Fresh and healthy three-year-old T. hemsleyanum plotted plants from Zhejiang Agriculture and Forestry University (Hangzhou Zhejiang, N 30°15’32’’, E 119°43’23.14’’) were selected as the experimental material. Sixty-five pots of healthy, disease-free, similarly sized T. hemsleyanum were selected and randomly divided into 13 groups. With five pots in each treatment group, the UV-A (320 nm–400 nm) radiation intensity was 10 w/m², UV-B (280 nm–320 nm) was 10 w/m² and UV-C (200 nm–280 nm) was 3 w/m², divided into two time periods: light treatment and light treatment + darkness treatment. UV irradiation for 1 h and 3 h (1 h, 3 h), and increased darkness treatment for 23 h and 21 h (1 h + 23 h, 3 h + 21 h). Those who were not exposed to UV light were used as controls and recorded as CK. Samples were collected at the end of the treatment, sealed, and immediately frozen.

Freeze-dried samples were crushed and passed through sieve no. 4. 1.5 mL of 80% ethanol was added to 0.1 g of finely powdered, then sonicated for 30 min. The filtrate was used as the test solution. A total of 23 control solutions were prepared by adding methanol to form a mixed reference solution for each group, which was grouped as follows. Group I: Naringenin, Naringenin, Isorhamnetin, Isorhamnetin-3-O-glucoside, Afodoside, Origenin; Group II: Apigenin, Apigenin-7-O-glucoside, Asiaticoside, Paclitaxel; Group III: Lignan, Lignan, Isomucoid, Hesperidin; Group IV: Mullein isoflavone glycoside, Quercetin, Isoquercitrin; Group V: Isoorientin, Hypericin, Lenoside, Lenoside; Group IV: Mullein, Quercetin, Isoquercitrin, Rutin, Sageol. The mixed control solutions were stored frozen at 4°C.

Acquity UPLC BEH C18 column (100 mm × 2.1 mm, 1.8 μm) with a constant temperature of 40°C was used. The mobile phase consisted of 0.1% aqueous formic acid (A) and 0.1% aqueous formic acid-acetonitrile (B) in gradient elution at a flow rate (0 min, 5% B; 1 min, 25% B; 3.5 min, 40% B; 4.5 min, 60% B) of 0.6 mL/min. The injected sample volume was 1 μL. Ion source: Turbo V, Ionization mode: ESI-, Volume flow of Curtain Gas (CUR): 30 L/min, IonSpray Voltage (IS): −4,500 V, Volume flow of Ion Source Gas1 (GS1): 55 L/min, Volume flow of Ion Source Gas2 (GS2): 55 L/min. Acquisition method: multiple reaction monitoring mode (MRM) and ionization temperature (TEMP): 550°C. Linearity was analyzed by diluting each of the above-mixed control solutions into seven different mass concentrations of the control solution, and then determining the peak area to analyze the conditions. Linear relationships are presented in Table 1 .

Determination of PAL activity: Prepared boric acid-borax buffer pH 8.8 was added to 0.5 g of T. hemsleyanum leaves or root tubers. A mixture of PVP, EDTA-Na, ascorbic acid, and mercaptoethanol was ground into a homogenate (9 mL) and centrifuged at 4°C for 20 min at 1,000 rpm. The resulting supernatant was used as the crude enzyme. In the control group, 1 mL of phenylalanine solution, 2 mL of distilled water, and 1 mL of boric acid-borax buffer were added sequentially. In the treatment group, 1 mL of phenylalanine solution, 2 mL of distilled water, and 1 mL of crude enzyme were added sequentially. The reaction was terminated by adding 0.2 mL of 6 mol/L hydrochloric acid solution to the tubes after spiking and placing them in a water bath at 30°C for 30 min. Absorbance was measured at 290 nm. CHS, CHI, F3’H, and other enzyme activity assays were performed according to the kit, and enzyme levels were determined by ELISA.

Leaf samples were collected from the CK, 1 h, 1 + 23 h, 3 h, and 3 + 21 h groups and snap-frozen in liquid nitrogen. The samples were stored in a refrigerator at −80°C for 24 h and then sent to Nanjing Personal Gene Technology Co. for RNA extraction and transcriptome sequencing. Oligo (dT) was used to enrich the mRNA. An Agilent 2100 Bioanalyzer was used for library quality control. After RNA extraction, purification, and library construction, the libraries were sequenced using next-generation sequencing (NGS) on the Illumina sequencing platform for paired-end (PE) sequencing. Unigene expression levels were calculated for each sample by functional annotation, SSR assay, and filtering out low quality, junctional contamination, and high unknown base N content reads to obtain clean reads and assemble Unigene. For multiple samples, differentially expressed genes were detected between samples according to the requirements, and in-depth clustering and functional enrichment analyses of differentially expressed genes were performed.

The data were analyzed using two-way ANOVA analysis and Duncan’s multiple range test (*p <0.05, **p <0.01, ***p <0.001) using Graphpad Statistics. Three independent experiments were performed, and the data points represent the mean ± SEM of three biological replicates. Error bars represent 95% t confidence intervals. Cluster heatmap analysis was performed using the TBtools software. Principal component analysis was performed using Origin 2021.

UV-B treatments after 3h of irradiation in darkness treatment increased the content of dihydroflavones in T. hemsleyanum roots. The highest content was naringenin ( Figure 1 ), which was 16.30 times higher in the 3L/21D h group (0.069 µg/g) than in the 3 h group (0.0042 µg/g). These changes were particularly pronounced in the 1L/23D h group of UV-B, with flavonoids showing trends similar to those of the monomer. The changes were particularly prominent in the 1L/23D h group of UV-B, and the sum of the flavones showed similar trends to those of the monomers. Among them, flavone (238.36 µg/g), orientin (0.25 µg/g), luteoloside (13.47 µg/g), and vitexin (224.64 µg/g) in 1L/23D h group were increased by 3.46 times (825.42 µg/g), 4.43 times (1.09 µg/g), 1.70 times (22.86 µg/g), and 3.57 times (801.47 µg/g) respectively, compared with 1 h group. UV-B treatment also increased flavonol content, and the levels were elevated by the addition of darkness treatment. The flavonoid content reached 35.95 µg/g in the 1L/23D h group, which was 12.19 times greater than CK group (2.95 µg/g) and 2.01 times higher than that in 1 h group (17.87 µg/g). While other monomer concentrations remained comparable across groups, it is noteworthy that rutin which was undetectable in CK was measured at a level of approximately 2.04 µg/g following UV-B treatment for 3 h. However, luteolin was not detected in any of the tubers analyzed.

The content of flavones in the leaves was far higher than that in the root tubers, and the sum of their contents showed an increasing trend after UV irradiation. UV-A and UV-B irradiation had an inhibitory effect on the content of dihydroflavones in the leaves, and the contents were mostly reduced after darkness treatment. In contrast, UV-C treatment significantly increased the content of dihydroflavone, with the highest level in 3L/21D h group (1.49 µg/g), which was 2.33 times higher than CK group (0.64 µg/g), and 2.11 times compared to 3 h group (0.71 µg/g). Meanwhile, darkness treatment after 3 h of irradiation favored reaccumulation. The most dramatic of them was 3L/21D h group (10,052.06 µg/g) of UV-C irradiation, which was 4.16 times more abundant than that of CK group (2,417.88 µg/g), and 1.58 times more than 3 h group (6,369.90 µg/g). In particular, the 3L/21D h group showed a significant increase compared with the CK group. The total flavonol content of the leaves was significantly lower than that of the tubers. In the leaves, the four monomers (astragaline, isorhamnetin-3-O-glucoside isoquercitrin, and isorhamnetin) showed similar changes to the sum of flavonols, i.e., the content decreased or changed slightly after treatment. In contrast, the rutin reached 3.11 µg/g after UV-A irradiation for 1 h, which was 67.5 times of the CK group (0.046 µg/g).

Activities of Chalconesynthase (CHS), chalcone isomerase (CHI), flavonol synthase (FLS), Flavanone 3-hydroxylase (F3H), Flavanone 3’-hydroxylase (F3’H), and Flavanone 3’-5’-hydroxylase (F3’5’H) increased after UV-A and UV-B irradiation in T. hemsleyanum leaves. Under UV-C treatment, the activities of both CHS and F3H were elevated, while the activities of the remaining enzymes were lower than those of the CK group, reaching the lowest in the 1 h group, notably F3’H in the leaves ( Figure 2 ) [0.52 times (54.52 U/L) lower than that of the CK group (105.36 U/L)]. After darkness treatment, most of the enzyme activities were higher than those during UV irradiation. The greatest increase in F3’5’H by UV-A treatment was observed in the 3L/21D h group (97.25 U/L), which was 1.81 times higher than that in the CK group (53.73 U/L) and 1.24 times more than 3 h group (78.12 U/L). Activity of aforementioned enzymes showed a similar trend in root tubers as in leaves (especially CHI activity in root tubers dropped after UV-A and UV-B treatments and peaked at 3 h (453.89 U/L), which was 0.81-fold lower than that of CK group (557.59 U/L)) and was higher than those in leaves (increase of 16.82%–41.41%, except for F3H and F3’H). CHS was 2.13 times (65.55 U/L) with 3L/21D h treatment more active than CK group (30.74 U/L) in UV-C. This indicated that UV-A and UV-B have a strong effect on promoting enzyme activities in both leaves and root tubers, while UV-C irradiation has an inhibitory effect on enzyme activity but the addition of darkness treatment promoted enzyme activity.

In order to better delineate the effects of UV irradiation and darkness treatment on enzyme activities in leaves and root tubers, we normalized the results and obtained a clustering heatmap. Based on the clustering of the vertical coordinates, the enzymes were split into two major clusters ( Figure 3H ). The first cluster contained CHI-R(R-root), CHS-L (L-leaf), and CHS-R, which were the closest to each other in the measured parameter responses, indicating high activity under UV-C irradiation. The remaining enzymes belonged to the second category, which included 11 enzymes, all of which showed elevated activity upon exposure to UV-A and UV-B, especially after UV-A and darkness treatments. FNS was the most notable enzyme, with lower expression in both the leaves and root tubers than in the CK group.

To acquire a comprehensive understanding of the response of chemical components to ultraviolet irradiation, a principal component analysis (PCA) model was constructed based on the contents of flavonoids with different durations and darkness treatments. Principal component 1 (PC1) and principal component 2 (PC2) accounted for 37.1% and 24.3% of the total variation, respectively ( Figures 3A, B ). Most flavonoid monomer contents were positively correlated with PC1, indicating that the differences in PC1 were mainly controlled by different durations of UV irradiation and darkness treatment. In the PCA scatter plot, treatments were classified into four groups, namely the distribution of CK, UV-A, UV-B and UV-C treatments. In leaves, the UV-A and UV-B groups were closely distributed, while the CK and UV-C groups were distributed in the other two quadrants. This suggests that the results of UV-C and darkness treatments on leaves differ from those of the UV-A and UV-B treatments. The first two principal components accounted for 65.2% of the total variation in the flavonoid monomer content (PC1, 51.1%; PC2, 14.1%). Most of the monomers were distributed in the UV-B groups, indicating a promising correlation between UV-B treatments and monomer content in root tubers.

Based on the aforementioned results, it was determined that the effect of UV-B irradiation and darkness treatment on T. hemsleyanum leaves was more pronounced; thus, further transcriptome analysis was conducted on these leaves. The comparison between groups revealed 96 differentially expressed genes (DEGs), as illustrated in the Venn diagram ( Figure 3C ). Specifically, there were 1,369 DEGs uniquely expressed in CK vs 1 h, 753 DEGs in CK vs 1 + 23 Dh, 2,220 DEGs in CK vs 3 + 21 Dh, and 1,190 DEGs in CK vs 3 h. There were 647 genes associated with the biosynthesis of other secondary metabolites. We employed DESeq for differential analysis of gene expression and screened 15,262 differentially expressed genes with the condition of expression difference multiplicity |log2FoldChange| >1 and significance P-value <0.05. The results indicated both up-regulation and down-regulation among these DEGs; Most DEGs were identified between 3h and 3L/21D h with 3,663, of which 1,719 were upregulated DEGs and 1,944 were downregulated DEGs.

Unigenes were compared to six functional databases for annotation, and finally 64,093 (NR: 50.35%), 31,591 (GO: 24.82%), 20,014 (KEGG: 15.72%), 23,085 (Pfam: 18.14%), 59,954 (eggNOG: 47.10%), and 37,438 (SwissProt: 29.41%) Unigene obtained functional annotations ( Figures 3D–H ). By matching all unigenes to the KEGG database, genes were classified into five branches based on the KEGG metabolic pathways: Metabolism, Genetic Information Processing, Environmental Information Processing, Cellular Processes, and Organismal Systems. Unigenes at levels 1 and 2 of the KEGG database were annotated, and KEGG functional distribution statistics were plotted. Among them, the largest number of genes related to environmental adaptation in Organismal Systems was 1,271. Plant–pathogen interactions (ko04626), sesquiterpenoid and triterpenoid biosynthesis (ko00909), and plant hormone signal transduction (ko04075), among others, were enriched. Notably, the circadian rhythm plant (ko04712), photosynthesis (ko00195), and flavonoid biosynthesis (ko00941) were also included, and the greatest enrichment was observed in anthocyanin biosynthesis (ko00942).

T. hemsleyanum was compared with the Plant Transcription Factor Database (PlantTFDB) database to predict the transcription factor and the family to which the transcription factors belong. In total, 10,676 TFs were identified in the five treatment groups, representing 57 different families, among which the Trihelix, bHLH, ERF, MYB, MYB-related, NAC, and WRKY families were transcription factors with a greater number of differential genes. The MYB family comprises 646 transcription factors, while 562 from the NAC family responded to stress conditions alongside 58 for AP2.

The accumulation of flavonoids and their biosynthetic pathways within T. hemsleyanum were constructed based on metabolomic data alongside KEGG pathway analyses, as well as previous research findings ( Figure 4 ). In this study, we analyzed the changes in key metabolite content (naringenin, naringin, eriodictyol, orientin, vitexin, luteolin, luteoloside, astragaline, isorhamnetin, isorhamnetin-3-O-glucoside, isoquercitrin, and rutin) and key enzyme activities (CHI, CHS, FNS, F3’H, F3’5’H, FLS, and F3H) in flavonoid biosynthesis in T. hemsleyanum leaves after UV-B and dark treatments. Among them, naringenin is the key metabolite in flavonoid synthesis, catalyzed by different enzymes, the naringenin produces four branches, one branch produces naringin, the second branch produces eriodictyol, luteolin and luteoloside, the third branch produces orientin and vitexin, and the fourth branch produces astragaline, isorhamnetin, isorhamnetin-3-O-glucoside, isoquercitrin and rutin.