Salinity is an increasingly important agricultural problem. The metabolism and physiological integrity of plants are affected by salt stress and in recent years many studies have been devoted to understanding the molecular mechanisms of plant salt tolerance. Studies have confirmed that resistance to salt stress is associated with the use of compatible solutes by plants , and with ion transporters . miRNAs are a class of non-coding sRNAs that are implicated in many developmental processes and in responses to various abiotic stresses, and which play pivotal roles in plant adaptation . To date, in excess of 24,000 hairpin sequences and 30,000 mature sequences have been identified from plants, animals, unicellular organisms and viruses (miRBase); however, scant attention has been paid to the identification of miRNAs in broccoli. In the present study, we identified several million sRNA sequences in salt-stressed plantlets of broccoli, using high-throughput sequencing methods to identify miRNAs associated with physiology and metabolism under salt stress. A total of 42 putative known and 39 putative candidate miRNAs that were differentially expressed between control and salt-stressed broccoli were discovered according to their sequencing-reads numbers and confirmed by real-time RT-PCR. Finally, analysis of the predicted targets of the miRNAs using the GO and KO databases indicated that a range of metabolic pathways and biological processes known to be associated with salt stress were up-regulated in broccoli treated with salt.
The study of miRNAs using traditional methods is complex, and high-throughput sequencing methods now provide a rapid and efficient additional approach by which to identify and profile populations of sRNAs at different stages of plant development. Large numbers of non-conserved or species-specific miRNAs often accumulate in plants at lower levels than conserved miRNAs, and are therefore not easily revealed using traditional sequencing methods [3,30]. In the present study, the results indicated a range of sRNAs, of length 16–29 nt, in broccoli, with most of the unique sequence reads being 24 nt in length. Several plant species, including A. thaliana, Citrus sativus and C. sinensis, had been shown to contain substantially more 24-nt sRNAs than 21-nt sRNAs [31,32]; on the other hand, sRNAs populations with more 21-nt members than 24-nt were reported in B. juncea and in Japanese apricot with imperfect flower buds [32,33]. B. juncea, B. napus and broccoli all belong to the same genus; however, when broccoli sRNAs are compared to those of B. juncea, a striking difference exists, whereas little difference is observed when broccoli sRNAs are compared to those of B. napus . Thus, we can conclude that the sRNA transcriptome is complex, and can be highly variable even between closely related plant species.
Generally speaking, length distribution analyses of sRNAs provide a helpful way to assess the composition of sRNA samples. Previous studies have shown that DCL1 mainly produces sRNAs that are 18–21 nt long. In contrast, the products of DCL2, DCL3 and DCL4 are 22 nt, 24 nt and 21 nt long, respectively. Furthermore, miRNAs are normally 21 nt or 22 nt long, whereas small interfering RNA (siRNA) are 24 nt long . By this reasoning, therefore, our results imply that the most abundant sRNAs are miRNAs in broccoli and siRNAs that have been cleaved by DCL1 and DCL4.
Sequence information for broccoli miRNAs is entirely absent from miRBase, and furthermore there are fewer than 50 B. rapa miRNAs in this database. Therefore, the sRNA sequences were aligned instead against miRNA precursors/mature miRNAs of the Viridiplantae in miRBase. A search for conserved miRNAs revealed that the majority of these molecules, known already from A. thaliana and other species, were detectable in broccoli and were expressed at relatively high abundance. The patterns of expression of conserved miRNAs are variable in different plants. Previous studies had shown that miR159, miR166a, miR164, miR171f and miR168 were all detectable in B.rapa, and with relatively high numbers of reads, whereas members of the miR169 family showed low read numbers [30,34]. The findings of the present study were quite different; miR166a, miR168a and miR157d showed the highest values for copy number in control broccoli.
A previous report suggested that miR403, which was initially identified in A. thaliana and later found in Populus trichocarpa, was a dicot-specific miRNA and that miR437 and miR444 might be monocot-specific miRNAs . On the other hand, Sunkar et al. indicated that miR390 was present in both monocots and dicots . Our results provided a large-scale database against which to examine these suggestions. We detected both miR403 and miR390 in broccoli, whereas miR437 and miR444 were not detectable, thereby confirming the previous reports. Furthermore, the miR158 and miR391 sequences have been considered to be A. thaliana-specific ; however, we detected miR158a, miR158b, and miR391 in broccoli, each at an abundance of about 30,000 reads.
Although many plant miRNAs have been found to be conserved – for example, miR319, miR156/157, miR169, miR165/166 and miR394 have been found in more than 40 plant species  – studies have indicated that most miRNAs can be induced by environmental stresses. For example, miR399 is induced under low-phosphate conditions , miR395 increases upon sulfate starvation  and miR393 is up-regulated in response to cold, dehydration, NaCl and abscisic acid (ABA) stress . A recent study showed that miR417 was transiently up-regulated in response to osmotic stress . On the other hand, miR398 is down-regulated in response to oxidative stress. Several differentially regulated miRNAs have been identified in salt-stressed plants. For example, miR530a, miR1445, miR1446a-e and miR1447 were down-regulated during salt stress in P. trichocarpa, as measured by microarray analysis of known miRNAs, whereas miR482.2 and miR1450 were up-regulated . In A. thaliana, miR156, miR158, miR159, miR165, miR167, miR168, miR169, miR171, miR319, miR393, miR394, miR396 and miR397 were all up-regulated in response to salt stress, whereas miR398 was down-regulated. Furthermore, miR169g and miR169n were also reported to be induced by high salinity . Recently, a study of maize roots using miRNA microarray hybridization indicated that members of the miR156, miR164, miR167, and miR396 families were down-regulated by salt shock, whereas miR162 and miR168 were up-regulated . Our result in broccoli indicated that, amongst the conserved miRNAs, miR393 and miR855 were the most strongly down-regulated in response to salt stress, whereas conserved miR838-5p and miR396a were the most strongly up-regulated. Of the putative novel miRNAs, miR37 showed the greatest degree of up-regulation in response to salt stress, whereas miR3 and miR34 showed the greatest degree of down-regulation. The differences between our findings and the reports relating to salt stress in P. trichocarpa, A. thaliana and maize roots  might be explained in part, of course, by differences in behavior between species but they might also reflect the use of different experiment methods: microarray analysis was used in some of the previous studies, whereas in the present study we used sequencing, which is more effective than microarray analysis for the detection of miRNAs, including novel miRNAs, that are expressed at low levels.
In the case of the human genome, more than a third of the genes have been predicted to be miRNA targets, and these targets would appear to be involved in a wide range of biological functions. In A. thaliana, miRNAs may target transcripts that encode proteins and transcription factors . Our results indicated that a total of 836 and 527 mRNAs were targets of putative conserved miRNAs and putative novel miRNAs, respectively. These targets regulate a range of cellular processes, within the broad categories of cell-cycle regulation, plant hormonal signal transduction, and metabolism. However, there are several miRNAs without target genes; these could be the result of erroneous target predictions, or they might be low-abundance miRNAs with limited or no activity. It is also possible that miRNAs might exist that have no targets. Nevertheless, the KO and GO analyses revealed that many of the genes targeted by miRNAs in broccoli are related to salt stress, supporting the hypothesis that miRNAs play an important role in the response of broccoli to salinity.