The vector pCambia1301-35SN is used for overexpression and can be classified as an A. tumifaciens Ti-plasmid vector because of the existence of the left border (LB) and right border (RB) (Figure 1A). The Ti-DNA region of this vector harbors hygromycin phosphotransferase (HygR) and LacZα as the selection marker genes and β-glucuronidase (GUS) as the reporter gene, which is interrupted by a catalase intron. The transcription of GUS gene and HygR gene are both driven by the CaMV 35S (35S) promoter and ended by the 35S’ terminator. The exogenous genes are integrated into the multiple cloning sites (MCS) at the restriction enzymes SalI and SpeI recognition sites. The 35S promoter and nitric oxide synthase terminator (Nos3′) are used to control the open reading frame (ORF) of the exogenous gene. In our experiment, the DNA fragments containing the SG gene and SpeI and SarI restriction sites were cut by the restriction enzymes SpeI and SarI. The purified SG fragments were inserted into the linearized pCambia1301-35SN with the T4 DNA ligase (NEB, Kunming, China).

We attempted to identify four efficient promoters for gene overexpression as follows: The Medicago truncatula hypothetical protein (MtHP) promoter (Xiao et al., 2005), a promoter isolated from the cassava vein mosaic virus (CVMV) (Verdaguer et al., 1996), a promoter isolated from mirabilis mosaic virus (MMV) (Dey and Maiti, 1999), and the peanut chlorotic streak caulimovirus (PCISV) promoter (Maiti and Shepherd, 1998). These were cloned into the pCambia1301 vector including GUS marker gene (Figure 1B), and then infected the protocorms by Agrobacterium-mediated transformation. Finally, we compared the activity of GUS driven by different promoters with 35S promoter to find out which one can work efficiently in the transgenic system of D. officinale.

To create a D. officinale CRISPR/Cas9 genome editing system, an RNA-guided genome editing vector line p1300DM-OsU3 (Hind III-AarI)-Cas9 (Figure 1C) was used for expressing engineered sgRNA and Cas9 in D. officinale cells. This vector also belongs to the pCambia vector family. In this vector every element is essential except the extra spacer between two Aar1 restriction sites in sgRNA expression cassettes which are used for the integration of sgRNA fragments. As shown in Figure 1C, the sgRNA expression cassette is driven by the OsU3 promoter. This promoter is a small nuclear RNA (snRNA) promoter that initiates transcription at an adenine nucleotide and drives the high expression of sgRNAs (Shan et al., 2013; Ma et al., 2015). The Cas9, which is modified with plant-optimized codons (Cong et al., 2013; Mao et al., 2013), has an SV40 nuclear localization signal (SV40 NLS) (Brandén et al., 1999) that can guide the mature CRISPR/Cas9 complex into the cell nucleus. It is driven by the 35S promoter. Another copy of 35S promoter also drives the expression of HygR, and our lab’s previous experience of successful application of the same CRISPR/Cas9 system in rice has excluded the concern that using two 35S promoters for both Cas9 and HygR expression might result in gene silencing.

The abundance of lignocellulose is a key factor affecting the taste and popularity of D. officinale in the natural health product market. So we were interested in finding out whether we could reduce the content of lignocellulose by using the CRISPR/Cas9 system to knock out genes in the lignocellulose biosynthesis pathway. Based on the public genome sequence of D. officinale (Yan et al., 2015) and knowledge of lignocellulose biosynthesis (Bonawitz and Chapple, 2010), five genes (Table 1), COUMARATE 3-HYDROXYLASE (C3H) (Schoch et al., 2001; Ralph et al., 2006), CINNAMATE4-HYDROXYLASE (C4H) (Sewalt et al., 1997; Bonawitz and Chapple, 2010), 4-COUMARATE:COENZYME A LIGASE (4CL) (Lee et al., 1997), CINNAMOYL COENZYME A REDUCTASE (CCR) (Goujon et al., 2003; Jackson et al., 2008; Bonawitz and Chapple, 2010) and IRREGULAR XYLEM5 (IRX) (Taylor et al., 2003), participating in the lignocellulose biosynthesis process, were selected as target genes to explore whether our CRISPR/Cas9 system could work in D. officinale efficiently. We first obtained the sequences of five genes in Arabidopsis thaliana from the NCBI database and then found their orthologous genes by BLAST against the V2 genome assembly of D. officinale as reported by (Yan et al., 2015).

Three pairs of oligonucleotides targeted to different sites were designed for each gene to increase the possibility that the right gene was targeted (Table 1). The DNA sequences with the form AN18-20NGG were selected as the candidates and the NGG means the protospacer adjacent motifs (PAM). The specificity of these candidates are examined by BLAST search against the genome sequence of D. Officinale. In order to make the CRISPR/Cas9 constructs, two DNA oligos were chemically synthesized for each gRNA. The two oligos were complementary for the sequence corresponding to the spacer, and the sequences GGC and AAAC were added to the 5′ end of the forward and reverse oligos, respectively, to allow for the formation of cohesive ends of AarI restriction sites following annealing. (See Table 1 for oligo sequences). These coupled oligos were ligated into the vector p1300DM-OsU3 (Hind III-AarI), linearized by Aar I restriction enzyme digestion.

The constructed vectors for overexpression, promoters test and genome editing were first amplified by transferring into E. coli competent cells (DH5α), incubated in the LB medium, which use kanamycin as a selection reagent. Then we used Agrobacterium-mediated method to transfer the plasmids containing the targeting expression cassettes into the prepared protocorms. The EHA105 A. tumifaciens strain was provided by Takara Biomedical Technology and the pCambia1301-35SN is kindly provided by Shanghai Institute of Plant Physiology and Ecology, Chinese Academy of Sciences. The transformation protocol used for the Dendrobium orchid was developed by previous study (Men et al., 2003; da Silva et al., 2011; Hsing et al., 2016; Phlaetita et al., 2015; da Silva et al., 2016).

To optimize the growth of the genetically modified plants, different media were used at different stages throughout the experiment (see detailed information about the media in the Supplemental Table 1). The medium Z1 used for seed germination is based on the MS basic medium (Murashige and Skoog, 1962; Robinson et al., 2009). The medium B3-3 used for proliferation of protocorms, the co-culturing medium G and the reproduction-screening medium B3-TS are all based on the basic medium B5 (Gamborg et al., 1968), which contains mineral salts, sucrose, the B-vitamins, and 2,4-dichlorophenoxyacetic acid (1 mg/L) (Robinson et al., 2009). There are two screening media for the positive transformants: The first stage screening medium SH4-TS is different from the SH basic medium (Reil and Berger, 1996; Schenk and Hildebrandt, 1972) and the second stage screening medium FD5-TS is derived from the MS basic medium, but the content of every component is half of the original MS medium, so this basic medium is also called 1/2 MS. The rooting medium SB2-TS used for rooting of the screened positive plants is based on the B5 basic medium.

Mature D. officinale capsules were sterilized with 75% ethyl alcohol, 0.1% (w/v) mercuric chloride, then washed with sterile distilled water. After sterilization, a hole was cut at one end of the treated capsule, and seeds were sown on the Z1 medium by holding the capsule with tweezers at the other end and shaking it slightly (Figure 2A1–3). All cultures were maintained in an environmentally controlled room with a 12-h photoperiod at 2000Lx provided by cool white fluorescent lights. The day/night temperatures were 25 ± 3°C. In general, protocorms have been used as targets because of their higher regeneration capacity (Kanchanapoon et al., 2012). So after the seeds sprouted and developed into the protocorm stage, the protocorms were cut into small pieces around 3–5 mm in diameter and put on the B3-3 medium to produce more protocorms (Figure 2A4). All procedures were performed in a sterilized environment.

The constructed vectors were transferred into the competent Agrobacterium cells by electroporation method. The Agrobacterium with vectors was inoculated in the YM liquid suspension medium with kanamycin, and incubated on a shaker at 220 rpm, at 28°C for 12 h until the density of OD600 was between 0.6 and 0.8. Next, the Agrobacterium cells were collected by centrifugation at 4000 rpm for 12 min at 4°C and resuspended in a moderate B5 liquid medium (Gamborg et al., 1968) to a final density of OD600 = 0.6∼0.8. The resuspended medium was the same as the basic medium, which is used for proliferation of protocorms, co-culturing and reproduction-screening. The prepared protocorms (0.5 cm in diameter) were mixed with a 50 ml bacteria liquid resuspension medium containing 100 μmol/L AS (acetosyringone) (Dutt and Grosser, 2009), which can promote exogenous gene transformation (Figure 2B1). Then, the mixtures were incubated on a shaker at 110 rpm for 40 min at 28°C to keep the protocorms in close and constant contact with Agrobacterium. After shaking, the protocorms were put on the co-cultivation medium G (Figure 2B2).

After three days of co-cultivation, the infected protocorms were transferred to the B3-TS medium and incubated for 30 days, and then transferred from B3-TS medium to the SH4-TS medium to incubate for 2 weeks. We cut the protocorms into small pieces of 3–5 mm in diameter. These green tissues growing on SH4-TS medium were picked up, and then transferred to the FD5-TS medium to incubate for 30 days. Finally, the surviving plantlets growing on the FD5-TS medium were put on SB2-TS for 60 days for rooting (Figure 2B3–5). The conditions for all of the above cultures was as follows: temperature of 25 ± 3°C, illumination time of 12 h, illumination intensity of 2000lx, cool white fluorescent light.

The GUS reporter gene encodes the β-glucuronidase, which is a hydrolase that catalyzes the cleavage of a wide variety of β-glucuronides. The transformed plants with GUS expression were stained with 5-bromo-4-chloro-3-indoyl β-D-glucuronide (X-Gluc). After being co-cultured with the Agrobacterium containing the vector pCAMBIA1301-35SN, the protocorms were immersed in X-Gluc solution (Jefferson et al., 1987) for 10 min in a mild vacuum, and then incubated overnight at 37°C. The positive transgenic protocorms should turn blue after staining.

In order to confirm whether the exogenous genes for overexpression have been successfully integrated into the plant genome, a polymerase chain reaction (PCR) is needed. Total DNA are isolated from 100 mg seedling leaves of different transformants using the GenElute^TM Plant Genomic DNA Miniprep Kit (Sigma-Aldrich, St. Louis, USA). The primers used in our experiment are shown in Table 2. First, we used the HygR primers to screen the DNA from positive transformants by PCR, which can confirm that the marker cassette has been transfected into tranformants successfully. Then we screened the DNA with SG primers, which can prove that the expression cassettes have integrated into the genome of positive transformants. Finally, in order to prove that the results are not due to PCR failure, we used the actin primer (Table 2) and SG primer to test the negative samples and positive samples. The PCR program was set as initial denaturation at 95°C for 5 min followed with 35 cycles of three-step cycle: denaturation at 95°C for 30 s, annealing at 62°C for 60 s and extension at 72°C for 40 s.

We designed the PCR primers of each target gene of CRISPR/Cas9 (Table 3) to screen the transformants. The forward primer annealed to the upstream sequence of the target area and the reverse primer annealed to the downstream sequence. We extracted the total DNA of the transformants to use as the template to perform PCR amplification with the primers shown in Table 3. The PCR products were sequenced and compared with the target gene sequences to calculate the mutagenic efficiency. Because the HygR expression cassette is fused with CPISPR/Cas9 expression cassettes, we could use HygR as a marker to detect whether the CPISPR/Cas9 expression cassettes inserted into the D. officinale genome by PCR amplification of total DNA of transformants. Using the same method as mentioned above, the HygR sequences were amplified by using the HygR primer shown in Table 2.

The fluorescence protein SG is used as the marker to confirm whether an exogenous gene is expressed successfully in D. officinale. By observing the fluorescence emission, we can roughly detect the expression of the protein. Twenty positive SG transformants and one negative control were tested through observing blue light emission under fluorescence microscopy using a Zeiss Discovery V12 microscope equipped with a Zeiss Rhodamine cube KSC 295–815D with excitation wave length at 485 nm. The units of different anatomical levels from whole leaves to cells were detected under the fluorescent microscope. Images were collected using an Optronics digital camera system (Nikon D60) with manual exposure settings. For all materials, the images under bright field were recorded using the same exposure times (Zhang et al., 2015).

After obtaining the total DNA, the target locus of sgRNAs were amplified with the primers shown in Table 3. The positive PCR products were sequenced and aligned with the target gene sequence. Those samples with mutated sequences were selected for further analysis.

In the vector of overexpression, the transcription of the GUS gene and the HygR gene is driven by the 35S promoter and ended by the 35S′ terminator. We chose 48 transformants to perform PCR amplification with HygR primers. The results (Figure 3A) showed that 45 of them were positive. Then PCR was conducted using DNA from 23 of the positive samples as templates with SG gene primers. The result is shown in Figure 3B, 91.3% (21 in 23 samples) of the candidates are the positive transformants with 714bp amplicons, which have the same length as the SG coding sequence. To exclude any false negative due to failure in PCR, we chose two negative samples and two positive samples to perform the PCR with SG and actin primer. As shown in Figure 3C, the SG band of amplification can be observed only in positive samples while the actin band of amplification can be observed in both positive and negative samples, which confirmed that those negative samples are true negatives. The PCR result confirms that the gene SG has been successfully inserted into the genome of D. officinale and the Agrobacterium-mediated gene transformation can work efficiently.

Superfolder green fluorescent protein (SG) is a well-folded variant of green fluorescent protein (Pedelacq et al., 2006), which can emit green light under the excitation light with wavelength at 485nm. The fluorescent signals at different anatomical levels from whole leaves to cells were detected under a fluorescent microscope. According to Figure 4, because of the existence of photosynthetic pigments that can emit red light under the excitation light, the controls and the samples all have strong red fluorescence background. When superfolder green fluorescent protein co-localizes with the red fluorescent pigment, a yellow spot will show in the merged image. Compared with the controls, all of the 20 samples coming from the SG transformants showed the expression of superfolder green fluorescent protein gene (data not shown). This result confirms that the transformed genes can transcribe and translate into functional proteins, and the established transformation system in D. officinale can work efficiently.

We used the GUS gene as a reporter to screen the positive transformants at an early stage after transformation. The protocorm tissues, which were successfully transfected with transgene expression cassettes, can be stained to blue with X-Gluc. As shown in Figure 5), when treated with X-Gluc, the Agrobacterium infected protocorms turn blue in some areas. The early expression of GUS indicates that the expression cassettes have been transformed into the protocorm cells.

To identify efficient promoters, the four promoters MtHP, CVMV, MMV, and PCISV were tested and the results were compared with the 35S promoter. The results are shown in Figure 6. The positive ratio of each promoter was calculated based on examining each of the protocorms in Figure 6A in different directions under a microscope, shown in Table 4. The results showed that MMV, CVMV, and PCISV are as effective as the 35S promoter with strong expression in all protocorms examined, while MtHP is less effective than the 35S promoter. Because only tiny spots on protocorms, and not whole protocorms, were stained blue from GUS transformants driven by MtHP, it is not easy to differentiate them from negative controls based on the picture in Figure 6. However, MtHP still showed weak expression in 17.8% of the protocorms probably due to MtHP promoter leakage in the orchid. Our results proved that MMV, CVMV, and PCISV can be used as efficient promoters in the transgenic system of D. officinale.

Here we chose five genes as the targets to explore the functional CRISPR/Cas9 system in D. officinale. Because the HygR gene is fused with the CRISPR/Cas9 expression cassettes, it can be used as a marker to detect the rate of successful insertion of CRISPR/Cas9 expression cassettes into transformants’ genomes. We randomly chose 47 samples to demonstrate the positive rate. As shown in Supplemental Figure 1, 93.6% of the all detected transformants are positive plants (44 out of 47 samples), and only three transformants are negative plants or reflect weak positive characteristics. To further check whether the CRISPR/Cas9 system can edit the target genes in the host cells, the target loci were PCR amplified with the primers shown in Table 3 and then sequenced. We analyzed the sequencing result of each target gene to characterize the mutation introduced by using CRISPR/Cas9. The results are shown in Table 5 and Figure 7. The original sequencing results are shown in supplemental Figure 2.

As shown in Figure 7, the sequences of each gene from the corresponding samples confirm lesions in target regions. There are two mutated loci (F2, F3) in C3H gene: the first locus (F2) has nucleotide substitution in one sample; the second locus (F3) has nucleotide substitution in three samples and nucleotide insertion in one sample. Two mutated target loci (F1, F2) were found in C4H gene. For F1 locus, we found nucleotide substitution and nucleotide deletion occurred in the first sample simultaneously. Nucleotide substitution and nucleotide insertion occurred in the second sample simultaneously, and nucleotide substitution occurred in the third sample. For F2 locus, nucleotide substitution occurred in all three samples. In 4CL gene, we detected two mutated target loci (F2, F3). Nucleotide substitution in five samples and nucleotide deletion in one sample occurred in locus F2. The F3 locus has nucleotide substitution in four samples. Only one mutated target locus was found in both CCR and IRX. In CCR gene, nucleotide substitution, insertion or deletion occurred in ten samples, while in IRX gene, nucleotide substitution occurred in two samples. The percentages of mutations are shown in Table 5, which varied between 10 and 100%.

These results provided evidence that the established CRISPR/Cas9 system can work efficiently although the phenotype of the mutant plants developed in this study has not been checked yet due to the slow growth of D. officinale. We can further use this system to modify other pathways to study the effects of gene functions on the development and metabolism of D. officinale.