Background

Panax notoginseng (Burk.) F.H. Chen, which is popularly known as Sanqi or Tienchi Ginseng, is a species of the Panax genus in the Araliaceae family [1]. P. notoginseng has been cultivated for about 400 years in China. It was previously considered a variety of Panax pseudo-ginseng, but in 1948 it was defined as an independent species of the Panax genus by Chen Feng-Huai and is now officially named Panax notoginseng (Burk.) F.H. Chen. Nowadays, more than 85% of the P. notoginseng in the worldwide market is produced in the city of Wenshan, Yunnan Province, China.

P. notoginseng is present in several famous traditional Chinese medicinal products, such as Yunnan Bai Yao (a remedy for injury induced by trauma and bleeding) and Pian Zai Huang (a remedy for relieving pain and detoxification). It is also famous for its haemostatic properties [2]. The classification of P. notoginseng from the American Herbal Products Association is Class 2b and it is indicated in pregnancy because of possible haemostatic effects. It was reported that P. notoginseng extract administered to rats after cerebral ischaemia reduced infarct volume and inhibited inflammatory inhibitors such as inducible nitric oxide synthase and cyclooxygenase 2 via blocking of the NF-κB pathway [3], suggesting a neuroprotective effect. Moreover, saponins of P. notoginseng extract were able to modulate the expression of caspases and attenuate apoptosis in rats following focal cerebral ischaemia-reperfusion [4]. In KK-Ay diabetic mice injected with P. notoginseng extract, significantly lowered fasting blood glucose levels, improved glucose tolerance and lighter body weights were observed [5]. Besides the roots of P. notoginseng, total saponins extracted from caudexes and leaves have been commonly used for improving mental function, treating insomnia, and alleviating anxiety [6]. The flower buds of P. notoginseng are also used in clinics for treating hypertension, vertigo, tinnitus and acute faucitis in China [7].

Chemically, the main bioactive compounds found in P. notoginseng are saponins, which have diverse biological activities such as membrane-permeabilising, immunostimulating, hypocholesterolemic, anti-carcinogenic, and anti-microbial activities [8-10]. The P. notoginseng-derived triterpene saponins include 20(S)-protopanaxadiol and 20(S)-protopanaxatriol, which exhibit opposing wound healing and anti-tumour actions on the vascular system [11]. Notably, P. notoginseng also shares many similar chemical constituents with Asian ginseng (P. ginseng C.A. Mey) and American ginseng (P. quinquefolius L.) [12]. These Panax species have species-specific saponin constituents, e.g. pseudo-ginsenoside F11 is unique to American ginseng whereas ginsenoside Rg3 is only present in Asian ginseng [13]. More than 60 chemotypes of P. notoginseng classified according to the accumulation of different ginsenosides in roots, leaves and flowers have been reported [11,14]. P. notoginseng contains significantly higher amounts of ginsenosides Rg1 and Rb1 compared with other ginseng species, and the ratios of Rg1/Re and Rg1/Rb1 in P. notoginseng are the highest among ginseng species. In particular, notoginsenoside R1 has been identified in P. notoginseng but is absent in other ginseng species.

Several previous studies have suggested that the precursor molecules for triterpene saponin biosynthesis are isoprenoids, which are synthesized via the mevalonic acid (MVA) pathway, leading to the biosynthesis of 2,3-oxidosqualene [15]. This central molecule is then modified through various biochemical reactions of its triterpene skeleton, resulting in the production of various ginsenosides. Notably, ginsenosides can be isolated from different parts of P. notoginseng, e.g. the underground parts of P. notoginseng are rich in protopanaxatriol- and protopanaxadiol-type saponins, while leaves and flowers contain protopanaxadiol-type saponins only. On the other hand, the ginsenosides Rc, Rb2 and Rb3 are relatively abundant in aerial parts, compared with the underground parts of P. notoginseng [11,16]. Although considerable research has been done on the pharmacological activities of ginsenosides, to date very little is known about the ginsenoside biosynthetic pathway. Some candidate genes likely to be involved in hydroxylation or glycosylation of aglycones for triterpene saponin biosynthesis are cytochrome P450s (CYP450) and glycosyltransferases (GT), but no candidate has been identified for the cyclization step [17].

In addition, P. notoginseng is a shade plant and is commonly cultivated in mountain areas of Wenshan at altitudes of 1200–2000 m around 23.5°N, 104°E [18]. Because of the humid and warm environment, P. notoginseng is easily infected by pests and diseases, especially in the roots. Alkaloids, which have been identified in more than 4,000 plant species, play a role in protecting plants from pathogen and pest damage [19]. However, alkaloid-related genes in P. notoginseng have not been reported previously.

Transcriptomic and genomic data for P. notoginseng are very limited despite the pharmacological importance of this plant. Considering there are thousands of genes in its genome, only 435 mRNA sequences originating from P. notoginseng could be retrieved from the nucleotide databases of the National Centre for Biotechnology Information (NCBI). Over the last decade, next-generation DNA sequencing technology has provided a rapid and economical way to study the gene expression profiles of plant species. In this study, we established transcript databases for leaves, roots and flowers from 3-year-old P. notoginseng. Moreover, we identified genes encoding enzymes involved in triterpene saponin and alkaloid biosynthesis. Differentially expressed CYP450s and GTs in the three tissues are also reported.

Results and discussion

Sequencing and de novo assembly

To study the transcriptomes of P. notoginseng, leaves, roots and flowers were collected from 3-year-old plants. Total RNA was extracted from each part and then mRNA was isolated. Each sample was sequenced using the Illumina HiSeq™ 2000 platform. Sequencing yielded approximately 213 million 90-base pair (bp) paired end raw reads, or approximately 19 Gbp in total. We filtered out adapter sequences and reads that were shorter than 50 bp, and ultimately generated 5.8, 5.8 and 6.1 Gbp of high-quality (HQ) reads for leaves, roots and flowers, respectively. All of the HQ sequencing reads from the three organs were deposited in NCBI and can be accessed in the Sequence Read Archive (SRA) under the BioProject accession number PRJNA228978. The HQ reads of each sample library were assembled using the Trinity software [20] and the TGI Clustering Tool (TGICL) [21], followed by the Phrap assembler [22] to remove redundant Trinity-generated contigs. Finally, 128,665, 94,258 and 124,888 unigenes were obtained for leaves, roots and flowers, respectively. We also pooled the reads from all three organs together and repeated the above steps, resulting in 205,000 contigs and 107,340 unigenes with a mean length of 781 bp and 1,039 bp, respectively. The length distribution of contigs and unigenes is shown in Additional file 1. A summary of sequencing and assembly results is shown in Table 1.

Table 1 Summary of Illumina sequencing and assembly of P. notoginseng
Full size table

Compared with a separate study of 4-year-old P. notoginseng [17], which generated 30,852 unigenes (14,005 contigs with mean length 581 bp and 16,847 singletons with mean length 343 bp), we present three times more and much longer unigenes from P. notoginseng roots (107,340 unigenes with mean length 1,039 bp). This is probably because of a 48-fold increase in sequencing throughput in this study. The large number of novel unigenes should cover the majority of genes in the P. notoginseng genome and provide a useful resource for future studies on this pharmacologically important plant.

Annotation and differential expression of transcripts in different tissues

All unigenes from leaves, roots and flowers were annotated separately using BLAST searches against the NCBI non-redundant protein (Nr), UniProt protein, The Arabidopsis Information Resource (TAIR), tomato genome (ITAG) and PlantCyc public databases. Detailed counts of the annotated unigenes are presented in Additional file 2. In total, there were 71,212, 58,182 and 75,404 annotated unigenes for leaves, roots and flowers, respectively, with at least one significant match in the aforementioned public databases. Figure 1 shows the number of unigenes annotated by all databases. The annotation percentage for P. notoginseng unigenes (74,839 out of 107,340, 69.72%) was much higher than that for P. ginseng unigenes (94,535 out of 178,145, 53.06%) [23]. When the unigenes from different P. notoginseng tissues were compared, we found that 41,373 unigenes were shared by all three tissues (Figure 2). On the other hand, 8,858, 5,814 and 10,649 unigenes were specifically found in leaves, roots and flowers, respectively, with flowers having the highest number of unique unigenes.

Figure 1
figure 1

Venn diagram of unigenes from leaves, roots and flowers annotated using various databases. The Venn diagram shows the overlapping unigenes annotated in the Nr, Uniprot, TAIR, ITAG and PlantCyc databases in leaves (A), roots (B) and flowers (C).

Full size image
Figure 2
figure 2

Venn diagram of the unigenes in the leaves, roots and flowers of P. notoginseng . The diagram shows the overlapping unigenes in the leaves, roots and flowers. A total 41,373 unigenes were found in all three tissues, while some unigenes only were found in specific tissues.

Full size image

We next investigated the transcriptomic similarities and differences between P. notoginseng and P. ginseng using BLAST searches of P. notoginseng unigenes against P. ginseng unigenes, with an E-value threshold of 1e-10 [23]. P. ginseng transcriptome contigs were obtained under accession number GAAG00000000 from the NCBI. The results showed that 77,470 unigenes of P. notoginseng had at least one match to the contigs of P. ginseng, suggesting that many of our unigenes are P. notoginseng-specific. Next, proteins encoded by our unigenes were predicted by the Trinity software and orthologous groups were identified using OrthoMCL [24], which groups orthologous proteins based on sequence similarity. A total of 64,742 protein sequences from 107,340 unigenes were clustered into 9,949 orthologous groups for P. notoginseng. For P. ginseng, a total of 29,289 protein sequences from 67,786 contigs were clustered into 8,959 orthologous groups. We found that 8,424 groups were common to both species, with 1,525 orthologous groups specific to P. notoginseng, and 535 groups specific to P. ginseng.

To evaluate the abundance of transcripts in specific organs, the high quality reads were mapped back to the transcriptome generated from all three plant tissues. Among the top 10 most expressed unigenes in the root transcriptome (Additional file 3), one of the most abundant genes was dammarenediol-II synthase (DS), which is an important player in triterpene saponin biosynthesis in P. notoginseng [25]. In addition, unigenes encoding cytochrome P450 CYP716A47, which is involved in the conversion of protopanaxadiol from dammarenediol-II, were also present at a very high level. This is consistent with a previous report about the co-expression of DS and CYP716A47 in P. ginseng [26]. Last but not least, reticuline oxidase-like protein-like isoform 1 is a predicted protein with sequence similarity to reticuline oxidase, which is involved in forming benzophenanthridine alkaloids as a pathogenic attack response.

Identification of genes involved in triterpene saponin biosynthesis

The Kyoto Encyclopaedia of Genes and Genomes (KEGG) is a tool for functional classification and pathway assignment based on gene-associated biochemical pathways. In total, 9,908 unigenes from all three tissues having enzyme commission numbers were assigned to 135 KEGG pathways (Additional file 4). The cluster for metabolism represented the largest group, with most unigenes involved in carbohydrate metabolism, amino acid metabolism, nucleotide metabolism and metabolism of cofactors and vitamins. Among them, 616 unigenes were involved in the biosynthesis of various secondary metabolites (Table 2), with unigenes involved in phenylpropanoid biosynthesis forming the largest group, followed by terpenoid backbone biosynthesis.

Table 2 Number of unigenes related to secondary metabolites in P. notoginseng
Full size table

Triterpene saponins are synthesized by terpenoid backbone biosynthesis, followed by sesquiterpenoid and triterpenoid biosynthesis. According to the putative pathway, specific CYP450s and GTs are involved in the formation of various ginsenosides (Figure 3). From our annotation results, 270 unigenes were identified to encode all of the known enzymes involved in triterpene saponin biosynthesis. There were multiple unigenes annotated to the same enzyme, which may represent different members of the same gene family. Table 3 shows the reads per kilobase of transcript per million reads mapped (RPKM) values of genes encoding enzymes involved in triterpene saponin biosynthesis in leaves, roots and flowers. The RPKMs of all annotated isoforms for the same gene were summed as the RPKM of that gene. We identified 12 out of 14 full-length cDNA encoding enzymes involved in triterpene saponin biosynthesis and their corresponding accession numbers in NCBI nucleotide databases are listed in Table 4. A heat map of differentially expressed genes involved in triterpene saponin biosynthesis is shown in Figure 3. Significant differential expression of these genes could be recognized in different P. notoginseng tissues, and most of the genes involved in triterpene saponin biosynthesis showed higher expression in roots compared with leaves or flowers. The expression levels of selected genes (8 out of 14) in each part were validated by real-time PCR (Figure 4).

Figure 3
figure 3

Putative ginsenoside biosynthesis in P. notoginseng . Enzymes found in this study are shown between the reactions. The expression of genes encoding these enzymes in the leaves (L), roots (R) and flowers (F) is shown by heatmap. The genes were mapped using RPKM values, colour coded by increasing relative expression. The broken arrow represents a putative ginsenoside biosynthesis step involving CYP450s and GTs. Abbreviations: AACT, acetyl-CoA acetyltransferase; HMGS, HMG-CoA synthase; HMGR, HMG-CoA reductase; MK, mevalonate kinase; PMK, phosphomevalonate kinase; MDD, mevalonate diphosphate decarboxylase; IPI, isopentenylpyrophosphate isomerase; GGPS, geranylgeranyl pyrophosphate synthase; GGR, geranylgeranyl diphosphate synthase; FPS, farnesyl diphosphate synthase; SS, squalene synthase; SE, squalene epoxidase; AS, β-amyrin synthase; DS, dammarenediol-II synthase; CYP450, cytochrome P450; GT, glycosyltransferase.

Full size image
Table 3 Discovery of unigenes involved in triterpene saponin biosynthesis in P. notoginseng
Full size table
Table 4 Genes involved in triterpene saponin biosynthesis in P. notoginseng
Full size table
Figure 4
figure 4

Real-time PCR analysis of selected genes involved in triterpene saponin biosynthesis. Real-time PCR was used to validate the expression levels of selected genes revealed by RNA-seq. Abbreviations: AACT, acetyl-CoA acetyltransferase; HMGS, HMG-CoA synthase; MK, mevalonate kinase; MDD, mevalonate diphosphate decarboxylase; IPI, isopentenylpyrophosphate isomerase; FPS, farnesyl diphosphate synthase; SS, squalene synthase; SE, squalene epoxidase.

Full size image

Luo et al. [17] previously analysed the root transcriptome of 4-year-old P. notoginseng and discovered many partial cDNAs encoding enzymes involved in triterpene saponin biosynthesis. However, hydroxymethyl glutaryl CoA synthase (HMGS), a key enzyme of triterpene saponin biosynthesis, was absent in their transcriptome. In contrast, the HMGS gene was abundant in our transcriptomes, probably because of the much greater sequencing depth achieved in our study. Therefore, compared with other published transcriptomes, the transcriptomes reported in this study will be more useful for cloning important genes involved in secondary metabolite biosynthetic pathways of P. notoginseng. Moreover, β-amyrin synthase (AS), a key enzyme for oleanane-type ginsenoside biosynthesis, was also found in this study. Notably, AS was discovered in all three tissues of 3-year-old P. notoginseng (Table 3).

Identification of genes related to biosynthesis of different ginsenosides

There are published reports showing that different parts of P. notoginseng differ in the synthesis of triterpene saponins. Roots are rich in protopanaxatriol- and protopanaxadiol-type saponins whereas leaves and flowers contain protopanaxadiol-type saponins only [11]. It has been suggested that putative candidate genes involved in triterpene saponins biosynthesis are mainly CYP450s and GTs, which may account for the synthesis and accumulation of triterpene saponins in specific organs [27,28]. In this study, 350 and 342 members of the CYP450 and GT gene families, respectively, were identified (Additional files 5 and 6).

It has been reported that the cytochrome P450 CYP716A53v2 participates in the formation of protopanaxatriol from protopanaxadiol in P. ginseng [29]. In our transcriptome, one unigene annotated as CYP716A53v2 was found and subsequent sequence analysis showed that it was the full-length homologue in P. notoginseng. Figure 5 shows the alignment of the predicted protein sequences of CYP716A53v2 from P. notoginseng, P. ginseng and P. quinquefolius. The accession numbers of the sequences are JX036031 for P. ginseng and KC190491 for P. quinquefolius in the GenBank database. All three CYP450 genes had 469 amino acid residues and showed 98% identity between P. notoginseng and P. ginseng, 96% identity between P. notoginseng and P. quinquefolius, and 97% identity between P. ginseng and P. quinquefolius. It was also very encouraging to find that the expression level of the putative CYP716A53v2 in the root (RPKM value 414.91) was much higher than that in the leaf (RPKM value 1.83, P < 0.001) or flower (RPKM value 4.62, P < 0.001) of P. notoginseng (Figure 6). To confirm the differential expression of the putative CYP716A53v2 gene, we analysed its expression level in different tissues by real-time PCR. The result obtained was consistent with the RPKM values, showing that the expression of the putative CYP716A53v2 in roots was significantly higher than in leaves or flowers (Figure 6). This may explain, at least partially, why the root has a much higher concentration of protopanaxatriol-type saponins. Moreover, this result also indicates that the abundance of target genes in our transcriptomes closely reflects the actual gene expression level.

Figure 5
figure 5

Alignment of CYP716A53v2 amino acid sequences from P. notoginseng , P. ginseng and P. quinquefolium . Alignment results showed 98% identity between P. notoginseng and P. ginseng, 96% identity between P. notoginseng and P. quinquefolius, and 97% identity between P. ginseng and P. quinquefolius.

Full size image
Figure 6
figure 6

Expression levels of target CYP450s in different P. notoginseng tissues. Comparison of relative abundance of target CYP450 unigenes in the leaf, root and flower according to real time PCR (A) and RPKM values (B).

Full size image

Identification of candidate genes involved in alkaloid biosynthesis

Alkaloids, which are found in about 20% of plant species, are a diverse group of low-molecular-weight compounds acting as poisonous agents in the defence of plants against herbivores and pathogens [19]. P. notoginseng grows sub-optimally in direct sunlight and so is often planted under tree canopies, but the shady and humid growing conditions favour infection by numerous phytopathogens, which can cause root rot, black spot or round spot diseases [18].

In this study, three pathways involved in alkaloid biosynthesis were found in our transcriptomes (Table 2), including isoquinoline alkaloid biosynthesis (KEGG pathway entry ko00950), indole alkaloid biosynthesis (ko00901) and tropane, piperidine and pyridine alkaloid biosynthesis (ko00945). The annotation results from the three transcriptomes were used to identify genes encoding enzymes involved in various alkaloid biosynthetic pathways. In total, 72 unigenes were assigned to six enzymes involved in alkaloid biosynthesis (Additional file 7). According to the RPKM values of the genes encoding enzymes involved in alkaloid biosynthesis in leaves, roots and flowers shown in Table 5, most of the enzymes were expressed at the lowest level in roots, in particular polyphenol oxidase (PPO), which has a role in plant resistance to stress and pathogens. It is notable that wounding and herbivore attacks have also been shown to induce PPO activity [30]. PPO activation is thought to involve proteolytic processing, but many mature PPOs appear to remain in a latent form [31]. Root rot disease is the most damaging disease that can plague P. notoginseng over its long growth period, leading to production loss and quality reduction. It is mainly caused by fungal pathogens such as Cylindrocarpon destructans, Cylindrocarpon didymum and Fusarium solani [32]. In this study, the expression levels of aspartate transaminase, strictosidine synthase and histidinol-phosphate transaminase in different tissues were validated by real-time PCR (Figure 7). Further investigation of the relationship between alkaloid biosynthesis genes and plant defence should provide novel insights into the complex disease resistance mechanisms of P. notoginseng.

Table 5 Discovery of unigenes involved in alkaloid biosynthesis in P. notoginseng
Full size table
Figure 7
figure 7

Real-time PCR analysis of genes involved in the alkaloid pathway. The expression levels of selected genes in the alkaloid pathway were validated by real-time PCR.

Full size image

Conclusions

P. notoginseng is a widely used medicinal herb. In this study, leaf, root and flower transcriptomes from P. notoginseng are presented. The resulting unigene dataset will provide a large number of transcripts for gene discovery and genetic analyses in this medicinal plant. Notably, many genes involved in triterpene saponin biosynthesis were identified in this study. More importantly, a large number of unigenes were annotated as CYP450s and GTs. In summary, this study provides comprehensive information on the transcriptional regulation of functionally important genes in P. notoginseng.

Methods

Plant materials

Growing 3-year-old P. notoginseng was randomly collected from the field of a P. notoginseng commercial planting base in Wenshan County, Yunnan Province, China on September 2, 2012. The highest/lowest temperature on that day was 28/20°C, and the relative humidity was 78%. The soil of the field was a sandy loam soil with a pH value of 5.5–7.0. After cleansing, the leaves, roots and flowers were collected separately, cut into small pieces, immediately frozen in liquid nitrogen, and stored at −80°C until further processing.

RNA extraction

mRNA isolation, cDNA library construction and sequencing were performed by the Beijing Genomics Institute (BGI) (Shenzhen, China). Briefly, total RNA was extracted from each tissue using TRIzol reagent (Invitrogen, Burlington, ON, Canada) and digested with DNase I (Takara, Dalian, China) according to the manufacturer’s protocol. Next, Oligo (dT) magnetic beads were used to isolate mRNA from the total RNA. By mixing with fragmentation buffer, the mRNA was then broken into short fragments.

cDNA synthesis and sequencing

The cDNA was synthesized using the mRNA fragments as templates. The short fragments were purified and resolved with EB buffer for end repair and single nucleotide A (adenine) addition, and then connected with adapters. Suitable fragments were selected for PCR amplification as templates. During the quality control steps, an Agilent 2100 Bioanalyzer (Agilent Technologies, Redwood City, CA, USA) and ABI StepOnePlus Real-Time PCR System (Life Technologies, Grand Island, NY, USA) were used for quantification and qualification of the sample library. Each cDNA library was sequenced in a single lane of the Illumina HiSeqTM 2000 system using paired end protocols according to the manufacturer’s instructions at the Beijing Genomics Institute (BGI) (Shenzhen, China). The amount of reads generated per sample was 5–8 Gb to obtain deep coverage of transcripts for de novo assembly [33]. As determined by SAMtools [34], the average depth of the generated P. notoginseng transcriptomes was over 130 × .

De novo assembly and sequence annotation

The raw reads dataset was first processed to remove the reads with adaptors or containing more than five unknown (‘N’) nucleotides. Next, the low quality reads (defined as reads having more than 20% of bases with quality ≤ 10) were trimmed. We used the Trinity software (release-20121005) [20] for de novo assembly of the high-quality cleaned reads, and then used TGICL [21] followed by the Phrap [22] assembler to remove the redundant Trinity-generated contigs.

The resulting unigenes were annotated by BLASTx [35] searches against the NCBI nr (ftp://ftp.ncbi.nih.gov/blast/db/FASTA/nr.gz), UniProt protein (http://www.uniprot.org/downloads), TAIR (http://www.arabidopsis.org/index.jsp), ITAG (http://solgenomics.net/organism/Solanum_lycopersicum/genome) and PlantCyc (ftp://ftp.plantcyc.org/Pathways/BLAST_sets/) databases with a cut-off value of 1e-5. The top hit was extracted for each unigene.

Groups of orthologous proteins were identified using the OrthoMCL algorithm [24]. To gain an overview of gene networks, KEGG pathway information was assigned to each unigene, based on similarity with the KEGG database [36] using a BLAST search with a cut-off value of 1e-5. The KEGG analysis output included enzyme commission (EC) numbers and KEGG orthology (KO) numbers.

Digital gene expression profiling

The high-quality reads were aligned to the assembled unigenes with the BWA program [37]. An RPKM value was calculated for each unigene in each tissue of P. notoginseng. The RPKMs of all annotated isoforms for the same gene were summed as the RPKM of that gene. Differential expression of unigenes was calculated with a threshold of P value < 0.001 and 2-fold change.

Real-time PCR analysis

RNA was isolated from different P. notoginseng tissues and reverse-transcribed to single-strand cDNA using the Super Script™ III First-Strand Synthesis System (Invitrogen™, USA). Quantitative reactions were performed on the Real-Time PCR Detection System (ABI 7500, Applied Biosystems, USA) using SYBRR Premix Ex Taq™ II (Takara Biotechnology, China). The reaction mixture (20 μL) contained 2× SYBRR Premix Ex Taq™ II, 0.4 μM each of the forward and reverse primers, and 2 μL of template cDNA. PCR amplification was performed under the following conditions: 95°C for 30 s, followed by 40 cycles of 95°C for 5 s and 60°C for 34 s, and with a dissociation stage of 95°C for 15 s, 60°C for 60 s and 95°C for 15 s. All primers used in this study are listed in Additional file 8. The relative gene expression was calculated with the ΔΔCT method. For each sample, the mRNA levels of the target genes were normalized to that of the actin mRNA. These experiments were repeated using three biological replications.

Ethical statement

There is no conflict of interest. An exemption from requiring ethics approval has been granted from the Joint Chinese University of Hong Kong - New Territories East Cluster Clinical Research Ethics Committee.

Availability of supporting data

All RNA sequences and raw reads data are available under accession numbers KJ804166 to KJ804179, KF935232 and BioProject accession PRJNA228978.