Putative Genes Identi ed in Alkaloids Biosynthesis in Dendrobium o cinale by Correlating the Contents of Main Bioactive Metabolites with Genes Expression between Protocorm-like Bodies and Leaves

Zhaojian Wang Anhui University of Traditional Chinese Medicine Shihai Xing (  xshshihai@163.com ) Anhui University of Traditional Chinese Medicine Weimin Jiang Hengyang Normal University Yingying Liu Anhui University of Traditional Chinese Medicine Xiaoxi Meng University of Minnesota Xinglong Su Anhui University of Traditional Chinese Medicine Mengyang Cao Anhui University of Traditional Chinese Medicine Liping Wu Anhui University of Traditional Chinese Medicine Nianjun Yu Anhui University of Traditional Chinese Medicine Daiyin Peng Anhui University of Traditional Chinese Medicine


Background
Dendrobium o cinale Kimura et Migo is one of the most famous species in the genus Dendrobium, which is a precious perennial epiphytic herb in China and other Asian countries [1]. Stem of D. o cinale has a long history as a Traditional Chinese Medicine (TCM) since Tang dynasty about 1,300 years ago with effects on tonifying stomach and nourishing uid, clearing heat and nourishing "yin", and improving immunological function [2]. Furthermore, D. o cinale plays an important role in treating atrophic gastritis, diabetes, cancer, cardiovascular, disease-cataract and delaying aging [3]. Its bioactive components include major D. o cinale polysaccharides (DOPs) [4] and alkaloids [5], and minimal avonoids [6]. DOPs have many pharmacological activities such as curving hypoglycemia [7], antioxidation [8], immunomodulation [9] and anti-tumor. Alkaloids of D. o cinale exhibit a variety of pharmacological activities such as antioxidant, anticancer, and neuroprotective activities [10], and avonoids of D. o cinale possess good functions on anti-cytotoxicity and antioxidant [11].
Polysaccharides are major medicinal components in D. o cinale [12]. The biosynthesis pathway of polysaccharides in medicinal plants can be divided into three main stages [13,14]. At the rst stage, sucrose produced by through chloroplast photosynthesis in leaves is converted into two kind of monosaccharides such as glucose 1-phosphate (Glc-1P) and fructose-6-phosphate (Fru-6P) under a series of enzymes. In the second stage, these two monosaccharides are extended into various nucleotidediphospho-sugars (NDP-sugars, and NDP includes UDP and GDP) by enzymatic reactions. Finally, all of the NDP-sugars are extended into macromolecular polysaccharides by successive catalysis of glycosyltransferases (GTs) [15]. Searches against KEGG database enabled prediction of the biosynthesis pathway of DOPs as shown in Additional le 1 [16,17]. The conversion of NDP-sugars into many types of DOPs in Golgi Complex was catalyzed by different kinds of glycosyltransferases such as glucosyltransferases, fucosyltransferases, mannosytransferases and xylosyltransferases in D. o cinale [12,18]. Water-soluble polysaccharides, including major mannose and minimal glucose were con rmed as the dominant DOPs in D. o cinale stems according to the previous study [19]. Mannan polysaccharide was under the action of mannan synthases which were encoded by Cellulose synthase-like A (CslA) family members [20,21] (Additional le 1).
Alkaloid was another type of major medicinal components in D. o cinale, most of which pertain to terpenoid indole alkaloids (TIAs) [22]. Strictosidine backbone was formed in a relatively conserved upstream biosynthetic pathway of TIAs in plants as the precursor for various types of speci c downstream alkaloids [23]. Then, strictosidine is catalyzed by enzymes to form enormous TIAs of D. o cinale [24].
In Dendrobium, there might be a series of P450 monooxygenases superfamily and aminotransferases following strictosidine in the downstream biosynthetic pathway of TIAs [25]. Cytochrome P450s are involved in the oxidation and hydroxylation reaction, and aminotransferases transfer amino acids to form alkaloids which is the nitrogenous amino acid derivatives [26]. Study had found that polyneuridinealdehyde esterase (EC: 3.1.1.78) gene (PNAE) participated in 16-Epivellosimine biosynthesis in D. o cinale [12]. Based on KEGG database and chemical structure of the compounds we propose a putative alkaloids biosynthesis pathway of D. o cinale (Additional le 2) [18,22,27]. Among the putative downstream speci c alkaloid biosynthetic pathway of D. o cinale, strictosidine is converted to strictosidine aglycone catalyzed by a speci c β-D-Glucosidase (EC3.2.1.21) (SG), and the conversion of strictosidine aglycone to 4,21-dehydro-geissoschizine, which can be reciprocally transformed to geissoschizine by geissoschizine synthase (GS), is catalyzed by successive unknown enzymes. Vinorine might be one of alkaloids predicted by us in D. o cinale, which is synthesized under the catalysis of vinorine synthase (VS) from 16-Epivellosimine.
Flavonoids are all originated from two intermediates (L-tyrosine and L-phenylalanine) produced by the shikimate pathway in plants, and these intermediates can generate p-coumarinyl-CoA via p-cinnamic acid and cinnamoyl-CoA by enzymes catalysis, respectively [28]. Then, different types of avonoids, such as avone, avanone, avonol and anthocyanin are produced in plants by various types of enzymatic reactions from p-coumaryl-CoA [29]. Among these avonoids, naringenin is an important avonoid which has much clearly effects on human body, such as inhibiting the in ammatory response of many cell types [30], treating hypotensive [31], anti-brotic and anti-cancer, as well as the effect on hepatoprotective [32]. Naringin is generated by chalcone isomerase (CHI) through a ring closure reaction from naringin chalcone, which is condensed from three molecules of malonyl-CoA and one molecule of p-coumaryl-CoA by chalcone synthase (CHS) [33]. The putative biosynthesis pathway of avonoids in D. o cinale was shown in Additional le 3 [34].
D. o cinale has already been endangered because of overexploiting and habitat destruction since its high medicinal value. More importantly, the wild D. o cinale is now on the verge of extinction and on the IUCN Red List of Threatened Species [34]. Therefore, it is imperative to conserve wild resources of D. o cinale and seek alternative ways to propagate it. Tissue culture technology may be an available method for D. o cinale breeding and the induction of protocorms from embrogenic tissues and protocorm-like bodies (PLBs) from non-embryogenic tissues culture in D. o cinale has been studied [35]. It was found the content of the main bioactive compound content of D. o cinale in its PLBs varied with the medium formulas and the effective elicitors as additives [36].
In this study, we aimed to establish a system for protocorms and PLBs induction; proliferation for D. o cinale in solid medium; and compare the contents of main metabolites such as polysaccharides, alkaloids, avonoids and naringenin among PLBs and organs of the original plant. Signi cant differences in contents of these metabolites were revealed between PLBs and organs of plant, suggesting that the PLB and leaf were good experimental materials for analyzing the expression levels of candidate enzyme-encoding genes involved in the biosynthetic pathways of these speci c metabolites in D. o cinale. Based on the differential expression of genes (DEGs) from transcriptome sequencing, the putative genes encoding enzymes involved in biosynthetic pathways of the main metabolites have be identi ed or veri ed. This study will help to conserve D. o cinale resources and deeply analyze the biosynthesis and regulation of main metabolites of D. o cinale.

Protocorms and PLBs Induction of D. o cinale
The optimal explants for PLBs induction and the optimum formula for PLBs propagation were screened in order to construct a comprehensive system for PLBs proliferation and protocorms/PLBs induction of D. o cinale. Five different explants were selected to induce protocorms or PLBs. Induction rate and the growth status of protocorms or PLBs determined the best explants. The results showed that seed is most suitable for inducing protocorms or PLBs, and capsules approaching maturity were collected and punctured to release seeds in sterile condition because of the tiny seeds (Table 1). An L 16 (4 5 ) orthogonal experiment was performed for screening the best protocol for PLBs propagation.
The results showed that the best formula for PLBs propagation of D. o cinale was A3B3C2D1(2)E2, and factor C (6-BA content) is the most important factor because of its highest R value. A veri cation experiment with formula A3B3C2D2E2 was conducted with 40 duplicates, and the propagation coe cient is up to 4.37 ( Table 2, Fig. 1a). Based on what is stated above, the optimal formula for PLBs propagation  Polysaccharides, avonoids and alkaloids were determined in various organs of D. o cinale, including the whole plant, root, stem, leaf, and PLB (Fig. 1b). Signi cant differences in contents of these three components among samples were detected (Fig. 2).

Differential Contents of Total Polysaccharides (DOPs)
The regression equations of the D-glucose standard curve obtained by UAE and HWE methods of polysaccharides are Y = 0.0491X -0.0035 (R 2 = 0.9997) and Y = 0.0684X − 0.0323 (R 2 = 0.9998), respectively (Additional le 4), with determination wavelength at 488 nm. The contents of polysaccharides in PLBs and organs measured by two different extraction methods had the same content distribution of polysaccharides, but the extraction rate of HWE method was higher than that of the UAE method ( Fig. 2a and b). It can be concluded that the HWE method is better for DOPs extraction.
Different distributions of DOPs in the organs of plant are as follows: stem > the whole plant > leaf > root.
The stem of D. o cinale has the highest content of polysaccharides in all organs at 402.93 mg·g − 1 (Dried weight). While, PLBs has the lowest content of polysaccharides than organs with average amount of 75.30 mg·g − 1 (Fig. 3).

Differential Contents of Total Alkaloids
The equation of standard curve with dendrobine as reference is Y = 0.1154X-0.0048 (R 2 = 0.9990) with the absorbance detected at 620 nm (Additional le 4). Signi cant difference of total alkaloids content among different parts of D. o cinale was not observed (all around 0.3 mg·g − 1 (dried weight)). While, the total alkaloids contents in the PLBs is up to 0.56 mg·g − 1 (dried weight), which is almost twice as high as that of D. o cinale plant organs (Fig. 2c). PLBs can be used as alternative materials for dendrobine extraction.

Differential Contents of Total Flavonoids and Naringenin
The regression equation of the standard curve for total avonoids measured with rutin as the reference is Y = 0.0114X-0.0009 with R 2 = 0.9992, and the absorbance is determined at 510 nm. The results showed that the content of total avonoid in the stem of D. o cinale was the lowest with only 4.07 mg·g − 1 , which was consistent with most ndings reported previously [34]. While, leaf had highest content of total avonoids among organs at 11.74 mg·g − 1 followed by root (Additional le 4). The content of total avonoids in PLBs is higher than that in the whole plant and stem, while lower than that in the leaf and root (Fig. 2d). Considering the lower biomass of leaf and root in D. o cinale, PLBs would be the most suitable materials for extracting total avonoids of D. o cinale.
Naringenin, an important avonoid, had been determinate by HPLC-UV and the chromatogram showed the retention time (RT) of naringenin standard was at 44.8 min (Fig. 3a). The regression equation of the standard curve for naringenin with its standard is Y = 2624X + 0.6714, R 2 = 0.9999 (Fig. 3b). The result showed that naringenin could only be quanti ed only in the stem of D. o cinale, which is consistent with the previous study [37], and the average content is about 0.029 mg·g − 1 . And no naringenin could be quanti ed in the PLB of D. o cinale, which indicates PLB can't be an alternative material used for naringenin extraction (Fig. 3c). Based on similarity in naringenin content between the whole plant about (0.021 mg·g − 1 ) and the stem (about 0.029 mg·g − 1 ), we could conclude that naringenin derives mainly from the stem. We also stablished that the biomass of D. o cinale was mainly from stem (Additional le 4).

Sequencing, Assembly and Unigenes Functional Annotation
Totally, 39.11 Gb transcriptomic data was generated by the platform of BGISEQ-500. Clean reads were assembled by Trinity software, and then cluster the transcripts to remove redundancy and obtain unigenes by Tgicl. Finally, 157, 901 unigenes were obtained (Additional le 5 Among these DEGs, 39, 597 unigenes were up-regulated and 31, 710 unigenes were down-regulated in PLB (Fig. 4a). Based on a Poisson distribution, 27, 242 unigenes were speci cally expressed in leaves, 11, 976 unigenes were expressed only in PLB, and 98, 096 shared unigenes were identi ed in both of these two samples (Fig. 4b, Additional le 7). The large number of identi ed DEGs exhibited signi cant difference between leaves and PLBs in various aspects, which directly re ected from the critical difference on appearance.

Differential Expression of Genes Involved in Polysaccharides Biosynthesis Pathway
From above, polysaccharide is more abundant in leaves, with much lower abundance in PLBs.

Differences in expression of genes encoding enzymes involved in polysaccharides biosynthesis between
PLBs and leaves were further revealed (Fig. 5). Some of genes were expressed much higher in leaves such as genes encoding INV, CslA, FBPase, PFP, MPI, scrK, KK, USP, UGPA, PMM, GMPP and UXS1 etc. The higher level of gene expression in leaves is positively correlated with the higher content of polysaccharide content in the leaves. Compared to PLBs, INV gene was highly expressed and SuS was mildly increased in leaves, which indicated that more sucrose was necessarily required to synthesize more downstream polysaccharides. The underexpression of gene encoding CslA was positively correlated with the lower polysaccharides (mainly composed by mannose monosaccharides) content of in PLB (Fig. 5a).
Meanwhile, there was no signi cant difference in the expression of some genes encoding enzymes such as PGM, HK, TK, UXE, PFK, TSTA3 and UER1 between leaf and PLB, suggested the catalysis reactions by these enzymes are not the restrictive steps for DOPs biosynthesis. Genes encoding enzymes such as UAE, UGD, XK, AR and GMDS are negatively correlated with total DOPs content, and the enzymes are all bifunctional and in closed loop pathway (Fig. 5a). This might indicate that these bifunctional enzymes have greater degradation ability than their biosynthesis effect in polysaccharides biosynthesis.
Glycosyltransferases (GTs) are a very widespread group of carbohydrate-active related enzymes participating in glycan and glycoside biosynthesis in higher plants [38]. Totally, 553 glycosyltransferase genes, including 305 glucosyltransferase genes, 29 fucosyltransferase genes, 110 mannosyltransferase genes and 106 xylosyltransferase genes, were identi ed using BLASTX in the transcriptomes. From which, 22 full-length glucosyltransferases genes, 5 fucosyltransferase genes, 9 mannosyltransferase genes and 8 xylosyltransferase genes were isolated. The expression levels of differential gene encoding enzymes between leaf and PL, such as CGT, UGT76C1_2, SGT1, ALG11, were in line with the expression pattern of DOPs, suggesting those could be candidate genes encoding enzymes involved in DOPs biosynthesis in D. o cinale (Fig. 5b, Additional le 9).

Analysis of Differential Genes Expression in Alkaloids Biosynthesis
Previous study showed alkaloids in D. o cinale are mainly terpenoid indole alkaloids (TIAs). The difference in expression of genes encoding enzymes in upstream of MVA, MEP and Shikimate pathway between leaves and PLBs was analyzed. The result showed signi cant differences between leaves and PLBs, but no consistent regulatory patterns were detected among these genes (Fig. 6a). Disorder of genes expression indicates that the intermediates in upstream of TIAs biosynthesis are not the restrictive products, which was further veri ed by the expression of two genes encoding LAMT and TDC involved in synthesizing loganin and tryptamine, respectively. Total alkaloids content is higher in PLBs than in leaves, but the expression level of genes encoding enzymes LAMT and TDC was the opposite of alkaloids content. There's no signi cant difference in expression of gene encoding SCS which catalyzed the conversion of loganin into secologanin between the PLB and leaf. From what is stated above, we can conclude that most intermediates, especially loganin and secologanin, as direct substates for strictosidine, the precursor of TIAs are su cient for downstream speci c alkaloids biosynthesis. The expression quantity of gene encoding enzyme STR is much higher in PLBs than in leaves, which means higher content of precursor strictosidine is necessary for producing higher content of total alkaloids.
For the downstream biosynthesis pathway of speci c alkaloids in D. o cinale, the expression level of putative genes encoding enzymes SG (Strictosidine beta-glucosidase), GS (Geissoschizine synthase) and PNAE (Polyneuridine-aldehyde esterase) are much higher in PLBs than in leaves, and the expression level of gene encoding VS (Vinorine synthase) has a small increase in PLB (Fig. 6a). Genes encoding SG, GS and VS are rst predicted in D. o cinale alkaloids biosynthesis. Moreover, these predicted genes involved in TIAs biosynthesis in D. o cinale indicates total alkaloids of D. o cinale might include strictosidine aglycone, geissoschizine and vinorine.
Some of aminotransferases and P450s superfamily enzymes are necessary for alkaloids biosynthesis.
By comparing the expression level between PLBs and leaves, we found some of aminotransferases and P450s were higher expressed in PLBs than in leaves (Fig. 6b, c, Additional le 9). We speculate that some highly expressed aminotransferases and P450s might participate in alkaloids biosynthesis in D. o cinale.

Differential Expression of Genes in Flavones Biosynthesis Pathway
Expression quantity of predicted genes encoding enzymes in avones biosynthesis between PLB and leaf was measured by the value of diffexp_log2fc_Cas-vs -Ctr (Cas is PLB, Ctr is leaf) of high amino acid sequence identity unigenes from RNA-seq database. There's signi cant difference in gene expression between PLBs and leaves (Fig. 7). TAL and PAL were higher expressed in leaves than in PLBs, which accorded with the higher content of avonoids in leaves. ADT and PDT were expressed much more than

Detection of Transcription Factor Families
Transcriptional factors (TFs) play an important role in the regulation of secondary metabolites by coordinating the expression of biosynthetic genes and controlling the production of secondary metabolites in plants [39]. In this study, by comparing to the NCBI database, a total of 1,113 unique putative transcripts encoding TFs were identi ed as differentially expressed genes, and these TFs belong to 51 known TF families (Additional le 10). The major types of these differentially expressed TFs are MYB, AP2/EREBP, bHLH and NAC etc. The number of upregulated and downregulated TFs from each type was shown in Fig. 8, and a striking number of differentially expressed TFs were identi ed in between PLBs and leaves.

Discussion
It had been reported that the extraction method has little effect on the types of monosaccharide, but in uences the monosaccharide composition of heterogeneous polysaccharides [40]. By using the HWE method, the diffusion rate of polysaccharides could increase dramatically due to the raised temperature of water, hence the improvement of extraction e ciency [41]. Our result indicates that the HWE method is more suitable for DOPs extraction, which is consistent with the above conclusion.
Among all organs of D. o cinale, the stem has the highest content of polysaccharides, which is consistent with the conclusion from previous researchers [22]. While, PLBs has the lowest content of polysaccharides, which indicates that the PLBs couldn't substitute for the plant or organs of D. o cinale in production of DOPs.
In spite of the relatively low content of total alkaloids in D. o cinale, more and more attention has been paid to their broad pharmacological activities, such as antipyretic, analgesic, and anti-tumor activities [12]. PLBs were much more suitable for producing alkaloids in our study, which provided a new idea for the future large-scale production of alkaloids in D. o cinale.
KEGG enrichment analysis indicated that comparative RNA-Seq between leaves and PLBs was consistent with the signi cant difference in content of DOPs, alkaloids and avonoids between PLBs and leaves of D. o cinale. It also demonstrates our reasonable prediction of the putative biosynthesis pathway of DOPs, alkaloids and avonoids in D. o cinale.
The characterization of the correlation between the content of DOPs, alkaloids and avonoids and comparative transcriptome sequence results in discovering many putative genes encoding enzymes involved in biosynthesis of DOPs, alkaloids and avonoids in D. o cinale. Next, we are going to work on the clone of the genes encoding SG, GS, PNAE and VS from D. o cinale, and the construction of expression vectors to make the proteins expressed. The experiment on enzyme activities in vitro will be carried out to validate their function, and the substrate speci city of these putative enzymes and their subcellular localization also need to be studied.

Previous reports had indicated that TFs played important roles in regulating alkaloids biosynthesis, such
as TFs of apetala2/ethylene response factors (AP2/ERFs) [42,43], C2H2 zinc ngers [44], WRKYs [45] and basic helix-loop-helix (bHLH) [46] that are involved in regulating TIAs biosynthesis. ORCA3, an AP2/ERF TF, induces the expression of a number of TIAs biosynthetic genes such as G10H, CPR, STR, AS, TDC, and DXS in the vindoline or tryptamine branch pathway [47,48]. CrMYC belongs to bHLH family, activates ORCA3, and then induces the expression of several TIAs biosynthetic genes, such as TDC and CPR [49]. CrBPF1 is a MYB transcription factor that regulates the expression of STR in C. roseus [50]. And in our study, some TFs such as MYB, AP2/EREBP, bHLH and NAC were found involved in alkaloids biosynthesis in D. o cinale.

Plant Materials and Reagents
Plants of D. o cinale were grown at 25℃ during day and 23℃ at night with 60-70% relative humidity and a light/dark cycle of 14/10 h in the herbal garden of Anhui University of Chinese Medicine, Hefei, China.

Protocorms and PLBs of D. o cinale Induction and Propagation
Capsules approaching mature, leaves, stem tips, stem fragments, and stems with nodes of D. o cinale were collected, and sterilized to induce protocorms and PLBs (Capsules were punctured to release seeds as explants on medium under sterile condition), each explant was performed more than 30 duplicates.
They were inoculated in 1/2 MS medium + 2, 4-D 0.5 mg·L − 1 + 30 g·L − 1 sucrose + 7 g·L − 1 agar at pH 5.6 to 5.8, and placed in the tissue culture rooms at 70% relative humidity, 1600 lx illumination and a light/dark of 16/8 h for 60 d, to screen the optimal explants for protocorms and PLBs induction. combination. It was found that the materials in tissue culture with natural additives and fruit juice grow much better than those without natural additives and fruit juice [52,53], in the PLBs proliferation processing potato juice was selected as additive to accelerate PLB growth and propagation. 200 g fresh potato was chopped into small pieces, added was mixed with puri ed water to 1 L and boiled to a mush. The mushy potato was passed through the gauze to collect the juice as additive. The concentration gradients of potato juice (Factor E) were 50 g·L − 1 , 100 g·L − 1 , 150 g·L − 1 and 200 g·L − 1 . All of the full test media with pH 5.6 ~ 5.8 for PLBs proliferation were included basic medium, PRGs combination, potato juice, sucrose (30 g·L − 1 ) and agar (7 g·L − 1 ), and the design of orthogonal experiment table head was showed in Additional le 11. Each group has 10 bottles with 4 pieces, and the culture condition was light/dark for 16/8 h per day, the illuminance and the culture temperature were at 1600 lx and (25 ± 1) °C.
Observations and statistics were carried out to analyze the propagation coe cient two months later since inoculation. The formula of propagation coe cient was calculated as follow: Propagation coe cient (%) = (FW 1 -FW 0 )/ FW 0 × 100%, FW0 and FW1 mean the fresh weight before inoculation and after culture, respectively.

Ultrasound-Assisted Extraction (UAE) or Hot Water Extraction (HWE) of Total Polysaccharides
Stems, leaves, roots and the whole plants from about 3 years old D. o cinale and PLBs were harvested, cleaned, and dried in an oven at 50 ℃ to a constant weight. Samples were ground to a ne powder using the pulverizer and sieve the powder through a 40 meshes sieve. The water-soluble polysaccharides in D. o cinale were extracted by UAE method [41] and HWE method [8] with a few modi cation and nally better one was selected for the extraction of polysaccharides in D o cinale from these two methods.
The modi ed UAE method was as follows: 10 mL of double distilled water was added into each 0.2 g pulverous sample, and the samples were homogenized in the ultrasonic cleaner at 40 ℃ ,40 Hz for 0.5 h, then ltered through lter-paper and the ltrate was obtained. Repeat above steps one more time and collect all ltrate. The ltrate was concentrated at 55℃ until its volume was down to 10 mL. 40 mL of ethanol was added into the concentrated ltrate, centrifuged at 4,000 rpm for 5 min, and discarded the supernatant. The samples were dissolved in 25 mL of pure water. 25 mL of savage reagent was added to remove impurities such as protein and nucleic acid, centrifuged at 1000 rpm for 5 min. 20 mL of the supernatant was transferred into a 50 mL volumetric ask and mixed with pure water to a constant volume of 50 mL for polysaccharides extraction use. The method HWE with a few modi cations was as follows: rstly, 0.3 g samples were placed in a round-bottomed ask, and add 200 mL of water. Then the sample solution was heated and re uxed for 3 h. Let the re ux extract cool off to room temperature before ltering it. The ltrate was transferred into a 250 mL volumetric ask, and mixed with double distilled water to a constant volume of 250 mL. 2 mL of sample solution was taken into a 15 mL centrifugal tube and 10 mL of solute ethanol was added, shake it and refrigerate it for 1 h followed by centrifugation at 1000 rpm for 20 min. Discard the supernatant and wash the precipitate twice with 8 mL of 80% ethanol. The nal precipitate was dissolved in heated water and transferred to a 25 mL volumetric ask. Let the solution cool off before mixing it with pure water to a constant volume of 25 mL for polysaccharides extraction use. Thus, the polysaccharide extracts by the two methods have been prepared, respectively.

Determination of Total Polysaccharides
1 mL of each polysaccharide extract was transferred into a 10 mL test tube, then add 1 mL of 5% phenol and vortex quickly. The solution was mixed thoroughly and 5 mL of concentrated sulfuric acid was added, shaken, and placed in a water bath at 100℃ for 20 min. Then the solution was placed in ice bath to cool for 5 min. The absorbance of the sample solution was measured at wavelength of 488 nm using ultraviolet visible spectrophotometer with 1 mL of water as a blank and the parallel detection was conducted three times. The standard curve was prepared from the D-glucose reference. Each sample was assayed in three times and the content showed as the weight of polysaccharides to the dried weight of materials.

Extraction and Determination of Total Alkaloids
The sample powder (0.5 g) was put into a 125 mL ground-mouth ask and mixed with 30 mL of petroleum ether. The sample mixture was placed in a constant temperature water bath (35℃) to degrease for 30 min, then remove the supernatant and evaporate the petroleum ether. The pH value was adjusted with a proper amount of ammonium hydroxide and add 10.0 mL of chloroform. Then the solution was re uxed for 2 h in a water bath at 80 ℃. Let it cool for 20 min in room temperature before ltering. The ltrate was obtained as total alkaloids solution. 5 mL of total alkaloids solution was mixed with 5 mL of chloroform, then 5.0 mL of pH = 4.5 buffer and 2.0 mL of 0.04% bromocresol green solution were added sequentially. Shake the mixture vigorously for 3 min and let it stand for 30 min before ltering it. And 5.0 mL of ltrate was mixed with 1.0 mL of 0.01 M NaOH-ethanol solution and vortex it. The absorbance of the samples was measured at 620 nm using UVvisible spectrophotometer with chloroform as blank. The standard curve was prepared from the Dendrobine as reference (> 98% purity) bought from Chengdu Push Bio-technology CO., Ltd. (0.857, 1.714, 2.571, 3.429, 4.286, and 5.143 µg·mL − 1 ). The parallel detection was carried out three times and the content showed as the weight of total alkaloids to the dried weight of materials.

Extraction and Determination of Total Flavonoids
The samples powder (1.00 g) was dissolved in 50 mL of 70% ethanol and re uxed at 60℃ for 2 h, then ltered. The ltrate was mixed with 70% ethanol again to a constant volume of 50 mL as total avonoids extracts.
1 mL of total avonoids extract was mixed with 5 mL of 70% ethanol. Then add 1 mL of 5% NaNO 2 and mix it well. Stand for 6 min and add 1 mL of 10% Al(NO 3 ) 3 and mixi it well. Let it stand for 6 min before adding 10 mL of 1 M NaOH and mix it with 70% ethanol to 25 mL, and stand for 15 min. The absorbance of the sample solution was measured at 510 nm using ultraviolet visible spectrophotometer. The standard curve was prepared from Rutin as reference (2.208, 4.416, 6.624, 8.832, and 11.040 µg·mL − 1 ). Each sample was assayed in three times and the content showed as the weight of total avonoids to the dried weight of materials.

Extraction and Quanti cation of Naringenin
Naringenin was extracted by the optimized methods: Solvent, methanol (20 mL) mixed with 20% hydrochloric acid (5 mL); Particle size of dry powder of D. o cinale < 0.355 mm; temperature, 80℃; Condensation re ux extraction time, 90 min; 20 mL of supernatant was evaporated using the rotating evaporator until it was dry. 5 mL of methanol was added to dissolve naringenin and the solution was ltered through a 0.22 mm Nylon membrane lter before HPLC analysis.
HPLC analysis was performed on an Agilent HPLC system, including quaternary solvent management, sampler manager, separation system, detection systems. An Agilent C 18 column (4.6 mm × 250 mm, Naringenin stock solution was prepared and diluted to an appropriate concentration for the preparation of calibration curve. The calibration curve was prepared according to the linear plots of the naringenin concentration versus the corresponding chromato-graphic peak area.

Total RNA Extraction, Construction of cDNA Libraries and RNA-Seq
Ethanol precipitation protocol and CTAB-PBIOZOL reagent was used to purify total RNA from PLBs and leaves according to the manual instructions, and each sample has three biological duplicates. 80 mg tissue samples were ground into powder in liquid nitrogen and the powdered samples were transferred into 1.5 mL of preheated 65℃ CTAB-pBIOZOL reagents. The sample solution was incubated by thermostatic mixer at 65 °C for 15 min to completely dissociate the nucleoprotein complexes. The supernatant was obtained by centrifuging the solution at 12000 × g, at 4℃ for 5 min. The supernatant was mixed with 400 uL of chloroform (per 1.5 mL of added CTAB-pBIOZOL reagent), and the sample mixture was centrifuged at 12000 × g, at 4˚C for 10 min. The supernatant was transferred to a new 2 mL tube and added 700 µL of acidic phenol and 200 µL of chloroform into it followed by centrifugation at 12000 × g, at 4˚C for 10 min. The aqueous supernatant was collected and an equal volume of chloroform was added into it followed by centrifugation at 12000 × g, at 4℃ for 10 min. An equal volume of isopropyl alcohol was added into the supernatant and the mixture was placed at -20℃ for 2 h to precipitate before centrifuging it at 12000 × g, at 4℃ for 20 min, and the supernatant was removed. The RNA pellet was washed with 1 mL of 75% ethanol before being air-dried in a biosafety cabinet and dissolved in 50 uL of DEPC-treated water. Subsequently, quali cation and quanti cation of total RNA were performed using a Nano Drop and Agilent 2100 bioanalyzer (Thermo Fisher Scienti c, MA, USA).
Oligo(dT)-attached magnetic beads were used to purify mRNA. The puri ed mRNA was split up into small pieces with fragment buffer at an appropriate temperature. The random hexamer-primed reverse transcription was performed to synthesize the rst-strand cDNA followed by the second-strand cDNA synthesis. Then, A-Tail Mix and RNA Index Adapter were added to end the repair. The cDNA fragments obtained from above mentioned steps were ampli ed by PCR. The products were puri ed by Ampure XP Beads and dissolved in EB solution. The product was veri ed for quality control on an Agilent 2100 bioanalyzer. The double-stranded PCR products were denatured and cycled by the splint oligo-nucleotide sequence to obtain the nal library, and the single-stranded circular DNA (ssCir DNA) was formatted into the nal library. The established library was ampli ed with phi29 to obtain DNA nanoball (DNB), each molecular of which had over 300 copies. The DNBs were added into the patterned nanoarray and 100 base pairs of reads generated on the BGIseq500 platform.

Functional Annotation of Unigenes by Reference Genome
The original sequencing data (original polymerase reads) produced from raw data by Paci c Bioscience Sequel were processed by the SMRT analysis package version 2.3.0 according to the IsoSeq protocol (Paci c Biosciences, https://www.pacb.com/products-and-services/analytical-software/smrt-analysis).
ROIs (reads of insert) were generated by original Polymerase reads, which had full passes > 0 and the predicted consensus accuracy > 0.75. According to whether the primers are 5'primers, 3'primers or poly-A tails, ROIs with a minimum length of 300 bp were divided into non-full-length and full-length transcribed sequences. The full-length sequences were processed to De novo consensus isoforms by ICE (Iterative Clustering for Error Correction) algorithm and then polished by Quvier quality-aware algorithm. The De novo consensus isoforms of high quality (the expected Quiver accuracy ≥ 0.95) from each library were combined and then rid redundancy by using CD-HIT [54] founded on the sequence similarity to get nal unique full-length isoforms.
The nal complete subtype was mapped to SwissProt (manually annotated and reviewed protein sequence database), NR (NCBI non-redundant protein sequence), KEGG (Kyoto Encyclopedia of Genes and Genomes), NT (NCBI non-redundant nucleotide sequence) and the KOG (Clusters of Eukaryotic Orthologous Groups) database through Blast software (version 2.2.23) [55] with default parameters (under a threshold E-value ≤ 10-5) to obtain the isoform annotations. GO (Gene Ontology) annotations and functional classi cations were acquired using Blast2GO program (version 2.5.0, E-value ≤ 10-5) [56] according to NR annotations. InterProScan5 software (version 5.11-51.0) [57] was used to acquire annotations from InterPro database.

Identi cation of Differentially Expressed Genes (DEGs)
All clean reads were mapped to the reference full-length transcriptome using BLAST software. Gene expression levels were determined by the number of full-length transcriptions, which belong to a cluster after ice clustering and CD-HIT process. Then use the total isoforms' counts to normalize the counts in each sample. Differentially expressed genes (DEGs) were acquired using DEseq2 with Q value (adjust P value) < 0.001 and fold change (FC) ≥ 2 or ≤ -2 [58]. These DEGs were then carried into GO and KEGG enrichment with Phyper in R package using Q value ≤ 0.05 as default.

Comparative Analysis of Putative Genes Expression in Metabolites Biosynthesis Pathway
The amino acid sequences of putative genes involved in polysaccharides, alkaloids and avonoids biosynthetic pathway were searched in National Center for Biotechnology Information (NCBI) database (Additional le 12). All amino acid sequences of putative genes were mapped to the protein information from RNA-Seq using BLAST software to get the average FPKM value of each genes in each sample.
Heatmaps of putative genes involving in polysaccharides, alkaloids and avonoids biosynthesis were drawn based on FPKM value.

Statistical analysis
The experimental data are expressed as the mean ± standard deviation of three independent biological replicates. The statistical differences between samples were analyzed using two-way analysis of variance by SPSS (version 17.0). Values at P < 0.01 were considered statistically signi cant. Abbreviations