Transcriptome and phytochemical analyses provide new insights into long non-coding RNAs and characteristic secondary metabolites of oolong tea (Camellia sinensis) in solar-withering

Background Oolong tea, a semi-fermented tea, was deeply loved by consumers. Among tea processing, withering is the rst indispensable process for improving the unique avor. However, the roles of long non-coding RNAs (lncRNAs) and characteristic secondary metabolites in withering of oolong tea remain unknown. Results Using phytochemical analyses, the total avonoid, total catechins, EGC, CG, GCG, ECG, and EGCG were all present at signicantly lower levels in solar-withered leaves (SW) than in fresh leaves (FL) and indoor-withered leaves (IW). However, terpenoid, JA, and MeJA were present at a higher level in SW than FL and IW. By analyzing the transcriptome data, we obtained a total of 32,036 lncRNAs. On the basis of Kyoto Encyclopedia of Genes and Genomes (KEGG) analysis, avonoid metabolic pathway, terpenoid metabolic pathway, and JA/MeJA biosynthesis and signal transduction pathway were the representative pathways in this study. And a total of 63 differentially expressed-lncRNAs (DE-lncRNAs) and 23 target genes were identied in the 3 pathways. Analysis of the expression proles of DE-lncRNAs and their target genes in SW compared with IW, we found that 4 up-regulated genes (FLS, CCR, CAD, and HCT), 7 up-regulated lncRNAs, 4 down-regulated genes (4CL, CHI, F3H, and F3'H), and 3 down-regulated lncRNAs in avonoid metabolism; 9 up-regulated genes (DXS, CMK, HDS, HDR, AACT, MVK, PMK, GGPPS, and TPS), 3 up-regulated lncRNAs, 6 down-regulated lncRNAs in terpenoid metabolism; 6 up-regulated genes (LOX, AOS, AOC, OPR, ACX, and MFP2), 4 up-regulated lncRNAs, and 3 down-regulated lncRNAs in JA/MeJA biosynthesis and signal transduction Conclusions These results suggested that the expression of DE-lncRNAs and their targets involved in the 3 pathways maybe related to the low content of total avonoid, total catechins, EGC, CG, GCG, ECG, and EGCG, and high content of terpenoid in SW. alanine, aspartate and glutamate metabolism, amino sugar and nucleotide sugar metabolism, brassinosteroid biosynthesis, plant hormone signal transduction, pentose phosphate pathway, monoterpenoid biosynthesis, terpenoid backbone biosynthesis, stilbenoid, diarylheptanoid and gingerol biosynthesis, glucosinolate biosynthesis, photosynthesis, DNA replication, avonoid biosynthesis, diterpenoid biosynthesis, metabolic pathways, glycosphingolipid biosynthesis, fructose and mannose metabolism, folate biosynthesis, and lysine biosynthesis. LTCONS_00060939 (targeting 4CHI gene), LTCONS_00056216 (targeting F3H gene), LTCONS_00044497 (targeting F3'H gene), LTCONS_00031811 (targeting FLS gene), LTCONS_00001863 (targeting CCR gene), LTCONS_00000233 (targeting CAD gene), LTCONS_00090121 (targeting CAD gene), LTCONS_00030131 (targeting HCT gene), and LTCONS_00101116 (targeting HCT gene). The results showed that the expression patterns of the detected lncRNAs and their target genes were consistent with the transcriptome dataset, demonstrating that lncRNAs participated in the regulation of avonoid metabolism by affecting the expression levels of related genes. And it may be an important factor in obvious differences in the content of avonoid metabolite between IW and SW. Using and we found that SW exhibited lower of total and EGCG) and (Z)-3-Hexenal, (E)-2-Hexenal, as well as higher of terpenoid volatiles and β-Ionone compared with IW and FL. Additionally, we also analyzed the expression proles of DE-lncRNAs and their target genes involved in avonoid metabolism, terpenoid metabolism, and JA/MeJA biosynthesis and signal transduction pathway that were differentially among FL, IW, and SW. The results suggested that the expression of DE-lncRNAs and their target genes involved in the three pathways might be related to the low content of total avonoid, total catechins (EGC, CG, GCG, ECG, and EGCG), and high content of terpenoid metabolites in SW. And lncRNAs regulatory mechanism play a key role in affecting the accumulation of related secondary metabolites in solar-withered leaves. Moreover, solar light, high content of JA and MeJA could have an accumulative effect on the accumulation of terpenoid. Therefore, the high transcript level of terpenoid metabolism-related genes and high content of terpenoid metabolites in SW are both signicantly higher than IW.

Three independent biological replicates were collected. Each replicate was collected from more than 10 randomly selected tea plants.
To initial study the taste and aroma of FL, IW, and SW, the three sample after 24 h of freeze-drying were all submitted to sensory evaluation. Then the samples were evaluated in accordance with previous method [48].
Determination of total avonoids, catechins, volatiles, JA, and MeJA contents The total avonoids content of three tea samples were extracted and detected using the aluminum chloride colorimetric method [49]. The weight (0.1 g) of tea samples was diluted with methanol until 100 mg/mL. The 2.0 mL diluted extract was added to extraction solution (0.1 mL of 10% aluminum chloride and 0.1 mL of 0.1 mmol/L potassium acetate solution) and incubated at room temperature for 30 min. Then, the absorbance of the extracts was read at 415 nm using a ultraviolet (UV)-spectrophotometer. And the contents of total avonoids were calculated and quanti ed according to the previous method [49]. Moreover, the catechins were extracted and determined from the samples according to previous method [50]. Catechins were analyzed using a Waters 2695 high performance liquid chromatography (HPLC) system equipped with a 2489 UV-visible detector. The detection wavelength was set to 278 nm and column temperature were maintained at 25 °C. Authentic standards of catechin (C), gallocatechin (GC), epicatechin (EC), epigallocatechin (EGC), catechin gallate (CG), gallocatechin gallate (GCG), epicatechin gallate (ECG), and epigallocatechin gallate (EGCG) were purchased from Solarbio (Beijing, China). Each sample was detected in triplicate.
To detect volatiles released from FL, IW, and SW, volatiles were extracted and determined following the previous method [51]. A Clarus SQ 8 gas chromatograph-mass spectrometer (GC-MS; PerkinElmer, New York, NY, USA) and a Turbmatix Headspace System (PerkinElmer) was used to detected the volatile compounds. Each sample was detected in triplicate. The GC-MS analysis was performed based on the TurboMass 6.1 software (PerkinElmer). And separated compounds were identi ed according to their retention index and the National Institute of Standards and Technology (NIST) Mass Spectral Library. And ethyl decanoate was used as an internal standard. The contents of volatiles in these three tea sample were calculated and quanti ed following the previous criteria [51].
JA were extracted and determined from three tea samples according to the previous method [38]. The each extracted sample (10 µL) subjected to a Waters 2695 HPLC system equipped with a 2489 UV-visible detector, and eluted JA were detected at 230 nm. In order to determine MeJA, equal weights (0.3 g) of tea leaves collected from three tea samples were immersed in 80% ethyl alcohol solution for 24 h. After centrifugation at 12,000 g for 10 min, the supernatant was ltered through a 0.22 μm organic membrane. Then each sample (10 µL) also subjected to the HPLC system equipped with the UV-visible detector. The detection wavelength was set to 210 nm and column temperature were also maintained at 30 °C. Each sample was detected in triplicate. Authentic standards of JA and MeJA were purchased from Solarbio (Beijing, China) and added to the extracted mixture as internal standards. The contents of JA and MeJA were quanti ed by calculating the area of each individual peak against the peak area of the authentic standard.
Total RNA extraction, library construction and high-throughput sequencing For high-throughput sequencing, total RNA was extracted separately from FL, SW, and IW using a Trizol Reagent Kit (TransGen Biotech, Beijing, China). And the RNA integrity checked by gel electrophoresis and micro-ultraviolet spectrophotometry (Nanodrop). After RNA integrity checked, ribosomal RNA (rRNA) was depleted using the Ribo-Zero rRNA Removal Kit (Illumina, San Diego, CA, USA). Then, strand-speci c cDNA was synthesized to construct these nine libraries (each sample with three biological replicates) using the TruSeq Stranded Kit (Illumina, San Diego, CA, USA), DNA polymerase I, and Ribonuclease H (RNaseH). Each library was sequenced pair end on the Illumina HiSeq X Ten platform (BGI; Shenzhen, China).

Transcriptome assembly and lncRNAs identi cation
After removing rRNA, low quality reads, adapter sequences, and contaminating reads, the clean reads were obtained. Then, all clean reads were aligned to the tea reference genome [47] using the hierarchical indexing for spliced alignment of transcripts (Hisat) software. And transcriptome of each sample was assembled independently using StringTie software [52,53]. The expression level of transcript was calculated and normalized using FPKM value [54].
In order to check the assembled quality, transcripts that overlapped with known genes in tea reference genome were discarded. Then, transcripts with short sequence length (length ≤ 200 bp) and low FPKM (FPKM < 0.5) were also removed. The candidate lncRNAs selected for further investigation were those that satis ed all the criteria in Coding Potential Calculator (CPC, score < 0) [55], txCdsPredict (score < 500) [56], Coding-Non-Coding Index (CNCI, score < 0) [57], and the transcript cannot be aligned to Pfam database [58].

Analysis of differentially expressed lncRNAs and their targets
Differentially expressed lncRNAs (DE-lncRNAs) and differentially expressed genes (DEGs) was detected by DEGseq software [59]. And |Fold change| ≥ 2 and false discovery rate (FDR) ≤ 0.001 were de ned as the threshold to identi ed the DE-lncRNAs and DEGs in these samples. Pearson correlation coe cient (PCC) and Spearman correlation coe cient (SCC), these two correlation coe cients were used to identify the potential target protein-coding genes of lncRNAs (|PCC value| ≥ 0.6 and |SCC value| ≥ 0.6). Moreover, the regulation of lncRNAs on its target genes can be divided into cisand trans-acting based on the coordinates of lncRNAs and its target genes [60,61]. When lncRNAs located 100 kb upstream and downstream of their target genes, it was de ned as cis-regulation. If lncRNAs exceeded this range, transregulation identi cation was performed according to the binding energy (value ≤ 30) of lncRNAs to mRNAs. To further investigate the potential functions of DE-lncRNAs and their related target genes, all target genes were searched against non-redundant protein (NR), Swiss-Prot, gene ontology (GO), and kyoto encyclopedia of genes and genomes (KEGG) database by using Blast [62] (E-value < 1.0E -5 ). In KEGG enrichment analysis, FDR ≤ 0.01 were de ned as the threshold to identi ed the signi cantly enrichment pathway. On the basis of FPKM value, the expression levels of DE-lncRNAs and DEGs were visualized by TBtools software [63].

Prediction of the eTMs and targets of miRNAs
To explore the relationships between lncRNAs and miRNAs, lncRNAs as eTMs of miRNAs in tea withering processing were identi ed by TAPIR software, and the criteria details were referred to Bonnet et al [64]. Furthermore, some lncRNAs and genes may serve as miRNA targets, which were directly regulated by miRNAs. The lncRNAs and genes as targets of miRNAs were predicted by Plant Small RNA Target Analysis Server (psRNATarget) software, respectively. The lncRNAs and genes with the expected value greater than 5 will be ltered, and the remaining lncRNAs and genes will be identi ed as potential targets of miRNAs. The miRNAs dataset in FL, IW, and SW were obtained through small RNA sequencing.
Relative expression analyses of the selected lncRNAs, mRNAs, and miRNAs Total RNA of FL, SW, and IW was reverse transcribed into rst-strand cDNA for mRNAs and lncRNAs qRT-PCR using a Transcript First-Strand cDNA Synthesis SuperMix (TransGen Biotech, Beijing, China). In addition, rst-strand cDNA of miRNA was synthesized using a Transcript miRNA First-strand cDNA synthesis SuperMix (TransGen Biotech, Beijing, China). The mRNAs, miRNAs and lncRNAs qRT-PCR was performed by the LightCycler 480 platform (Roche Applied Sciences, Basel, Switzerland) with Transtart Tip Green qPCR SuperMix (TransGen Biotech, Beijing, China), respectively. The setting of qRT-PCR procedure and reaction system were both referred to the method of Chen et al [12]. The glyceraldehyde-3phosphate dehydrogenase (GAPDH) and β-actin genes were used as reference genes for mRNAs and lncRNAs normalization. And 5.8S ribosomal RNA (5.8S rRNA)and U6 small nuclear RNA (U6 snRNA)were used to normalizemiRNAs. The relative expression levels was calculated using 2 -ΔΔCT method, and all primers used for qRT-PCR are designed using the automatic primer designer tool at Tea Plant Information Archive (Additional le 1: Table S1) [65]. All qRT-PCR analyses were performed in 3 biological replications, respectively.

Statistical analyses
All data are expressed as the means ± standard deviations (SD). Group differences was performed by one-way analysis of variance (ANOVA) followed by Tukey's post-hoc test. And signi cant differences among various groups are represented by different letters. Lowercase letter means signi cant difference (P<0.05); uppercase letter means extremely signi cant difference (P<0.01). The data were analyzed by SPSS 25 software.

Results
Analysis of total avonoids, catechins, volatiles, JA, and MeJA contents in fresh leaves, indoor-withered leaves, and solar-withered leaves The phenotypes of tea leaves of FL, IW, and SW were recorded ( Figure 1). The tea leaves of FL were glossy, straight, and green color. Compared with FL, the leaves of IW showed slight curling and deformation, while the leaves of SW showed obvious curling and the color of leaves became dull green. After sensory evaluation, we found that the three samples also have differences in taste and aroma, the leaves of FL and IW both have strong bitter and astringent tastes, and strong grass aroma, while the leaves of SW have weak astringent and mellow tastes, light grass avor, and slightly oral and fruity aromas.
To explore the changes of characteristic tea compounds in FL, IW, and SW, we determined the contents of total avonoids, catechins, and volatiles in these samples. The total avonoids contents were 57.02, 53.28, and 47.98 mg/g in FL, IW, and SW, respectively ( Figure 2). And the total avonoids content in SW was signi cantly lower than that in FL and IW. Moreover, the total catechins were also present at signi cantly lower content in SW than in FL and IW. Further analysis of individual catechins found that EGCG were the most abundant catechins in all three sample. The content of EGCG showed both signi cantly higher in FL (61.37 mg/g) and IW (57.55 mg/g) than SW (55.35 mg/g). Similarly, other four individual catechins, including, EGC, CG, GCG, and ECG, showed signi cantly higher content in FL and IW than in SW. However, the contents of C, GC, and EC have no signi cant difference in FL vs. SW, FL vs. IW,IW vs. SW. In conclusion, total avonoids, total catechins, and ve individual catechins, especially galloylated catechins (CG, GCG, ECG, and EGCG), were both present a signi cantly lower content in SW compared to FL and IW.
Phytohormones, especially JA and MeJA, have been reported to act as signal substances to promote the synthesis of related volatiles [66,67]. In order to understand the accumulation of JA and MeJA in withering processing, we also determined the content of JA and MeJA in fresh leaves and withered leaves. The JA contents in FL, IW, and SW were 0.92, 1.81, and 3.14 μg/g, respectively. Statistical analyses showed that JA content was present at a signi cantly higher level in SW (3.14 μg/g) than FL (0.92 μg/g) and IW (1.81 μg/g). Similarly, The MeJA content also showed signi cantly higher in SW (0.45 μg/g) than that in FL (0.08 μg/g) and IW (0.21 μg/g). Combined with volatiles contents in FL, IW, and SW, the difference in jasmonates contents among these three samples may be related to the accumulation of volatiles.
Sequencing and assembly of transcriptome data To investigate the regulation mechanism of lncRNAs in withering processing of oolong tea, the FL, IW and SW were sequenced from three independent biological replicates using Illumina HiSeq X Ten platform.
After removing rRNA, low quality reads, adapter sequences, and contaminating reads, the average of clean data were obtained per sample is 12.66 Gb (Additional le 2: Table S2). Then, all clean data were aligned to tea reference genome. A total of 82,077 transcripts were detected, including 32,036 lncRNAs and 50,041 mRNAs. Further analysis of the length distribution and exon number of lncRNAs and mRNAs, it was found that most lncRNAs (72.36%) were shorter than 1000 bp in length and only 3.94% were longer than 3000 bp. In contrast, the proportion of mRNAs in the range of 0~1000 bp was 40.46%, and the proportion of mRNAs exceeding 3000 bp was 10.07%. Moreover, 68.22% of lncRNAs contain only one exon, while most mRNAs (85.98%) contain more than two exons. This result is consistent with the ndings in maize [3], kiwifruit [68] and longan [12], the lncRNAs contains fewer exons than annotated mRNAs and the majority of the lncRNAs in plants are single exon transcripts.
On the basis of transcriptome datase, the quantitative expression analysis revealed that 28,677 lncRNAs were expressed. And 26,740 expressed lncRNAs were identi ed in FL, 26,508 expressed lncRNAs were identi ed in IW, and 26,479 expressed lncRNAs were identi ed in SW. Further analysis of lncRNAs expressions revealed that the expression levels of majority lncRNAs in FL, IW, and SW were both at a low level (FPKM ≤ 1), while the number of highly expressed (FPKM ≥ 10) lncRNAs did not exceed 5% of the total number of lncRNAs ( Figure 3). Furthermore, the function of lncRNAs are mainly achieved by acting on target genes by cisor transregulation. Thus target genes prediction was performed on these DE-lncRNAs. The results showed that 2,892 DE-lncRNAs were identi ed to contain target genes, including 2,180 cis-regulated target genes and 1,096 trans-regulated target genes. Among these, 1,694 DE-lncRNAs only targeted one coding gene and only 14 DE-lncRNAs had more than 6 target genes. Further analysis of target genes revealed that 1,904 target genes were regulated by only one DE-lncRNA, while 15 target genes were corresponded to more than 6 DE-lncRNAs.

Functional annotation of target genes of DE-lncRNAs based on GO and KEGG
To determine the speci c function of target gene, all target genes were aligned to the GO databases by using the blast program. The GO classi cation annotation of DE target genes, showing 3,543 DE target genes in FL vs. IW were annotated to three major categories (Figure 4a), namely biological process, cellular component, and molecular function. The top 3 subgroups in biological process were cellular process, metabolic process, and biological regulation. In cellular component, most DE target genes were annotated into membrane, membrane part, and cell subgroups. In the classi cation of molecular function, most of the DE target genes were classi ed into three subgroups, binding, catalytic activity, and transporter activity.
In FL vs. SW, 2,085 target genes were also enriched in these 3 major categories (Figure 4b). At the same time, the top 3 subgroups of the cellular process, metabolic process, and biological regulation in FL vs. SW were consistent with the results in FL vs. IW.
In IW vs. SW, only 890 target genes were annotated to cellular process, metabolic process, and biological regulation categories (Figure 4c). In cellular process, the most enriched 3 subgroups were cellular process, metabolic process, and response to stimulus. And the results of the top 3 enriched subgroups in the remaining two categories were also consistent with the enrichment results in FL vs. IW and FL vs. SW.
Then we performed the KEGG annotation analysis on these target genes of DE-lncRNAs. In the FL vs. IW, 3,543 DE target genes were enriched in 129 pathways ( Figure 5a). The top 20 enriched pathways were focused on plant-pathogen interaction, sesquiterpenoid and triterpenoid biosynthesis, alanine, aspartate and glutamate metabolism, amino sugar and nucleotide sugar metabolism, brassinosteroid biosynthesis, plant hormone signal transduction, pentose phosphate pathway, monoterpenoid biosynthesis, terpenoid backbone biosynthesis, stilbenoid, diarylheptanoid and gingerol biosynthesis, glucosinolate biosynthesis, photosynthesis, DNA replication, avonoid biosynthesis, diterpenoid biosynthesis, metabolic pathways, glycosphingolipid biosynthesis, fructose and mannose metabolism, folate biosynthesis, and lysine biosynthesis.
In FL vs. IW and FL vs. SW, 7 pathways in the top 20 enriched pathways were consistent (Figure 5a and 5b). These 7 pathways included plant-pathogen interaction, plant hormone signal transduction, stilbenoid, diarylheptanoid and gingerol biosynthesis, avonoid biosynthesis, sesquiterpenoid and triterpenoid biosynthesis, glucosinolate biosynthesis, and monoterpenoid biosynthesis. Considering the effects of avonoids on the tea taste and terpenoids on the tea aroma, we paid more attention to the avonoids and terpenoids metabolic pathways. Plant hormone signal transduction pathway is also involved, in which linoleic acid metabolism is related to the biosynthesis of JA, and MeJA, while jasmonate in plants hormone signal transduction is bene cial to induce the formation of terpenoid volatiles.
Then, further analysis of transcriptome differences between IW and SW (Figure 5c), 13 of the top 20 enrichment pathways were consistent with FL vs. IW and FL vs. SW. The terpenoids metabolism, including sesquiterpenoid and triterpenoid biosynthesis, monoterpenoid biosynthesis, and diterpenoid biosynthesis, were found to be a crucial metabolic pathway in our study. In addition, the pathways related to avonoid metabolism and plant hormone signal transduction were also worthy of further study.
The expression trends of the 124 transcripts involved in avonoid metabolic pathway of transcriptome data were compared based on FPKM value ( Figure 6). In short, all transcripts of 6 genes such as C4H, 4CL, F3H, F3'H, FLS, and HCT, showed the higher expression in FL and IW than SW; most transcripts of 6 genes, including PAL, CHS, CHI, ANS, ANR,and DFR genes, also showed the higher expression in FL and IW than SW. Moreover, 9 transcripts belonging to the CCR gene had the highest expression in SW. And more than half of the transcripts belong to the CAD gene, which also showed the higher expression in SW compared to FL and IW. In addition, 83 transcripts in the avonoid metabolic pathway had the higher expression levels in IW compared to SW. These results implied that the avonoid metabolism in SW might had a weak metabolic ux than that in IW. Therefore, solar-withering could inhibit the accumulation of avonoids in the withered leaves.
To decipher the regulation mechanism of lncRNAs in avonoid metabolism, DE-lncRNAs and their target genes related to avonoid metabolism were analyzed. Among them, 31 DE-lncRNAs and 8 differentially expressed target genes involved in avonoid metabolic pathway were identi ed from FL, IW, and SW transcriptome dataset (Additional le 4: Table S4). These identi ed target genes included 4CL, CHI, F3H, F3'H, FLS, CCR, CAD, and HCT genes. In addition, lncRNAs do not have a one-to-one correspondence to the regulation of their target genes. We de ned a lncRNA to regulate a mRNA as a pair of regulatory relationship. And a total of 43 pairs of regulatory relationships were identi ed from these DE-lncRNAs and their target genes ( Figure 7). Moreover, the number of lncRNAs that regulated HCT and CAD genes was the highest, and these two genes were each regulated by 10 lncRNAs, while 4CL, CHI, F3H and F3'H genes were regulated by only one lncRNA, respectively.
To further study the expression pattern of lncRNAs related to avonoid metabolism, the expression patterns of these 31 DE-lncRNAs were analyzed in FL, IW, and SW based on FPKM values ( Figure 8). The expression levels of lncRNAs in FL and IW was similar. Among the regulatory relations, there was a positive regulatory relationships between lncRNAs and their target genes in the 36 pairs of regulatory relationships, while other 7 pairs of regulatory relationships were negatively regulated. Besides, we found that some target genes were also regulated by multiple lncRNAs, and some lncRNAs can simultaneously regulate multiple target genes. Thus, it indicated that lncRNAs may be involved in the regulation of avonoids by affecting the expression level of genes related to avonoids metabolism through complex regulatory mechanisms.
In addition, we further analyzed the JA/MeJA biosynthesis and signal transduction pathway. In this pathway, a total of 48 DE-transcripts were assigned to 7 genes, including LOX (lipoxygenase), AOS (allene oxide synthase), AOC (allene oxide cyclase), OPR (oxophytodienoic acid reductase), ACX (acyl-CoA oxidase), MFP2 (enoyl-CoA hydratase/3-hydroxyacyl-CoA dehydrogenase), and JAR (jasmonic acidamino synthetase) (Additional le 6: Table S6). All the 7 genes contained more than 2 transcripts, respectively. Among these, the ACX gene had the largest number of transcripts. On the basis of these identi ed DEGs, a proposed terpenoids metabolic pathway was constructed ( Figure 10). On the basis of FPKM values, we also compared the expression trends of the 48 transcripts involved in JA/MeJA biosynthesis and signal transduction pathway between fresh leaves and withered leaves ( Figure 10). In the expression patterns of these genes, most transcripts showed higher transcript levels in SW than that in IW. Among these, all transcripts corresponding to LOX, AOS, AOC, OPR, MFP2, and JAR genes showed highest transcript levels in SW. And 17 transcripts corresponding to ACX genes also showed higher expression levels in SW than that in IW. It suggested that the high transcript level of these 6 JA/MeJA biosynthesis-related genes in SW may affect the accumulation of JA and MeJA.
In short, high expression levels of most transcripts corresponding to the 17genes involved in terpenoids metabolic pathway may be positively correlated with the content of terpenoids. Also, the high expression levels of 7 genes involved in JA/MeJA biosynthesis and signal transduction pathway may promote the accumulation of jasmonate. And jasmonate is also bene cial for the synthesis of terpenoid volatiles [66].
Thus, these results suggested that the transcript levels of these DEGs may be related to the terpenoids contents in FL, IW, and SW.
To further analysis the regulation mechanism of lncRNAs in terpenoids metabolic pathway, and JA/MeJA biosynthesis and signal transduction pathway, we also predicted the DE-lncRNAs and their target genes involved in these 2 pathways. We found a total of 10 DE-lncRNAs owning 14 cis-target genes and other 4 DE-lncRNAs owning 5 trans-target genes in terpenoids metabolic pathway (Additional le 7: Table S7). Further analysis revealed that GGPPS and TPS genes were each regulated by 4 lncRNAs, while 4 genes (DXS, CMK, AACT, and PMK)only targeted by onelncRNA, respectively ( Figure 11).
In JA/MeJA biosynthesis and signal transduction pathway, 18 DE-lncRNAs and 18 differentially expressed target genes were identi ed from FL, IW, and SW (Additional le 8: Table S8). Of which, a total of 6 genes involved in JA/MeJA biosynthesis and signal transduction pathway were regulated by lncRNAs, including LOX, AOS, AOC, OPR, ACX, and MFP2 genes. In addition, we also identi ed the speci c lncRNA-mRNA regulatory relationship. In brief, a total of 22 pairs of lncRNA-mRNA regulatory relationships were identi ed from these DE-lncRNAs and their target genes involved in JA/MeJA biosynthesis and signal transduction pathway (Figure 11). were similar between the qRT-PCR and transcriptome dataset analyses ( Figure 13). Together, these results con rmed that there are signi cant differences in the transcription levels of terpenoid metabolism-related lncRNAs and target genes in FL vs. IW, FL vs. SW, and IW vs. SW.
Relationships between the lncRNAs and miRNAs involved in avonoid metabolism, terpenoid metabolism, and JA/MeJA biosynthesis and signal transduction To analyze relationships between lncRNAs and miRNAs involved in avonoid metabolism, terpenoid metabolism, and JA/MeJA biosynthesis and signal transduction, the psRNATarget program was used to predict the potential regulatory relationships between lncRNAs and miRNAs. In total, 41 lncRNAs involved in these three metabolic pathways were predicted to be targets of 76 miRNAs in 35 families. And 133 lncRNA-miRNA interaction pairs were identi ed (Additional le 9: Table S9). Among them, 28 lncRNAs could be targeted by multiple miRNAs. However, 13 lncRNAs are regulated by only one miRNA. Similarly, we found that 36 miRNAs have only one target lncRNA; while the remaining 40 miRNAs have more than two target lncRNAs. This indicated that most lncRNAs identi ed in present study were regulated as targets of multiple miRNAs.
As a novel regulatory factor, eTMs can effectively inhibit miRNAs functions by binding to miRNAs, which indirectly affects the expression level of miRNA targets. And some lncRNAs can be involved in the regulation of miRNAs as eTMs. In the present study, two lncRNAs in transcriptome dataset were predicted to be potential eTMs (LTCONS_00026271 and LTCONS_00020084) for two miRNAs (novel_miR44 and miR169d-5p_1) involved in JA/MeJA biosynthesis and signal transduction pathway, respectively ( Figure   14a). Moreover, we found that novel_miR44 targets LOX, and miR169d-5p_1 targets ACX (Additional le 10: Table S10).
To further explore the regulatory networks of eTMs with miRNAs and target genes, the expression patterns of these eTMs (LTCONS_00026271 and LTCONS_00020084), miRNAs (novel_miR44 and miR169d-5p_1), and target genes (LOX: MTCONS_00093153 and ACX: MTCONS_00008303) were detected in the qRT-PCR analyses. According to qRT-PCR analyses (Figure 14b), the expression levels of LLTCONS_00026271 and LOX gene both signi cantly increased in SW compared to FL and IW, whereas the expression level of novel_miR44 in SW is signi cantly lower than that of FL and IW. In addition, TCONS_00020084 and ACX gene were also up-regulated in SW than that in FL and IW, while miR169d-5p_1 was down-regulated in SW than that in FL and IW. On the basis of these results, we found negative relationships between eTMs and miRNAs, as well as miRNAs and targets. Conversely, the expressions of eTMs and target genes were positively correlated. And the lncRNAs-miRNAs-mRNAs regulatory network in JA/MeJA biosynthesis and signal transduction pathway was also constructed (Figure 14c). Together, these results suggested that lncRNAs may affect the expressions of related mRNAs through miRNAs.

Discussion
The expression levels of identi ed DE-lncRNAs and their target genes involved in avonoid metabolism were associated with the low contents of total avonoid and catechins in the solar-withered leaves In the previous studies [28], tea avor is greatly in uenced by the manufacturing process. Withering is the rst indispensable processing for improving avors in oolong tea. During this processing, the content of some tea avor-related compounds changed. Among these avor-related compounds, avonoid and catechins are the the representative factors to evaluate the avor quality of oolong tea [34,35]. However, the contents of avonoid and catechins in fresh leaves and withered leaves of oolong tea had not been determined. Thus, we detected the total avonoid and total catechins, including 8 individual catechins, in FL, IW, and SW, respectively.
In this study, the leaves of SW had lower levels of total avonoid, total catechins, and individual catechins in FL vs. SW, and IW vs. SW. Further analysis found that 5 individual catechins (EGC, CG, GCG, ECG, and EGCG) were all present the lowest contents in SW than that in FL and IW. However, the molecular mechanism underlying with the different contents of total avonoid and catechins in solar-withered leaves and indoor-withered leaves remains unclear. Moreover, many research reported that lncRNAs play important regulatory roles at the transcript level of genes [69][70][71]. Conversely, studies on lncRNAs in tea plants are still in their infancy, and there is no research on the regulatory mechanisms of lncRNAs in tea processing. Thus, identifying and characterizing the lncRNAs that are present in withering processing are bene t to understand their roles for solar-withering.
To explore the regulatory mechanisms of lncRNAs in withering processing, we performed KEGG pathway Moreover, it has been reported that some lncRNAs can negatively regulate the expression of target genes [72][73][74]. In our study, 3 pairs of negatively regulated lncRNAs-mRNAs were also found in the avonoid metabolic pathway. High expression of LTCONS_00056216 and LTCONS_00044497 suppressed the expression of F3H and F3'H gene CAD gene in SW, while low expression of LTCONS_00000233 in SW promote the highly expressed in CAD gene.
Among these genes in avonoid metabolism, 4CL is a key gene in the avonoid metabolic pathway of plant, which is responsible for catalyzing coumarate and its derivatives, providing the substrate for the synthesis of avonoid and other substances in downstream of this metabolism [75,76]. Many studies have shown that high expression of 4CL gene would promote the avonoid biosynthesis [77,78], and down-regulation of 4CL gene would reduce the accumulation of avonoid [79]. In our research, the low expression of LTCONS_00054003 inhibit 4CL expression in SW. Similarly, the expression levels of 4CL gene in Populus tomentosa [80] and Paulownia tomentosa [81] are also regulated by related lncRNAs. Besides, previous research has shown that 4CL is a key gene affecting the production of catechins in tea [82]. It indicated that the expression level of 4CL was positively correlated with the content of total avonoid and catechins, and the low expressions of LTCONS_00054003 and 4CL may be not conducive to accumulation of these compounds in SW. In the avonoid metabolic pathway, CHI is located downstream of 4CL, and the main role of CHI gene is to catalyze the intramolecular cyclization of chalcone into naringenin in plants [83,84]. In various tissues of grapevine, the expression pattern of CHI was positively correlated with the total avonoid accumulation [85]. In addition, it has been reported that withering is conducive to decreasing the expression of CHI and accumulation of catachins [37]. Furthermore, recent research showed that EGCG content was suppressed in tea leaves under drought stress, and the biosynthesis of EGCG were positively associate with the expression of CHI gene in response to drought [86]. Consistent with this, we observed that the transcript levels of CHI and EGCG content were both present at a lower level in SW than in IW. Besides, studies have found that some lncRNAs have the regulatory effects on the transcript level of CHI gene [71]. Also, we found that the LTCONS_00060939 and CHI were both present at lower expression level in SW than in IW. It inferred that high expression level of LTCONS_00060939 is a key factor in inhibiting the expression of CHI gene. During the solar-withering processing, the withered leaves will suffer from different types of stress, including drought, heat, and UV/light radiation, while the indoor withered leaves are mainly affected by drought stress [29,87]. The solar-withering accelerates the dehydration process of tea leaves, and the degree of drought stress in solar-withered leaves were deeper than that in indoor-withered leaves. Thus, we suggested that solar-withering is bene cial to reduce the expression of CHI gene and EGCG content. Other 2 genes in the avonoid metabolism were F3H and F3'H. Previous studies have shown that the catechins content in tea plant were positively correlated with the transcript levels of F3H and F3'H [88,89]. In this research, there was a negative correlation between the expression levels lncRNA and these two genes. Therefore, the high expression of LTCONS_00056216 and LTCONS_00044497 in SW were not conducive to the normal expression of F3H and F3'H compared to SW, which also decreased the content of total avonoid and catechins in SW.
Downstream of the avonoid metabolic pathway, FLS is a branching gene responsible for converting dihydro avonols to avonols. At the same time, it competes with DFR for the substrate dihydro avonols, and DFR is bene cial to promote the synthesis of catechins. Meanwhile, it has a competitive relationship with catechins biosynthesis. A previous research showed that FLS expression was negatively correlated with catechins biosynthesis, especially with EGCG [90]. Thus, we inferred that low expression of FLS in IW had little effect on catechins biosynthesis, while high expression of FLS in SW inhibited catechins accumulation. And highly expressed lncRNA (LTCONS_00031811) may positively regulate FLS gene, which may be responsible for the high expression of FLS in SW. The remaining three genes (CCR, CAD, and HCT) belong to the lignin biosynthesis pathway, and their abundance directly affects the lignin accumulation in plants [91]. In phenylpropanoid metabolism, lignin biosynthesis pathway has parallel status with avonoid metabolic pathway. Many studies have found that the content of lignin is positively correlated with the expression of CCR, CAD, and HCT genes, and up-regulation of these genes may bene t to the increase of lignin production [92][93][94]. Similar to that in Arabidopsis thaliana [95], lignin metabolism may also compete for same precursors substance with avonoid metabolism in tea plant. Therefore, compared with IW, the higher transcript level of CCR, CAD, and HCT in SW may transfer the metabolic ux to lignin biosynthesis and suppressed avonoid biosynthesis. This would reduce the amount of precursor substance for catechins biosynthesis in SW, and it may explain that high levels of total avonoids and catechins can occur in SW, but not in IW.
Together, the above results suggest that high transcript levels of FLS, CCR, CAD, and HCT, and seven upregulated lncRNAs (LTCONS_00056216, LTCONS_00044497, LTCONS_00031811, LTCONS_00001863, LTCONS_00090121, LTCONS_00030131, and LTCONS_00101116), and low transcript levels of 4CL, CHI, F3H, and F3'H, and three down-regulated lncRNAs (LTCONS_00054003, LTCONS_00060939, and LTCONS_00000233) in SW closely linked with the low contents of total avonoid and catechins in the solar-withered leaves. Thus, we suggested that, due to the regulation of related lncRNAs, the geneexpression related to avonoid biosynthetic pathway was inhibited and the expression of related genes involved in lignin biosynthetic pathway was increased, thereby reducing total avonoid and catechins in SW, especially the accumulation of galloylated catechins. Combined with our sensory evaluation, the reduction of galloylated catechins and avonoid content is bene cial to reduce bitter and astringent tastes and moderately improve the palatability of tea infusion. Therefore, the solar-withering is conducive to the formation of weak astringent and mellow tastes of oolong tea.
The expression levels of identi ed DE-lncRNAs and their target genes involved in terpenoid metabolism were connected with the high contents of terpenoid volatiles in the solar-withered leaves The terpenoid is one of the important components in oolong tea, and its content is positively correlated with the aroma quality of oolong tea [28]. Some gene families involved in the terpenoid metabolic pathway were expanded in the tea genome, which indicated that terpenoid was a crucial factor in aroma formation of tea [46]. On the basis of volatile determination, the terpenoid volatiles, including β-ocimene, limonene, γ-terpinene, α-farnesene, and β-myrcene, were also signi cantly increased in SW than that in IW.
To investigate the differences in terpenoid volatiles between IW and SW, we further analyzed the transcript level of DE-lncRNAs and their target genes related to terpenoid metabolism via qRT-PCR. In this research, we identi ed 9 regulatory relationships of lncRNAs and mRNAs in terpenoid metabolism, including 3 lncRNAs-mRNAs pairs (LTCONS_00093140-DXS, LTCONS_00012676-CMK, and LTCONS_00092790-PMK) with the positive regulatory relationships and other 6 lncRNAs-mRNAs pairs (LTCONS_00002173-HDS, LTCONS_00078708-HDR, LTCONS_00039845-GGPPS, LTCONS_00025739-AACT, LTCONS_00091745-MVK, and LTCONS_00043160-TPS) with the negative regulatory relationships. Then, the qRT-PCR results con rmed that LTCONS_00093140, LTCONS_00012676, and LTCONS_00092790 were up-regulated in SW compared with IW. Also, their related target genes (DXS, CMK, and PMK) were present at a higher transcript level in SW than in IW. The other 6 lncRNAs (LTCONS_00002173, LTCONS_00078708, LTCONS_00039845, LTCONS_00025739, LTCONS_00091745, and LTCONS_00043160) in SW were all present at a low transcript level, and related target genes in SW also showed a high transcript level than that in IW. Therefore these lncRNAs have the negative regulatory effect on the expression of their target genes (HDS, HDR, GGPPS, AACT, MVK, and TPS).
Among these genes in terpenoid metabolism, DXS, CMK, HDS, and HDR genes belong to methylerythritol phosphate (MEP) pathway, other 3 genes (AACT, MVK, and PMK) belong to mevalonate (MVA) pathway [37,42]. In the MVA pathway, acetyl-CoA acts as a substrate, while pyruvate and glyceraldehyde-3phosphate (GA-3P) were used as the metabolic substrate in the MEP pathway. However, the nal products of these two pathways are isopentenyl diphosphates (IPP). Moreover, the biosynthesis of IPP will directly affect the yield of downstream terpenoids [96]. Additionally, several studies have shown that the expression levels of these genes involved MEP pathway are positively correlated with the yield of terpenoids [30,97,98]. Consistent with this, transcript levels of these target genes related to terpenoid metabolic pathway showed that the high expression of 7 upstream genes (DXS, CMK, HDS, HDR, AACT, MVK, and PMK) in SW than in IW was regulated by related lncRNAs. Thus, we inferred that high content of terpenoid metabolites such as γ-Terpinene, α-farnesene and β-myrcene in SW may be due to the transcript level of these lncRNAs and their target genes related to terpenoid metabolism, which enhanced the metabolic ux of terpenoid biosynthesis and their precursors, especially IPP. The remaining two target genes, GGPPS and TPS, are located downstream of the terpenoid metabolism. GGPPS is known to be a key gene that catalyze the biosynthesis of carotenoid from geranylgeranyl diphosphate (GGPS) [99,100].
And carotenoids can be further degraded into aromatic compounds such as β-ionone, which has a signi cant contribution to the avor of tea because of its low olfactory threshold [66,101,102]. In our study, the content of β-ionone in SW was signi cantly higher than that in IW. After being negatively regulated by LTCOS_00039845, the transcript level of GGPPS gene was also up-regulated in SW compared to IW. It indicated that the expression level of GGPPS gene is positively correlated with the content of β-ionone, while the expression level of LTCOS_00039845 is a factor that negatively regulates the content of β-ionone. Thus the high level of GGPPS gene may function as a profound role in biosynthesis of β-ionone. In addition, sunlight and UV treatments were bene cial for the accumulation of β-ionone in plants [103][104][105]. Consistent with sunlight and UV caused by solar-withering, we also suggested that solar-withering was also a factor in which the contents of β-ionone in SW is higher than IW. And β-ionone is one of the characteristic aroma components of oolong tea, its high content in SW also indicated that solar-withering is more conducive to the formation of high quality aroma in oolong tea. TPS is an indispensable gene directly catalyzing the formation of the most abundant and structurally diverse terpenoids in tea aroma, and high expression of TPS facilitates the conversion of the precursors into various terpenoids [106,107]. In present study, highly expressed LTCONS_00043160 inhibits the expression of TPS gene in IW. However, LTCONS_00043160 was in low expression level in SW, which had little effect on the expression level of TPS gene. Therefore, the expression level of TPS gene in SW was higher than IW. Meanwhile, the content of terpenoid volatiles were present at a high level in SW than in IW.
We inferred that the content of terpenoid volatiles will be affected by the expression level of the TPS gene and its related lncRNA. Above all, SW had the 9 up-regulated genes (DXS, CMK, HDS, HDR, AACT, MVK, PMK, GGPPS, and TPS), 3 up-regulated lncRNAs (LTCONS_00093140, LTCONS_00012676, and LTCONS_00092790), 6 downregulated lncRNAs (LTCONS_00002173, LTCONS_00078708, LTCONS_00039845, LTCONS_00025739, LTCONS_00091745, and LTCONS_00043160) involved in terpenoid metabolism compared with IW. In addition, lncRNAs in plants are recognized as emerging regulatory components involved in most secondary metabolism [108][109][110][111]. Consistent with previous research, terpenoid metabolism in withering processing were both not only affected by the expression level of related genes, but also by the changes in the expression level of related lncRNAs. And it is indicated that the synergism between lncRNAs and target genes was related to terpenoid metabolism in oolong tea withering. After the solar-withering processing, the oolong tea products have the oral and fruity aroma. However, after indoor withering, the oolong tea products has the characteristics of calm and dull aroma [33]. Combined with the expression levels of lncRNAs and their target genes related to terpenoid metabolism, high contents of terpenoids may bene cial for the formation of high quality aroma in SW than in IW. Moreover, compared with IW, the content of β-ionone in SW increased signi cantly, while the content of (Z)-3-hexenal and (E)-2-hexenal decreased signi cantly. And β-ionone has oral aroma and is one of the characteristic aroma components of oolong tea. However, (Z)-3-hexenal and (E)-2-hexenal are the most important components of grass avor, which is an undesirable aroma in tea [36]. Therefore, we concluded that after the solarwithering processing, high levels of terpenoid and β-ionone, as well as low levels of (Z)-3-hexenal and (E)-2-hexenal, are bene cial for SW to have the light grass avor, and slightly oral and fruity aromas in SW. And this would be one of the reasons for why the solar-withering is more conducive to the formation of higher quality aroma than the indoor-withering. Solar light, high contents of JA and MeJA, and eTMs regulatory mechanism are bene t to the accumulation of terpenoid in solar-withered leaves Light is a major environmental factor affecting the accumulation of terpenoid metabolites, and expression patterns of numerous genes involved in terpenoid metabolic pathway are regulated by light [112][113][114]. On the basis of previous research [115][116][117][118], light has a positive effect on the transcript levels of DXS, CMK, HDS, and HDR genes. In present study, the transcript levels of DXS, CMK, HDS, and HDR were signi cantly increased in SW than that in IW. It can be seen that solar light has the ability to induce high expression of these genes in the tea withering. Meanwhile, there are a large number of light-speci c cisacting elements in the promoter region of most genes in the MEP pathway [118]. After solar light stimulation, the expression of DXS, CMK, HDS, and HDR genes is activated and the expression level of these genes and related terpenoid content are up-regulated in solar-withering. At the same time, the solarwithering has stronger light intensity than the indoor-withering, which promotes the up-regulation of related genes in the MEP pathway and enhances the metabolic ow of terpenoids, thereby increasing the content of terpenoid volatiles in SW. It can therefore be concluded that the whole MEP pathway in withering processing is light-stimulated. Additionally, light can also induce signi cant expression of AtGGPPS11 downstream of terpenoid metabolism in A. thaliana seedlings [96], and the expression of TPS in tea plant were regulated by light [32]. Consistent with it, we also observed that the GGPPS and TPS were present a high transcript level in SW than IW, and β-ionone and terpenoid contents were highly co-associated with the transcript level of GGPPS and TPS, respectively. The lack of su cient light illumination in the indoor-withering, the expression level of related genes in terpenoid metabolism is present at a low level in IW than SW, which is not conducive to the accumulation of related terpenoid volatiles.Thus, solar-withering is bene t for up-regulating the transcript level of DXS, CMK, HDS, HDR, GGPPS,and TPS and increasing the accumulation of the β-ionone and terpenoid volatiles in SW, which may be the reason for the oral and fruity aroma of oolong tea after solar-withering processing.
Phytohormones are important bioactive molecules, which act as signal chemicals to regulate the synthesis of related volatiles that have been demonstrated [38,119]. Over the past decade, some studies have shown that jasmonate is bene t to the formation of terpenoid volatiles [120][121][122]. However, little is known about the speci c role of jasmonate in tea processing. To investigate the effects of JA and MeJA on the tea aroma in withering processing, we detected the contents of JA and MeJA in FL, IW and SW, respectively. In present study, the content of JA and MeJA both showed the a signi cantly higher level in SW compared with IW and FL. Moreover, 7 regulatory relationships of lncRNAs and mRNAs involved in JA/MeJA biosynthesis and signal transduction pathway were identi ed. Among these, 4 lncRNAs-mRNAs regulatory relationships (LTCONS_00040667-LOX, LTCONS_00032547-OPR, LTCONS_00064473-ACX, and LTCONS_00061187-MFP2) that had positively regulated and other 3 regulatory relationships (LTCONS_00087608-AOS, LTCONS_00035664-AOC, and LTCONS_00087182-ACX) had negatively regulated between lncRNAs and their target genes. In JA/MeJA biosynthesis, a series of genes are involved, including LOX, AOS, AOC, OPR, ACX, and MFP2 [123].Among these genes, LOX is the rst step of linolenic acid utilization for JA/MeJA biosynthesis, and up-regulated LOX gene in plants will induce the accumulation of jasmonate in response to various stresses [124]. In our research, highly expressed LTCONS_00040667 promote the expression of LOX gene in SW, and JA and MeJA were also present at a high level in SW. It is indicated that the transcript level of LTCONS_00040667 and LOX gene were highly associated-with the contents of JA and MeJA, and high expression of LTCONS_00040667 and LOX gene were bene t to JA/MeJA biosynthesis in SW.
In the allene oxide synthase (AOS) pathway, 5 genes (AOS, AOC, OPR, ACX, and MFP2) played the key roles in JA/MeJA biosynthesis, and high expression level of AOS, AOC, OPR, ACX, and MFP2 were contribute to the accumulation of JA and MeJA [40,123,125,126]. In tea withering processing, highly expressed LTCONS_00032547, LTCONS_00064473, and LTCONS_00061187 promote up-regulation of OPR, ACX, and MFP2 gene in SW by positive regulation, respectively. The other 3 lncRNAs (LTCONS_00087608, LTCONS_00035664, and LTCONS_00087182) in SW were all present at a low transcript level, and they have a negative regulatory effect on the expression of their target genes (AOS,AOC,and ACX), so these target genes also showed a high transcript level in SW than that in IW. Therefore, these lncRNAs promote the up-regulation of JA/MeJA-related genes in SW through multiple regulatory mechanisms. Previous studies have demonstrated that jasmonate biosynthesis pathway in plants could be induced under abiotic stress, and a large number of jasmonate-related metabolites play the important roles in plants defense and plant-environment interactions [127,128]. Similarly, the solarwithered leaves will suffer from different types of stress, including drought, heat, and UV/light radiation, while the indoor withered leaves are mainly affected by drought stress. Therefore, the degree of abiotic stress on the withered leaves in the solar-withering is higher than that in the indoor-withering. In addition, the accumulation of jasmonate in plants is increased by activating the expression of genes related to the JA/MeJA biosynthetic pathway in response to the damage caused by biotic or abiotic stress [129,130]. In present study, JA/MeJA biosynthesis-related genes were positive correlation with the content of JA and MeJA. In SW, we inferred that up-regulated JA/MeJA biosynthesis-related genes would promote the accumulation of JA and MeJA, and then high levels of jasmonate activate downstream terpenoid metabolic pathway and increase the synthesis of related terpenoid volatiles.
Previous study has shown that some lncRNAs can act as eTMs to bind to miRNAs, thereby inhibiting miRNAs from cleavage of target genes, and this reveal a new mechanism by which lncRNA regulates miRNAs in plants [10]. However, the lncRNAs as eTMs have not been identi ed in tea plant, and the regulation of miRNAs and mRNAs by eTMs is still unclear. On the analysis of transcriptom dataset, 2 lncRNAs (LTCONS_00026271 and LTCONS_00020084) are found to be potential eTMs for 2 miRNAs (novel_miR44 and miR169d-5p_1) involved in the JA/MeJA biosynthesis and signal transduction pathway, respectively. Further analysis of the target genes of these 2 miRNAs revealed that novel_miR44 targets the LOX gene and miR169d-5p_1 targets the ACX gene. Compared with IW, the LTCONS_00026271 and LOX gene were present at a higher transcript level in SW, while the expression level of novel_miR44 were down-regulated in SW. Similarly, the expression level of miR169d-5p_1 in SW was signi cantly lower than that of IW, and the LTCONS_00020084 and ACX gene consistently showed signi cant high expression levels in SW compared to IW. Consistent with previous research [11,131], Highly expressed lncRNA acts as a sponge to bind a large number of miRNAs, thereby protecting coding-genes from the cleavage of miRNAs. In tea withering processing, high transcript level of LTCONS_00026271 and LTCONS_00020084 acting as the eTMs to promote the expression levels of LOX and ACX genes by decoying relevant miRNAs. And the high transcription level of LOX and ACX genes are bene t to the accumulation of jasmonic acid, and activated the terpenoid metabolic pathway through JA/MeJA signal transduction to promote the increase of terpenoid volatiles in SW. Thus, the eTMs regulatory mechanism (LTCONS_00026271-novel_miR44-LOX and LTCONS_00020084-miR169d-5p_1-ACX) in SW were also a crucial factor in increasing the content of terpenoid.
In total, solar light, high contents of JA and MeJA, and eTMs regulatory mechanism are bene t to terpenoid metabolism during tea withering. Among them, it is speculated that solar light, high content of JA and MeJA could have an accumulative effect on the accumulation of terpenoid. The content of MeJA and JA in IW are both signi cantly higher than FL, so the content of related genes and metabolites in terpenoid metabolism were present at a higher level than FL. Compared with IW, SW not only has higher content of JA and MeJA, but also has higher intensity of solar light. Therefore, the high transcript level of terpenoid metabolism-related genes and high content of terpenoid metabolites in SW are both signi cantly higher than IW. Additionally, eTMs regulatory mechanism in SW were also a crucial factor in increasing the content of JA, MeJA, and terpenoid. Therefore, solar light, high contents of JA and MeJA, and eTMs regulatory mechanism in SW are all contribute to the accumulation of aroma-related terpenoid.
And this also con rms the view of "Kàn qīng shài qīng" that the solar-withering is more contribute to the high quality aroma of tea compared to the indoor-withering.

Conclusions
Using transcriptome and phytochemical analyses, we found that SW exhibited lower contents of total avonoid, catechins (EGC, CG, GCG, ECG, and EGCG) and (Z)-3-Hexenal, (E)-2-Hexenal, as well as higher contents of terpenoid volatiles and β-Ionone compared with IW and FL. Additionally, we also analyzed the expression pro les of DE-lncRNAs and their target genes involved in avonoid metabolism, terpenoid metabolism, and JA/MeJA biosynthesis and signal transduction pathway that were differentially expressed among FL, IW, and SW. The results suggested that the expression of DE-lncRNAs and their target genes involved in the three pathways might be related to the low content of total avonoid, total catechins (EGC, CG, GCG, ECG, and EGCG), and high content of terpenoid metabolites in SW. And lncRNAs regulatory mechanism play a key role in affecting the accumulation of related secondary metabolites in solar-withered leaves. Moreover, solar light, high content of JA and MeJA could have an accumulative effect on the accumulation of terpenoid. Therefore, the high transcript level of terpenoid metabolismrelated genes and high content of terpenoid metabolites in SW are both signi cantly higher than IW.
Further analyses of lncRNAs revealed the eTMs regulatory mechanism (LTCONS_00026271-novel_miR44-LOX and LTCONS_00020084 -miR169d-5p_1-ACX) in SW were also a crucial factor in increasing the content of terpenoid. Combined with the the expression of DE-lncRNAs and their target genes, our study revealed low contents of total avonoid, catechins (EGC, CG, GCG, ECG, and EGCG) and (Z)-3-Hexenal, (E)-2-Hexenal, as well as high contents of terpenoid volatiles and β-Ionone in SW were contribute to the formation of the weak astringent and mellow tastes, light grass avor, and slightly oral and fruity aromas. This ndings also con rm the view of "Kàn qīng shài qīng" that the solar-withering is more contribute to the high quality avor of oolong tea compared to the indoor-withering ( Figure 15). Abbreviations nt: nucleotide; lncRNA: long non-coding RNA; miRNA: microRNA; mRNA: messenger RNA; FPKM: expected fragments per kilobase of transcript per million fragments mapped; qRT-PCR: quantitative real-time polymerase chain reaction; bp: base pair.

Declarations Ethics approval and consent to participate
The eight-year-old tea plants (C. sinensis cv. Tieguanyin) was cultivated at Fujian Agriculture and Forestry University, Fuzhou, Fujian Province, China (E 119°14′, N 26°05′). No speci c permits were required for plant collection. The study did not require ethical approval or consent as no endangered or protected plant species were involved.

Consent for publication
Not applicable.

Availability of data and materials
All data presented in this study are provided either in the manuscript or additional les.

Competing interests
The authors declare that they have no competing interests.

Funding
The fees for high-throughput sequencing and data processing were supported by the Natural Science Authors' contributions YQG, ZXL, and CZ designed the work; YQG, CZ, and STZ performed the experiments and wrote the paper; YQG, CZ, SSZ, HFF, XJC, XZL, and YLL analyzed the data; CZ, STZ, CZZ, and LC helped to perform the sequence analysis and revised the paper carefully. All authors have read and approved the manuscript. Additional le 10: Table S10. Summary of mRNA-miRNA pairs related to avonoid metabolic pathway, terpenoid metabolic pathway, and JA/MeJA biosynthesis and signal transduction pathway in FL, IW, and SW. Figure 1 Leaf phenotype of fresh leaves, indoor-withered leaves, and solar-withered leaves.       The relative expression levels of the differentially expressed-lncRNAs (DE-lncRNAs) related to avonoid metabolic pathway   The lncRNA-mRNA relationships involved in JA/MeJA biosynthesis and signal transduction pathway and terpenoid metabolic pathway. Solid line represents direct action; dashed line represents potential action.

Figures
Solid line with arrow represents positive regulation; "T" type solid line negative regulation.

Figure 12
The relative expression levels of differentially expressed-lncRNAs (DE-lncRNAs) related to JA/MeJA biosynthesis and signal transduction pathway and terpenoid metabolic pathway.
in JA/MeJA biosynthesis and signal transduction pathway. The solid line with the arrow represents direct action; dashed line represents indirect negative regulation; "T" type solid line negative regulation.

Figure 15
Schematic diagram of key lncRNAs, genes and metabolites metabolites of avonoid metabolic pathway, terpenoid metabolic pathway, and JA/MeJA biosynthesis and signal transduction pathway in SW. Red font indicates lncRNA and gene with higher expression level or metabolite with higher content in SW than in IW; green font indicates lncRNA and gene with lower expression level or phytohormones and metabolite with lower content in SW than in IW. Solid line represents direct action.