The complete chloroplast genome of Dendrobium nobile, an endangered medicinal orchid from north-east India and its comparison with related Dendrobium species

Sample collection, DNA extraction and sequencing

Fresh leaves of D. nobile were collected from plants growing in greenhouses of National Research Centre for Orchids, Sikkim, India and voucher specimen was deposited in Botanical Survey of India as well as in the Department of Botany, North-Eastern Hill University, Shillong. The high molecular weight cpDNA was extracted using a modified CTAB buffer, and treated according to a standard procedure for next generation sequencing on Illumina HiSeq 2500-PE150. The quality and quantity of the genomic DNA was assessed through agarose gel electrophoresis, Nanodrop and Qubit detection method. The experiments included both paired-end and mate-pair libraries. Tagmentation was carried out with ∼4 µg of Qubit quantified DNA and the tagmented sample was washed using AMPURE XP beads (Beckman Coulter #A63881) and further exposed to strand displacement. The strand-displaced sample of 2–5 kb and 8–13 kb gel was size selected and taken for overnight circularization. The linear DNA was digested using DNA Exonuclease. Further the circularized DNA molecules were sheared using Covaris microTUBE, S220 system (Covaris, Inc., Woburn, MA, USA) for obtaining fragments in the range 300 to 1,000 bp. M280 Streptavidin beads (ThermoFisher Scientific, Waltham, MA) was used to cleanse the sheared DNA fragments with biotinylated junction adapters. The bead-DNA complex was subjected to End Repair, A-Tailing and Adapter ligations.

Data processing

The data quality assessment for Illumina WGS raw reads was carried out using FastQC tool. Perl scripts were written for adapter clipping and low quality filtering. Chloroplast genomes of D. officinale, D. huoshanense and D. strongylanthum retrieved from NCBI-RefSeq database was used as reference for the assembly. BWA-MEM algorithm with default parameter settings was used for aligning the adapter clipped and low quality trimmed processed reads with the Dendrobium cp genomes (Li & Durbin, 2009). SPAdes-3.6.0 program was used for k-mer based (k-mer used 21, 33, 55 and 77) de-novo assembly with the aligned reads and the quality of the assembled genome was gauged using Samtools and Bcftools (read alignment and genome coverage calculation)  (Bankevich et al., 2012) (https://samtools.github.io/bcftools/bcftools.html). The cp genome of D. nobile was also generated through reference-assisted assembly using the high quality paired-end libraries by NOVOPlasty (Dierckxsens, Mardulyn & Smits, 2017) for further validation. It is specifically designed for de novo assembly of mitochondrial and chloroplast genomes from WGS data with the aid of a reference or seed sequence. The seed sequence can correspond to partial or complete sequence of chloroplasts of closely to distantly related species. The cpDNA RefSeq sequence of D. officinale was used as a seed sequence to perform reference-assisted assembly.

Genome annotation and codon usage

Basic Local Alignment Search Tool (BLAST; BLASTN, PHI-BLAST and BLASTX)  (Altschul et al., 1997), chloroplast genome analysis platform (CGAP) (Cheng et al., 2013) and Dual Organellar GenoMe Annotator (DOGMA) (Wyman, Jansen & Boore, 2004) was used to annotate protein-coding and ribosomal RNA genes. The boundaries of each annotated gene with putative start, stop, and intron positions were manually determined by comparison with homologous genes from other orchid cp genomes. Further tRNA genes were predicted using tRNAscan-SE  (Lowe & Eddy, 1997) and ARAGORN (Laslett & Canback, 2004). RNA editing sites in the protein-coding genes (PCG) of D. nobile were predicted using Plant RNA Editing Prediction & Analysis Computer Tool (PREPACT) (http://www.prepact.de). For this analysis, D. nobile cp genome was BLAST aligned against Nicotiana tabacum, Oryza sativa Japonica Group, Phalaenopsis aphrodite subsp. Formosana, Physcomitrella patens subsp. patens and Zea mays with a cutoff E-value set to 0.08. The circular genome map was drawn in OrganellarGenomeDRAW (Lohse et al., 2013) followed by manual modification. The sequencing data and gene annotation were submitted to GenBank with accession number KX377961. MEGA 7 was used to analyze and calculate GC content, codon usage, nucleotide sequence statistics and relative synonymous codon usage (RSCU) (Kumar, Stecher & Tamura, 2016).

Gene Ontology annotation and assignment of GO IDs

Gene Ontology (GO) annotation of D. nobile chloroplast genes was carried out in Blast2GO (Conesa et al., 2005) by blast aligning the gene sequences from the GenBank annotation files to Orchidaceae sequences in non-redundant (nr) database with an e-value cutoff of 1e⁻⁵ and queried in InterProScan (Jones et al., 2014). GO mapping and annotation of genes followed this from blast results and were subsequently merged with GO IDs from InterProScan. The merged GO annotations were validated based on True-Path-Rule by removing redundant child terms for each gene sequence. The GO annotations were slimmed down using plant-slim option. 

Simple sequence repeats analysis

MISA (http://pgrc.ipk-gatersleben.de/misa/misa.html), a tool for identification and location of perfect microsatellites and compound microsatellites was used to search for potential simple sequence repeats (SSRs) loci in the cp genomes of different Dendrobium species. The threshold point for SSRs identification was set to 10, 5, 4, 3, and 3 for mono-, di-, tri-, tetra-, and penta-nucleotides SSRs, respectively. All the repeats found were manually curated and the redundant ones were removed.

Phylogenetic reconstruction with whole genome alignment and rearrangement analysis

For phylogenetic reconstruction, we included D. nobile cp genomes from India and China along with 32 other Dendrobium cp genomes retrieved from GenBank. Four Goodyera species were taken as outgroup. The cp genome sequences were aligned with MAFFT v7.0.0 (Katoh & Standley, 2013) and manually curated by visual inspection. PCGs as well as whole cp genomes were used for Bayesian phylogenetic reconstruction using MRBAYES 3.2.6 (Huelsenbeck & Ronquist, 2001). To further validate our results we employed “k-mer Based Tree Construction” in CLC Genomics Workbench that uses single sequences or sequence lists as input and creates a distance-based phylogenetic tree. For visualization and testing the presence of genome rearrangement and inversions, gene synteny was performed using MAUVE as implemented in DNASTAR 12.3 with default settings. Comparative analysis of intra nucleotide diversity (Pi) within the Dendrobium cp genomes was performed using MEGA 7.

Single nucleotide polymorphism identification and phylogenetic analysis without genome alignment

Phylogenetic tree was constructed based on the Single Nucleotide Polymorphisms (SNPs) identified in the whole cp genomes using kSNP3.0 with default settings except for k-mer size (Gardner, Slezak & Hall, 2015). SNPs were identified with k-mer size set to 23, based on which, approximately 79% of the k-mers generated from median-length genome were unique.

Genome organization and features

The complete cp genome of D. nobile was determined from the data generated out of a whole genome project initiative of the same species by Paired-end and Mate pair data from Illumina HighSeq with 150*2 and Illumina NextSeq500 with 75*2 respectively. Further the aligned Illumina reads were separated and assembled using CLC Main Workbench Version 7.7.1 into the single longest scaffold. The D. nobile cp genome is a typical circular double-stranded DNA with a quadripartite structure; it is 152,018 bp in size and consists of Large Single Copy (LSC) (1..84,944; 84,944 bp), Small Single Copy (SSC) (111,230..125,733; 14,504 bp), and two Inverted Repeat (IR) regions of 26,285 bp: IRA (84,945..111,229) and IRB (125,734..152018). In total 134 unique genes (79 PCGs, 8 rRNA genes, 7 pseudogenes and 38 tRNA genes) were successfully annotated, of which 12 genes {rps16, atpF, rpoC1, ycf3, rps12 (2), clpP, petB, rpl2 (2), ndhB (2)} are reported with introns (Fig. 1). We could identify a total of 20, 81 and 11 genes duplicated in the IR, LSC and SSC regions respectively in the D. nobile cp genome. There were a total of 49 RNA editing sites predicted in 23 genes of D. nobile cp genome. The whole chloroplast genome alignment included 34 Dendrobium species and four species from the genus Goodyera as outgroup. Each genome’s panel contained its name, sequence coordinates and a black coloured horizontal centre line with coloured block outlines appearing above and below it. Homology between the cp genomes is represented by each block with the genes, internally free from genomic rearrangement, connected by thin lines to similarly coloured blocks depicting comparative homology between the genomes (Fig. 2). The positions of LSC/IRA/SSC/IRB borders revealed similar structures at the IR/LSC junction in the overall alignment of Dendrobium whole cp genomes (Fig. 3).

Gene ontology mapping and annotation

We further analyzed the D. nobile coding cp genome sequences using the Blast2GO suite and annotated the sequences for three GO terms (biological process, molecular function, and cellular component). In case of GO term there were a total of 231 annotations in biological process (P), molecular function (F) and cellular compartment (C) level. In the category of biological processes a large number of these sequences are annotated for translation, photosynthesis, metabolic processes, and ribosome biogenesis. Similarly, for the GO term molecular function, the top GO categories include functions related to structural molecule activity, catalytic activity, ion and rRNA binding, transporter and transferase activity. Finally, terms including membrane, ribosome and thylakoid were annotated GO categories for cellular compartment with most of the sequences. These results are summarized along with the information on RNA editing sites in Table 1.

Simple sequence repeat identification

SSRs were identified in MISA perl scripts with a minimum of 10 bp repeats among all the Dendrobium species. Of all the SSRs, the mononucleotide A/T repeat units occupied the highest proportion. A higher proportion of di-, tri- repeats are reported rather than tetra- and penta-nucleotide repeats across Dendrobium cp genomes (Fig. 4).

Phylogenetic analysis

Phylogenetic analyses of chloroplast PCGs from Dendrobium species were performed with or without partitions of sequences. Both Bayesian and K-mer based trees (Figs. 5 and 6) recovered a monophyly of the Dendrobium species, irrespective of whether or not the partitions of sequences were incorporated in the analysis supported by strong bootstrap values. The phylogenetic analyses based on complete cp genomes, suggested that five major subgroups within the genus Dendrobium evolved in a nested evolutionary relationship. D. aphyllum, D. parishii, D. loddigesii and D. primulinum are the most recently evolved species that nested into a single monophyletic sub group within the Dendrobium clade. D. chrysotoxum and D. salaccense were a bit primitive on the evolutionary ladder in the phylogenetic tree. Goodyera species emerged as the outgroup that claded separately in the over all tree topology. Similar results were also obtained in the alignment free phylogenetic tree with SNPs (Fig. 6).

Potential RNA editing sites

RNA editing is involved in plastid posttranscriptional regulation and thus provides an effective way to create transcript and protein diversity (Chen & Bundschuh, 2012; Knoop, 2011). In Orchidaceae, RNA editing sites were identified in 24 protein-coding transcripts in P. aphrodite (Zeng, Liao & Chang, 2007). Earlier studies indicate RNA editing sites from the same subfamily to be more conserved than those from different subfamily (Luo et al., 2014). However, orchids and other angiosperms have relatively less common editing sites. For example, orchids and Cocos nucifera share 10 potential RNA editing sites; comparisons among Nicotiana tabacum, Arabidopsis thaliana and orchid RNA editing sites have shown low conservation of editing sites (one common editing site in rpo B) (Luo et al., 2014). Our studies congruently predicted all 49 RNA editing sites (Table 1) in 23 genes of D. nobile from at least 75% of the reference organisms (Nicotiana tabacum, Oryza sativa Japonica Group, Phalaenopsis aphrodite subsp. Formosana, Physcomitrella patens subsp. Patens and Zea mays) and resulted in amino acid substitutions. All the RNA-editing sites were non-silent and edited C to U. Of the 49 RNA editing sites 89.8% (44) editing sites appeared in the second position of triplet codon, 10.2% (five) editing sites appeared in the first position of triplet codon whereas no editing sites appeared in the third base of triplet codon. The genes ndhD, rpoB, rpoC1 had eight, six, and four RNA editing sites, respectively. All the 49 RNA editing sites led to changes in the amino acid. The most frequent amino acid conversion was hydrophilic to hydrophobic (S to L, 22 occurrences and S to F, eight occurrences), followed by hydrophobic to hydrophobic conversions (P to L, 12 occurrences). Seven conversions were found to be hydrophilic to hydrophilic (H to Y, five occurrences and T to M, two occurrences).

Comparison with other chloroplast genomes within the genus Dendrobium

We compared thirty-four chloroplast genomes representing different species within the genus Dendrobium (Table 2). The length of the Dendrobium species cp genomes ranged from 148,778 to 153,953 bp, with D. chrysotoxum being the largest cp genome and D. moniliforme the smallest. The cp genomes have acquired the familial angiosperm plastome organization comprising of a LSC, an SSC and a pair of IR regions each. Dendrobium cp genomes are also AT-rich (62.26–62.39%) quite similar to other orchid cp genomes (Zhitao et al., 2017). Differences in the cp genome size of these species are primarily due to the variations in the length of LSC, SSC and IR regions. Synteny comparison revealed a lack of genome rearrangement and inversions, thereby, substantiating for the highly conserved nature in the genomic structure, including gene number and gene order in these cp genomes. However, structural variation was predominant in the LSC/IR/SSC boundaries (Fig. 2), which could be harnessed for predicting potential biomarkers for species identification.

IR regions are generally considered to be highly conserved regions in the chloroplast genome. IR expansion or contraction is determined by the variability of genes flanking IR/SC junctions (Huelsenbeck & Ronquist, 2001). In the evolutionary ladder, SSC and IR border regions experience expansion and contraction that overall contribute to the variation in chloroplast genome length (Wang et al., 2008; Li et al., 2013). At the IR/LSC boundaries, most IRs of non-orchid monocots exhibit trnH-rps19 gene clusters, excluding Ψrpl22 genes, leading to more-progressive expansion of IRs compared to non-monocot angiosperms (Yang et al., 2010; Goulding et al., 1996). Contrarily, the orchid chloroplast genomes have distinct characteristics at the IR/SSC junction and are classified into four types based on the organization of genes flanking the IR_B/SSC junction (J_SB). In type I structure, J_SB is located upstream of the ndhF-rpl32 cluster and is primarily seen in Cypripedium and Dendrobium species. Type II junction is found in Cymbidium species in which J_SB is located within Ψycf1 and ndhF genes. Type III is reported in Oncidium, Erycina, and Phalaenopsis equestris, in which J_SB is located inside the Ψycf1-rpl32 cluster, with the loss of ndhF gene. The type IV structure is characterized by the incorporation of the entire ycf1 into the SSC, with J_SBinside trnN-rpl32 (Gardner, Slezak & Hall, 2015). In the present study, the positions of LSC/IRA/SSC/IRB borders were examined in the overall alignment of Dendrobium whole cp genomes and all of them were found to have similar structures at the IR/LSC junction akin to type I structure (Fig. 3). Previous studies emphasize that IR expansion or contraction may not correlate with the taxonomic relationships (Chen & Bundschuh, 2012). More molecular data is required for enhancing our present understanding of the genes flanking IR/SSC junctions and their underlying variations.

A comparative nucleotide sequence statistics (counts of annotations, AT/GC counts, nucleotide frequency in codon positions etc.) for all the Dendrobium species including representatives from outgroup are outlined in Tables 3, 4 and 5. The relative synonymous codon usage is given in parentheses following the codon frequency (averages over all taxa) involved (Table 6). Maximum Likelihood analysis of natural selection codon-by-codon was carried out. For each codon, estimates of the numbers of inferred synonymous (s) and nonsynonymous (n) substitutions are presented along with the number of sites that are estimated to be synonymous (S) and nonsynonymous (N) (Table S1). These estimates were calculated using the joint Maximum Likelihood reconstructions of ancestral states under a Muse-Gaut model (Muse & Gaut, 1994) of codon substitution and Felsenstein 1981 model  (Felsenstein, 1981) of nucleotide substitution. For estimating ML values, a tree topology was automatically computed. The test statistic dN-dS was used for detecting codons that have undergone positive selection, where dS is the number of synonymous substitutions per site (s/S) and dN is the number of nonsynonymous substitutions per site (n/N). A positive value for the test statistic indicates an overabundance of nonsynonymous substitutions. In this case, the probability of rejecting the null hypothesis of neutral evolution (p-value) was calculated  (Kosakovsky Pond & Frost, 2005; Suzuki & Gojobori, 1999). A value of p less than 0.05 was considered significant at a 5% level and was highlighted (Table S2). Normalized dN-dS for the test statistic is obtained using the total number of substitutions in the tree (measured in expected substitutions per site). The analysis involved 38 nucleotide sequences. Codon positions included were 1st+2nd+3rd+non-coding and all positions containing gaps and missing data were eliminated. There were a total of 108,594 positions in the final dataset.

Gene ontology analysis

The GO annotation revealed majority of the chloroplast genes are involved in the process of translation, photosynthesis, ion transport and transcription (Table 1). The molecular functions of the genes are majorly binding—RNA, metal ion, DNA, ion and electron transport, RNA polymerase activity and various other enzymatic activities. Enzyme classification showed seven genes to be translocases, four as transferases, two as oxidoreductases, and one each as hydrolase, lyase and ligase. A majority of the genes encode proteins localizing in chloroplast thylakoid membrane, ribosome and few are transported to the mitochondria. The ndhB gene is involved in photosynthesis, while rpoB and rpoC1 are involved in biosynthetic process.

Characterization of simple sequence repeats

Previous studies have documented prevalence of mononucleotide and dinucleotide SSRs in atleast 15 Dendrobium species from 92 syntenic intergenic and intronic loci. Of all these loci, 10(mutational hotspots: psbB-psbT, rpl16-rps3, trnR-atpA, trnL intron ndhF-rpl32, rpl32-trnL, trnT-trnL, clpB-psbB, rps16-trnQ and trnE-trnT) are reported to be the fastest evolving and are termed as top ten hotspots (Chen & Bundschuh, 2012). The SSRs lying in these regions could be further investigated for identifying potential markers that can aid in barcoding analysis.

Phylogenetic analyses

In the present study, we employed two different approaches for phylogeny reconstruction. First we aligned the whole cp genomes and exported the alignment matrices for creating a Bayesian tree (Fig. 5). Two independent MCMC chains were run with first 25% of the cycles removed as burn-in, coalescence of substitution rate and rate model parameters were also examined and average standard deviation of split frequencies was carried out and generations added until the standard deviation value was lowered to 0.01. Secondly we performed a phylogenetic tree construction using an alignment free approach. In this case we identified the SNPs from the cp genomes and utilised them in constructing the phylogenetic tree (Fig. 6). A total of 13,839 SNPs were identified in the 38 genomes analyzed, of which 2,203 were homoplastic SNPs i.e., SNPs that do not correspond to any node in the parsimony tree. The fraction of k-mers present in all genomes is 0.482. The numbers at the nodes in the phylogenetic tree indicate the number of SNPs that are present in all of the descendants of that node and absent in others (Fig. 6). The numbers at the tips indicate the number of SNPs unique to each particular species.

The two different methods that employed both alignment and alignment-free approach resulted in highly reliable identical phylogenetic trees within each data set. Different analyses based on the two datasets generated largely congruent topologies (Figs. 5 and 6) with Dendrobium species forming one clade and Goodyera species forming another clade as an outgroup.