Dendrobium officinale plants, potted in a substrate of shattered fir bark, were maintained in a greenhouse in the South China Botanical Garden, Guangzhou, China under natural conditions. About 13-month-old plants which sprouted in April were used to determine water-soluble polysaccharide, monosaccharide and starch content, as well as the localization of polysaccharides.

In this study, A. thaliana ecotype Columbia (sustained in our laboratory) plants served as the wild type (WT) and were used for transgenic experiments. Plants were grown in a growth chamber under a 16-h photoperiod (100 μmol m^-2 s^-1) at 22°C. To screen transgenic plants, seeds were sown and germinated on Murashige and Skoog (1962) medium with 1.5% (w/v) sucrose and 0.8% (w/v) agar, and supplemented with 30 mg/L hygromycin B. Plants were potted in a substrate of topsoil and vermiculite (1:3, v/v), and periodically watered with liquid Hyponex fertilizer (N:P:K = 6-10-5, diluted 1,000-fold; Hydroponic Chemicals Co., Findlay, OH, USA).

Stems from about 13-month-old D. officinale were harvested (two stems from each pot, and at least 100 pots), cleaned, and dried in an oven at 105°C until constant weight. Samples were powdered to a fine powder by a DFT-50 pulverizer (Xinno Instrument Equipment Inc., Shanghai, China) and used to analyze water-soluble polysaccharide content and monosaccharide composition. To extract the water-soluble polysaccharides, the powder (0.3 g) was pre-extracted in 80% ethanol for 2 h at 80°C and filtered through Whatman filter paper No. 1. The residue was extracted with double-distilled water for 2.5 h at 100°C. Double-distilled water was added to the supernatant and made up to 250 mL after the residue was filtered out by Whatman filter paper No. 1. This stock was deemed as the polysaccharide solution and was used for the analysis of water-soluble polysaccharide content by the phenol-sulfuric acid method according to Dubois et al. (1956) and He et al. (2015). Briefly, 200 μL of polysaccharide solution was mixed with 1800 μL of double-distilled water, added 1 mL of 5% phenol and rapidly vortexed, then mixed with 5 mL of concentrated sulfuric acid. The reaction solution was placed in a 100°C bath for 20 min. The absorbance of the sample solution was measured at 488 nm with a UV-6000 spectrophotometer (Shanghai Metash, Shanghai, China) when the reaction solution had cooled down to room temperature. The reaction solution, when added to 2000 μL of distilled water, was used as the calibration standard. Glucose was used to calculate a standard curve (10, 20, 40, 60, 80, and 100 μg/mL). Each sample was assayed as three replicates.

For the analysis of monosaccharides from D. officinale stems, 0.12 g of powder described above of each sample was used to extract water-soluble polysaccharides that were analyzed by high performance liquid chromatography (HPLC) according to The State Pharmacopoeia Commission of People’s Republic of China (2010) and He et al. (2015). Briefly, the powder of each sample was pre-extracted with 80% ethanol at 80°C for 2 h. This process was repeated four times to remove monosaccharides, oligosaccharides and ethanol-soluble materials, then water-soluble polysaccharides were extracted with double-distilled water at 100°C for 2.5 h. The extraction was hydrolyzed by 3.0 M HCl, derivatized with 1-phenyl-3-methyl-5-pyrazolone (PMP) and monosaccharide content was analyzed by HPLC according to He et al. (2015).

For the analysis of mannose from A. thaliana, the above-ground parts (leaves, flowers, and stems) from 2-month-old A. thaliana ecotype Columbia and transgenic lines were harvested, cleaned and grounded to a fine powder with liquid nitrogen using a mortar and pestle, then dried in an oven at 80°C until constant weight. To analysis mannose content, 0.3 g of powder was pre-extracted with 80% ethanol for 2 h, then extracted with double-distilled water for 4 h at 100°C. Four volumes of 100% ethanol were added to the extracted solution, mixed and kept at 4°C overnight, then centrifuged at 10,000 rpm for 20 min. The residue was re-dissolved in 20 mL of double-distilled water to form the polysaccharide solution. This polysaccharide solution was hydrolyzed and derivatized, and the mannose content was analyzed by HPLC, as described above.

The powdered samples used in the analysis of mono- and polysaccharides were also used to determine starch content. Starch extraction and determination were performed according to McCready et al. (1950). Briefly, 0.200 g of powdered sample was wet with a few drops of 80% alcohol in a 50 mL centrifuge tube, 5 mL distilled water was added followed by 25 mL of 80% ethanol. This mixture was vortexed thoroughly with a vortex mixer (Scilogex, Berlin, NH, USA). After left to stand at room temperature for 5 min, the mixture was centrifuged by a universal 32R (Hettich, Tuttlingen, Germany) at 2,500 rpm for 5 min. The residue was pre-extracted by 30 mL of hot 80% ethanol until a test with anthrone (Morris, 1948) proved negative. To extract starch, 5 mL of distilled water and 30 mL of 52% perchloric acid were added to a centrifuge tube that contained the residue described above, and vortexed thoroughly by a Scilogex vortex mixer for 10 min and centrifuged by universal 32R at 2,500 rpm for 10 min. The supernatant was collected into a 100 mL volumetric flask. The extraction was repeated and the supernatant was collected into a volumetric flask. The combined solutions were diluted to 100 mL, filtered through Whatman filter paper No. 1, and the first 5 mL of the solution was discarded. The starch solution was diluted so that it contained 20 to 100 μg of starch per 1 mL. Starch solution (2 mL) was transferred to a 10 mL test tube, 6 mL of anthrone-sulfuric acid solution (2 g of anthrone per 1 L of 95% sulfuric acid) was added, vortexed thoroughly, cooled in water for 2 min and placed in a 100°C bath for 5 min. The absorbance of the sample solution was measured at 630 nm with a UV-6000 spectrophotometer (Shanghai Metash, Shanghai, China) after cooling to room temperature. The reaction solution, which was added to 2 mL of distilled water, replaced the starch solution and was used as the calibration standard. Each sample was assayed as three replicates.

The stems from 13-month-old D. officinale were cut into 5 mm long transects and fixed in a solution of 2.5% glutaraldehyde and 2% paraformaldehyde in 0.1 M sodium phosphate buffer (pH 7.2). Samples were cut longitudinally to 4–8 mm² cross-sections under a SZX7 stereoscopic microscope (Olympus America Inc., Center Valley, PA, USA) and immersed in the same fixative while cutting. Segments were collected into sampling bottles filled with fixative then vacuum infiltrated for at least 30 min to facilitate penetration of the fixative and then kept at 4°C for about 7 days. After fixation, samples were washed six times with 1% sodium phosphate buffer, 30 min each time, and post-fixed in 1% osmium tetroxide (OsO₄) in 0.1 M sodium cacodylate buffer for 4 h (pH 7.2). A graded series of ethanol (30, 50, 75, 85, 95, 100%, v/v) was used to wash and dehydrate samples for 30 min in each step. For osmosis, segments were treated in a graded series buffer (acetone: Epon812, 3:1, 1:1, 1:3) for 30 min in each step, then immersed in Epon812 overnight. On the second day, segments were placed in embedding molds (Beijing Zhongjingkeyi Technology Co., Ltd., Beijing, China) with Epon812 and baked in an oven at 60°C for 2 days. Serial cross-sections of embedded material were cut to 1 μm thickness with an LKB-11800 ultramicrotome (LKB, Bromma, Sweden) and the PAS reaction was performed to stain sections, as described by Tütüncü Konyar et al. (2013). Cross-sections were photographed with a Leica S8 APO stereomicroscope (Leica Microsystems Ltd., Heerbrugg, Switzerland). For transmission electron microscope observations, materials were cut into ultrathin sections (50–70 nm) by a Leica-EM-UC6 ultramicrotome (Leica Microsystems GmbH, Wetzlar, Germany), and then examined and photographed with a JEOL-JEM-1010 transmission electron microscope (Jeol Ltd., Tokyo, Japan) at 100 kV.

Eight DoCSLAs that were likely involved in the biosynthesis of mannan polysaccharides were identified in our previous study (He et al., 2015). To comprehensively analyze the evolutionary relationships of the CLSA family between D. officinale and other plant species, amino acid sequences of CSLA proteins from a dicot (A. thaliana) and a monocot (Oryza sativa L.) were used to construct an unrooted tree with the Neighbor-Joining method (Saitou and Nei, 1987).

The genomic sequences of these DoCSLAs were downloaded from whole genome assemblies of D. officinale (DDBJ/EMBL/GenBank accession code: JSDN00000000, Zhang et al., 2016). Genomic and mRNA sequences were used as queries to generate a gene structure diagram with the Gene Structure Display Server¹ (Hu et al., 2015). The motifs in the amino acid sequences of DoCSLAs were identified using MEME 4.11.2².

Total RNA was extracted from D. officinale stems with Column Plant RNAout2.0 (Tiandz, Inc., Beijing, China) and reverse transcribed for first-strand cDNA by using M-MLV reverse transcriptase (Promega, Madison, WI, USA) according to the manufacturer’s protocol. The cDNA was used as a template to amplify the DoCSLA6 gene with a specific set of primers (DoCSLA6OxF/DoCSLA6OxR; Supplementary Table 1) and cloned into the NcoI site of the binary vector pCAMBIA-1302 by an In-Fusion^® HD Cloning Kit (Takara Bio Inc., Dalian, China) according to the manufacturer’s instructions. Expression of the DoCSLA6 gene was under the control of the CaMV35S promoter. The 35S:DoCSLA6 construct was introduced into Arabidopsis plants (ecotype Col) by an Agrobacterium-mediated (Agrobacterium tumefaciens, EHA105 strain) method described by Clough and Bent (1998).

Leaves from about 1-month-old Arabidopsis plants were collected and kept in liquid nitrogen to extract total RNA. The TRIzol RNA isolation method (TRIzol, Invitrogen, Carlsbad, CA, USA) was used for total RNA extraction as described by Meng and Feldman (2010). Two microgram of total RNA was used for reverse-transcription reactions by the M-MLV Reverse Transcriptase Kit (Promega) according to the manufacturer’s protocol. For the PCR reaction, the Taq DNA Polymerase Kit (Takara Bio Inc.) was used with the following amplification protocol: 94°C for 3 min; 30 cycles (25 cycles for AtUBQ10) of 94°C for 30 s, 55°C for 30 s, 72°C for 1 min; a final elongation step at 72°C for 10 min. PCR products (5 μL) were evaluated by electrophoresis on a 1% agarose gel in TAE buffer and photographed with Gel Documentation System GenoSens 1880 (Shanghai Qinxiang Scientific Instrument Co. Ltd., Shanghai, China). The A. thaliana ubiquitin10 gene (AtUBQ10) was used as an internal control based on the recommendation of Zhao et al. (2015). The primers (DoCSLA6F/DoCSLA6R and AtUBQ10F/AtUBQ10R) used for qRT-PCR are listed in Supplementary Table 1.

Data were analyzed using SigmaPlot12.3 software (Systat Software Inc., San Jose, CA, USA) by a t-test. P < 0.05 was considered to be statistically significant.

Generally, starch is the major type of storage polysaccharide in higher plant species. To understand the type of polysaccharides in D. officinale stems, water-soluble polysaccharides, the monosaccharide fraction of water-soluble polysaccharides, and starch were analyzed. Water-soluble polysaccharides were abundant, about 367 mg/g, in the stems of D. officinale (Table 1). The main monosaccharide within the water-soluble polysaccharides was mannose, about 257 mg/g (Table 1), indicating that the main polysaccharides were mannan polysaccharides. Glucose was the second most common monosaccharide, found at 55 mg/g (Table 1) in the water-soluble polysaccharide fraction. Galactose was also found in the water-soluble polysaccharide fraction, but it had a very low content, about 1 mg/g (Table 1). Although starch is perceived to be the major type of non-structural polysaccharide in plants, the starch content in D. officinale stems was about 93 mg/g. These results indicate that mannan polysaccharides are the main polysaccharides in D. officinale stems, in agreement with Xing et al. (2014) and Wei et al. (2016).

Having understood that the stems of D. officinale contain a high content of mannan polysaccharides and a low starch content, we investigated the localization of polysaccharides in stems by anatomical observations. Transverse sections from fresh stems showed that a number of vascular bundles were dispersed throughout the stem, similar to other monocots (Figure 1A). Ground tissue was composed of a mass of parenchyma cells among which vascular bundles were embedded (Figures 1A,B). The ground tissue stained with PAS was strongly labeled, but weak labeling of the sheath and a weak signal in the walls of cortical cells and parenchyma cells (Figure 1B). Surprisingly, polysaccharides formed granules in parenchyma cells stained an intense purple with polysaccharide stains (Figures 1B,C).

In order to identify the localization of polysaccharide granules at the subcellular (organelle) level, ultrathin sections were made and analyzed. There was no discernible nucleus, vacuole or cellular organelles in parenchyma cells, but numerous polysaccharide granules were clearly visible (Figure 2A). A complicated membrane system, in which the polysaccharide granules were embedded, was present in parenchyma cells (Figure 2A). The polysaccharide granules had various forms with unequal size and were localized in plastids (Figures 2B,C). The membrane structure of plastids was clearly visible, and wrapped several polysaccharide granules in a single plasmid (Figures 2B,C). The stems contained a considerable amount of polysaccharides that were stored in the plastids.

Mannan polysaccharides were the main polysaccharides in D. officinale stems, accounting for about 58.3% of crude polysaccharides. In the phylogenetic tree, the CSLA family was divided into two branches: clusters I–III in one branch and cluster IV in another branch, indicating that two ancestral genes were the origins of CSLA in both dicots and monocots (Figure 3). In addition, the phylogenetic tree clearly showed that CSLA members were separated into four clusters: cluster I included only A. thaliana; cluster II included proteins of both the dicot (A. thaliana) and monocot (rice); clusters III and IV included proteins from D. officinale and rice but not from Arabidopsis (Figure 3). DoCSLA6 was included in cluster II, and had a close relationship with cluster I (Figure 3).

Most of the CSLAs possess nine exons and eight introns as was observed in rice and Arabidopsis (Richmond and Somerville, 2000; Hazen et al., 2002). In order to gain information about the gene structure of DoCSLAs, genomic regions of D. officinale corresponding to DoCSLAs were identified and used to analyze the architecture of introns and exons. Most members shared similar intron/exon structures but the length of their genomic region differed (Figure 4A). Most DoCSLAs (excluding DoCSLA7) contained nine exons and eight introns similar to other CSLA family members (Figure 4A). However, DoSCLA7 showed variation in intron/exon organization, and contained 10 exons and nine introns (Figure 4A). The length of the genomic region was also different. For example, the genomic region of DoCSLA1 was no longer than 3 kb, but DoCSLA3, DoCSLA5 and DoCSLA7 were longer than 15 kb (Figure 4A).

To better understand the similarity and diversity of motifs in the protein sequences of DoCSLAs, the conserved motifs in proteins were investigated. Among the 12 distinct conserved motifs identified in all of the DoCSLAs, motifs 1–4 collectively comprised the catalytic subunit (Figure 4B).

Only four out of nine CSLA genes, namely AtCSLA2, AtCSLA3, AtCSLA7, and AtCSLA9, are known to produce mannan polysaccharides in A. thaliana (Sandhu et al., 2009; Dhugga, 2012). A phylogenetic tree analysis showed that DoCSLA6 had a close relationship with AtCSLA9 and AtCSLA2, and may play a similar role to these genes. Consequently, over-expression (OE) lines of DoCSLA6 were generated and analyzed. The transcription of DoCSLA6 was detected in the OE lines but not in the WT plant, suggesting that the DoCSLA6 gene were successfully transformed with a normal transcript in A. thaliana (Figure 5A). The HPLC-UV chromatograms are shown in Supplementary Figure 1. The OE lines showed no distinct phenotype compared with WT plants (Figures 5B,C). However, the mannose content was significantly higher in the OE lines with 0.478, 0.4997, and 0.4105 mg/g DW in lines #1, #2, and #3, respectively, while the WT plant only contained 0.294 mg/g DW (Figure 5D). This result suggests that DoCSLA6 contributed to the synthesis of mannan polysaccharides.