High-throughput sequencing of M. truncatula small RNAs
High-throughput sequencing has been used to study miRNAs at whole genome level in several plant species, including Arabidopsis [39, 40], rice , wheat , soybean , cotton , grapevine , Medicago [42, 51, 52], Brachypodium , tomato , populus , citrus , peanut , Porphyra . However, most of the studies on high-throughput sequencing miRNAs have focused on miRNAs under non-stressed, normal growth conditions, and only a few studies have compared miRNAs under control conditions to those under conditions of abiotic stresses by high-throughput sequencing plant miRNAs [20, 43, 44]. For instance, Sunkar et al. (2008) reported a total of 714,202 reads in rice seedlings from three independent libraries (control, drought and salt stress libraries) . Twenty-eight and 12 miRNAs in response to cold stress in Brachypodium [20, 44] and heat stress in wheat  have been identified by high-throughput sequencing, respectively. In the present study, we constructed two sRNA libraries, non-stressed, control and drought stressed in legume model plant, M. truncatula and identified known and new miRNAs that were responsive to drought stress.
The successful application of high-throughput sequencing technology to systemically identify plant miRNAs has greatly advanced our knowledge on the functions of miRNAs in plants in recent years. There have been several reports on the identification of miRNAs in M. truncatula in the literature. For instance, 26,656 and 844,110 sRNA reads were reported in two recent studies from M. truncatula by Jagadeeswaran et al. (2009) and Lelandais-Briere et al. (2009), respectively, with 97,028 unique sequences being obtained by the latter studies [51, 52]. However, in both cases, sRNAs were sequenced using Roche 454 sequencer, and the total reads and unique sequences obtained from these studies are much less than we reported in the present study (Table 1). In a study by Szittya et al. (2008), the authors obtained 3,948,871 reads and 1,563,959 unique sequences from two sRNAs libraries of M. truncatula using the same high-throughput sequencing technology (Illumina-Solexa) . In contrast to these reported findings, we got a total of 13,683,619 reads and 4,878,445 unique sequences from two libraries of M. truncatula. Therefore, our results contain more reads than those previously reported for miRNAs in M. truncatula in the literature. Moreover, the high-throughput sequencing technology used in the present study allows us to identify the miRNAs with low abundance, thus accounting for the greater unique sequences found in our studies. In addition to the greater amounts of miRNAs sequencing data, the use of most updated miRBase also contributes to our identification of more known miRNAs in M. truncatula. The available database for M. truncatula contains more miRNAs than those for other plant species. For example, in the database, there are 19 miRNAs and 41 miRNAs for Brachypodium and wheat, respectively, while there are 375 miRNAs in the database for M. truncatula (miRBase 17). In addition, to the best of our knowledge, we are the first one to use the M. truncatula genome sequence Mt3.0 to analyze miRNAs in this species. The more genomic information in Mt3.0 than Mt2.0 may also contribute to predicting more new miRNAs in M. truncatula.
Plant miRNAs have a strong propensity to target genes associated with development, particularly those genes encoding transcription factors and F-box proteins . In the present study, we found that the known or predicted targets of miR164, miR169, miR171, miR396, miR1510 and miR5558 were either transcription factors or F-box proteins (Table 3, 4). Under drought stress, an increase in root/shoot ratio was found in M. truncatula (data not shown). miR164 has been reported to regulate root development by a homeostatic mechanism to clear NAC1 mRNA, leading to down-regulation of auxin signals. It has been shown that decreases in NAC1 mRNA levels due to inducible expression of miR164, thus resulting in reduction in lateral root emergence in Arabidopsis . Therefore, it is conceivable that the suppression of miR164 expression may contribute to the increase in root/shoot ratio under drought stress.
Several studies have revealed that miR169 is responsive to abiotic stresses such as drought, cold and salinity in different species [14, 15, 17, 22, 33]. Under drought stress, miR169 exhibited different expression patterns among different species. For example, in Arabidopsis, miR169 is down-regulated by drought stress through an ABA-dependent pathway, resulting in accumulation of NFYA5 with high affinity and sequence specificity for the CCAAT box, which is crucial for the expression of a number of drought-responsive genes . In contrast to Arabidopsis, miR169g in rice is up-regulated by drought [17, 18]. In rice, DREs (dehydration-responsive element) as the upstream of MIR169g are regulated by cold and drought stress, thus leading to up-regulation of miR169g . Trindade et al. (2010) reported that the expression of miR169 in leaves of M. truncatula is not responsive to drought stress . In contrast, our results from both the high-throughput sequencing and RT-qPCR showed that miR169 was down-regulated under drought stress in M. truncatula (Figure 3b and 3c, Additional file 4). This discrepancy may result from the difference in degree of drought stress exposed to plant materials in the two studies. For instance, in the present study, the drought stressed samples were collected after exposing of plants to drought for varying period (6, 8, 10 and 12 d after withholding water), thus our samples included plant materials suffering from a wide range of drought stress. This type of stress has been used in study of molecular response of plants to drought stress [66, 67]. In the studies of Trindade et al. (2010), water stress was exposed by suppressing water supply with relative water content in leaves being approx. 50% and 30%, respectively . In the present paper, the drought stressed leaves with relative water content of 87.4%, 78.8%, 75.2% and 68.3% were used for analysis of miRNAs. Therefore, our samples were mainly those suffering from mild and moderate water stress compared to those of Trindale et al. (2010). If miR169 is responsive to mild and/or early drought stress exclusively, the responsiveness of miR169 may not be detected by the more severe drought stress used by the authors . In addition to withholding water supply, several studies on effect of drought on miRNAs also treated plants with PEG or mannitol for varying period [17, 33]. It is difficult to compare the natural drought stress with PEG-induced and/or mannitol osmotic stress as the two treatments may differ in induction of water stress in terms of rapidity and severity.
Accumulation of reactive oxygen species (ROS) is a common phenomenon in response of plants to abiotic stress. The accumulated ROS damage nucleic acid, oxidize proteins and cause lipid peroxidation [68, 69]. Superoxide dismutases (SODs) detoxify superoxide radicals. The targets of miR398 are two Cu/Zn superoxide dismutases (cytosolic CSD1 and chloroplastic CSD2), and miR398 expression was reported to be down-regulated transcriptionally by oxidative stresses . Oxidative stress often occurs concurrently with drought stress. In the present study, we found that miR398 was down-regulated under drought stress. This would lead to increases in activities of SODs. The drought-induced down-regulation of miR398 in M. truncatula is consistent with the results in maize , but it is contrast to the results reported by Trindale et al. (2010)  and Kantar et al. (2011) . The differences in the expression of miR398 among different studies may results from differences in species, extent and duration of drought stress in different studies.
Plants suffering from water-deficit often display reduced uptake of mineral nutrients. In this context, miR399 negatively regulates the concentration of inorganic phosphate (Pi) by targeting PHO2, a type of E2 conjugase, and overexpression of miR399 in Pi-replete conditions represses expression of E2 conjugase, leading to an increase in Pi concentration in leaves in Arabidopsis . Expression of miR399 is reduced under Pi-deprived conditions to facilitate accumulation of Pi in plants. It has been verified that miR2111 is up-regulated by Pi starvation . In our study, the expression miR399 and miR2111 was similar under drought stress. These results indicate that the function of miR2111 may be as important as miR399 in regulation of nutrient acquisition.
Exposure of plants to a moderate stress induces resistance to other stresses, a phenomenon known as cross adaptation, which has been found in different combinations of stresses . For example, osa-miR821 isolated from virus-infected rice tissues is also expressed in roots of salt-stressed plants, while it is not expressed in healthy, non-stressed plants . Under drought stress, we found that there was up-regulation of miR2089 and miR2118, whose targets may be proteins associated with disease resistance. It is envisaged that these miRNAs may enhance the ability of drought tolerance through unknown mechanisms associated with cross adaptation in plants. Future work aiming at functional elucidation of these miRNAs is warranted by over-expressing these miRNAs in M. truncatula.