Traditional processing increases biological activities of Dendrobium offificinale Kimura et. Migo in Southeast Yunnan, China

The orchid Dendrobium officinale grows throughout southeast China and southeast Asian countries and is used to treat inflammation and diabetes in traditional Chinese medicine. Tie pi feng dou is a well-known traditional Chinese medicine made from the dried D. officinale stems. Processing alters the physicochemical properties of TPFD; however, it is unclear how processing affects the quality and medicinal value of this plant. Here, we analyzed and compared the chemical composition of fresh stems of D. officinale and TPFD and explored possible explanations for the enhanced medicinal efficacy of processed D. officinale stems using qualitative and quantitative methods. To identify the components of FSD and TPFD, we used ultra-high-performance liquid chromatography combined with mass spectrometry in negative and positive ion modes and interpreted the data using the Human Metabolome Database and multivariate statistical analysis. We detected 23,709 peaks and identified 2352 metabolites; 370 of these metabolites were differentially abundant between FSD and TPFD (245 more abundant in TPFD than in FSD, and 125 less abundant), including organooxygen compounds, prenol lipids, flavonoids, carboxylic acids and their derivatives, and fatty acyls. Of these, 43 chemical markers clearly distinguished between FSD and TPFD samples, as confirmed using orthogonal partial least squares discriminant analysis. A pharmacological activity analysis showed that, compared with FSD, TPFD had significantly higher levels of some metabolites with anti-inflammatory activity, consistent with its use to treat inflammation. In addition to revealing the basis of the medicinal efficacy of TPFD, this study supports the benefits of the traditional usage of D. officinale.

Sample preparation. Similarly sized samples were selected for analysis. Each sample was prepared in quadruplicate. First, all samples were thoroughly ground. For the analysis, 80 mg of FSD (Sample No.: DF-1-1 to DF-1-4) and TPFD (Sample No.: DF-F-1 to DF-F-4) was transferred into a 1.5-mL microfuge tube. Twenty microliters of internal standard (L-2-chlorophenylalanine, 0.3 mg/mL; methanol), 1 mL methanol-water (V:V = 7:3), and two small steel balls were added. The samples were chilled to − 20 °C for 2 min and then ground at 60 Hz for 2 min, extracted with ultrasonic waves for 30 min in an ice-water bath, and incubated at − 20 °C for 20 min. The samples were centrifuged at 4 °C and 13,000 rpm for 10 min. Then, a glass syringe was used to collect 150 μL of supernatant, which was filtered through microfilters (0.22 μm). The filtrate was transferred into LC vials, which were stored at − 80 °C until analysis.
For quality control (QC), pooled samples were prepared by mixing aliquots of all the samples. Ultimate 3000 RS) with a mass spectrometer (Q-Exactive plus quadrupole-Orbitrap) equipped with a heated electrospray ionization (ESI) source (Thermo Fisher Scientific) was used to analyze the metabolic profiles in ESI positive and negative ion modes. The column (ACQUITY UPLC HSS T3, 1.8 μm, 2.1 × 100 mm) was employed in positive and negative modes. The elution reagents were (A) water with 0.1% (v/v) formic acid and (B) acetonitrile with 0.1% (v/v) formic acid, and the gradient was as follows: 0 min, 5% B; 1 min, 5% B; 2.5 min, 30% B; 6 min, 50% B; 7 min, 70% B; 10 min, 80% B; 12 min, 100% B; 14 min, 100% B; 14.2 min, 5% B; and 16 min, 5% B at 0.35 mL/min and a column temperature of 40 °C. Samples were maintained at 4 °C during analysis. The injection volume was 5 μL. The mass range was detected between 100 and 1200 mass-to-charge ratio (m/z). A resolution of 70,000 was used for full MS scans, and 17,500 was used for higher-energy collisional dissociation (HCD) MS/MS scans, with a collision energy of 10, 20, and 40 eV. The mass spectrometer was operated as follows: spray voltage, 3800 V(+) and 3200 V(−); sheath gas flow rate, 40 arbitrary units; auxiliary gas flow rate, 15 arbitrary units; capillary temperature, 320 °C; auxiliary gas heater temperature, 350 °C; and S-lens RF level, 55. Every four samples, a QC sample was injected to assess repeatability. Statistical analysis. Progenesis QI V2.3 (Nonlinear Dynamics, Newcastle, UK) was used for baseline filtering, peak identification, integral retention time correction, peak alignment, and normalization of raw LC-MS www.nature.com/scientificreports/ data, with 5 ppm precursor tolerance, 10 ppm product tolerance, and 5% product ion threshold. Compound identification was based on comparing the precise m/z values, secondary fragments, and isotopic distribution with the Human Metabolome Database (HMDB) for qualitative analysis. The data were further processed by removing peaks with missing values (ion intensity = 0) in more than 50% of groups by replacing zero values with half of the minimum value and by screening according to the qualitative results of the compound. Compounds with scores below 36 (of 60) were deemed to be inaccurate and removed.
A data matrix was generated from the positive and negative ion data and used for principal component analysis (PCA) in R. Orthogonal partial least squares discriminant analysis (OPLS-DA) and partial least squares   www.nature.com/scientificreports/ discriminant analysis (PLS-DA) were used to identify the metabolites that differed between groups. Seven-fold cross-validation and 200 response permutation tests were used to prevent overfitting and evaluate the quality of the model. Variable importance of projection (VIP) values from OPLS-DA were used to rank the contribution of each variable to the discrimination of groups. A two-tailed Student's t-test was used to verify the significance of the differences in metabolite abundance between the groups. Metabolites with VIP > 1.0 and P < 0.05 were selected as differentially abundant. FSD and TPFD samples contained all the identified metabolites, but the relative contents of individual compounds were remarkably different between the two groups ( Fig. 2). However, metabolite contents were similar in the four biological replicates of an individual sample. The PCA plots between FSD and TPFD samples also showed clear differences ( Fig. 2A); for example, PC1 was clearly separated between FSD and TPFD and represented 69.7% of the difference in their chemical compositions. Figure 2B presents the PLS-DA of the two groups. The R2 value of the PCA and Q value of the PLS-DA (Fig. 2) show that compound abundances between FSD and TPFD samples were statistically significantly different. OPLS-DA, another supervised method, was used to highlight the quantitative variation in the metabolites between TPFD and FSD samples (Fig. 2C). Cross-validation with 200 permutations supported the reliability of this OPLS-DA model, with R2 and Q2 intercepts of 0.932 and 0.129, respectively (Fig. 2D). These results show that TCM processing techniques lead to significant changes to the metabolite contents of D. officinale.

Identification of metabolite diversity in
Characterization of five categories of differentially abundant metabolites. Pairwise comparisons of metabolite abundances in FSD and TPFD using the OPLS-DA model identified differentially abundant metabolites based on the VIP value. Next, all identified and annotated metabolites were screened for different abundances between FSD and TPFD samples (Fig. 3A). Using the set criteria (VIP > 1; P < 0.05), 370 metabolites were found to be significantly differentially abundant between TPFD and FSD, the majority of which were organooxygen compounds, prenol lipids, flavonoids, carboxylic acids and their derivatives, and fatty acyls (Fig. 3B).
First, organooxygen compounds were significantly more abundant in TPFD than in FSD samples. These compounds, especially carbohydrates and carbohydrate conjugates, directly contribute to the physiological activity of D. officinale products. Carbohydrates and carbohydrate aggregates from D. officinale have antioxidant,   www.nature.com/scientificreports/ anti-tumor, immune-enhancing, and anti-inflammatory effects; they protect the liver and nerves; and they are useful for the treatment of diabetes and the intestinal microbiome 22 . A total of 77 organooxygen compounds were significantly differentially abundant between FSD and TPFD samples in this study, accounting for 20.8% of all the differentially abundant metabolites. Of these, 74 compounds were carbohydrates or carbohydrate aggregates. Sixty-one carbohydrates and carbohydrate aggregates were significantly more abundant in TPFD samples, while 13 carbohydrates and carbohydrate aggregates were significantly less abundant (Table 1). In particular, isopropyl apiosylglucoside, D-erythro-D-galacto-octitol, cyclodopa glucoside, 6'-apiosyllotaustralin, and maltohexaose ( Fig. 4) were an average of 5.80 × 10 11 times more abundant in TPFD samples than FSD, likely due to the TCM processing. By contrast, 2-phospho-D-glyceric acid and 3-O-alpha-D-glucopyranuronosyl-D-xylose (Fig. 4) were an average of 3.72 × 10 −12 times less abundant in TPFD than in FSD. In addition, 44 carbohydrates and carbohydrate aggregates were at least twice as abundant in TPFD as in FSD, but only 9 compounds were half as abundant.
These results suggest that TCM processing positively affects the accumulation of carbohydrates in TPFD. Second, Prenol lipids are naturally occurring and are formed by the condensation of isoprene subunits 23 . Prenol lipids have critical roles not only as structural components of cell membranes, but also as essential signaling   26 .
Third, fruits and vegetables contain abundant quantities of flavonoids, which contribute to plant color and protect against microbial infection 27 . The properties of flavonoids depend on the arrangement of hydroxyl, methoxy, and glycosidic side groups and the conjugation between the A-and B-rings 28 . A total of 40 flavonoids, including common flavonoids such as naringin and rutin (Fig. 4), were significantly differentially abundant between TPFD and FSD samples (Table 3). After TCM processing, 26 flavonoids were significantly more abundant in TPFD than in FSD samples, while 14 flavonoids were significantly less abundant. Flavonoids possess anti-inflammatory and antioxidant activities and are considered potential therapeutic agents 28,29 . The content differences of these flavonoids therefore may influence the therapeutic characteristics of TPFD.
Finally, carboxylic acids, their derivatives, and fatty acyls are also major categories of compounds that are differentially abundant between TPFD and FSD samples. Here, 35 carboxylic acids and derivatives and 29 fatty acids were significantly differentially abundant between the samples (Tables 4 and 5). Carboxylic acids and derivatives are important substances in animal and plant metabolism and are used commercially in the synthesis of pesticides, herbicides, and insect repellents. Mounting evidence suggests that carboxylic acids and derivatives also have considerable pharmacological activities; for example, pentacyclic triterpenoid carboxylic acids have strong antioxidant, anti-inflammatory, antibacterial, anti-diabetic, and anti-tumor activities [30][31][32] . TPFD is traditionally used for the treatment of diabetes, cancer, and inflammation, among other conditions, suggesting that changes in carboxylic acids and their derivatives may determine the efficacy of TPFD. Short-chain fatty acids contribute to the flavor of D. officinale, and long-chain fatty acids can be degraded and transformed into various active flavor components through oxidation reactions 33 . The content difference of these fatty acids may therefore affect the flavor characteristics of TPFD.

Correlation analysis of biological activities.
The chemical composition of TPFD affects its efficacy. As mentioned above, 370 metabolites are significantly changed in the traditional processing of FSD into TPFD for TCM, the majority of which are organooxygen compounds, flavonoids, prenol lipids, fatty acids, and carboxylic acids and their derivatives. These metabolites have different activities and therefore may affect the efficacy of TPFD; therefore, future research should examine these 370 differentially abundant metabolites as potential chemical markers.
To provide visual evidence of the distinct nature of TPFD samples, the above OPLS-DA models were used to construct an S-plot and loading analysis ( Fig. 5A and B), which provided a graphical projection of specific compounds. In these plots, metabolites close to the origin make a small contribution to the separation of the samples. A total of 43 metabolites (Fig. 5C) had a VIP score ≥ 4.0 in the OPLS-DA model, and a t-test revealed that they significantly differed (P < 0.05) between FSD and TPFD. In the S-plot and loading analysis, these compounds were farthest from the origin (in the positive and negative directions), indicating that they make a greater contribution to the distinction between samples. These 43 metabolites (listed in Table 6 with their activities) could therefore be used as chemical markers to assess whether the biological activity of D. officinale is altered through traditional processing. www.nature.com/scientificreports/ Of the 43 chemical markers, 29 were more abundant in TPFD than in FSD samples, while 14 were less abundant. Their medicinal properties include anti-inflammatory, anti-mutagenic, analgesic, neuroprotection and anti-Alzheimer's, anti-tumor, antibacterial, anti-toxicity, antioxidant, anti-nociceptive, anti-hypertension, anti-diabetic, anti-depressant, lipase-inhibiting, immune-enhancing, cis-diaminedichloroplatinum nephrotoxicity-preventing, cytoprotective, Fanconi syndrome-attenuating, cardiotoxicity-preventing, anti-fatigue, and anti-tyrosinase activities. Anti-inflammatory activity is the most common function of the significantly upregulated metabolites (Table 6). Among the 43 chemical markers, [6]-dehydroshogaol and capsaicin (Fig. 4) showed the greatest difference in abundance between TPFD and FSD samples, with log2 fold-change values of 13.16 and 11.88, respectively. Imm et al. established that capsaicin and [6]-dehydroshogaol inhibited the production of nitric oxide (NO) in LPS-stimulated cells in a dose-dependent manner 34 . These chemicals are also likely to have anti-inflammatory and antioxidant effects by inactivating the eukaryotic transcription factor NF-kB 35,36 . www.nature.com/scientificreports/ Furthermore, the log2 fold-change values of the anti-inflammatory compounds N2-(3-hydroxysuccinoyl)arginine, naringenin, citronellyl beta-sophoroside, methyl beta-D-glucopyranoside, and hydroxysafflor yellow A (Fig. 4) were > 3 (Table 6). Thus, TPFD has better anti-inflammatory properties than FSD, which is beneficial for its applications in TCM. In addition, the 43 marker metabolites included compounds with anti-tumor and anti-diabetic activities. Arlatin, naringenin, and methyl beta-D-glucopyranoside (Fig. 4) showed significant anti-tumor activity, while citroside A (Fig. 4) possesses significant anti-diabetic activity. The contents of all these compounds were significantly higher in TPFD than in FSD samples. These results are consistent with the reported effects of TPFD in TCM, providing scientific evidence of the efficacy of the traditional application of D. officinale.
We also detected some compounds in D. officinale with potential therapeutic effects on neurological diseases; for example, norcapsaicin (Fig. 4) has neuroprotection and anti-Alzheimer's activities, while osmanthuside B (Fig. 4) shows anti-depressant activity. Future studies should explore the relationship between the concentrations of these compounds present and the therapeutic effects and health-promoting properties of D. officinale products.

Discussion
The significant differences on metabolites between FSD and TPFD. We integrated UHPLC coupled with Q Exactive plus quadrupole-Orbitrap MS in the positive and negative ion modes, combined with a HMDB and multivariate statistical analysis for qualitative analyses, to screen the different constituents of FSD and TPFD. The result of this study reveals that five type compounds including organooxygen compounds, prenol lipids, flavonoids, carboxylic acids and their derivatives, and fatty acyls show the significant differences among 370 differential metabolites which were unambiguously detected or tentatively identified. FSD and TPFD sam- Table 4. Information of carboxylic acids and derivatives with significant changes.

Metabolites
Compound ID m/z Retention time (min) VIP P-value log2(FC) www.nature.com/scientificreports/ ples contained all the identified metabolites, there is a large amount of metabolites data available on FSD and TPFD under different processing. But the relative contents of individual compounds were remarkably different between the two groups.
The possible mechanism of substance and its bioactivities. Chemical structures of typical compounds and related bioactive in TPFD have been identified by OPLS-DA methods, For example, 43 metabolites were identified as chemical markers that could be used to distinguish FSD and TPFD samples according to qualitative research, and the contents of some compounds with anti-inflammatory and antioxidant, such as [6]-dehydroshogaol, capsaicin, arlatin, naringenin, methyl beta-D-glucopyrano-side, and citroside A, significantly increased after processing according to quantitative research. This finding shows that traditional processing could possibly changed the contents of partial active compounds and improved the efficiency of D. officinale.
In order to promote further research into FSD and TPFD, attentions should be paid to the following work in the future: (1) Although current metabolites technologies have been used to study TPFD, there are limitedly comprehensive metabonomics studies. Obviously, the negative or positive ion model method such as metabolomics cannot satisfy the necessary deep research into D. officinale 21 . (2) it is important to comprehensively investigate why differences exist in active compounds between FSD and TPFD which can provide propitiate conditions for metabolite accumulation. (3) its TPFD applications has been rarely described, and TPFD improvements are still required for its industrial applications. (4) the functions of most these compositions need to confirm through functional investigation.

Conclusion
For thousands of years, D. officinale has been processed to enhance its medicinal value for use in TCM. The chemical composition of this material after processing is key to its efficacy. Traditional processing of D. officinale produces a difference in the contents of key metabolites. Moreover, combining metabolomics and multivariate statistical analysis methods can accurately identify markers that differentiate processed and raw materials. Thus, we revealed the basis of the improvement in efficacy of TPFD compared with FSD, enabling the identification of active substances with functions not included in the traditional use of TPFD. These results indicate the need for the further exploration of TPFD in the treatment of additional, previously untested conditions. This systematic study of FSD and TPFD provides a useful analytical strategy for rapidly screening and identifying the constituents of other TCMs and TCM formulas. In addition, the results of this research provide a theoretical basis for quality control.    Figure 6. Chemical constructure of 15 significant increased compounds.