oolong tea, which is famous for its unique rich flavor and orchid-like aroma (Chenet al.2013).Jinxuan (JX, no.MS2011002) is an excellent tea plant cultivar with a unique floral and creamy aroma.Moreover,Wuyi Rock tea is renowned for its rich flavor and longlasting fragrance (Maet al.2013), with Fujian Shuixian(FJSX, no.GS13009-1985) as its primary cultivar.

In recent years, widely targeted metabolomics has proven to be a high-precision, wide-coverage metabolite detection method, and its combination with highthroughput transcriptomic technology has been widely used to explore tea flavor-related genes and metabolites(Chen X Jet al.2020; Wanget al.2020b; Wuet al.2020;Fanet al.2021; Zhenget al.2021).In this study, liquid chromatography–electrospray ionization–tandem mass spectrometry (LC–ESI–MS/MS) was used to detect the non-volatile metabolites of seven tea plant cultivars.In addition, we quantified nine catechins, three purine alkaloids, and 21 free amino acids in those seven tea plant cultivars by targeted metabolomics.The weighted gene co-expression network analysis (WGCNA)-based system biological strategy was employed to dissect the transcriptional regulatory network of the metabolic pathways related to the characteristic metabolites of the seven tea cultivars.The results reveal the metabolic and transcriptional regulatory mechanisms of different tea cultivars and provide new insights for mining improved genetic resources of elite tea plants.

2.Materials and methods

2.1.Tea plant materials and chemicals

In April 2021, the young shoots (one bud and two leaves)of the tea plants ‘Tieguanyin’ (TGY), ‘Jinxuan’ (JX), ‘Fujian Shuixian’ (FJSX), ‘Fudingdabai’ (FDDB), ‘Baihaozao’(BHZ), ‘Longjing 43’ (LJ43), and ‘Shuchazao’ (SCZ) were collected from the tea germplasm plantation of Wuyi University (Wuyishan City, Fujian, China; 27°73′17′′N,118°00′18′′E) for detection of the released volatiles and transcriptome analysis.All the tea plants were grown under the same cultivation practices, and three independent biological replicates were set up.The collected samples were immediately frozen with liquid nitrogen and stored in a freezer at –80°C.

The standards of nine catechin components, 21 amino acids, three purine alkaloids, LC grade solvents,methanol, acetonitrile, formic acid, ammonium formate,and the companies where they were purchased refer to the method of Wanget al.(2022).The standards and reagents used in the widely targeted metabolome detection were supported by MetWare (Wuhan, China).

2.2.Analysis of non-volatile metabolites by widely targeted metabolomics

The freeze-dried samples were extracted as previously reported (Linet al.2021), crushed by a grinder, dissolved in a 70% methanol solution, and placed in a refrigerator at 4°C overnight.During this period, the mixture was vortexed six times and the supernatant was aspirated,filtered, and stored in a sample bottle for subsequent analysis.A quality control sample (mix) was inserted into each of the two test samples to monitor the repeatability.

Metabolite profiling was analyzed using an LC–ESI–MS/MS system (UPLC, SHIMADZU Nexera X2; MS,Applied Biosystems 4500 QTRAP).The liquid phase was Agilent SB-C18 chromatographic column (1.8 μm,2.1 mm×100 mm); mobile phase A was ultrapure water (add 0.1% formic acid), mobile phase B was acetonitrile (with 0.1% formic acid); the gradient programs referred to our previous description (Linet al.2021); the other conditions were: flow rate, 0.35 mL min–1; column temperature,40°C; and injection volume, 4 μL.Mass data acquisition was conductedviaelectrospray ionization (ESI), and the operating parameters were set as follows: ion source temperature, 550°C; positive ion spray voltage, 5 500 V,negative ion spray voltage, 4 500 V; ion source gas I was set at 55 psi, gas II was set at 60 psi, the curtain gas was set at 25 psi; and the collision-induced ionization parameter was set to high.

The substances detected in the samples were qualitatively analyzed by mass spectrometry.Metabolites were identified by comparing their fragmentation patterns, retention times, and accuratem/zvalues to the standards in the MetWare Database (MWDB).The data were processed using Analyst 1.6.3 Software.Principal component analysis (PCA) of the identified metabolites was performed using the R package (https://www.r-project.org/).Based on the variable importance in project (VIP)scores obtained by the OPLS-DA model, metabolites with VIP≥1.0 and fold change (FC)≥1.5 or FC≤0.67 were defined as significantly changed metabolites (SCMs).

2.3.Determination of catechins, free amino acids,and purine alkaloids

The samples were pretreated by freeze-drying and then pulverized through a 60-mesh sieve, and the experiment was repeated three times.The content was measured as the fresh weight in fresh tea leaves.The catechin components were determined by HPLC with reference to the method of Wanget al.(2022).Referring to Chenet al.(2021), UPLC–MS/MS was used to determine the purine alkaloid components.By referring to the method reported by Zhouet al.(2019), free amino acids were determined by AQC-derived LC/MS.The data were analyzed for significance using SPSS 26.0 Software (P<0.05).

2.4.Analysis of RNA isolation and transcriptome sequencing data

Total RNA was extracted from the samples using an RNAprep Pure Plant Kit (DP441, TIANGEN, Beijing,China) according to the manufacturer’s protocol.A total of 21 libraries (seven samples with three biological replicates) were sequenced on the Illumina NovaSeq 6000 platform (Shanghai Majorbio Bio-pharm Technology Co., Ltd., China).

Transcriptome data analysis was performed with reference to previous methods (Wang P Jet al.2021a).Briefly, the Fastq files of the raw reads were qualitycontrolled, filtered and trimmed by SeqPrep (https://github.com/jstjohn/SeqPrep) and Sickle (https://github.com/najoshi/sickle) with default parameters.Then, clean reads were separately aligned to the tea plant (TGY) genome(Zhang X Tet al.2021) using TopHat 2 (Kimet al.2013).The gene expression levels of each sample were analyzed using RNA-Seq by Expectation-Maximization (RSEM)(http://deweylab.biostat.wisc.edu/rsem/) software.The gene expression levels were measured using transcripts per million reads (TPM).Read counts based on gene expression level analysis were used for differential gene expression analysis with DESeq2 Software.Genes with |log2FC|≥1 andP-value<0.05 were considered to be differentially expressed genes (DEGs).The DEGs were further subjected to Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analyses using clusterProfiler Software (Yuet al.2012).TBtools Software was used to make a heatmap for visualization of the DEGs(Chen C Jet al.2020).

2.5.Co-expression analysis

WGCNA was performed using the WGCNA R package(Langfelder and Horvath 2008).Genes with TPM>1 and coefficient of variation (cv)>0.1 were used to construct the co-expression network.After filtering, the abundances of 11 222 genes and 20 metabolites were used to build a signed co-expression network by calculating the Pearson correlations.The soft-thresholding power of the correlation network was set to 14, the minimum module size was set to 30, and the minimum height for merging modules was set to 0.5.The module networks were visualized using Cytoscape 3.9.0 Software (Kohlet al.2011).

2.6.Real-time quantitative PCR (qRT-PCR) analysis

To validate the reliability of the transcriptome results, 12 genes were selected for qRT-PCR analysis to examine their different expression profiles.cDNA synthesis and qRT-PCR were performed according to a previously reported method (Wanget al.2019).CsGAPDH(GE651107) was used as a reference control, and the primers of validated genes were designed using Primer3Plus (https://www.primer3plus.com/).The primer information is listed in Appendix A.All samples were analyzed in three biological replicates.The relative expression level of each gene was calculated using the 2–ΔΔCtmethod (Livak and Schmittgen 2001).The histograms were made by GraphPad Prism 8.0.Analysis of variance and significant difference analysis were performed by SPSS 26.0 (P<0.05).

3.Results

3.1.Widely targeted metabolite analysis of the seven tea cultivars

By observing the phenotypes of seven tea cultivars (Fig.1-A),we found that the leaves of TGY were green with purple,while the leaves of the other six tea cultivars were simply green.The PCA results showed that the metabolite profiles of the seven tea cultivars were clearly separated (Fig.1-B), and hierarchical clustering analysis showed that these samples had different metabolic patterns (Fig.1-C).Thus, the PCA and cluster analysis together indicated that the cultivars have distinct metabolite profiles.Furthermore, a total of 1 162 non-volatile metabolites were detected in the seven tea cultivars (Fig.1-D), including 62 nucleotides and derivatives, 87 amino acids and derivatives, 205 phenolic acids, 257 flavonoids, 93 organic acids, 146 lipids, 32 terpenoids, 74 alkaloids, 40 lignans and coumarins, 21 vitamins, 70 saccharides and alcohols, 30 tannins, and 45 other metabolites.

To more clearly assess the fold changes in metabolite levels across the seven cultivars, a heatmap of 14 metabolite categories was generated (Fig.1-E; Appendix B).The heatmap indicated that phenolic acids, terpenoids,flavonoids and tannins accumulated in JX, while the levels of procyanidins and alkaloids were lower, and the contents of free fatty acids and nucleotides in TGY were lower.Amino acids and their derivatives influence the flavor and aroma of tea, and they were significantly accumulated in FJSX.The levels of saccharides and alcohols in FDDB were higher, but flavonoids were lower.Organic acids,which play a key role in determining sourness and fruity taste, accumulated significantly in BHZ.The procyanidins in SCZ were higher, while the levels of organic acids,lignans, and coumarins were lower.Nucleotides and their derivatives are reported to contribute to the umami taste of tea, and they were higher in LJ43.

3.2.Differential metabolite analysis of the seven tea cultivars

In the present study, significantly changed metabolites(SCMs) were identified in six tea cultivars, but no SCMs were identified in BHZ.These SCMs could be generally grouped into 12 categories (Fig.2; Appendix C).A total of 76 SCMs were identified in JX (55 up- and 21 downregulated), 44 SCMs in TGY (11 up- and 33 downregulated), 41 SCMs in FDDB (17 up- and 24 downregulated), 29 SCMs in LJ43 (20 up- and 9 downregulated), 21 SCMs in FJSX (10 up- and 11 downregulated), and 12 SCMs in SCZ (10 up- and 2 downregulated).

Flavonoids are essential determinants of tea quality and flavor.A total of 22 flavonoids were significantly increased in JX, including dihydromyricetin, kaempferol,quercetin glycosides, and kaempferol glycosides.The levels of tricin, phloretin, phloretin-2′-O-glycosides,3-O-methylquercetin, and quercetin glycosides were lower in TGY, and those of myricetin, eriodictyol, and apigenin glycosides were reduced in FJSX.Myricetin and its glycosides, acacetin-7-O-glycosides, acacetin-7-Ogalactoside, and quercetin glycosides were higher in LJ43.Moreover, 24 flavonoids were reduced in FDDB, mainly including quercetin glycosides, isovitexin glycosides, and vitexin glycosides, while the levels of apigenin glycosides and luteolin glycosides were higher.

In addition to flavonoids, the SCMs in JX were also abundant, among which 14 phenolic acids, six terpenoids,and eight tannins were heavily accumulated, mainly gallic acid, methyl gallate, galloyl methyl gallate, ellagic acid,and 3-O-methylellagic acid.Procyanidins had a positive correlation with higher bitterness and an astringent taste, with seven procyanidins and five indole alkaloids reduced in JX, including theaflavin-3-gallate, theaflavin-3′-gallate, procyanidin A6, procyanidin B2, procyanidin B3, procyanidin C1, procyanidin B4, 3-indoleacrylic acid,indole-3-carboxaldehyde, methyl dioxindole-3-acetate,methoxy-indole-acetic acid, and 1-methoxy-indole-3-acetamide.

Chlorogenic acids had obvious differences among the cultivars, among which chlorogenic acid and chlorogenic acid methyl ester were significantly higher in TGY, while isochlorogenic acid was higher in FDDB, and nerochlorogenic acid was higher in LJ43.Moreover, the levels of four nucleotides, two saccharides, and four free fatty acids were lower in TGY,including uridine, xanthosine, adenosine, thymidine,DL-xylose, D-galactose, methyl linolenate, and 9-oxo-12Z-octadecenoic acid.Interestingly, l-ascorbic acid was significantly higher in FJSX.We also found some significantly more abundant free fatty acids, mainly linoleic acid and 17-hydroxylinolenic acid in LJ43, 15(R)-hydroxy linoleic acid,α-linolenic acid andγ-linolenic acid in FDDB,and palmitoleic acid and 10-heptadecenoic acid in SCZ.

Fig.1 Multivariate statistical analysis of the metabolites of seven tea cultivars: Tieguanyin (TGY), Jinxuan (JX), Fujian Shuixian(FJSX), Fudingdabai (FDDB), Baihaozao (BHZ), Longjing 43 (LJ43), and Shuchazao (SCZ).A, phenotypes of one bud and two leaves from seven tea cultivars.B, PCA score plot of metabolites identified in seven tea cultivars.C, clustering heatmap of metabolites from seven tea cultivars.D, pie chart of the numbers of different types of 1 162 metabolites.The values shown for each represent the number of specific types of metabolites and the percentage of total metabolites.E, heatmap of the changes in the 14 metabolite categories of seven tea cultivars.

3.3.Analysis of catechins, purine alkaloids, and free amino acids

As shown in Table 1, we detected and quantified the catechins, purine alkaloids, and free amino acids of the seven tea cultivars.The contents of nine catechins were mainly high in TGY, JX, and FJSX.The contents of C, GC, and EGCG′′3Me in JX were as high as 11.00,15.84, and 22.07 mg g–1, respectively.The EC and EGC contents in FJSX were 18.40 and 35.53 mg g–1,respectively, but the content in JX was significantly lower than the other six cultivars.The ECG and EGCG contents in TGY were 31.07 and 114.85 mg g–1, respectively.The total contents of catechins were significantly higher in both TGY and FJSX.The caffeine content in TGY was 35.85 mg g–1, which was significantly higher than the levels in the other cultivars, and the theobromine contents in JX and TGY were higher, at 6.66 and 4.91 mg g–1,respectively.

Fig.2 Heatmap of the significantly changed metabolites (SCMs) in the different tea cultivars: Tieguanyin (TGY), Jinxuan (JX),Fujian Shuixian (FJSX), Fudingdabai (FDDB), Longjing 43 (LJ43), and Shuchazao (SCZ).The color bars represent the normalized fold change values.The SCMs were strictly filtered by variable importance in project (VIP)≥1, fold change (FC)≥1.5 or FC≤0.67,and were significantly upregulated or downregulated compared to the other six tea cultivars.

Table 1 Contents of catechins, purine alkaloids, and free amino acids in seven tea cultivars

The contents of theanine in LJ43, JX, and FDDB were significantly higher than those in the other cultivars,which were 36.55, 36.13, and 33.52 mg g–1, respectively.Furthermore, the contents of GABA,β-ABA, arginine,alanine, and glutamic acid were significantly higher in FDDB.The threonine and serine contents were significantly higher in LJ43 and FDDB.The aspartic acid,asparagine, and glutamine contents were significantly higher in JX.The content of GABA was significantly lower in TGY, while the content of citrulline was significantly higher.

3.4.Transcriptome characteristics of the seven tea cultivars

To identify differentially expressed genes among the seven tea cultivars, a transcriptome sequencing analysis was conducted (Appendix D).A total of 1 221 621 898 raw reads were obtained, and the average Q20 and Q30 values of those samples were 94.38 and 98.22%,respectively.After filtering for very low-expression transcripts, 42 825 transcripts were finally obtained.All RNA-seq data are publicly available in the BIG Data Center (https://bigd.big.ac.cn/) under project number PRJCA009753.Pearson analysis (Fig.3-A) indicated that there was high repeatability.The PCA of the samples found that PC1 and PC2 of the gene expression data accounted for 19.43 and 17.23% of the total variance of expressed genes, respectively, which clearly distinguished seven groups of samples (Fig.3-B) and was consistent with the metabolite PCA results.

Fig.3 Multivariate statistical analysis of seven tea cultivars: Tieguanyin (TGY), Jinxuan (JX), Fujian Shuixian (FJSX), Fudingdabai(FDDB), Baihaozao (BHZ), Longjing 43 (LJ43), and Shuchazao (SCZ).A, correlation analysis of the average expression levels of 21 samples (n=3).B, PCA score plot of genes identified from seven tea cultivars.

To verify the accuracy of the transcriptome data,12 DEGs, including two l-ascorbate photosynthesisrelated DEGs, one flavonoid biosynthesis-related DEG, two caffeine biosynthesis-related DEGs, two free fatty acid biosynthesis-related DEGs, one chlorophyll biosynthesis-related DEG, and four TFs involved in these pathways, were selected and their expression levels were detected by qRT-PCR.As shown in Fig.4, the relative expression of the 12 DEGs was consistent with the trend of transcriptome sequencing, indicating that the transcriptome sequencing results could be reliable.

3.5.Differential expression of DEGs in the seven tea cultivars

To study the functions of the primary differentially expressed genes among the seven tea cultivars, a total of 15 939 DEGs were subjected to trend analysis(Fig.5-A).The results showed significant differences in the expression levels of genes in the 20 classes in the seven tea cultivars, among which classes 1, 4, 5, 13,15, 18, and 20 (with 1 143, 1 128, 930, 1 260, 1 252, 795,and 972 genes, respectively) had the highest expression levels in LJ43, FJSX, FDDB, JX, TGY, BHZ, and SCZ,respectively (Appendix E).

The DEGs of these seven classes were subjected to KEGG pathway enrichment analysis, and the results showed that only the top 10 metabolic pathways were enriched (Fig.5-B).JX and TGY were significantly enriched in the plant hormone signal transduction,phenylpropanoid biosynthesis, amino sugar, and nucleotide sugar metabolism pathways.Furthermore,TGY was significantly enriched in the flavonoid biosynthesis, flavone, and flavonol biosynthesis pathways.Remarkably, FJSX was also enriched in ascorbate and sucrose metabolism, which may be related to the higher l-ascorbic acid content.

FDDB was enriched in theα-linolenic acid metabolism and fatty acid degradation pathways, which may play a role in free fatty acid synthesis and metabolism in FDDB.Genes in BHZ were mainly significantly enriched in monoterpenoid biosynthesis and zeatin biosynthesis.SCZ was significantly enriched in sesquiterpenoid and triterpenoid biosynthesis, fatty acid elongation, and linoleic acid metabolism.LJ43 was enriched considerably in energy and carbohydrate metabolism, mainly including oxidative phosphorylation, pyruvate metabolism, carbon fixation in photosynthetic organisms, glyoxylate and dicarboxylate metabolism, butanoate metabolism, and starch and sucrose metabolism.

3.6.DEG analysis of the seven tea cultivars

Fig.5 Clustering of co-expressed genes of seven tea cultivars: Tieguanyin (TGY), Jinxuan (JX), Fujian Shuixian (FJSX), Fudingdabai(FDDB), Baihaozao (BHZ), Longjing 43 (LJ43), and Shuchazao (SCZ).A, clustering analysis of 15 939 differentially expressed genes (DEGs), where the DEGs were strictly filtered by |log2FC|≥2 and P-value<0.05 and were significantly differentially expressed among at least two cultivars.The seven red boxes correspond to the class with the highest expression in the seven tea cultivars.B, KEGG pathway enrichment for differential gene expression among seven tea cultivars.The red boxes indicate metabolic pathways associated with the synthesis of metabolites characteristic of tea cultivars.

To determine the mechanisms of up- or downregulation of the differential metabolites in the seven tea cultivars at the molecular level, we identified DEGs in relevant biosynthetic pathways.Normal green leaves are attributed to the synthesis of chlorophyll.Our analysis found that the level ofHemF(CsTGY08G0002487) was downregulated 2.20-fold, and Mg-protoporphyrin IX nonomethyl ester cyclase(MPEC) (CsTGY10G0001989)was downregulated 11.12-fold in TGY, which may inhibit chlorophyll accumulation in TGY (Fig.6-A).In addition,the high concentration of accumulated anthocyanins is the reason for the purple color of some shoots and leaves, and the DEGs highly expressed in TGY that are in the flavonoid pathway may jointly regulate anthocyanin synthesis (Fig.6-B).Thus, anthocyanin accumulation and chlorophyll degradation may account for the green and purple appearance of TGY shoots.

A total of 18 genes were involved in the flavonoid biosynthesis pathway (Fig.6-C), of which 15 DEGs (twoPAL, one4CL, oneCHS, oneCHI, twoF3′5′H, oneDFR, oneANS, oneLAR, oneSCPL1A, and threeUGTgenes) were more highly expressed in TGY; while four DEGs were more highly expressed in JX, including oneDFR, twoF3′5′Hand, oneUFGTgenes.Among them,F3′5′H(CsTGY12G0001787), a key enzyme in the flavonoid biosynthesis pathway, was only significantly expressed in TGY, andF3′5′H(CsTGY12G0001780)was significantly upregulated by 6.11- and 7.17-fold in TGY and JX, respectively.Moreover, we also found five hydroxycinnamoyl CoA quinate hydroxycinnamoyl transferase(HCT) genes involved in chlorogenic acid synthesis and six genes involved in caffeine synthesis.Among them, theHCT(CsTGY12G0001876 and CsTGY06G0003042) genes were upregulated by 1.09- to 2.04-fold in TGY.ThreeAPRTgenes were more highly expressed in TGY, particularlyAPRT(CsTGY02G0001969) andIMPDH(CsTGY08G0001307)which were upregulated by 2.42- and 2.21-fold in JX.Among the specifically expressed classes in FJSX(Fig.6-D), three genes were involved in the l-ascorbic acid biosynthesis pathway, of which one L-galactono-1,4-lactone hydrogenase(GalLDH) (CsTGY13G0000389) and two Myo-inositol oxygenase(MIOX) (CsTGY14G0001769,and CsTGY14G0001770) genes were significantly upregulated by 3.10- to 4.32-fold in FJSX, which may promote l-ascorbic acid accumulation in FJSX.

Fig.6 Differentially expressed genes (DEGs) and significantly changed metabolites (SCMs) of seven tea plant cultivars: Tieguanyin(TGY), Jinxuan (JX), Fujian Shuixian (FJSX), Fudingdabai (FDDB), Baihaozao (BHZ), Longjing 43 (LJ43), and Shuchazao (SCZ).TBtools Software was used to draw the heatmap, and the function “scale” was used to normalize the data by rows.A, chlorophyll biosynthesis pathway.B, flavonoid biosynthesis pathway.C, heatmap of DEGs related to the synthesis of catechins, caffeine,theobromine, and chlorogenic acid in the TGY and JX classes with high expression.D, heatmap of DEGs related to catechins and l-ascorbic acid synthesis in the FJSX class with high expression.E, free fatty acid biosynthesis pathway.Heatmap of DEGs related to the synthesis of free fatty acids in the FDDB, BHZ, LJ43, and SCZ classes with high expression.

Cultivars FDDB, BHZ, LJ43, and SCZ had higher levels of free fatty acids, and their genes were also enriched in lipid metabolism pathways.Therefore, we analyzed the genes in the free fatty acid biosynthesis pathway (Fig.6-E) and identified a total of six DEGs,of which oneFATBgene was more highly expressed in SCZ, and oneFAD8gene was highly expressed in FDDB and BHZ.TwoHADgenes showed higher expression in FDDB and SCZ but were significantly downregulated in TGY and FJSX.Remarkably,SAD(CsTGY09G0001385)andFATB(CsTGY14G0001990) genes were significantly upregulated by 1.47- and 4.99-fold in LJ43.These upregulated genes mainly promote the formation of free fatty acids in the cultivars.

3.7.Analysis of differentially expressed transcription factors (TFs)

Transcription factors play an essential regulatory role in the growth and development of tea plants.A total of 2 221 TFs were identified, and these TFs were divided into 46 families, each of which was differentially expressed in the total TPM values in the different cultivars (Fig.7-A).The TFs were mainly FAR1, MYB, ERF, bHLH, MYBrelated, and NAC (number>100).They included the TFs with higher total TPM values in FDDB, TGY, and FJSX,such as FAR1, C2H2, bHLH, bZIP, and MYB in FDDB,as well as GRAS and ZF-HD in TGY.Moreover, the total TPM values of the MYB-related, ERF, and LBD TFs in JX were higher.We then identified the top 20 TF families(Fig.7-B), and found that the highest expression levels of MYB-related TFs were upregulated 3-fold compared to the other TFs.The average expression levels of GATA TFs were higher, followed by the WD40 and C3H TFs.Furthermore, the MYB-related, bHLH, and bZIP TFs had essentially the same average expression levels.

Fig.7 Expression patterns of transcription factors (TFs) families.A, heatmaps of different TF families in seven tea cultivars:Tieguanyin (TGY), Jinxuan (JX), Fujian Shuixian (FJSX), Fudingdabai (FDDB), Baihaozao (BHZ), Longjing 43 (LJ43), and Shuchazao(SCZ).The colors represent the total transcripts per million reads (TPM) values normalized to all TFs of a particular TF family.The numbers in brackets indicate the total number of members in the TF family.B, boxplot of different TF families in seven tea cultivars.The boxplot shows the top 20 TF families in terms of quantity, and the ordinate is the average of the TPM values of the seven tea plant cultivars.C, heatmap of differentially expressed transcription factors (DETFs) in seven tea cultivars.The DETFs were strictly filtered by |log2FC|≥2 and P-value<0.05 and were significantly upregulated compared to the other six tea cultivars.

To obtain a clearer understanding of the differentially expressed TFs among different cultivars, we screened a total of 90 TFs that were significantly expressed in each cultivar (Fig.7-C).Among them, FAR1 was upregulated in all cultivars except SCZ, 11 ERFs were upregulated in JX, and three bHLH TFs were mainly upregulated in TGY.Furthermore, WRKY was mainly upregulated in JX,BHZ and SCZ, while MYB was mainly upregulated in TGY and SCZ, and NAC was upregulated in FJSX, FDDB, and BHZ.Some specifically expressed TFs in the cultivars were also found, such as AP2 and Dof in TGY, ARF in JX, and C3H in SCZ.These TFs may play important regulatory roles in differential metabolite accumulation.

3.8.Co-expression network analysis

Fig.8 Co-expression network analysis.A, hierarchical cluster tree showing the 14 modules obtained by weighted gene coexpression network analysis (WGCNA).The grey modules represent genes that were not assigned to specific modules.Each branch in the tree points to a gene.B, matrix of module-metabolite associations.The data of gene expression profiles for the different tea cultivars and the change patterns of significantly changed metabolites (SCMs) were combined to perform the WGCNA.The number of genes per module is shown in the left box.Correlation coefficients and P-values between modules and metabolites are shown at the row-column intersections.C, co-expression subnetwork analysis of the modules related to SCM accumulation.The top 50 nodes of the seven modules were selected to construct the network.The hub genes are shown in the seven modules,and the genes involved in SCMs are shown in these modules.

WGCNA is a highly robust method for classifying genesviahierarchical clustering of the gene co-expression network.To understand the metabolic significance of the gene modules, we correlated the 14 modules with the accumulation levels of 20 SCMs, and a total of 11 222 DEGs were used to conduct WGCNA (Fig.8-A).After merging similar modules, 14 modules were generated,each of which comprised 98 to 2 503 genes.We found that seven modules were significantly correlated (r≥0.6,P-value<0.05) with 20 SCMs (Fig.8-B; Appendix F).Among these genes, the black, blue, and green-yellow modules were significantly correlated with catechin and theobromine.The pink module was significantly correlated with glutamic acid, and the yellow module was significantly correlated with l-ascorbic acid.The brown and salmon modules were significantly correlated with linoleic acid and palmitoleic acid, respectively.

Module hub genes are generally considered representative of a given module in a biological network,and we constructed such a network as shown in Fig.8-C.Based on the eigengene connectivity (KME) values, the top 50 genes with the highest connectivity in the seven modules were selected and used to construct the coexpression network, and the TFs were chosen as the key hub genes.The black, blue, and green-yellow modules identified 15 TFs, including eight ERF, one WRKY, one bHLH, one MYB, two HB-other, one MADS-box, and one CPP TFs; and these TFs may play an important role in the formation of catechin and caffeine.The pink module had four genes selected as key hub genes, including one NAC, one HB-other, and one FAR1 TFs, and we noted that these TFs showed the strongest relationship to theFATBgene.Additionally, we identified one MYB and one FAR1 TFs in the brown and salmon modules that may be involved in regulating the formation of linoleic acid and palmitoleic acid; and the TFs in the yellow module, including one WRKY TF, may be involved in l-ascorbic acid formation.These results suggested that the TFs above may be involved in regulating metabolite accumulation in the different tea cultivars.

4.Discussion

The tea plant is an important leaf-use economic crop(Zhenget al.2020), and the chemical compositions of the tender shoots can significantly influence the quality of the tea (Chaturvedula and Prakash 2011).In this study, seven nationally representative tea cultivars were collected, and their shoots were subjected to a combination of high-throughput and high-sensitivity metabolomics and transcriptomics, including biochemical assays of important characteristic metabolites, to gain some insight into the differences in chemical components and gene regulation.

Flavanols are the primary functional polyphenols in tea, among which the catechin contents of oolong tea cultivars were more significant than those of green tea, black tea and white tea (Chenet al.2022).The results of our study clearly showed that the contents of nine catechins were higher in TGY, JX, and FJSX,which may be related to their fresh leaves being suitable for making oolong tea.Correspondingly, twoF3′5′H(CsTGY12G0001787 and CsTGY12G0001780) genes were found to be upregulated in TGY and JX.F3′5′H is considered an exclusive enzyme that catalyzes the 5′position hydroxylation of the B ring in flavonoid molecules,and higher expression ofF3′5′His closely related to the higher accumulation of catechins (Toguriet al.1993).Our previous study found that some genes, such asAPRT,TCS,ADSL, andIMPDH, play vital roles in caffeine synthesis (Chenet al.2021).These results showed that six genes in TGY and JX were upregulated, namelyAPRTandIMPDH.In particular,APRT(CsTGY02G0001969)andIMPDH(CsTGY08G0001307) were significantly upregulated in JX, and they may mainly promote the formation of caffeine.Among the modules associated with caffeine and catechins, we identified eightERFs,bHLH(CsTGY10G0002198),MYB(CsTGY15G0001211),andMADS-box(CsTGY02G0002489) TFs.MYB-bHLHWDR(MBW) ternary complexes comprise the essential regulatory machinery for catechin and anthocyanin biosynthesis (Lepiniecet al.2006; Xuet al.2015).Another study revealed that MADS-box and AP2/ERF TFs might be involved in different steps in catechin biosynthesis (Zhenget al.2019).We speculate that these TFs were significantly positively correlated with catechin and caffeine, suggesting a regulatory relationship between the catechin and caffeine pathways.

Compared to green-leaf tea cultivars, the concentrations of anthocyanins are known to be significantly higher in the new shoots of purple-leaf tea cultivars (Li C Fet al.2018).Flavonoid-related upregulated genes with high expression in TGY may contribute to the synthesis of anthocyanins.In addition,Mg-protoporphyrin IX nonomethyl ester cyclase (MPEC)is an essential enzyme in chlorophyll synthesis that catalyzes the conversion of magnesium protoporphyrin methyl ester to chlorophyllide a (Liu and Zheng 2008).We speculated that the 11.12- to 28.97-fold downregulation ofMPEC(CsTGY10G0001989) in TGY is the main reason for the reduction in the chlorophyll content.Additionally, pharmacological research has demonstrated that chlorogenic acid exhibits various pharmacological properties, including antiviral, antitumor, antibacterial,and antioxidative activities, and hydroxycinnamoyl CoA quinate hydroxycinnamoyl transferase (HCT) is a key enzyme in the chlorogenic acid synthesis pathway(Chaowuttikulet al.2017; Zhanget al.2017).Our results showed that the contents of chlorogenic acid and methyl chlorogenic acid were higher in TGY and that theHCT(CsTGY12G0001876 and CsTGY06G0003042) genes may be mainly involved in its accumulation.

L-Ascorbic acid (AsA) plays an essential antioxidant function in plants and a vital role in oxidative stress defense (Mellidouet al.2021).Our results found that the l-ascorbic acid content was significantly higher in FJSX.L-galactono-1,4-lactone hydrogenase (GalLDH) is a key enzyme in the final step of plant AsA biosynthesis (Wheeleret al.1998), and one study found that the content of AsA in plants is closely related to the expression level and activity of theGalLDHgene (Liuet al.2011).MIOX is a critical enzyme in the plant AsA biosynthesis pathway(Muniret al.2020), and a previous study proposed thatMIOXgene overexpression leads to a 2- to 3-fold increase in the AsA content of transgenicArabidopsisplants(Radzioet al.2003).Studies have shown that WRKY TF can promote the production of AsA by regulating the expression of stress-related genes to control the stress response (Liuet al.2013, 2019).In this study,GalLDH(CsTGY13G0000389) andMIOX(CsTGY14G0001769 and CsTGY14G0001770) genes in the l-ascorbate biosynthesis pathway were upregulated in FJSX, andWRKY(CsTGY11G0001197) plays an important role in regulating the expression of these genes and l-ascorbic acid accumulation.

The amino acid constituents of tea leaves greatly impact the taste and aromatic properties (Alcazaret al.2007).Most amino acids, especially theanine, impart umami to green tea and have a significant positive correlation with green tea quality (Chenet al.2022),which is one of the reasons why LJ43 and JX are suitable for green tea.Correspondingly, amino acids and soluble sugars are the critical components responsible for the brisk and sweet tastes (Cuiet al.2019).The levels of sweetness-related amino acids, sugars, and alcohols in FDDB were higher, but flavonoids were lower, which may be the reason for the umami and sweetness of processed white tea.Previous research has shown that NAC is not only involved in the synthesis of theanine by regulating theADSandGSgenes, but it is also a key central gene for the synthesis in the theanine, catechin, and caffeine biosynthesis pathways (Liet al.2015).Our results found thatNAC(CsTGY01G0001178) was strongly associated with other node genes in the pink module, which play an important role in regulating free amino acid accumulation.

Fatty acids are well-known aroma precursors in tea(Ramaswamy and Ramaswamy 2000), and they were mainly abundant in four cultivars in this study, especially linoleic acid in LJ43, palmitoleic acid in SCZ, andα-linolenic acid andγ-linolenic acid in FDDB.A previous study found that genes such asSAD,FATB,FAD, andACSLare positively associated with lipid accumulation and thatCsMYBmay be the hub gene regulating the effects of blue light on lipid metabolism (Wanget al.2020a).Previous studies have found that the MYB transcription factor is involved in regulating fatty acid metabolism inArabidopsisseeds, and MYB76 affects the accumulation of seed fatty acids by affecting fatty acid synthesis (Duanet al.2017).Tissue-specific transcriptome sequencing and lipid analysis of castor oil identified transcription factors such as bHLH, MYB, and NAC that play an important role in palmitoleic acid synthesis (Eastmond 2004).Our study found thatMYB(CsTGY14G0002344) may be the hub gene regulating palmitoleic acid accumulation.Acyl-ACP thioesterase (FAT) regulates the amount of monounsaturated fatty acids in the endoplasmic reticulum by hydrolyzing fatty acyl-CoA to generate free fatty acids,and overexpression of theArabidopsisFATB1gene in seeds can increase the C16:0 fatty acids in seeds(Joaqu??n and John 2002).Based on WGCNA analysis,NAC(CsTGY01G0001178) showed a strong relationship withFATB(CsTGY14G0001990) in the pink module, and we speculated thatNAC(CsTGY01G0001178) may play an important role in regulating the expression ofFATB(CsTGY14G0001990) and the formation of free fatty acids.

5.Conclusion

In this study, we conducted a comprehensive metabolomic and transcriptomic analysis of seven tea plant cultivars to investigate the cultivar characteristics of the different tea plants and their possible molecular mechanisms leading to metabolite accumulation.By mining the transcriptome data, characteristic metabolites and major regulatory genes that led to the accumulations in the seven tea plant cultivars were discovered.More importantly, we found that the purple-green color of TGY was mainly due to the downregulation of theMPEC(CsTGY10G0001989) gene that affects chlorophyll synthesis.Furthermore, based on the annotation of transcription factors and WGCNA analysis, we further found thatWRKY(CsTGY11G0001197) may be a key gene regulating l-ascorbic acid accumulation, andMYB(CsTGY14G0002344) may be the hub gene for the regulation of free fatty acid accumulation.Overall,these data provide new insights into the characteristic metabolites and their key regulatory genes in the seven tea cultivars, which offers valuable insights for further investigations on the cultivar selection and suitability of tea plants.

Acknowledgements

This work was supported by the Major Special Project of Scientific and Technological Innovation on Anxi Tea,China (AX2021001), the Fujian Agriculture and Forestry University Construction Project for Technological Innovation and Service System of Tea Industry Chain, China(K1520005A01), the earmarked fund for China Agriculture Research System (CARS-19), and the fund for Excellent Master’s Dissertations of Fujian Agriculture and Forestry University, China (1122YS01007).

Declaration of competing interest

The authors declare that they have no conflict of interest.

Appendicesassociated with this paper are available at https://doi.org/10.1016/j.jia.2023.02.009