Cloning and characterization of small RNAs from Medicago truncatula reveals four novel legume-specific microRNA families


Author for correspondence:
Ramanjulu Sunkar
Tel: +1 405 744 8496


  • • MicroRNAs (miRNAs) and small-interfering RNAs (siRNAs) have emerged as important regulators of gene expression in higher eukaryotes. Recent studies indicate that genomes in higher plants encode lineage-specific and species-specific miRNAs in addition to the well-conserved miRNAs. Leguminous plants are grown throughout the world for food and forage production. To date the lack of genomic sequence data has prevented systematic examination of small RNAs in leguminous plants. Medicago truncatula, a diploid plant with a near-completely sequenced genome has recently emerged as an important model legume.
  • • We sequenced a small RNA library generated from M. truncatula to identify not only conserved miRNAs but also novel small RNAs, if any.
  • • Eight novel small RNAs were identified, of which four (miR1507, miR2118, miR2119 and miR2199) are annotated as legume-specific miRNAs because these are conserved in related legumes. Three novel transcripts encoding TIR-NBS-LRR proteins are validated as targets for one of the novel miRNA, miR2118. Small RNA sequence analysis coupled with the small RNA blot analysis, confirmed the expression of around 20 conserved miRNA families in M. truncatula. Fifteen transcripts have been validated as targets for conserved miRNAs. We also characterized Tas3-siRNA biogenesis in M. truncatula and validated three auxin response factor (ARF) transcripts that are targeted by tasiRNAs.
  • • These findings indicate that M. truncatula and possibly other related legumes have complex mechanisms of gene regulation involving specific and common small RNAs operating post-transcriptionally.


There are two major classes of endogenous small RNAs in plants: microRNAs (miRNAs) and small-interfering RNAs (siRNAs). Based on their origin, biogenesis, and potential targets, endogenous plant siRNAs are further divided into trans-acting siRNAs (tasiRNAs), natural antisense transcript-derived siRNAs (nat-siRNAs) and repeat-associated siRNAs (rasiRNAs) (Vaucheret, 2006).

The miRNAs are c. 21-nt long noncoding RNAs, which result from processing of imperfectly folded hairpin-like single-stranded RNAs by the Dicer-Like1complex (Jones-Rhoades et al., 2006; Ramachandran & Chen, 2008). The 21- to 24-nt siRNAs are processed by the other members of the Dicer family of proteins (DCL2, DCL3 and DCL4) from long, perfectly paired double-stranded RNAs (dsRNAs). The dsRNAs result from the transcription of inverted repeats, or convergent transcription of sense–antisense gene pairs or due to activity of RNA-dependent RNA polymerases (RDRs) on aberrant transcripts (Allen et al., 2005; Vaucheret, 2006).

Gene silencing occurs in both plants and animals when endonuclease complexes are guided by small RNAs to target RNAs (Bartel, 2004). The target genes are repressed post-transcriptionally when miRNAs direct cleavage of target mRNAs or prevent protein production (Jones-Rhoades et al., 2006; Mallory & Vaucheret, 2006; Brodersen et al., 2008). Endogenous siRNAs are capable of regulating gene expression at either transcriptional (heterochromatic siRNAs) or post-transcriptional levels (tasiRNAs and nat-siRNAs) (Arazi et al., 2005; Vaucheret, 2006).

Identification of small RNAs and their target messenger RNAs are essential for understanding small RNA-mediated gene regulation of development and stress responses (Arazi et al., 2005; Jones-Rhoades et al., 2006; Sunkar et al., 2007). Accurate annotation of small RNAs into miRNAs, tasiRNAs, nat-siRNAs and rasiRNAs requires genome sequence information. As a result, plants with sequenced genomes such as Arabidopsis, rice (Oryza sativa), Populus and Physcomitrella, are the most intensively investigated plant systems for the discovery of small RNAs (Jones-Rhoades & Bartel, 2004; Arazi et al., 2005; C. Lu et al., 2005; Rajagopalan et al., 2006; Barakat et al., 2007; S. Lu et al., 2008). The common themes emerging from these studies are that c. 20 miRNA families are highly conserved across monocots and dicots, but plants also express lineage-specific and species-specific miRNAs (Arazi et al., 2005; Sunkar et al., 2005; Lu et al., 2006; Talmor-Neiman et al., 2006a; Axtell et al., 2007; Fattash et al., 2007; C. Lu et al., 2008; Moxon et al., 2008; Sunkar & Jagadeeswaran, 2008). Conserved miRNAs appear to have conserved biological functions whereas lineage-specific miRNAs and species-specific miRNAs may have lineage-specific and species-specific roles, respectively. This observation means cataloging of small RNAs is required to accurately trace the evolution of lineage-specific and species-specific miRNAs in plants.

Currently, leguminous plants account for one-third of the world's primary crop production and are critical to meet large quantities of food and feed demands by humans and animals (Benedito et al., 2008). Legumes are also an interesting group of plants because they can fix atmospheric nitrogen. However, most cultivated legumes are polyploid with complex genomes and are therefore not amenable for genomic studies (Benedito et al., 2008). Most cultivated legumes are also resistant to common genetic manipulation tools such as transformation. Medicago truncatula, however, possesses a simple diploid genome, and is relatively easy to transform (Trinh et al., 1998; Chabaud et al., 2003; Crane et al., 2006). As a result, sequencing of the M. truncatula genome is nearly complete, and it has become a model legume for functional genomics research (Benedito et al., 2008). Sequence conservation has allowed scientists to predict miRNAs in the legumes, M. truncatula and soybean (Glycine max) (Zhang et al., 2006; Sunkar & Jagadeeswaran, 2008). Recently, sequencing a small RNA library from soybean identified three novel miRNAs (Subramanian et al., 2008). Complete genome information is needed to facilitate confident annotation of these and other small RNAs as miRNAs or siRNAs in soybean (Subramanian et al., 2008).The near-complete genome sequence of M. truncatula will allow more accurate characterization of small RNAs.

In the present study, we have sequenced a M. truncatula small RNA library and identified 20 conserved miRNA families. More importantly, eight novel small RNAs have been identified in M. truncatula. Four of these are annotated as legume-specific miRNAs because these miRNAs and their hairpin structures are conserved in related legumes. The remaining four appear to be candidates for M. truncatula-specific miRNAs. Three novel transcripts encoding TIR-NBS-LRR disease-resistance proteins are validated as targets for the novel miRNA, miR2118 in M. truncatula. We also validated one representative target for most of the conserved miRNA families using the 5′-rapid amplification of cDNA ends (RACE) assay. Biogenesis and target gene validations were also determined for TAS3-siRNAs.

Materials and Methods

Cloning of M. truncatula small RNAs

Total RNA was isolated from the frozen seedlings and flowers with TRIzol (Invitrogen) according to the manufacturer's instructions. Cloning of the miRNAs was performed as described (Sunkar et al., 2008). The final PCR product was sequenced using a 454 sequencer (Roche) at the University of Oklahoma, USA.

Plant materials and growth conditions

Medicago truncatula Gaertner cv. Jemalong plants were grown in a controlled growth chamber (22–24°C) with a 16-h photoperiod and 300 µmol m−2 s−1 light intensity. Tissue samples from different organs were harvested and flash frozen and stored at −80°C. For stress treatments, 3-wk-old seedlings grown on hydroponic cultures were transferred to the same medium without sulfate, or phosphate or copper. Root and shoot tissues were harvested separately and stored.

Method for identifying new candidate miRNAs

Our computational methods for analysing 454 small RNA libraries was reported previously (Sunkar et al., 2008). Briefly, all small RNA reads without perfect matches to the most proximal 11 nt of both adaptor sequences were first removed. Reads corresponding to repeats were removed using the einverted and etandem programs in the emboss (2000) package, respectively. The unique small RNAs were aligned to repbase (version 13.04, obtained from and known noncoding RNAs (rRNAs, tRNAs, snRNAs, snoRNAs, etc., obtained from with National Center for Biotechnology Information (NCBI) blastn. Then the small RNAs were mapped to the reported miRNAs in the miRBase (version 11, obtained from Small RNAs that matched known miRNAs of M. truncatula or other plant species resulted in identification of conserved miRNA homologs in M. truncatula. The unique small RNAs were aligned to the genome sequence of M. truncatula (downloaded from (, release 2.0), and for those sequences that matched with the genome, the fold-back structures were predicted using the rnafold program (Hofacker, 2003). This resulted in identification of 41 small RNAs with 145 loci. Most nonconserved and species-specific miRNAs have single locus in Arabidopsis and rice (S. Lu et al., 2008; Fahlgren et al., 2006), and we applied this feature in our analysis. This resulted in identification of 21 candidate sequences. Out of these, only five were considered for further analysis, because these could be detected using small RNA blot analysis. The unique small RNAs that could not be mapped to the available M. truncatula genome were queried against the expressed sequences tag (EST) database of the NCBI to find homologs in other plants, including legumes. Surprisingly, three of the unique small RNAs that could not be mapped to the M. truncatula genome, were found to have perfect matches with the ESTs from M. truncatula and other related legumes. Fold-back structures could be predicted for these ESTs. This resulted in identification of novel legume-specific miRNAs.

miRNA target prediction and validation

The known M. truncatula open reading frames (ORFs), downloaded from the Medicago Genome Annotation Database, were used for miRNA target predictions. For selecting putative miRNA–target pairs, only three mismatches were allowed between a mRNA target and miRNA in our prediction (Rhoades et al., 2002; Jones-Rhoades & Bartel, 2004). A modified 5′-RACE assay was performed using the GeneRacer Kit (Invitrogen) to validate the predicted targets. Briefly, the RNA was ligated with a 5′ RNA adapter and a reverse transcription was performed. The resulting cDNA was used as template for PCR amplification with GeneRacer 5′ primer and a gene-specific primer. A second nested PCR was performed using nested primers (GeneRacer 5′ nested primer and a gene-specific nested primer). The amplified products were gel purified, cloned and sequenced. Gene specific primers used are provided as in the Supplementary Information, Table S3.

Identification of TASi locus and tasiRNAs in M. truncatula

To predict TAS genes and tasiRNAs in M. truncatula, we examined Hitsensor scores (Zheng & Zhang, 2008) and inter-site distances of miR390 binding sites of Arabidopsis TAS3 genes. AtTAS3 genes have two Hitsensor sites for miR390 (of which the 5′ site with critical mismatches in position 10–11), with Hitsensor scores from 200 to 300. Based on these distances (228 nt for AtTAS3a, 201 nt for AtTAS3b and 181 nt for AtTAS3c) we examined 250 nt upstream and 250 nt downstream from M. truncatula unique RNA reads using Hitsensor to identify possible binding sites of miR390. Sequences without two miR390-binding sites were removed.

Small RNA blot analysis

Total RNA was isolated from different tissues of M. truncatula with TRIzol reagent following the manufacturer's instructions (Invitrogen). Low-molecular-weight (LMW) RNA was isolated from total RNA using polyethylene glycol (PEG) precipitation. Twenty micrograms of LMW RNA was used for detection of miRNAs or candidate miRNAs, whereas 50 µg was used for detection of miRNA*. Small RNA blot analysis was performed as previously reported (Sunkar et al., 2008).


Sequence analysis and annotation

To identify miRNAs and other endogenous small RNAs expressed in M. truncatula seedlings, we generated and sequenced a small RNA library using pooled RNA isolated from seedlings and flowers. After removal of the 5′ and 3′ adapter sequences, a total of 26 656 raw sequence reads ranging in size between 18 nt and 26 nt were obtained. Of these, 22 353 reads could be mapped to the existing M. truncatula genome, and the remaining sequences (c. 4237) that could not be mapped thus, were discarded. Of the 22 353 genome-matched reads, redundant sequences were noted and only unique sequences were used for further analysis. Small RNAs matching the rRNA, tRNA, snRNA and snoRNAs from these unique sequences were removed as reported previously (Sunkar et al., 2008).

The cloning frequency of different sized (18–26 nt) small RNAs revealed two predominant sizes: 21 nt and 24 nt, which is consistent with previous reports (Lu et al., 2005, 2006; Sunkar et al., 2005; Fahlgren et al., 2007). Small RNAs of the 24-nt size class represented the largest category of sequence reads (9333 of 22 353 reads, 42%). Most 24 nt sequences appeared only once in our sequence reads suggesting the 24 nt small RNAs are extremely diverse in M. truncatula. Similar results were found in Arabidopsis and rice (Lu et al., 2006; Nobuta et al., 2007). This study focuses on the 21-nt size class of small RNAs, representing miRNAs and tasiRNAs.

Identification of four novel legume-specific miRNAs

Annotating novel miRNAs generally requires the dcl1 knockout mutant (Ambros et al., 2003). In addition, studies using Arabidopsis showed that dcl4 may also process certain miRNA precursors (Rajagopalan et al., 2006), suggesting both the dcl1 and dcl4 knockout mutants are needed for confident annotation of novel species-specific miRNAs in plants. In the absence of these genetic tools, it was recently suggested that sequencing of a miRNA* is required for annotating a small RNA as a new miRNA in plants (Meyers et al., 2008). In addition, showing conservation as determined by bioinformatics or experimentation (e.g. small RNA blot analysis) strengthens confidence in annotating novel small RNAs as miRNAs (Meyers et al., 2008). Using these guidelines, we assigned a small RNA as a novel miRNA if: (1) a miRNA* is detected using small RNA blot analysis; (2) its conservation is confirmed by using bioinformatics (small RNA sequence as well as predicted fold-back structure for the precursor sequence is conserved in related plant species); and (3), if the small RNA could be detected in M. truncatula and other related legumes using small RNA blot analysis. These criteria identified four small RNAs (miR2118, miR2199, miR1507 and miR2119) as novel miRNA families in M. truncatula (Table 1, Fig. 1). To strengthen our annotation assignments, we analysed the expression of these M. truncatula miRNA* sequences using small RNA blots. The antisense probes corresponding to the predicted miRNA* gave discrete bands suggesting that the miRNA* sequence accumulates to detectable levels (Fig. 2b). Attempts to hybridize the same blots with antisense probes from precursor sequences flanking the mature miRNA showed no signal (data not shown). These findings suggest that only the miRNA and miRNA* are excised from their precursors of these four novel small RNAs in M. truncatula.

Table 1.  Identified novel small RNAs (four legume-specific and four candidate miRNAs) in Medicago truncatula
sRNA_idSequence (cloning frequency)Detected using blot analysisConservation
  1. The annotation of legume-specific miRNAs was based on the detection of miRNA and miRNA* as well as their conservation in related legumes.

miR2118UUACCGAUUCCACCCAUUCCUA (3)+Soybean, Chickpea, Peanut, Phaseolus, Vigna
miR2199UGAUACACUAGCACGGGUCAC (1)+Lotus japonicus
sRNA19284 (candidate)UAGGUUUGAGAAAAUGGGCAG (1)+ 
Figure 1.

Predicted fold-back structures for the legume-specific miRNAs identified in this study. (a) miR2118, (b) miR2199, (c) miR1507 and (d) miR2119. Ah, Arachis hypogaea; Gm, Glycine max; Lj, Lotus japonicus; Mt, Medicago truncatula; Pv, Phaseolus vulgaris; Vu, Vigna unguiculata.

Figure 2.

Characterization of four novel legume-specific miRNAs in Medicago truncatula. (a) Expression analysis of four novel miRNAs (miR2118, miR2199, miR1507 and miR2119) in related legumes along with Arabidopsis and rice using 20 µg of low molecular weight RNA. Blots were stripped and rehybridized with the indicated miRNA probes. The U6 probe served as a loading control. (b) Detection of miRNA* sequence for the four of the novel miRNAs (miR2118, miR2199, miR1507 and miR2119) using 50 µg of low molecular weight RNA for small RNA blot analysis. Small RNA blots from Medicago samples were hybridized with antisense probes of predicted miRNA star sequence. The U6 probe served as a loading control. (c) Validation of three targets for the newly identified miRNA, miR2118, by modified rapid amplification of cDNA ends (RACE) assays in Medicago truncatula. The mRNA target (top) and its corresponding miRNA (bottom) are shown in each alignment; matches are indicated with straight lines, mismatches with colons and G–U wobbles represented with a circle. Arrows show the site of cleavage and the fraction of cloned products that terminated at the miRNA complementary sites is indicated above.

Our sequence analysis resulted in identification of 21 candidate sequences, based on the genome matching and fold-back structure predictions. Of these, only five were considered for further analysis based on accumulations detected using small RNA blot analysis. Among these, miR1507 is conserved in related legumes and thus is annotated as a legume-specific miRNA, whereas the remaining four are regarded as candidate M. truncatula-specific miRNAs. The fully sequenced M. truncatula genome is currently unavailable for analysis. We suspected that some of the unique small RNAs without matches to the available genome of M. truncatula could still be miRNAs; these unique small RNAs were searched against the EST resources at NCBI to identify their precursors. Three of the unique small RNAs that could not be mapped to the M. truncatula genome, had perfect or nearly perfect matches (one or two mismatches) with the ESTs from M. truncatula and other related legumes. The blast analysis identified miR2118 homologues in soybean, pea (Phaseolus vulgaris) and black-eyed pea (Vigna unguiculata). Similarly, homologs for miR1507 and miR2199 are found in peanut (Arachis hypogaea) and Lotus japonicus, respectively (Table 1, Fig. 1). Another small RNA (miR2119) is conserved in soybean (Table 1, Fig. 1). Fold-back structures could be predicted using the genomic or EST sequences surrounding the novel small RNAs in the above mentioned leguminous plant species (Fig. 1). We then analysed the expression of these novel miRNAs in related legumes such as soybean, chickpea (Cicer arietinum), peanut, black-eyed pea along with the Arabidopsis and rice (Fig. 2a). miR2118 and miR2199 could be detected in all five legumes tested (Fig. 2a), whereas miR1507 and miR2119 could be detected in three legumes of the five tested (Fig. 2a). None of these four small RNA sequences exist in completely sequenced genomes of Arabidopsis, rice, Populus or Physcomitrella. Accordingly, we annotated these four novel small RNAs (miR2118, miR2199, miR1507 and miR2119) as legume-specific miRNAs based on detection of miRNA and miRNA* and conservation of small RNA sequence as well as their fold-back structure. These criteria have been used to categorize small RNA as a novel miRNA (Sunkar et al., 2005, 2008; S. Lu et al., 2008). From the data, this study has provided compelling evidence for annotation of four novel small RNAs as legume-specific miRNAs. One identified based on matching with the M. truncatula genome and the other three matching with the EST resources.

The remaining four novel small RNA homologs (sRNA7556, sRNA19284, sRNA15993 and sRNA21166) were not found in any other plant species (Table 1). Fold-back structures are predicted for their precursor sequences (see the Supporting Information, Fig. S1) and some of them could be detected using small RNA blot analysis (Fig. 3). However, these characteristics do not meet the criteria to annotate them as Medicago-specific miRNAs without cloning miRNA* sequence (Meyers et al., 2008). Thus, these four small RNAs are regarded as candidate miRNAs in M. truncatula. Recently, one of these novel small RNAs (sRNA19284) was also sequenced by another group (Szittya et al., 2008) and classified as a potential new miRNA in M. truncatula, supporting our current results.

Figure 3.

Detection of novel candidate miRNAs in Medicago truncatula. Expression analysis of newly identified candidate miRNAs in roots and shoots of M. truncatula. Blots were stripped and rehybridized with the indicated miRNA probes. The U6 probe served as a loading control.

Interestingly, more than 20 Toll/Interleukin 1 Receptor-nucleotide binding site-leucine-rich repeat (TIR-NBS-LRR) genes have been predicted as targets for the novel legume-specific miRNA, miR2118 in M. truncatula (Table S1). However, at least two TIR-NBS-LRR genes in M. truncatula do not possess complementary sites for the miR2118 (Table S1). The TIR-NBS-LRR gene family is highly conserved among higher plants and homologs of this gene family possesses the complementary site for miR2118 in Arabidopsis and rice, but miR2118 homologs are absent in Arabidopsis, rice and Populus. Using 5′-RACE assays, three predicted targets (AC202360_18.1, AC203224_17.1 and AC143338_38.2) belonging to the TIR-NBS-LRR gene family have been validated as genuine targets for miRNA, miR2118 (Fig. 2b).

Identification of conserved miRNAs from M. truncatula

Higher plants have at least c. 20 conserved miRNA families, and the latest miRBase release (11.0, September, 2008) lists 30 miRNAs belonging to nine conserved miRNA families in M. truncatula. Our small RNA sequence analysis identified miRNAs belonging to 13 conserved miRNA families, which includes nine families reported in the miRBase (Table 2). To detect the expression of the remaining conserved miRNA families (miR162, miR393, miR394, miR395, miR397, miR398 and miR399), we performed small RNA blot analysis using labeled antisense oligos. These experiments confirmed the expression of miR162, miR393 and miR398 in diverse tissues (Fig. 4). The other four miRNA families (miR395, miR399 and miR397/398) are induced specifically under sulfate- or phosphate- or copper-deprived conditions, respectively (Chiou, 2007; Sunkar et al., 2007) and could not be detected using RNA isolated from the M. truncatula seedlings grown hydroponically with optimal levels of nutrients. Using small RNA blot analysis, miR395, miR398 and miR399 were detected in M. truncatula seedlings grown hydroponically but without sulfate or copper or phosphate respectively (Fig. 5a–c). We also detected miR397 and miR408 (data not shown) in seedlings grown on medium without copper (Fig. 5c). When the miRNA sequences in our library (Table 2) are combined with the miRNA families detected in Figs 4 and 5, we confirmed the expression of 20 conserved miRNA families in M. truncatula. While our manuscript was in review, two recent reports identified the conserved miRNAs in M. truncatula (Szittya et al., 2008; Zhou et al., 2008). Using bioinformatics, Zhou et al. (2008) reported the identification of 11 conserved miRNA families, whereas using an experimental approach Szittya et al. (2008) reported the identification of c. 20 conserved miRNA families in M. truncatula. However, these reports do not analyse the expression patterns of conserved miRNAs in different tissues or in response to limiting nutrient (sulfate or phosphate or copper) availability.

Table 2.  Identified conserved miRNAs in Medicago truncatula using cloning approach and small RNA blot analysis
miRNAmiRNA sequenceCloning frequencyValidated by small RNA blot analysisValidated targetsValidated target gene family
miR160UGCCUGGCUCCCUGUAUGCCA 4+BQ148941, ES612384Auxin response factors
miR162UCGAUAAACCUCUGCAUCCAG+AC150443_32.2Dicer Like-1
miR164UGGAGAAGCAGGGCACGUGCA 2+AC203553_1.1NAC domain protein
miR167UGAAGCUGCCAGCAUGAUCUA 8+AC144478_44.4, CU326393_14.1Auxin Response Factor
miR168UCGCUUGGUGCAGGUCGGGAA 5+AW773594.1Argonaute-1
miR-170/171UGAUUGAGCCGUGUCAAUAUC 2+AC121238_43.2Scarecrow-like transcription factor
miR172AGAAUCUUGAUGAUGCUGCAU734+AL383429AP2 domain transcription factor
miR390AAGCUCAGGAGGGAUAGCGCC 2+Noncoding RNATAS3 pre-transcript
miR393UCCAAAGGGAUCGCAUUGAUCC+AC133780_22.2F-box protein
miR395AUGAAGUGUUUGGGGGAACUC+AC146721_16.4ATP sulfurylase
miR397UCAUUGAGUGCAGCGUUGAUG+AC203224_21.1, AC135467_30.2Copper-resistance protein
miR399aUGCCAAAGGAGAUUUGCCCAG+AC159143_20.4Ubiquitin-conjugating enzyme (E2 ligase)
Figure 4.

Expression analysis of conserved miRNAs in different tissues of the Medicago truncatula. Small RNA blots of low molecular weight RNA isolated from different tissues as indicated. The blots were probed with 32P-end-labelled oligonucleotides complementary to the miRNAs. Individual blots were used for each probe except for miR162, which was stripped and rehybridized with miR164; U6 served as a loading control.

Figure 5.

Characterization of nutrient deprivation-induced miRNAs in Medicago truncatula. (a–c) Small RNA blots of low molecular weight RNA isolated from M. truncatula seedlings grown on hydroponic medium were grown continuously on the same medium (control) or transferred to medium lacking the indicated nutrient (a), sulfate, (b) phosphate and (c) copper. The blots were probed with 32P-end-labelled oligonucleotides complementary to the miRNAs. The blots were stripped and re-probed with U6, which served as a loading control.

The frequency of conserved miRNAs varied between 1 and 734 reads in our library (Table 2). The miR172 family is the most abundant, represented by 734 reads, of which miR172b alone accounted for 444 (c. 60% of 734) reads. The second most-abundant miRNA family was miR159, represented by 234 reads (Table 2). A very high count of miR172 reads was expected, because miR172 regulates the AP2 transcription factor implicated in flower development (Aukerman & Sakai, 2003; Chen, 2004), and our library was generated using pooled RNA isolated from M. truncatula seedlings and flowers. However, tissue-specific expression analysis indicated that miR172 is abundant in leaves, stem, root and flowers (Fig. 4). These results suggest that miR172 has developmental regulatory roles, in addition to roles in flower development.

Sequenced small RNA libraries often contained miR* sequences, although at much lower abundance compared with miRNA (Lu et al., 2006; Rajagopalan et al., 2006). Not all conserved miRNAs were detected at high abundance in this library. Many conserved miRNA family members (miR160a, miR164b, miR167a, miR167b and miR167c) were represented by single reads (Table 2). Only miR396, miR160, miR169 and miR171 had their miR* sequences in our library.

Expression analyses of conserved miRNAs in M. truncatula

Understanding the temporal and/or spatial expression of a miRNA is an important initial step in probing its functional role in any organism. Many small RNAs are expressed only in certain tissues or cell types (Chen, 2004; Sunkar & Zhu, 2004; Combier et al., 2006; Boualem et al., 2008). Here, we determined the expression patterns of miRNAs from diverse tissues of M. truncatula, such as leaves (young and old), stems, roots, flowers and in 3-wk-old seedlings (Fig. 4). Mostly independent blots were used for each probe, thus the expression levels of each microRNA was monitored unambiguously.

Although most miRNAs were abundantly expressed in M. truncatula leaves, considerable variation in their abundance was noticeable. The exceptions to higher levels of expression in leaf tissues were miR160, miR168, miR170, miR390 and miR398, which showed reduced levels in leaves (Fig. 4). Similarly, flower tissue expressed all miRNAs tested with notable exceptions being miR156 and miR393, where expression was the least compared with other tissues analysed (Fig. 4). By contrast, miR169, miR170 and miR398 showed elevated expression in flowers. miR169, was distinctly abundant in flower tissue, while other organs showed only a faint expression. Both miR393 and miR398 had relatively low expression in the M. truncatula stem, whereas the other miRNAs showed strong expression (Fig. 4). Abundant expression of miR159, miR166 and miR167 was observed in roots (Fig. 4).

The spatial expression pattern of miR398 differed greatly between M. truncatula and Arabidopsis. In Arabidopsis, miR398 is expressed abundantly in leaves (cauline and rosette) but not in inflorescence (Jones-Rhoades & Bartel, 2004; Sunkar & Zhu, 2004; Sunkar et al., 2006). By contrast, miR398 in M. truncatula was expressed abundantly in flowers but only at lower levels in leaves (Fig. 4). These findings indicate that the expression patterns of conserved miRNAs vary greatly across plant species.

Characterization of nutrient deprivation-induced miRNAs in M. truncatula

Studies show that miR395, miR399 and miR398 are induced in response to sulfate, phosphate and copper deficiency, respectively, in Arabidopsis (Sunkar et al., 2007). Copper deprivation has been also shown to induce the expression of miR397 and miR408 in Arabidopsis and Brassica (Yamasaki et al., 2007; Abdel-Ghany & Pilon, 2008; Buhtz et al., 2008) To compare these data with miR395, miR397, miR398, miR399 and miR408 in M. truncatula, 3-wk-old seedlings grown on optimal hydroponic medium were transferred to identical medium lacking sulfate or phosphate or copper. miR395 was upregulated in shoots and roots, with the upregulation being very strong in roots of M. truncatula seedlings grown on the medium without sulfate (Fig. 5a). Similarly, miR399 was induced when grown on medium without phosphate (Fig. 5b). As shown in Fig. 5c, miR398 was upregulated both in shoots and roots of M. truncatula grown on medium without copper, with the accumulation greater in roots than shoots. A similar pattern of induction was observed for miR397 in M. truncatula seedlings grown on copper-deficient medium (Fig. 5c).

miRNA target validations

Plant miRNAs and their targets are highly complementary (18–21 nt), which facilitates target prediction with the use of blastn search or patscan search (Rhoades et al., 2002; Jones-Rhoades & Bartel, 2004). To predict miRNA-targeting mRNAs in M. truncatula, we downloaded the currently annotated coding sequences from the Medicago genome project ( and used miRNA sequences to search complimentary mRNA sequences as suggested previously (Rhoades et al., 2002; Jones-Rhoades & Bartel, 2004). This analysis has identified c. 92 transcripts as potential targets in M. truncatula (Table S2). To date, only four mRNAs have been validated as genuine targets of miRNAs in M. truncatula (Combier et al., 2006; Boualem et al., 2008). The miR169-guided cleaved fragments were detected for the MtHAP2-1 transcript (Combier et al., 2006). Similarly, miR166-guided cleaved fragments were found for three transcripts, MtCNA1, MtCNA2 and MtHB-8 in M. truncatula (Boualem et al., 2008). In this study, a total of 15 mRNAs (representing at least one target for most of the conserved miRNA families) were confirmed as targets in M. truncatula (Table 2, Fig. 6). Most of the cleaved fragments were mapped exactly at the predicted cleavage sites (between nucleotides 10 and 11 from the 5′ end). Some validated targets were cleaved at a slightly different position possibly owing to variations (5′ or 3′ shifts) in mature miRNA species. ATP sulfurylases, UBC-like-E2-ligase and plantacyanin are predicted targets for miR395, miR399 and miR408, respectively (Table 2). Our attempts to validate these transcripts as miRNA targets were unsuccessful using the Medicago seedlings grown on control medium. However, miRNA-directed cleaved fragments for these transcripts could be detected in seedlings grown on medium without sulfate or phosphate or copper. This can be attributed to nutrient-specific induced transcription of the miRNAs targeting these transcripts (Fig. 6). In Arabidopsis and rice, UBC transcript in its 5′-UTR (untranslated region) possesses four or five complementary sites of miR399 (Fujii et al., 2005; Bari et al., 2006). In M. truncatula, miR399 has five target sites located in the 5′-UTR of UBC and four of them were found to be cleaved (Fig. 6). These target validations are consistent with those of most conserved miRNA targets in Arabidopsis, rice and Populus (Jones-Rhoades et al., 2006).

Figure 6.

Validation of conserved miRNA targets in Medicago truncatula. Detection of miRNA-guided cleavage sites on target transcripts using modified rapid amplification of cDNA ends (RACE) assays in M. truncatula. Partial mRNA sequences from target genes were aligned with the miRNAs. In each alignment, the straight lines represent perfect matches, colons represent mismatches and circles represent G–U wobbles. The fraction of cloned products that terminated at the predicted cleavage site is indicated.

Identification and characterization of TAS3-tasiRNAs in M. truncatula

The tasiRNAs (trans-acting siRNAs) are a class of endogenous 21 nt siRNAs that downregulate target mRNAs at the post-transcriptional level, as do miRNAs (Peragine et al., 2004; Vazquez et al., 2004; Allen et al., 2005). They have been shown to regulate vegetative phase changes in Arabidopsis (Peragine et al., 2004; Vazquez et al., 2004; Allen et al., 2005; Gasciolli et al., 2005; Xie et al., 2005; Yoshikawa et al., 2005) and in moss (Arazi et al., 2005; Talmor-Neiman et al., 2006b). An initial DCL1-dependent, miRNA-guided cleavage of tasiRNA primary transcript defines the 5′ or 3′ end of the transcript to be converted into dsRNAs by RDR, and sets the phase for accurate formation of 21 nt tasiRNA by DCL4 in Arabidopsis (Peragine et al., 2004; Vazquez et al., 2004; Allen et al., 2005; Axtell et al., 2006). Thus, two DCLs, DCL1 and DCL4 and an RDR are required for tasiRNA biogenesis. In Arabidopsis, three miRNAs – miR173 miR390 and miR828 – have been found to target primary tasiRNA transcripts (Allen et al., 2005; Rajagopalan et al., 2006). Of these, only miR390 is conserved from moss to higher plants and miR828 appears to be conserved between Arabidopsis and Populus, whereas miR173 homologs have not been found outside Arabidopsis. Since not all Arabidopsis miRNAs that target tasiRNA primary transcripts are conserved, predicting computationally how many miRNAs target tasiRNA precursors and how many tasiRNA loci exist in M. truncatula is difficult. Only sequencing can help identify tasiRNAs more confidently.

To identify tasiRNA loci and tasiRNAs in Medicago, we searched for clusters of small RNAs that can be mapped to one locus but surrounded by two miR390 target sites. This resulted in identification of one authentic tasiRNA locus in Medicago (MtTAS3a) (Fig. 7a). In addition to the conserved dual miR390 target sites on the tasiRNA precursor, this transcript has another conserved region, which may be processed into a tas3-siRNAs, and are complementary to auxin response factor (ARF) homologs in M. truncatula (Fig. 7a–c). MtTAS3-tasiRNAs were in phase with the 3′ target site (Fig. 7a) as found in Arabidopsis and rice (Allen et al., 2005; Liu et al., 2007). In Arabidopsis, only the 3′ target site, not the 5′ target site, is subjected to miR390-guided cleavage (Axtell et al., 2006; Howell et al., 2007). To verify whether the interaction between Medicago miR390 and the MtTAS3 precursor is similar to what was observed in Arabidopsis, 5′-RACE assay was performed to detect cleavage events at the 5′ and 3′ target sites on the TAS3 precursor. We failed to detect the cleavage at the 5′ target site and these results are consistent with the previous suggestion that miR390 is unable to guide for a cleavage because of mismatches between 9 nt and 11 nt from the 5′ end of the miR390 (Axtell et al., 2006). Surprisingly, only a small fraction of clones (3/40) confirmed the cleavage at the 3′ target site using RNA isolated from seedlings. Instead, the sequenced PCR product corresponding to the 3′ site mostly yielded a cleavage site 33 nt upstream from the predicted site (Fig. 7a). In Arabidopsis, detection of a cleavage at 33 nt upstream from the predicted site on TAS3 precursor was reported as a major cleavage event, although a minor cleavage event was also found at the predicted site (Allen et al., 2005). The upstream cleavage event in Arabidopsis has been attributed to the existence of a hypothetical siRNA (–D2) generated from the TAS3 precursor (Allen et al., 2005), although the accumulation of such a siRNA has not been examined. Because a similar cleavage event was also found in this study, we tested for the accumulation of the hypothetical –D2 siRNA using a small RNA blot and an antisense probe corresponding to the cleaved site. This analysis confirmed the accumulation of –D2 siRNA (Fig. 7d). Accumulation of sense (+D2) siRNA could not be detected using a complementary probe, suggesting that –D2 siRNA corresponding to the cleaved site has the ability to direct cleavage of TAS3 precursor.

Figure 7.

Tas3-siRNA biogenesis, expression analysis and validation of its targets in Medicago truncatula. (a) Nucleotide sequence of MtTAS3 locus. The 5′ and 3′ miR390 target sites are shown as alignments and the predicted cleavage site at the 3′ target site is shown with red arrow. Predicted Dicer processing sites are shown with blue lines. MtTAs3 siRNAs that are complementary to the auxin response factors are indicated with letters (+7TAS3 and +8TAS3). Detected miR390-guided cleavage fragments of the MtTAS3a precursor in M. truncatula are shown with a blue arrow. Detected –D2 siRNA-guided cleavage fragments of the MtTAS3a precursor in M. truncatula are shown with a red arrow. The fraction of cloned products that terminated at the cleaved site is indicated. (b) Sequence alignment of the TAS3 siRNAs from Arabidopsis, rice and M. truncatula. (c) Expression analysis of TAS3 (+8 TAS3) siRNAs in M. truncatula. (d) Expression analysis of –D2 siRNAs in M. truncatula. (e) Detection of TAS3-siRNA-guided cleavage sites on three target transcripts by modified rapid amplification of cDNA ends (RACE) assay in M. truncatula. Partial mRNA sequences from target genes were aligned with the TAS3 siRNA. The fraction of cloned products that terminated at the predicted cleavage site is indicated.

Cleavage at the 3′ target site on the TAS3 precursor is essential to generate 21 nt phased, two tandemly arranged conserved TAS3 siRNAs targeting auxin response factor (ARF)-like genes in plants (Allen et al., 2005; Axtell et al., 2006). However, a cleavage exactly 33 nt upstream from the 3′ target site on TAS3 precursor was observed in this study and in the study of Allen et al. (2005). This cleavage would generate siRNAs completely out of phase such that instead of generating active TAS3-tasiRNAs, the TAS3 precursor is degraded. A consistent cleavage in two different plant species (M. truncatula and Arabidopsis) further strengthens the proposal suggested by Allen et al. (2005) that the siRNAs generated from the same locus will degrade the TAS3 precursor.

The expression of TAS3-siRNAs in different tissues of M. truncatula was confirmed using a small RNA blot analysis (Fig. 7c). TAS3-siRNAs target three ARF genes (ARF2, ARF3 and ARF4) in Arabidopsis (Allen et al., 2005; Williams et al., 2005) and five ARF genes (four ARF3 homologs and one ARF2 homolog) in rice (Liu et al., 2007). Our computational prediction found four ARF genes as the targets of MtTAS3-siRNAs (AC150891_17.2; AC152176_68.2; AC158497_40.2 and AC126794_50.2) in M. truncatula (Fig. 7e). Of these four genes, three (AC150891_17.2; AC152176_68.2 and AC158497_40.2) are close relatives of Arabidopsis ARF3. These three genes have two MtTAS3-siRNA complementary sites in their ORFs. Another gene, AC126794_50.2, is a homolog of the Arabidopsis ARF2 gene and it harbors only one TAS3-siRNA complementary site in its ORF. To verify whether MtTAS3-siRNAs directs cleavage of their target genes in M. truncatula, we used 5′-RACE assay to map the cleavage sites on four ARF3-like genes (Fig. 7e). This analysis confirmed two cleavages at the predicted sites on two ARF3-like genes (AC152176_68.2 and AC158497_40.2) while only one cleavage at the 5′ target site was found for AC150891_17.2. In order to confirm the cleavages at two different sites on the same target gene (AC152176_68.2 and AC158497_40.2), we designed two independent primers downstream from each of the predicted target site (Table S3). We were unable to confirm the predicted cleavage site on ARF2-like gene because the mRNA could not be amplified using RNA isolated from the seedlings. Nonetheless, these observations confirm that the MtTAS3-siRNAs are processed and can regulate multiple (3) ARF genes in M. truncatula.


The present study has characterized eight novel small RNAs in M. truncatula (Table 1). Of these, four are annotated as legume-specific miRNAs, because both miRNA and miRNA* are detected. These miRNA sequences and predicted fold-back structures are conserved in leguminous plants, but these sequences are absent in the completely sequenced genomes of Arabidopsis, rice or Populus. This was further confirmed using small RNA blot analysis (Fig. 2). The remaining four novel small RNAs could be detected using small RNA blot analysis and consequently annotated as candidate miRNAs in M. truncatula but their confident annotation as miRNAs or siRNAs requires deep sequencing. We have predicted more than 20 target transcripts encoding TIR1-NBS-LRR disease resistance proteins as targets for one of the newly identified legume-specific miRNA (miR2118) and validated three of them in M. truncatula.

In M. truncatula, TIR-NBS-LRR genes exist in an extensive cluster of R gene loci on the top arm of M. truncatula chromosome 4 (Ameline-Torregrosa et al., 2008). Recently, Yang et al. (2008) cloned RCT1 that encodes a TIR-NBS-LRR type R protein conferring broad-spectrum anthracnose resistance when transferred into the susceptible alfalfa lines. Interestingly, we validated RCT1 transcript (AC202360_18.1) as a target for one of the legume-specific miRNAs (i.e. miR2118). In addition to the RCT1, we validated two other members (AC203224_17 and AC143338_38) of the TIR-NBS-LRR gene family as targets for the same miRNA (miR2118). We predicted c. 20 genes belonging to this gene family as potential targets for this miRNA in M. truncatula (Table S1). However, at least two genes belonging to this gene family are unlikely to be targeted by miR2118 (Table S1), despite the fact that the peptide sequence corresponding to the target site in all these proteins is highly conserved. A search of public databases showed a near-perfect target site in TIR-NBS-LRR genes of several other plant genomes including Arabidopsis and rice, but not the miR2118 homolog. Because miR2118 targets transcripts encoding TIR1-NBS-LRR proteins implicated in disease resistance, it will be interesting to see whether the expression of miR2118 is regulated during pathogen infection. Our experimental validation of three TIR-NBS-LRR genes as genuine targets and our prediction that 20 other related genes are potential targets of miR2118 provide opportunities to explore miRNA-mediated plant defense responses in Medicago and other legumes.

The miRNA-dependent tasiRNA pathway is a plant-specific RNA silencing pathway that mimics the miRNA pathway for mRNA regulation. miR390 is conserved in all higher plants and in primitive land plants such as Physcomitrella and Selaginella (Arazi et al., 2005; Axtell et al., 2006, 2007). In Arabidopsis, miR390-guided cleavage occurs only at the 3′ target site of the TAS3 precursor while the 5′ target site is resistant to cleavage but is important for binding to the tasiRNA precursor transcript (Axtell et al., 2006; Howell et al., 2007). This characteristic feature is attributed to mismatches at nucleotides 9–12 from the 5′ end of the miRNA. These mismatches are conserved in the TAS3-primary transcript of M. truncatula and may be critical for the generation of TAS3-siRNAs in this species.

The Arabidopsis TAS3 is expressed on the adaxial side of early leaf primordia (Adenot et al., 2006; Garcia et al., 2006), which possibly regulates the expression of ARF3 and ARF4 in the adaxial domain and determines the dorso-ventral leaf polarity (Garcia et al., 2006). A similar role has been reported for TAS3-tasiRNAs in maize that target ARF3 and ARF4 orthologs (Nogueira et al., 2007). The observation that M. truncatula miR390 targets TAS3 tasi-precursors and that the TAS3-tasiRNAs derived from these precursors in turn target four of the ARFs (three validated in this study) suggests that the TAS3-siRNA pathway indeed plays important regulatory roles in M. truncatula. Future experiments will address the functional aspects of TAS3-siRNA-guided ARF regulation in M. truncatula and other legumes.

In summary, with a sequencing-depth of 26 656 reads the present study has uncovered the existence of four legume-specific miRNAs and four candidate miRNAs in M. truncatula. By applying more robust deep sequencing technologies such as Sequencing-By-Synthesis (Illumina, Haywood, CA, USA), there is ample scope for the discovery of several additional novel miRNAs. Indeed, while our manuscript was under review Szittya et al., reported eight new miRNA families in M. truncatula by using a deep sequencing approach (Szittya et al., 2008). Interestingly, despite examining approx. 4 million reads there are only two small RNAs (one legume-specific miRNA (miR1507) and one candidate miRNA (sRNA19284)) that overlap between the study of Szittya et al. (2008) and this study. This suggests the identification of miRNAs in M. truncatula is far from being saturated. Importantly, we have uncovered legume-specific miRNAs in this study which are not found in the deeply sequenced library (Szittya et al., 2008). Thus our study has complimented the published report (Szittya et al., 2008). Identification of small RNAs and their target genes is a highly useful resource for the large community of researchers working on gene regulation in M. truncatula and other legume crops. Further work is required to identify a near-complete set of miRNAs and other novel small RNAs in M. truncatula.


Support for this research was provided by the Oklahoma Agricultural Experiment Station to R.S., and by the National Science Foundation grants IIS-0535257 and DBI-0743797, a grant from the Alzheimer's Association and a grant from Monsanto to W.Z. We thank Drs Rao Uppalapati and Kiran Mysore (Samuel Roberts Noble Foundation, Ardmore) for providing us some of the tissues used in small RNA blot analysis. L.S. was a recipient of a BOYSCAST fellowship from the Department of Science and Technology, Government of India.