Comparative metabolomic analyses of Dendrobium officinale Kimura et Migo responding to UV-B radiation reveal variations in the metabolisms associated with its bioactive ingredients

Plant materials and UV-B treatment

Three-year-old D. officinale plants were used in this study. All the seedlings were planted in a growth chamber of Zhejiang Academy of Agriculture Science. The condition was set at a temperature of 25 ± 1 °C with a light/dark cycle of 12/12 h and 60%–70% relative humidity. For the UV-B treatment, 10 plants, as one group, were exposed to UV-B radiation for 12 h. The UV-B radiation was artificially generated by a UV-B lamp and wavelength of UV-B lower than 280 nm was filtered out by 3-mm transmission cutoff filters (Schott, Mainz, Germany). The irradiance of UV-B at the sampling area was set at 1.6 W m⁻². The irradiance of UV-B was measured by an ultraviolet intensity meter (Apogee Instruments). Another 10 seedlings were planted in normal condition as the controls. Ten biological repetitions were applied to each group. Fresh stem samples were collected from the two groups of D. officinale seedlings after the treatments

Metabolite isolation for untargeted metabolomic analyses

Stem samples from above two groups of D. officinale seedlings (50 mg each, N = 10) were harvested and put into different tubes. Each sample was added with 800 µL of pre-colded methanol (50%) and ground by a 2010 Geno/Grinder (SPEX SamplePrep, Metuchen, NJ, USA) at 1,900 strokes/min for 2 min. Then, the mixture was added with 500 µL of pre-colded chloroform/methanol/water (v:v:v, 1:3:1), shaken at 4 °C for 10 min, and put to ultrasonication for another 5 min. The supernatant was collected by 13,000× g centrifugation, vacuum-dried, and resuspended in methanol solution (50%). Control samples were arranged by mixing an equal volume of each experimental sample from different groups.

UPLC-MS/MS analysis for the metabolomes

For the UPLC-MS/MS analysis, all samples were chromatographic fractionated using an Applied Biosystems SCIEX UPLC system (Foster City, CA, USA) with a Waters reversed-phase ACQUITY BEH Amide column (2.1 × 100 mm, 1.7 µm particle size; Milford, MA, USA). The temperature of oven was kept at 35  °CC and the rate of flow was set 0.4 mL/min. Mobile phase consisted of solution A and B. The solution A is an aqueous solution with 25 mM CH₃COONH₄ and 25 mM NH₄H₂O, and the solution B is a mixture of IPA:CAN (v:v, 9:1) adding with 0.1% formic acid. The gradient elution was set according to the previous study (Yu et al., 2018a). The fractions eluted from the column was subjected to a high-resolution MS/MS SCIEX Triple-TOF-5600 plus system and analyzed in both positive and negative ion modes. The detail parameters as set according to the previous study (Yu et al., 2018a).

Bioinformatic analysis of the untargeted metabolomic dataset

Three softwares, including XCMSPlus software (https://sciex.com/products/software/xcms-plus-software), CAMMERA, and metaX toolbox (http://www.bioconductor.org/packages/2.4/bioc/html/CAMERA.html) were used to process the LC-MS/MS raw data (Smith et al., 2006). For MS data pretreatment, a series of operations, including peak and second peak grouping, peak picking, retention time (RT) correction, and annotation of isotopes and adducts, were carried out. Each ion was recognized by combining RT and m/z values. Intensity of each peak were calculated and a 3D matrix containing arbitrarily assigned peak indices (retention time-m/z pairs), sample names and ion intensity information was constructed.

The peak features that were detected in less than 50% of quality control (QC) samples or less than 80% of experimental samples were disregarded, the remaining peaks were processed using the K-nearest neighbor algorithm to improve their quality. Principal component analysis was carried out for outlier detection and batch effects evaluation. In addition, the relative standard deviations of the peak features were recorded across all the QC samples, and those more than 30% were deleted.

Annotation of identified metabolites

The online Kyoto Encyclopedia of Genes and Genomes (KEGG, https://www.kegg.jp/) database was applied to annotate the metabolites by matching the m/z value of each sample with the metabolites from online database. If a mass difference in metabolite between the detected and the database value was <10 ppm, the metabolite would be annotated according the database and the its molecular formula would further be checked by the isotopic distribution measurements.

Identification of differentially accumulated metabolites (DAMs) between groups

Wilcoxon tests were used to detect the differences in metabolite levels between two sample group. The P values were adjusted for multiple tests using an FDR with Benjamini–Hochberg method. Supervised partial least squares-discriminant analysis was applied through metaX software to discriminate the variables between two groups. A VIP cut-off value of 1.0 was used to screen key features. The differentially accumulated metabolites (DAMs) were selected with —fold change—>2 and with statistical significance (P value <0.05).

Quantitative analysis of total alkaloids, polysaccharides and flavonoids

Several new seedlings were used in quantitative analysis. For quantitative analysis, three replicates were used. Stems from the seedlings under the two groups were harvested. The polysaccharide contents in different samples were determinated using the phenol-sulfuric acid method (Masuko et al., 2005). The total alkaloid contents were determinated using the UPLC method (Xu et al., 2010). In briefly, all stem samples were washed three times with 75% ethanol for 5 min, and heated for 2 h. The extracts were concentrated and dissolved with 5% hydrochloric acid. Then, aqueous acid solution was extracted three times by petroleum. The total alkaline solution was extracted three times with chloroform for the UPLC analysis.

Total flavonoids were determined by an aluminum nitrate colorimetric assay. In briefly, stem tissues were homogenized in 75% ethanol at room temperature for 30 min. The supernatant was extracted by 12,000× g centrifugation. A mixture solution of Al(NO₃)₃ and NaNO₂ were added to the supernatant, and then coloration reaction was induced by addition of NaOH solution. Absorbance was determined at wavelength of 510 nm. Rutin was used as a standard.

Quantitative analysis of total alkaloids, polysaccharides and flavonoids was performed for five biological repeats and figures show the average values of five repeats.

Real-time PCR validation

Transcriptomes of D. officinale were download from the NCBI database (SRR2014227). According to the transcriptomic annotation, four glycolysis-related genes, four flavonoid-related genes, and four alkaloid-related genes were selected. Sequence information of all these genes were Table S1. In the experiment of UV-B induced expression confirmation, a D. officinale ACTIN gene was used as an internal standard to analyze the related mRNA level basing on the comparative cycle threshold (2^−ΔΔCt) values. Firstly, total RNAs were isolated from the samples used for the UV-B treatments. Total RNA of each sample was reverse transcribed to cDNA using a Thermo Scientific First Strand cDNA Synthesis Kit. Then, a SYBR Premix Ex Taq Kit and an ABI PRISM 7700 DNA Sequence Detection System were used for qRT-PCR analysis. Three independent samples of each treatment were used for the qRT-PCR analysis. The saw data and primer sequences are showed in Table S1. Differences in values between two groups were calculated using one-way ANOVA with Student’s t-test at P < 0.05.

Untargeted metabolomes of D. officinale under the control and UV-B treatment

To explore the effects of UV-B treatment on the accumulation of bioactive ingredients in D. officinale, an untargeted approach was applied, identifying 3,655 metabolites from 5,994 ion features (Table S2). To check the quality of the MS data, m/z width and RT width were analyzed, suggesting that instrument preparation reached the standards (Figs. 1A and 1B). A PCA showed that the percentages of value in the metabolite analysis of PC1 and PC2 were 52.14% and 8.08%, respectively (Fig. 1C). Based on the annotations, many metabolites were grouped into at least one KEGG category. The top five largest KEGG categories were ‘Global and overview maps’ (757 metabolites), ‘Biosynthesis of other secondary metabolites’ (315 metabolites), ‘Amino acid metabolism’ (211 metabolites), ‘Carbohydrate metabolism’ (177 metabolites), and ‘Metabolism of terpenoids and polyketides’ (158 metabolites) (Fig. S1). Furthermore, 1403 metabolites were grouped into 10 primary metabolic categories, including alkaloids (176 metabolites), amino acids (330 metabolites), carotenoids (17 metabolites), lipids (102 metabolites), flavonoids (169 metabolites), hormones (52 metabolites), phenylpropanoids (112 metabolites), saccharides (168 metabolites), terpenoids (267 metabolites), and vitamins (10 metabolites) (Fig. 1D and Table S3).

Identification of DAMs between the control and UV-B treatment groups

To provide a comprehensive overview of metabolic variations under the UV-B treatment, two quality control parameters, coefficient of variation (CV) and normalized intensity, were checked. Our data showed a obvious separation between the two groups (Figs. S2 and S3). After filtering, 4827 high quality metabolites were selected to analyze the DAMs between the two groups (Table S4). Statistical analysis showed 1423 significant DAMs, including 640 up- and 783 down-regulated metabolites (Figs. 2A, 2B and Table S5). A total of 690 DAMs were grouped into different primary metabolic categories, including 176 alkaloids, 303 amino acids, 102 lipids, 169 flavonoids, 52 hormones, 25 phenylalanines, 87 phenylpropanoids, 168 saccharides, and 267 terpenoids (Fig. 2C). For several primary metabolic categories, the number of up-regulated metabolites was larger than that of down-regulated metabolites. For example, 121 up- and 54 down-regulated amino acids, 44 up- and 13 down-regulated lipids, 118 up- and 24 down-regulated terpenoids, and 46 up- and 31 down-regulated saccharides were identified under the UV-B radiation (Fig. 2D).

Putative metabolites associated with alkaloid biosynthesis of D. officinale

The primary style of alkaloids in D. officinale was TIA, and therefore the metabolites assigned into the TIA biosynthesis pathway were screened. In our study, the metabolites classified into the putative upstream elements of alkaloid biosynthetic pathway, including ‘’terpenoid backbone biosynthesis’, ‘monoterpenoid biosynthesis’, and ‘intermediate products’, were identified. In detail, five metabolites, including 2-C-methyl-D-erythritol 4-phosphate, mevalonate 3-phosphate, pyruvate, 2-C-methyl-D-erythritol-2,4-cyclodiphosphate, and isopentenyl phosphate, were targeted to the terpenoid backbone biosynthesis pathway. Among the metabolites in terpenoid backbone biosynthesis pathway, three metabolites were predominantly accumulated under the UV-B treatment (Fig. 3A). For the monoterpenoid biosynthesis pathway, eight metabolites, including 7-dexyloganate, iridotrial, deoxylogannin, 7-dexyloganate, (6E)-8-oxogeranial, loganin, loganate, and secologanate, were identified, five of which were significantly up-regulated by the UV-B treatment (Fig. 3B). Moreover, four key intermediate products, such as tryptamine, tryptophan, secologanin, and strictosidine, were identified. Interestingly, three of the four key intermediate products were highly accumulated under the UV-B treatment (Fig. 3C).

Putative metabolites associated with polysaccharide biosynthesis of D. officinale

Based on their KEGG annotation, three metabolic pathways, providing fundamental building blocks for polysaccharide biosynthesis, were identified in D. officinale. For the glycolysis, nine metabolites, including seven UV-B up-regulated metabolites, were identified (Fig. 4A). For the starch and sucrose metabolism pathway, eight metabolites, including iso-maltose, α, α-trehalose, 3-ketosucrose, 4-ketosucrose, cellobiose, maltose, levanbiose, and 5-ketosucrose, were identified. Half of the metabolites involved in the starch and sucrose metabolism pathway were up-regulated by the UV-B treatment (Fig. 4B). For the fructose and mannose metabolism pathway, ten metabolites, including D-fructose, 2,4-diketo-3-deoxy-L-fuconate, D-mannose, L-rhamnono-1,4-lactone, D-allose, D-sorbitol, mannitol, α-D-glucose, and 2-O-(α-D-mannosyl)-D-glycerate, were identified. Among these fructose and mannose metabolism-related metabolites, only three metabolites (mannitol, D-mannose, and 2-O-(α-D-mannosyl)-D-glycerate) were significantly up-regulated by the UV-B treatment (Fig. 4C).

Putative metabolites associated with flavonoid biosynthesis of D. officinale

In our study, many metabolites related to flavonoid metabolism were identified in D. officinale, including two products of CHALCONE SYNTHASE (CHS), nine products of CHALCONE ISOMERASE (CHI), and five products of FLAVANONE 3-HYDROXYLASE (F3H). As the downstream products of CHS, 2′,4,4′,6′-tetrahydroxychalcone was down-regulated and aureusidin was induced by the UV-B treatment (Fig. 5A). As the downstream products of CHI, five metabolites, including vitexin, eriodictyol, apigenin, pinostrobin, and neohesperidin, were significantly accumulated in the seedlings under UV-B treatment (Fig. 5B). As the downstream products of F3H, three metabolites, including cyanidin, fustin, and pelargonidin, were greatly up-regulated by the UV-B treatment (Fig. 5C).

Determination of total polysaccharides, total alkaloids and total flavonoids

The contents of total alkaloids, total polysaccharides, and total flavonoids were determinated in the control and UV-B treatment groups. The contents of total alkaloids were induced from 0.035% to 0.065% by the UV-B treatment (Fig. 6A). The contents of total polysaccharides were up-regulated from 32% to 42% by the UV-B treatment (Fig. 6B). Moreover, the contents of total flavonoids were increased from 0.17% to 0.34% by the UV-B treatment (Fig. 6C). The saw data was listed in Table S6.

Validation of the expression levels of several genes associated with bioactive ingredients

To check the changes in the expression levels of several genes associated with bioactive ingredient biosynthesis, a qRT-PCR assay was performed. For the glycosylsis, four genes, including pyrophosphate-fructose 6-phosphate 1-phosphotransferase (PFP), mannose-6-phosphate isomerase (MPI), xylose isomerase (XI), GDP-L-fucose synthase (FS), were selected; for alkaloid biosynthesis, four key genes, including 1-deoxy-D-xylulose-5-phosphate synthase (DXS), 1-deoxy-D-xylulose-5-phosphate reductoisomerase (DXR), tryptophan synthase (TS), and strictosidine synthase (STR), were selected; for flavonoid biosynthesis, four key genes, including CHS, CHI, F3H, and flavanone 4-reductase (F4R), were selected; and for polysaccharide biosynthesis, four genes, including cellulose synthase-like 1 (CSL1), cellulose synthase-like D5 (CSLD5), cellulose synthase-like A1 (CSLA1), and cellulose synthase-like D3 (CSLD3), were selected. Our data showed that the selected flavonoid biosynthesis- and alkaloid biosynthesis-related genes were siginificantly induced by the UV-B treatment. For the glycosylsis-related genes, the PFP and MPI genes were up-regulated and the XI and FS genes were reduced by the UV-B treatment. For the polysaccharide biosynthesis-related genes, CSL1, CSLD5, CSLA1 and CSLD2 were significantly up-regulated by the UV-B treatment (Fig. 7).