Measurement of metabolite variations and analysis of related gene expression in Chinese liquorice (Glycyrrhiza uralensis) plants under UV-B irradiation

Plants respond to UV-B irradiation (280–315 nm wavelength) via elaborate metabolic regulatory mechanisms that help them adapt to this stress. To investigate the metabolic response of the medicinal herb Chinese liquorice (Glycyrrhiza uralensis) to UV-B irradiation, we performed liquid chromatography tandem mass spectrometry (LC-MS/MS)-based metabolomic analysis, combined with analysis of differentially expressed genes in the leaves of plants exposed to UV-B irradiation at various time points. Fifty-four metabolites, primarily amino acids and flavonoids, exhibited changes in levels after the UV-B treatment. The amino acid metabolism was altered by UV-B irradiation: the Asp family pathway was activated and closely correlated to Glu. Some amino acids appeared to be converted into antioxidants such as γ-aminobutyric acid and glutathione. Hierarchical clustering analysis revealed that various flavonoids with characteristic groups were induced by UV-B. In particular, the levels of some ortho-dihydroxylated B-ring flavonoids, which might function as scavengers of reactive oxygen species, increased in response to UV-B treatment. In general, unigenes encoding key enzymes involved in amino acid metabolism and flavonoid biosynthesis were upregulated by UV-B irradiation. These findings lay the foundation for further analysis of the mechanism underlying the response of G. uralensis to UV-B irradiation.

. PCA score plot of metabolites in Glycyrrhiza uralensis leaves from control seedlings and UV-B irradiated seedlings at four time points. Solid squares, diamonds, triangles, and dots represent leaf samples from seedlings irradiated by UV-B for 12, 24, 48, and 96 h, respectively. Open squares, diamonds, triangles, and circles represent leaf samples from control seedlings at 12, 24, 48, and 96 h, respectively.  Table S2) were subjected to the targeted MS 2 mode to obtain fragmentation patterns, and subsequently a MS/MS spectral tag (MS2T) library containing 170 high-quality (s/n > 10), highly reproducible and (almost) non-redundant metabolite signals with the product ion spectra (MS 2 ) was created (Supplementary Table S3). The database was then used to identify/annotate metabolites based on the accurate m/z values, retention time (RT), and fragmentation patterns. In terms of the annotation, metabolites with available commercial standards were identified by comparing the m/z values, RT, and fragmentation patterns of metabolites with those of the standards, which were analyzed by employing the same profiling procedure used for the plant extracts. For the metabolites with no available standards, their fragmentation patterns were compared with what have been published in literatures or databases (MassBank, KNApSAcK and METLIN), best matches were then searched in KEGG for possible structures.
We putatively identified more than 54 metabolites by the procedures mentioned above (Table 1) Table S4).

Variations in identified/annotated metabolites over time in G. uralensis leaves under UV-B radiation.
In this work, we examined the variation of the identified/annotated metabolites over time, including amino acids and flavonoids. Analysis of the normalized fold changes in amino acid contents performed by hierarchical cluster analysis (HCA) showed that among the 17 amino acids detected in G. uralensis leaves, the levels of two aliphatic amino acids (Val and Ile), two aromatic amino acids (Tyr and Trp), and two basic amino acids (His and Lys) increased in seedlings under various periods of UV-B irradiation compared to the controls ( Supplementary Fig. S3). By contrast, Asp and Met levels decreased at all four time points compared to the control ( Supplementary Fig. S3). The levels of some amino acids changed at specific time points after UV-B radiation, such as Phe, Thr, and Glu. The contents of these amino acids increased at 12 and 24 h of UV-B irradiation but decreased at 48 and 96 h ( Supplementary Fig. S3). These results suggest that short-term UV-B irradiation might activate the biosynthesis of Phe, Thr, and Glu in G. uralensis seedlings, whereas these amino acids might be degraded or used to synthesize other metabolites in response to UV-B radiation after more than 24 h of treatment. To better understand the variation in individual amino acids within the metabolic network of G. uralensis leaves in response to different lengths of UV-B radiation, we allocated the amino acids identified in our study to common amino acid biosynthetic pathways. A diagram of the changes in amino acid contents in primary metabolic pathways in G. uralensis under UV-B irradiation at all four time points compared to the control is shown in Fig. 3; amino acids with significantly altered levels are indicated in the diagram (P < 0.05, Supplementary Table S5).
To further investigate the variation in flavonoid levels in G. uralensis leaves under different periods of UV-B irradiation, we normalized the fold changes (UV treated/controls) of various flavonoids and subjected them to hierarchical cluster analysis. Flavonoids with distinct modification groups exhibited diverse variation patterns during the time course of UV-B irradiation (Fig. 4). The contents of flavonoids without any glycosylation, including apigenin and naringenin ( Fig. 4-I), were higher in UV-B-treated leaves compared to the controls. The levels of apigenin in leaves treated with 12 h irradiation were ten-fold more than those of the controls (Supplementary Table S5), suggesting that increased amounts of apigenin are synthesized within a short period in G. uralensis under UV-B radiation. Flavonoids with glycosylation detected in our results can be divided into several kinds according to their modification groups, including C/O monoglycosylation, di-glycosylated flavonoids with aromatic acylated and di-glycosylated flavonoids without any acylation. For malonylated flavonoids with O-monoglycosylation ( Fig. 4-II), their levels exhibited two kinds of changes. In the first group, comprising the O-malonylhexosides of tricetin, quercetin, and naringenin, the levels increased after UV-B irradiation. In the second group, the levels of other O-malonylhexosides, including tricin O-malonylhexoside, eriodictyol O-malonylhexoside, and luteolin O-malonylhexoside, decreased to some degree during the course of UV-B irradiation. And for flavonoids of O-monoglycosylation without malonylation ( Fig. 4-III), including chrysoeriol 5-O-hexoside, chrysoeriol 7-O-hexoside and resokaempferol 7-O-hexoside, their contents changed not much except that increased levels were observed under treatment of 96 h. For flavonoids with C-monoglycosylation ( Fig. 4-V), such as eriodictyol C-hexoside and luteolin 6-C-glucoside, the levels did not change much after UV-B Analysis of differentially expressed genes involved in amino acid metabolic pathways in G. uralensis leaves under UV-B irradiation. Among the functionally annotated unigenes, differentially expressed genes (DEGs) related to amino acid metabolism were identified from the Kyoto Encyclopedia of Genes and Genomes (KEGG) annotation based on two criteria: P < 0.05 and |log 2 (fold change)| > 1.0. The total number of DEGs involved in various amino acid metabolic pathways increased dramatically after 12 h or more of UV-B irradiation (Fig. 5, Supplementary Table S6), indicating that different periods of UV-B irradiation induce responses related to amino acid metabolism at the transcriptomic level in G. uralensis. As shown in Fig. 5, there were more DEGs involved in ' Arg and Pro metabolism' and 'Cys and Met' than in other categories. Also, there were more DEGs involved in Phe metabolism, which is closely related to secondary metabolism, than in most other metabolic pathways under UV-B treatment. These results suggest that UV-B irradiation activates the amino acid metabolism in G. uralensis leaves, as hundreds of DEGs responded to UV-B irradiation.
In addition, the expression levels of members of the Lys biosynthesis pathway were highest among amino acid metabolic pathway genes at 12, 24, and 48 h of UV-B treatment (Fig. 6), suggesting that DEGs involved in Lys biosynthesis were highly enriched compared to other pathways, despite the relatively small number of genes. Also, expression of a unigene encoding aspartate kinase (AK), whose activity leads to Lys biosynthesis, increased during UV-B irradiation (Fig. 7). In addition, two unigenes encoding the enzyme Lys-ketoglutarate reductase (LKR)/saccharopine dehydrogenase (SDH), which is involved in Lys catabolism, were upregulated after 48 and 96 h of UV-B irradiation (Fig. 7). Among genes in the Lys degradation pathway, unigenes encoding citrate synthase (CS), catalyzing the formation of citrate, a TCA cycle intermediate, were upregulated under UV-B treatment (Fig. 7). Another amino acid closely related to the TCA cycle is Glu, whose metabolism plays an important role in plants. Among genes in the Glu metabolism pathway, we identified three GAD (glutamate decarboxylase) genes, encoding an enzyme that specifically catalyzes Glu to γ-aminobutyric acid (GABA), and two GSS (glutathione synthase) genes, which are involved in the production of GSH from Glu. GAD and GSS genes were both upregulated throughout treatment (Fig. 7), implying that Glu is most likely converted into GABA and GSH under UV-B irradiation in G. uralensis leaves. We also identified DEGs encoding glutathione peroxidase (GPX) and glutathione-disulfide reductase (GSR), which catalyze the conversion reactions between GSH and GSSG. The expression of the GPX genes was elevated at all five time points and that of GSR genes was elevated after 24 h or more of UV-B treatment (Fig. 7), suggesting that GSH plays a positive role in G. uralensis leaves after exposure to UV-B radiation.

Analysis of DEGs involved in phenylpropanoid and flavonoid biosynthesis pathways in G. uralensis leaves in response to UV-B radiation.
In the present work, we identified 18 candidate unigenes involved in phenylpropanoid and flavonoid biosynthesis, as shown in Table 2. Five DEGs encoding phenylalanine ammonia lyase (PAL), the rate-limiting enzyme in the phenylpropanoid biosynthesis pathway, were upregulated by UV-B treatment, as were four DEGs encoding cinnamate 4-hydroxylase (C4H), which catalyzes the conversion of cinnamic acids into p-coumaric acid (Fig. 8). Among genes in the flavonoid biosynthesis pathways, seven CHS (chalcone synthase) genes and one CHI (chalcone isomerase) gene were more highly expressed in G. uralensis leaves after UV-B irradiation compared to the controls; these enzymes are involved in reactions that form flavanones such as naringenin (Fig. 8). Finally, the expression of one unigene encoding flavonol synthase (FLS) increased in G. uralensis plants after UV-B radiation for 24 h or more ( Fig. 8).
We also detected five unigenes annotated as UGTs (UDP-glycosyltransferase), which might encode enzymes catalyzing the glycosylation reaction of various flavonoids ( Table 3). The five UGTs identified by transcriptome analysis were upregulated under UV-B treatment (Fig. 8). Taken together, these results indicate that all DEGs related to the phenylpropanoid and flavonoid biosynthesis pathways identified in this study were upregulated after exposure to UV-B.
Quantitative Real-Time PCR (qRT-PCR) Validation of DEGs from RNA-Seq. The differential expression levels of DEGs were confirmed by qRT-PCR, and 10 unigenes involved in amino acid metabolism (6 DEGs) and flavonoid biosynthesis pathways (4 DEGs) were selected to validate the RNA-Seq results, including , and UGT72E2 (Unigene0034774). The qRT-PCR results showed that seven of these genes showed up-regulation after 6 h UV-B irradiation, consistent with the RNA-Seq data (Fig. 9A). Eight and nine of the ten genes were up-regulated after 12 h and 24 h UV-B irradiation, respectively, which was consistent with the increased expression of these unigenes in the RNA-Seq data. Inconsistent results for a few unigenes, including LKR/SDH at 12 and 24 h, and FLS at 12 h of treatment, were likely due to the fact that they were not detected in the RNA-Seq data (Fig. 9B,C). All ten genes were upregulated after 48 h and 96 h UV-B irradiation, in line with the RNA-Seq data (Fig. 9D,E). Thus, the RNA-Seq results were reliable for the identification and measurement of expression of DEGs involved in amino acid metabolism and flavonoid biosynthesis pathways in G. uralensis leaves under different durations of UV-B radiation.

Discussion
G. uralensis is one of the oldest Chinese medicinal herbs, and has high levels of bioactive phytochemicals 22,23 and multiple pharmacological properties [24][25][26][27] . However, previous studies have mainly focused on the roots of this species [40][41][42][43] . In order to investigate the variations of metabolites in G. uralensis leaves under UV-B irradiation, we performed non-targeted metabolic analysis. We found that UV-B irradiation has a marked effect on the metabolic profiles of G. uralensis leaves and metabolite levels in the leaves changed with different durations of irradiation. In order to explore the mechanism underlying the varied metabolites in G. uralensis leaves in response to UV-B treatment, we examined the changes in primary and secondary metabolites and analyzed DEGs involved in corresponding metabolic pathways under UV-B irradiation. We found that unigenes related to amino acid metabolism and flavonoid biosynthesis pathways were activated in response to UV-B radiation and that these variations at the transcript level ultimately led to changes in metabolite profiles.
Amino acids are major primary metabolites in plants and, in addition to forming proteins, amino acids can be catabolized into intermediates of the TCA cycle for energy generation 44,45 . In addition, amino acids can function as the precursors of secondary metabolites that can protect plants from various stresses 46 . Amino acid metabolic pathways play important roles in regulating plant growth and development 44,47 as well as stress-related metabolism 48,49 . The conversion of amino acids to TCA cycle intermediates for energy generation 45,47 involved in stress response implies resource reallocation of plants in the face of stress 48 . In addition, stress-associated defensive metabolites synthesized from amino acid metabolic pathways demonstrate the vital roles of amino acid metabolism in response to stresses in plants 48,50 . Previous studies found that several amino acids, including Pro, Ser, Leu, Ile, and Glu were accumulated in plants following exposure to high-level UV-B irradiation 51 , and amino acid metabolism-related pathways were enhanced, such as shikimate metabolic pathway 10 and leucine/isoleucine related pathways 52 in leaves of C. terniflora exposed to UV-B irradiation. However, the specific mechanism underlying the amino acids metabolism that results in different accumulations of amino acids in G. uralensis leaves under UV-B irradiation has not yet been characterized. In this study, variations of amino acids that responded to different durations of UV-B irradiation were found in G. uralensis leaves, including Lys, Asp, Glu and Phe, which were closely linked with other metabolic pathways for regulating amino acid metabolism and deploying defense responses. Particularly, the contents of Lys increased significantly at four time points of UV-B irradiation (P < 0.05, Supplementary Table S5) compared to the controls, which implies that G. uralensis leaves tend to accumulate Lys when exposed to UV-B radiation. Lys has been suggested to be involved in plant stress responses 44 , which indicates that the accumulated Lys in G. uralensis leaves after UV-B treatment may be associated with UV responses. As the precursor of Lys, Asp levels decreased slightly in G. uralensis leaves after UV-B irradiation for different lengths of time. Moreover, the expression of unigenes encoding AK increased after exposure to UV-B irradiation (Fig. 7), suggesting that the metabolic flux was likely directed into Lys biosynthesis through the Asp-family pathway, which has been demonstrated to link with energy and stress regulation 44 . Then, the accumulated Lys in G. uralensis leaves after UV-B exposure might be degraded to replenish TCA cycle metabolites due to the expression of LKR/SDH, which encodes a bifunctional polypeptide converting Lys into the energy-associated TCA cycle metabolite acetyl-CoA 53 in response to energy starvation upon stress conditions 44 , was up-regulated after UV-B irradiation for more than 48 h. The enhanced expression of LKR/SDH, which is involved in Lys catabolism, in response to stress has been reported 46,54 . Accordingly, the UV-induced LKR/SDH in G. uralensis leaves likely contributes to Lys catabolism, which may play an important role in stress response under UV-B irradiation. Acetyl-CoA reacts with oxaloacetate to form citrate through CS, which was upregulated by UV-B irradiation (Fig. 7), in the first step of the TCA cycle. This further suggests that Lys catabolism into acetyl-CoA can likely contribute to the TCA cycle to supplement the energy lost in G. uralensis leaves in response to UV-B stress. It is suggested that stress-induced Lys catabolism generates Glu, which is the precursor of stress-related metabolites and therefore can contribute to plant responses to environmental stress 50,53 . Glu functions as a plant signaling molecule playing vital roles in regulating plant growth and defense responses 10,50-57 . In the current study, Glu levels increased after 12 and 24 h of UV-B irradiation compared to the controls, followed by a slight decrease at 48 and 96 h of treatment. The declining levels of Glu might play an important role in maintaining the redox state through conversion into GSH and GABA in G. uralensis leaves under UV-B irradiation. One the one hand, Glu combines with cysteine and glycine to produce GSH via enzymatic reactions catalyzed by the GSS enzyme. GSH plays a role in maintaining the redox balance in plants and helping them cope with oxidative stress [58][59][60][61] . The increased expression of unigenes encoding GSSs after UV-B exposure suggests that Glu generates GSH under UV-B irradiation in G. uralensis leaves. The conversion reaction between the reduced form of glutathione (GSH) and the oxidized form of glutathione (GSSG), which function as a redox couple in plants to help plants maintain a relatively stable GSH/GSSG ratio 62 , via GPX and GSR. The antioxidant marker gene GPX is upregulated under abiotic stress and is thought to be involved in ROS scavenging in Panax ginseng 63 . GSR activity was induced by oxidative stress caused by UV-B radiation in transgenic tobacco plants, thereby giving the plants increased resistance to UV-B stress [64][65][66] . The increased expression of unigenes encoding GPX and GSR indicates that ROS or other free radicals might be generated in G. uralensis leaves exposed to UV-B radiation. We detected increased levels of the oxidized form of  Table S5), corresponding to the increased expression of unigenes encoding the related enzymes. On the other hand, GABA responds to various abiotic stresses in plants including drought, salt, heat, cold, low oxygen levels, and ROS [67][68][69][70] . The enzyme GAD catalyzes Glu into GABA which is the first step of the GABA shunt and the activity of GAD affects the metabolism of the GABA shunt 71,72 . In our results, unigenes encoding GAD were upregulated in G. uralensis leaves under UV-B irradiation (Fig. 7), suggesting that the gradual decrease in Glu levels was likely due to its catalysis into GABA to mitigate the oxidative stress caused by UV-B irradiation. Accordingly, the UV-induced Glu derived from Lys catabolism plays a vital role in G. uralensis leaves in response to UV-B radiation through its catabolism into GABA and GSH in order to reduce oxidative stress caused by UV-B irradiation. In general, Lys, Asp, and Glu are closely related to each other to form a regulatory metabolic network in G. uralensis leaves in response to UV-B irradiation through replenishing energy-associated TCA cycle metabolites and forming antioxidant compounds (Fig. 10).
In addition to primary metabolism, several amino acids contribute to biosynthesis of secondary metabolites, such as flavonoids, upon UV-B irradiation 46 . Especially, Phe serves as the entry compound of phenylpropanoid metabolism, which leads to the biosynthesis of flavonoids in plants 73 . In our results, increased levels of Phe in G. uralensis leaves were observed after UV-B irradiation for 12 or 24 h, and slight decreases were observed after UV treatment for more than 48 h. Moreover, unigenes encoding PAL and C4H, two key enzymes involved in the phenylpropanoid biosynthesis pathway, were upregulated after exposure to UV-B in G. uralensis leaves (Fig. 8). The elevated expression levels of PAL and C4H might indicate that UV-B irradiation promotes the conversion of Phe into flavonoids by stimulating PAL and C4H in G. uralensis leaves under UV-B irradiation. These results imply that phenylalanine metabolism is pushed towards the biosynthesis of flavonoids, which will help protect G. uralensis from UV-B irradiation.
Extensive works have been carried out to study the accumulation of flavonoids in plants under UV and its function as absorbing compounds 74 or free radical scavengers to reduce the damage of plants [75][76][77][78][79] . Recently, UV-protective function of flavone O-glucosides were reported 80 , and glycosylated flavonols significantly increased after UV exposure in plants 9 . However, as of now, there is no report on either the effect of UV irradiation on  leaves, or its flavonoids, especially glycosylated flavonoids accumulations in G. uralensis leaves under UV stress. In the current study, we detected kinds of flavonoids in G. uralensis leaves and found that their variation patterns during the time course of UV-B irradiation were related to their respective modification groups. Many of them were glycosylated flavonoids such as mono-C/O glycosylated flavonoids and di-C, C-glycosyl or di-C, O glycosylated flavonoids. Flavonoids commonly occur in plants as glycosides, since glycosylation can increase the water solubility of flavonoids and protect their reactive groups from oxidation by free radicals 16,73 . In addition, glycosylated flavonoids can be further acylated by aliphatic or aromatic acetyl groups 6,81 , such as the malonylated flavonoids and those with feruloyl or coumaroyl modifications detected in our study. In particular, we found that flavonoid glycosides with modifications of distinct groups in G. uralensis leaves under UV-B irradiation displayed distinct variation patterns during the treatment periods (Fig. 4), suggesting that flavonoids with different substituent groups might play different roles in the response to UV-B irradiation in G. uralensis leaves. The diverse variation patterns (according to the different modification groups) might also indicate that the activities of flavonoids are affected by different types and positions of glycosylation 73 . The glycosylated flavonoids have been reported in species such as maize, wheat, and rice 39,[81][82][83] , and C-glycosides of flavones such as apigenin and/or luteolin were suggested to act as antioxidants, insect feeding attractants, and UV-protective pigments 82,84 . However, the functions of most glycosylated flavonoids detected in this work and the correlation between their functional mechanisms and substitutions of groups in G. uralensis leaves in the response to UV-B irradiation remain to be explored. Furthermore, flavonoids with different chemical structures differ in their antioxidant activity, which has been mainly ascribed to the hydroxyl groups substituent to them, and the ortho-dihydroxylated B-ring flavonoids have been suggested as important potential radical-scavenging compounds 85,86 . In this work, we observed an increase in the levels of flavonoids with two hydroxyl groups attached to B-rings, such as quercetin O-malonylhexoside. Numerous studies have demonstrated that ortho-dihydroxylated B-ring flavonoids such as quercetin and its derivatives, accumulate in plants under stresses, including UV irradiation, and function as inhibitors of ROS generation and as ROS scavengers [15][16][17][87][88][89][90][91] . The increased dihydroxy B-ring-substituted flavonoids detected in our study might scavenge the ROS in G. uralensis leaves to protect the plant from oxidative stress caused by UV-B irradiation. In particular, we found that the levels of apigenin and several of its glycosides exhibited an upward trend in this study, nevertheless the levels of apigenin glycoside and its derivatives changed little under increased irradiation in other species 16,92 . Apigenin is a common flavonoid, and along with several of its C-glycosylated derivatives, has anti-cancer and anti-inflammatory activities and inhibit RNA viruses [93][94][95][96][97] . In the present study, the level of apigenin increased sharply in G. uralensis leaves treated with UV-B, suggesting that G. uralensis preferentially synthesizes apigenin in leaves under UV-B irradiation. Therefore, G. uralensis leaves irradiated by UV-B may have increased pharmaceutical activity for prevention and treatment of human disease. In addition, levels of other flavonoids also increased significantly after UV-B irradiation, such as naringenin, tricetin O-malonylhexoside, di-C,C-hexosyl-methylluteolin, etc. The levels of these flavonoids were elevated in UV-B-treated G. uralensis leaves, but the exact roles of these compounds upon UV-B stress response remain to be investigated.
Unigenes encoding the key enzymes in flavonoid biosynthesis and glycosylation were upregulated in G. uralensis leaves in response to UV-B treatment, including CHS, FLS, and UGT. UGTs mediate a major modification of secondary metabolites, i.e. glycosylation 98 . Many flavonoids detected in our study were glycosylated by UGTs, although the exact functions of the respective UGTs have not been determined. The qRT-PCR results were consistent with the expression pattern of unigenes detected in RNA-Seq data. The upregulation of these genes implies that UV-B irradiation induced flavonoid biosynthesis pathways and the related modifications of glycosylation in G. uralensis leaves.
We performed comprehensive metabolic profiling to investigate the variations of primary and secondary metabolites in G. uralensis leaves under various durations of UV-B irradiation. We identified/annotated 54 metabolites in G. uralensis leaves during the course of UV-B irradiation by using an LC-MS/MS system. The altered levels of various metabolites, primarily including those involved in amino acid metabolism and flavonoid biosynthesis pathways, together with the variation in the transcript levels of related unigenes, shed light on the potential interaction of amino acid metabolism with plant energy and secondary metabolism in G. uralensis leaves in response to UV-B irradiation. Our results increase our understanding of the metabolic regulation of G. uralensis leaves in response to UV-B irradiation and pave the way for further dissection of metabolic pathways in the pharmaceutical plant G. uralensis under adverse conditions.

Methods
Plant materials. Glycyrrhiza uralensis (Fisch.) seeds were collected from the Liquorice Planting Base in Yanchi County, Ningxia province. Mature, plump seeds were surface sterilized in concentrated H 2 SO 4 for 25 minutes, washed for 2 h with tap water, and sown in pots (10 cm * 10 cm * 10 cm) filled with moistened sandy soil. The pots were placed in a growth chamber with 20-30% relative humidity, 145 μmol m −2 s −1 photosynthetically active radiation (PAR), and 27/24 °C during the light/dark (12 h/12 h) periods, respectively.
Experimental design and UV-B radiation treatment. At 60 days after germination, healthy seedlings were selected and randomly divided into four groups for metabolome analysis and six groups for transcriptome analysis. For metabolome analysis, each group had three replicates, each with two plants. The control groups of plants were exposed to continuous light supplied by T8      Chemicals. The solvents used for extraction, including acetonitrile, acetic acid, and methanol, were of HPLC-grade (Merck, Darmstadt, Germany). The water used in this study was doubly deionized using a MilliQ ULTRA purification system (Millipore, Vimodrone, Italy). Lidocaine, which was used as an internal standard, was purchased from Shanghai New Asiatic Pharmaceuticals Co., Ltd. (http://www.xinyapharm.com/). All authentic standards were purchased from Sigma-Aldrich, USA (www.sigmaaldrich.com/united-states.html). Standard stock solutions of flavonoids and amino acids were prepared using methanol and deionized water as the respective solvents and stored at −20 °C.
Sample preparation and extraction. All leaf samples were collected from 10:00 to 11:00 am, immediately frozen in liquid nitrogen, ground into fine powder in liquid nitrogen, and vacuum freeze-dried for 24 h. Approximately 100 mg of each sample was combined with 1.0 ml of solvent comprising 70% aqueous methanol  with 0.1 mg/l lidocaine and extracted overnight at 4 °C. After centrifugation at 10,000 g for 10 min, the supernatants were filtered (SCAA-104, 0.22 μm pore size; Anpel, Shanghai, China, http://www.anpel.com.cn/) before LC-MS analysis 35 .

Metabolite analysis by LC-MS/MS. Leaf samples were analyzed via HPLC-ESI-QTOF-MS/MS system
(6520B, Agilent, USA), and fragmentation patterns were obtained in the targeted MS 2 mode. The data were processed using MassHunter Qualitative Analysis software 35 (Agilent Technologies, Barcelona, Spain). Multiple reaction monitoring (MRM) was performed using the LC-ESI-Q TRAP-MS/MS (4000Q TRAP, ABI, USA) for quantification of metabolites. Data acquisition, curves calibration, peak integration, and calculation were implemented with Analyst 1.6 software (AB Sciex). Data processing and the analytical conditions were as described previously 35 . Qualitative and quantitative chromatographic parameters were the same: The HPLC column was a shim-pack VP-ODS C18 (pore size 5.0 μm, length 2 × 150 mm); the column temperature was set at 40 °C; the solvent system was selected as water (0.04% acetic acid added): acetonitrile (0.04% acetic acid added). The gradient program was as follows: 100 To quantify the amino acids, the area of each individual peak was calculated and compared to the standard curves obtained from authentic standard tryptophan (Supplementary Table S4). All standard curves were constructed using authentic standards at four different concentrations under the same analytical conditions.

Statistical analyses.
Principal component analysis (PCA) was performed with R (www.r-project.org/) using the Pareto scaling method to obtain grouping information for non-targeted metabolomics data. Seventeen amino acids and 36 flavonoids were subjected to hierarchical clustering analysis via HemI (http://hemi.biocuckoo.org/) to visualize changes in metabolite profiles.
Sample collection, RNA extraction, and RNA-sequencing for transcriptome analysis. G.
uralensis leaves treated with UV-B irradiation for 6, 12, 24, 48, and 96 h, as well as the control, were collected for RNA extraction. Each treatment was repeated with three biological replicates. Total RNA was extracted from seedlings using an RNAprep Pure Plant kit (Tiangen, China) according to the manufacturer's instructions. After enrichment with oligo(dT) beads, the fragmented mRNA was reverse transcribed into cDNA with random primers. Second-strand cDNA was synthesized using DNA polymerase I, RNase H, dNTP, and buffer. The cDNA fragments were purified with a QiaQuick PCR Extraction kit, end repaired, poly(A) tailed, and ligated to Illumina sequencing adapters. The ligation products were size selected by agarose gel electrophoresis, PCR amplified, and sequenced using an Illumina HiSeq. 4000 system from Gene Denovo Biotechnology Co. (Guangzhou, China). The metrics of RNA sequencing analysis are provided in Supplementary Table S7.
Quantitative real-time polymerase chain reaction (qRT-PCR). The total RNA was extracted from samples collected at the same time points (0, 6, 12, 24, 48, and 96 h; 0-h sample was used as the control) as the ones collected for RNA-Seq analysis, reverse-transcribed (Takara, http://ww.takara-bio.com/), and subjected to PCR in a BioRad CFX96 Real-Time System according to the manufacturer's instructions. The specific primers for the ten selected genes were designed after BLASTing the sequences from of our RNA-Seq with the CDS of Glycyrrhiza uralensis genome 100 (http://ngs-data-archive.psc.riken.jp/Gur-genome/blast.pl). Sequences of primers and the identities of the BLAST hits are provided in Supplementary Table S8. The G. uralensis actin gene (Guactin, GeneBank: EU 190972) was used as an internal reference standard, and the relative expression levels were calculated using the 2 −ΔΔCT method 101 .
Data availability. All data generated or analysed during this study are included in this published article and its Supplementary Information files.