The effect of Alu on other ncRNAs has rarely been studied. In recent years, as ncRNA research has progressed rapidly, more and more repeats have been found to be important for gene transcription. Dieci et al. comprehensively summarized the evidence that Pol-III may transcribe their downstream sequences through tRNAs, repeats or ncRNAs [7
]. On the basis of previous studies, we hereby propose the hypothesis that some short interspersed nuclear elements (SINEs) without terminator sequences may trigger the transcription of their downstream miRNAs. Guided by this hypothesis, we investigated the mechanism of miRNA transcription through Alu (Figure ). We not only predicted 60 Alu-related miRNAs that might be transcribed through Alu, including the previously reported miR-517a, but also identified 23 of them by Solexa sequencing. More importantly, using induction by heat shock and cycloheximide, we showed that the expression of Alu RNAs was consistent with the expression of three miRNAs downstream of those elements, providing clear-cut evidence that transcription of some miRNAs may be initiated by Alu transcription.
Recently, studies on miRNA transcription by two different groups have indicated that Pol-II or Pol-III is associated with miR-517a transcription [14
]. RNA Pol-I is known to be insensitive to α-amanitin; RNA Pol-II is very sensitive to it and RNA Pol-III moderately sensitive. In order to evade the ambiguous effects of α-amanitin, we used tagetitoxin as a specific Pol-III inhibitor [33
]. As expected, our results showed that tagetitoxin significantly repressed the expression of Pol-III-mediated miRNAs, but did not affect the transcription of the Pol-II-driven let-7a. Considered together, the inhibitors used for Pol-III may be a key factor in the differences among research results. Although only the example of Alu RNA has been investigated in this article, we cannot exclude the possibility that other SINEs play the same role as Alu, because some known and predicted miRNAs were also found downstream of MIR (Additional file 5
). Transcription of these miRNAs seems likely to depend on key sequences such as A/B boxes or other motifs embedded within Alu elements.
Other members of the SINE family, in addition to Alu, all originate from tRNAs and have a classical structure comprising the following three parts: first, a promoter transformed from tRNA; second, a sequence specific for the SINE; third, a sequence needed for reverse transcription, similar to the 3' end of LINE and simple repeats [8
]. While the other SINE family members have mostly evolved from tRNAs, they are different from Alu in structure but very similar in transcription. It is generally believed that in both tRNAs and SINEs, the A box and B box within the left arm are used to regulate transcription. Of these two boxes, the box B may play a decisive role [7
]. Therefore, the transcription of SINE elements is also very likely to trigger the expression of their downstream small RNAs.
However, among the three newly identified miRNAs that have transcriptional activity and are co-expressed with Alu elements, only AluJo-576611 and AluJo-135090 contain the classic B box, and AluJb-641 has no B box; neither has the upstream AluSg/x-1110454, which has been shown to be able to transcribe miR-517a. Even though AluSg/x-1110454 was mapped to the left arm of the uniform sequence of Alu in our multiple sequencing alignments, its B box was severely mutated. In the B box, the critical nucleotides are the first locus G and the third locus T, but the third locus of AluSg/x-1110454 is G. In the multiple sequencing alignments, one motif, GAGGCTGAGG, was found to be highly conserved. It is present not only in these four sequences but also in uniform sequences of all Alu family members. This motif may be of great significance for the transcription of Alu and its downstream miRNAs, or have a similar function to the B box in regulating Alu transcription. However, this possibility still awaits further study.
When miRNAs were discovered, they were originally thought to have nothing to do with repeats. For example, when miRNAs were predicted, Alu-related repeats were first filtered out. Nonetheless, it has been found over the past two years that miRNAs and repeats are closely interrelated. miRNAs are very likely to originate from repeats and may interact with repeats, or with mRNAs with related repeats embedded in their 3'UTR regions [36
]. Repeats are now known to be able to direct transcription of miRNAs. Considering these findings together, we believe that with further studies, a regulatory network constituted by miRNAs, repeats and target RNAs will be unveiled layer by layer, enhancing our understanding of the nature of life.