Identification, Molecular Cloning and Expression Analysis of Five RNA-Dependent RNA Polymerase Genes in Salvia miltiorrhiza

RNA-dependent RNA polymerases (RDRs) act as key components of the small RNA biogenesis pathways and play significant roles in post-transcriptional gene silencing (PTGS) and antiviral defense. However, there is no information about the RDR gene family in Salvia miltiorrhiza, an emerging model medicinal plant with great economic value. Through genome-wide predication and subsequent molecular cloning, five full-length S. miltiorrhiza RDR genes, termed SmRDR1–SmRDR5, were identified. The length of SmRDR cDNAs varies between 3,262 (SmRDR5) and 4,130 bp (SmRDR3). The intron number of SmRDR genes varies from 3 (SmRDR1, SmRDR3 and SmRDR4) to 17 (SmRDR5). All of the deduced SmRDR protein sequences contain the conserved RdRp domain. Moreover, SmRDR2 and SmRDR4 have an additional RRM domain. Based on the phylogenetic tree constructed with sixteen RDRs from Arabidopsis, rice and S. miltiorrhiza, plant RDRs may be divided into four groups (RDR1–RDR4). The RDR1 group contains an AtRDR and an OsRDR, while includes two SmRDRs. On the contrary, the RDR3 group contains three AtRDRs and two OsRDRs, but has only one SmRDR. SmRDRs were differentially expressed in flowers, leaves, stems and roots of S. miltiorrhiza and responsive to methyl jasmonate treatment and cucumber mosaic virus infection. The results suggest the involvement of RDRs in S. miltiorrhiza development and response to abiotic and biotic stresses. It provides a foundation for further studying the regulation and biological functions of SmRDRs and the biogenesis pathways of small RNAs in S. miltiorrhiza.


Introduction
RNA-dependent RNA polymerases (RDRs), which catalyze the conversion of single-stranded RNAs (ssRNAs) into doublestranded ones, play vital roles in the production of various small interfering RNA (siRNA) species in plants through collaboration with other proteins, such as Dicer-likes (DCLs) capable of cleaving double-stranded RNAs (dsRNAs) into 21-24 nt duplexes [1,2]. Based on the origins and biogenesis pathways, siRNAs generated from dsRNAs can be classified into several groups, such as transacting small interfering RNAs (ta-siRNAs), heterochromatic siRNAs (hsiRNAs) and natural antisense transcript-derived siRNAs (nat-siRNAs) [3][4][5]. These siRNAs may incorporated into the Argonaute (AGO)-containing RNA-induced silencing complexes (RISCs) to silence a variety of gene transcripts, repetitive sequences, sense transgenes, viruses and mobile elements through RNA cleavage, translational inhibition, DNA methylation and heterochromatin formation [6]. RDR proteins are characterized by the conserved RNAdependent RNA polymerase catalytic domain (RdRp) and are among the first components identified for plant small RNA biogenesis pathways [7][8][9][10]. They are present in fungi, viruses, plants and nematodes, but have not been found in insects and vertebrates [11,12]. The activity of RDR was detected in Chinese cabbage more than forty years ago [8], whereas the cDNA of RDR was first isolated from tomato in 1998 [13]. So far, RDR genes have been identified in various plant species, such as Arabidopsis thaliana, rice, maize, tomato and tobacco [14][15][16][17]. Similar to DCLs and AGOs, RDR genes exist as a family in plants and the number of RDR genes may be not the same in different species. For instance, there are six RDR members in Arabidopsis and Solanum lycopersicum [14], five in rice [15] and maize [16], and at least three in Nicotiana tobaccum and N. attenuate [17]. Each small RNA biogenesis pathway may involve different member of the RDR family.
The six members included in the A. thaliana AtRDR gene family were termed AtRDR1-AtRDR6, respectively [14]. AtRDR1, AtRDR2 and AtRDR6 play distinct and overlapping functions in various aspects, such as viral resistance, chromatin silencing and post-transcriptional gene silencing (PTGS) [19,20]. AtRDR1 and its ortholog in tobacco, NtRDR1, are induced by salicylic acid (SA) treatment and virus infection and involved in plant susceptibility to tobacco mosaic tobamovirus (TMV) and tobacco rattle virus (TRV) [21]. The underlying functional mechanism of AtRDR1 appears to produce and amplify exogenous and virus-derived siRNAs (vsiRNAs) in infected plants [19][20][21][22][23][24]. In addition to antiviral responses, RDR1 plays a significant role in plant resistance to herbivore attack [18]. AtRDR2 is involved in the production of the most abundant endogenous hsiRNAs that are mostly 24 nt in length and are associated with heterochromatic and repetitive regions, such as the pericentromeric regions and telomeres [25]. RDR2-dependent RNA-directed DNA methylation (RdDM) is responsible for siRNA-mediated DNA methylation and histone modifications at Arabidopsis telomeres, and is required for the maintenance of telomeric heterochromatin [25]. In addition, AtRDR2 is involved in the development of the female gametophyte [26]. AtRDR6, acts in various gene silencing pathways and plays important roles in the biogenesis of ta-siRNA and nat-siRNA. It is also well-known for the amplification of improper terminated and unpolyadenylated RNAs generated from transgenes or inverted repeats to trigger degradation of complementary RNA species [27]. The functions of AtRDR3-AtRDR5 are currently unknown [7].
Although RDR genes have been isolated from various plant species, to our best knowledge, there is no report on RDRs in medicinal plants. Salvia miltiorrhiza, well-known as danshen in Chinese, is an economically significant medicinal plant and is an emerging model medicinal plant for Traditional Chinese Medicine (TCM) studies [28]. It has been used for treating various human diseases, such as dysmenorrhoea, amenorrhoea and cardiovascular disease, for thousands of years [28][29][30]. The S. miltiorrhiza genome has been preliminarily decoded (Chen et al., unpublished data). We have previously analyzed the S. miltiorrhiza AGO gene families involved in small RNA production and action [31]. In order to characterize the RDR genes in S. miltiorrhiza, we performed a genome-wide search of RDRs against the working draft of the S. miltiorrhiza genome, followed by molecular cloning and molecular analysis. The results will be a basis for understanding the gene silencing pathways in S. miltiorrhiza.

Materials and Methods
Plant materials and stress treatment S. miltiorrhiza Bunge (line 993) were grown in a field nursery. Flowers, leaves, stems and roots were collected from two-year-old plants and stored in liquid nitrogen until use. Plantlets cultivated in vitro were grown at 25uC with a photoperiod of 16 h light and 8 h dark for six weeks as described previously [32]. MeJA treatment were carried out following the procedures reported previously [32,33]. Plantlets were treated for 12, 24, 36 and 48 h and then sampled. Sterile water-treated plantlets were used as controls. For cucumber mosaic virus (CMV) infection, the silicon carbide powder friction method was used. Briefly, leaves of six-week-old plantlets cultivated in vitro were dusted with silicon carbide powder and then inoculated with CMV subgroup I for 12, 24, 48 and 72 h. Leaves inoculated with phosphate buffered saline (PBS) were used as controls. All tissues collected were stored in liquid nitrogen until use. Three independent biological replicates were performed for each experiment.

Molecular cloning of SmRDR cDNAs
The 59-RACE and 39-RACE were carried out as described previously [31]. Briefly, total RNA extracted from the root of S. miltiorrhiza was purified using the oligotex mRNA mini kit (Invitrogen). 59 and 39 RACE was performed on mRNA using the GeneRacer kit (Invitrogen). Gene specific nesting and nested primers were designed and synthesized (Tables S1 and S2). PCR products were purified, cloned and sequenced. Based on the obtained 59 and 39 cDNA sequence, gene-specific forward and  reverse primers were designed and synthesized (Table S3). Fulllength SmRDR cDNAs were PCR-amplified, cloned and sequenced as described [31].

Bioinformatic analysis and phylogenetic tree construction
Bioinformatic analysis of SmRDR sequence features, such as intron/exon structures, molecular weight (MW), theoretical isoelectric point (pI) and conserved domain, were performed as described previously [31,36]. The conserved motifs of SmRDRs were analyzed using Multiple Expectation Maximization for Motif Elicitation (MEME) version 4.9.1 [37] with the following parameters. Optimum motif width was set to §6 and !50. The maximum number of motifs was designated to identify 20 motifs following the previous reported studies [17]. The conserved residues were analyzed by alignment of amino acid sequences using T-coffee [38]. Phylogenetic tree was constructed for protein sequences of sixteen RDRs from S. miltiorrhiza, Arabidopsis and rice using MEGA version 4.0 by the neighbor-joining method with 1,000 bootstrap replicates [39,40].

Quantitative real-time reverse transcription-PCR (qRT-PCR)
Expression of SmRDRs in roots, stems, leaves and flowers of 2year-old S. miltiorrhiza plants and in plantlets treated with MeJA and CMV was analyzed using the qRT-PCR method as described previously [31]. Briefly, Gene-specific forward and reverse primers were designed and synthesized (Table S4). About 10 ng cDNA reversely transcribed from total RNA was used as a template in a 20 ml volume. SmUBQ10 was used as a reference [31]. qPCR was carried out in triplicates for each biological sample using the BIO-RAD CFX system (Bio-Rad). Three fully independent biological replicates were performed. The specificity of amplification was assessed by dissociation curve analysis. Gene expression levels were determined using the 2 2DDCq method, where Cq represents the threshold cycle [41]. Relative amount of transcripts was calculated and normalized as described previously [42]. Average Cq were log transformed, mean centered and autoscaled [42]. Standard deviations of mean value from three biological replicates were calculated as described previously [42].

Results
Identification and molecular cloning of five S. miltiorrhiza RDR genes BLAST analysis of Arabidopsis and rice RDRs against the working draft of the S. miltiorrhiza genome (Chen et al., unpublished data) using tBLASTn [34,35] showed the existence of five SmRDR gene loci in the S. miltiorrhiza genome. Genomic DNA sequences were retrieved and predicted for gene models on the GENSCAN web server (http://genes.mit.edu/GENSCAN. html). The five gene models computationally predicted were BLAST-analyzed against the non-redundant protein sequence (nr) database (http://www.ncbi.nlm.nih.gov/BLAST) using BLASTx with default parameters and then manually corrected according to the alignment between SmRDRs and other plant RDRs. To further experimentally validate the predicted cDNA sequences of SmRDRs, molecular cloning of full-length SmRDR cDNA was carried out using RNA ligase-mediated rapid amplification of 59 (59 RACE) and 39 (39 RACE) cDNA ends and subsequent PCR amplification of coding regions. The deduced amino acid sequences of all five SmRDRs share high sequence identity with known plant RDRs and contain the conserved RdRp domain, suggesting they are authentic SmRDRs. The identifed SmRDR genes are termed SmRDR1-SmRDR5, respectively. The cloned cDNAs have been submitted to GenBank under the accession numbers KF872203-KF872207.

Gene structure and conserved domain analyses
Sequence feature analysis of SmRDRs suggests that the length of open reading frames (ORFs) of SmRDRs varies from 2,697 (SmRDR5) to 3,588 bp (SmRDR3) ( Table 1). The length of 59 and   Table 1). The SmRDR5 locus, which produces the shortest SmRDR, has seventeen introns ( Fig. 1, Table 1). SmRDR2 contains four introns including one is located within the 59 untranslated region (UTR) (Fig. 1, Table 1). The rest three, including SmRDR1, SmRDR3 and SmRDR4, have 3 introns (Fig. 1, Table 1). All of the introns of SmRDR1 and SmRDR4 are located in the coding regions; however, of the 3 introns of SmRDR3, only one is located in the coding region. The other two are located within the 59 UTR. These 59 UTR-located introns might enhance gene transcription and RNA stability [43][44][45]. Additionally, the size of introns varies significantly among SmRDRs (Fig. 1). It suggests the diversity of SmRDRs in sequence features and gene structures.
The search for conserved domains in SmRDR proteins against the NCBI Conserved Domain Database showed that all of the five SmRDRs contained the conserved RdRp domain (Fig. 2). It is consistent with the results from other plant RDRs [15][16][17]. SmRDR2 and SmRDR4 have an additional RNA recognition motif (RRM) in the region close to the N-terminus (Fig. 2). RRM in SmRDR2 starts from the 4 th amino acid and ends at the 66 th , while SmRDR4 RRM starts from the 14 th amino acid and ends at the 91 st . The actual function of additional RRM in RDR proteins remains to be elucidated [46].
Using the MEME motif search tool, we analyzed conserved motifs of SmRDRs (Fig. 3). The results revealed four motifs (1, 7, 10 and 12) conserved in all of the five SmRDRs. It suggests the conservation of SmRDRs. On the other hand, various less conserved motifs were found. For instance, motifs 11, 15, 16, 17 and 19 are specific to SmRDR1 and SmRDR2. Motifs 13 and 20 are specific to SmRDR1, SmRDR2 and SmRDR3. Additionally, among the five SmRDRs, SmRDR1 and SmRDR2 have 20 motifs, whereas SmRDR5 only contain 4. These less conserved motifs could be associated with gene-specific functions. Sequence alignment of RDR proteins from S. miltiorrhiza and Arabidopsis using T-coffee [38] showed that SmRDRs contained the DLDGD/ DFDGD signature, the partial sequence of motif 1 (Figs. 3 and 4).  DLDGD/DFDGD has been previously found in various other plant RDRs and seems to be part of the nucleotidyl transferase active site of RDR proteins [47]. It further confirms the role of identified SmRDRs in the conversion of ssRNAs into dsRNAs.

Phylogenetic analysis of RDR proteins in S. miltiorrhiza, Arabidopsis and rice
Previous study revealed that RDRs in eukaryotic organisms might be divided into three clades, RDRa, RDRb, and RDRc [47]. RDRaproteins exist in all three kingdoms, whereas the proteins included in the RDRb clades are present only in animals and fungi and RDRcproteins are found only in plants and fungi. Among the six Arabidopsis AtRDRs, AtRDR1, AtRDR2 and AtRDR6 belong to the RDRa clade, while AtRDR3, AtRDR4 and AtRDR5 are included in the RDRcclade [47]. To determine the evolutionary relationship among RDRs from S. miltiorrhiza, Arabidopsis and rice, an unrooted neighbor-joining tree was constructed for the full-length protein sequences of five SmRDRs, six AtRDRs and five OsRDRs. The results showed that the sixteen RDRs might be divided into four groups, termed RDR1, RDR2, RDR3 and RDR4, respectively (Fig. 5). It is consistent with previous results for RDRs from Arabidopsis, rice and Zea mays [15][16][17]. The RDR1, RDR2 and RDR4 groups contain SmRDR1, SmRDR2, SmRDR3 and SmRDR4 from S. miltiorrhiza, AtRDR1, AtRDR2 and AtRDR6 from Arabidopsis, and OsRDR1, OsRDR2 and OsRDR6 from rice. All of them belong to the RDRaclade and are characterized by the DLDGD signature (Fig. 4). SmRDR1 and SmRDR2 and the antiviral defense-associated AtRDR1 and OsRDR1 [15,16] are included in the RDR1 group. Members of the RDR2 group includes SmRDR4, Arabidopsis AtRDR2 and rice OsRDR2, of which AtRDR2 and OsRDR2 are involved in the production of the most abundant endogenous hsiRNAs and are responsible for siRNA-mediated DNA methylation and histone modifications at telomeres [25]. The RDR4 group contains SmRDR3, AtRDR6 and OsRDR6. AtRDR6 is associated with amplifying improper terminated and unpolyadenylated RNAs generated from transgenes or inverted repeats [27]. The RDR3 group includes AtRDR3, AtRDR4 and AtRDR5 from Arabidopsis, OsRDR3 and OsRDR4 from rice and SmRDR5 from S. miltiorrhiza. Members of the RDR3 group belong to the RDRcclade and are characterized by the DFDGD signature (Fig. 4). Although it is the biggest group, the function of RDRs in the RDR3 group is currently unknown.

Tissue-specific expression of SmRDR genes
To preliminarily elucidate the function of SmRDR genes, we analyzed the expression patterns of five identified SmRDRs in flowers, leaves, stems and roots of 2-year-old and field nurserygrown S. miltiorrhiza using the quantitative RT-PCR technology. The transcripts of all five SmRDRs could be detected in the tissues analyzed (Fig. 6), which is consistent with the vital roles of RDRs in plants. All of them showed the highest expression in roots and less in flowers and leaves (Fig.6). Further sequencing and analyzing the sRNAome in S. miltiorrhiza may help to elucidate the underlying mechanisms.

The response of SmRDRs to MeJA treatment and CMV infection
RDRs catalyze the conversion of single-stranded RNAs into double-stranded ones and are core components for the production of siRNAs involved in plant development and response to abiotic and biotic stresses. In order to investigate the expression pattern of SmRDR genes under abiotic treatments, the expression level of SmRDRs in leaves of plantlets treated with MeJA was analyzed using the quantitative RT-PCR method. Plantlets treated with sterile water were used as controls. The results showed that all five SmRDRs were suppressed under MeJA treatment (Fig. 7). MeJA usually up-regulates the expression of genes associated with the biosynthesis of secondary metabolites, which play significant roles in plant response to stress [32,33]. Down-regulation of SmRDR gene expression could be helpful to increase the production of stress-related secondary metabolites.
To examine the response of SmRDRs in biotic stress, S. miltiorrhiza plantlets were inoculated with CMV using the silicon carbide powder friction method. The level of SmRDR transcripts in leaves treated for 12, 24, 48 and 72 h was analyzed using the quantitative RT-PCR method. Leaves inoculated with phosphate buffered saline (PBS) were used as controls. As shown in Fig. 8, SmRDR1, SmRDR2 and SmRDR3 were up-regulated at different time points. The level of SmRDR1 showed a 1.9-fold-increase after CMV inoculation for 48 h (Fig. 8A). SmRDR2 showed 2.3-, 3.4and 4.2-fold-increase after being treated with CMV for 12, 48 and Figure 6. Expression of SmRDRs in flowers (Fl), leaves (Le), stems (St) and roots (Rt) of S. miltiorrhiza. The expression patterns were analyzed using the quantitative RT-PCR method. PCR was carried out in triplicates for each biological sample. Three independent biological replicates were performed. SmUBQ10 was used as a reference. Fold changes of SmRDR expression are shown. The levels in roots were arbitrarily set to 1 and the levels in other tissues were given relative to this. Error bars represent the standard deviations of the mean value of three biological replicates. doi:10.1371/journal.pone.0095117.g006 72 h, respectively (Fig. 8B). SmRDR3 was accumulated to 3.7-and 2.3-folds of controls after CMV infection for 12 and 72 h (Fig. 8C). However, no significant change was observed for the level of SmRDR4 and SmRDR5 after CMV infection (Figs. 8D and 8E).
The results indicate that SmRDR1, SmRDR2 and SmRDR3 may be involved in antiviral defense in S. miltiorrhiza.

Discussion
S. miltiorrhiza is an economically significant medicinal plant species belonging to the largest genus, Salvia, in the mint family. It is native to China and Japan and is widely distributed in China. S. miltiorrhiza has close phylogenetic relationships with other Asian and Mediterranean Salvia species, such as S. roborowskii and S. glutinosa [48]. The root of S. miltiorrhiza has been widely used in TCMs for hundred of years to treat dysmenorrhoea, amenorrhoea and cardiovascular diseases [28][29][30]. The main bioactive components in S. miltiorrhiza are lipophilic diterpenoid tanshinones and hydrophilic phenolic acids. Genes involved in the biosynthesis of these components have been intensely studied recently [32,33,[49][50][51]. Because of its relatively small genome size, short life cycle, undemanding growth requirements and significant medicinal value, S. miltiorrhiza is being developed to become a model medicinal plant for TCM studies [31][32][33]. Elucidation of small RNA biogenesis pathways in S. miltiorrhiza appears to be urgent, given the significant regulatory roles of small RNAs in plant development and growth. Results from Arabidopsis suggest that the core components of small RNA pathways include at least three gene families, DCL, AGO and RDR [2]. The AGO gene family in S. miltiorrhiza has been previously characterized by our research group [31].
RDRs are core components of various gene silencing pathways and play vital roles in plant development and antiviral defense through regulating gene expression at the transcriptional and posttranscriptional levels [52]. RDR genes have been identified in various plants, such as Arabidopsis [14], rice [16] and maize [17]. However, many of them were computationally predicted only. For instance, among the six Arabidopsis AtRDRs, only three, including AtRDR1, AtRDR2 and AtRDR6, were cloned [7]. Of the five rice OsRDRs, only OsRDR6 were experimentally isolated [53]. Through genome-wide analysis of the working draft of the S. miltiorrhiza genome, we identified a total of five SmRDRs in S. miltiorrhiza. The full-length cDNAs of all predicted SmRDRs were then cloned and characterized. To our best knowledge, it is the first set of full-length SmRDR cDNAs from S. miltiorrhiza. The results provide useful information for further elucidation of RDR functions in S. miltiorrhiza.
Using a comprehensive approach, which combines sequence feature, gene structure, conserved domain and phylogenetic analyses, we characterized the five identified SmRDRs. The genomic sequence of SmRDR3 generating the longest SmRDR cDNA contains three introns, all of which locate in the open reading frame (ORF); whereas the SmRDR5 gene that produces the shortest SmRDR cDNA has seventeen introns with sixteen in ORF, suggesting the wide range of intron number in SmRDRs. The results are consistent with those from Arabidopsis and rice. The numbers of introns in AtRDR and OsRDR ORFs vary from 1 (AtRDR6) to 17 (AtRDR3, AtRDR4 and AtRDR5) and 1 (OsRDR6) to 18 (OsRDR4), respectively [16]. Examination of intron numbers in the ORF of S. miltiorrhiza, Arabidopsis and rice RDR genes suggest that RDRs in each phylogenetic group have similar number of introns ( Fig. 1) [16]. For instance, all RDR2 group members, including SmRDR4, AtRDR2 and OsRDR2, have 3 introns. The members in the RDR3 group contain 16-18 introns. It suggests the close evolutionary relationship of RDRs in a phylogenetic group.
Phylogenetic analysis of RDR proteins in S. miltiorrhiza, Arabidopsis and rice showed that RDRs clustered into four distinct groups (Fig. 5). Members of the RDR1, RDR2 and RDR6 groups are RDRa-type proteins characterized by the DLDGD signature (Fig. 4). There is only one RDRa-type RDR in the most recent common ancestor of plants, animals and fungi [47]. However, there are four in S. miltiorrhiza, of which, two are in the RDR1 group, one belongs in the RDR2 group, and the other one is included in the RDR4 group (Fig. 5). It suggests the occurrence of gene duplication for the RDRa-type gene during S. miltiorrhiza evolution. Since these RDRa-type RDRs clustered into different groups, they could be functionally diversified. All of the RDRctype RDRs, which are characterized by the DFDGD signature ( Fig. 4), clustered in the RDR3 group (Fig. 5). It includes three Arabidopsis RDRs (AtRDR3-AtRDR5), two rice RDRs (OsRDR3 and OsRDR4); however, there is only one S. miltiorrhiza RDRs (SmRDR5). It suggests that gene duplication has occurred for the RDRc-type gene during Arabidopsis and rice evolution but not in S. miltiorrhiza evolution.
Gene expression analysis showed that SmRDRs were expressed in all of the tissues detected and responsive to MeJA treatment and CMV infection. All five SmRDRs were suppressed under MeJA treatment (Fig. 7), whereas SmRDR1, SmRDR2 and SmRDR3 were up-regulated after CMV inoculation (Fig. 8). The involvement of S. miltiorrhiza SmRDRs in anti-viral defense is consistent with previous results for Arabidopsis AtRDR1, an ortholog of SmRDR1 and SmRDR2, and AtRDR6, an ortholog of SmRDR3 [7]. Further genetic manipulation of SmRDRs will shed light on the small RNA biogenesis pathways and RDR functions in S. miltiorrhiza.  Figure 8. SmRDRs responsive to CMV infection. The expression patterns were analyzed using the quantitative RT-PCR method. PCR was carried out in triplicates for each biological sample. Three independent biological replicates were performed. SmUBQ10 was used as a reference. Fold changes of SmRDRs in leaves of S. miltiorrhiza plantlets infected with CMV for 12, 24, 48 and 72 h are shown. The level of transcripts in leaves inoculated with phosphate buffered saline (CK) was arbitrarily set to 1 and the level in tissues inoculated with CMV was given relative to this. Error bars represent standard deviations of mean value from three biological replicates. doi:10.1371/journal.pone.0095117.g008