SINEs are the numerically most abundant class of transposable elements in the genomes of higher metazoa and thus have a profound impact on the genomic landscape in these organisms. Most importantly, SINEs are frequently found within genes and are therefore transcribed as part of intronic sequences, UTRs, or even exons [36–38].
Recent research has demonstrated that the presence of SINEs in inverted orientation (iSINEs) in transcribed regions of genes can have a significant influence on gene expression through different proposed molecular mechanisms, including translational inhibition and nuclear retention [17, 21, 23].
We therefore studied whether iSINE-containing genes were found with equal frequency as dSINEs. Indeed, our data show that iSINE insertions into annotated transcripts are less abundant than dSINE insertions. Thus, either insertion of SINEs in inverted orientation is selected against as suggested by some [26] or, alternatively, duplicated dSINE insertions may be selected for, as suggested by others [27]. Clearly, although iSINEs have a negative impact on gene expression, they can be found in a considerable number of genes. This suggests that some iSINEs may have acquired regulatory functions that may have beneficial effects. Recently, it was shown that site-selective editing events are significantly increased in the vicinity of iSINEs [39]. Since ADAR enzymes bind to secondary structures, it has been suggested that iSINEs can act as a bait for ADAR enzymes and induce editing at sites located several hundred nucleotides from the Alu elements in the surrounding transcript [40]. Moreover, Ricci and colleagues [20] showed that iSINEs represent a major group of binding targets for STAUFEN1 and overexpression of STAUFEN1 mildly increases nucleocytoplasmic export of the respective mRNA. In general, different RNA binding proteins, including ADAR1, p54nrb, STAUFEN1, and PKR, have been shown to interact with iSINEs and subsequently affect mRNA modification, nuclear retention, mRNA transport, and translational repression, respectively [15, 21, 39, 41]. It seems that iSINEs may act as regulatory elements in the mRNA by providing a double-stranded RNA structure, which serves as a platform for double-stranded RNA-binding proteins. Based on the availability of double-stranded RNA-binding proteins in the cell, iSINEs might recruit them and trigger different cellular processes.
Capshew and colleagues [23] have suggested that the relative position of iSINEs in the 3? UTR can influence the impact of SINEs on gene expression. A minimal distance of 65 nucleotides from the iSINE to the stop codon would be required to repress gene expression [23]. However, the Znf708 iSINE is located close to the stop codon (50 bp) and still leads to a significant reduction in gene expression. Thus, at least in the context of the constructs used by us, we do not observe a position-dependent effect, suggesting that there may be other factors modulating the effects of iSINEs on gene expression.
Our data also indicate that the extent of double-stranded structures formed by iSINEs influences the strength of gene repression. This fact may explain why not all iSINEs in mRNA repress gene expression. Interestingly, artificial 3? UTRs that mimick the secondary structure of an iSINE but contain sequences that were not Alu-like did not reduce gene expression. This finding strongly suggests that, besides the formation of a double-stranded structure, the sequence would also be important.
While the impact of iSINEs on gene expression has been shown in other studies, our study shows a STAUFEN1-independent effect of iSINEs on RNA levels [17, 21, 23]. Similarly, our study also shows that other RNA-binding proteins that recognize double-stranded RNAs are not responsible for the observed reduction in gene expression of iSINE-containing genes. It should be noted, however, that our reporter constructs did show different levels of expression in cell lines deleted for Staufen, ADAR, PKR, or Dicer. This may indicate that these factors do have some effect on gene expression but may also reflect the fact that these cell lines show different degrees of differentiation and also originate from different mouse strains. Importantly, however, irrespective of the genetic background, iSINE-containing constructs consistently showed a reduced level of expression, suggesting a more general mechanism of gene repression conserved in all cell lines tested. Indeed, our Pol II ChIP experiments showed that the presence of iSINEs leads to a reduction in Pol II density distal to iSINEs in the 3? UTR. We conclude that iSINEs interfere with Pol II activity and thus have a negative impact on mRNA transcription. Interestingly, in bacteria, double-stranded structures in RNAs have also been shown to interact with the bacterial RNA polymerase exit channel, prolong RNA polymerase pausing, and consequently reduce the transcriptional elongation rate [42]. Importantly, the crystal structures of bacterial RNA polymerase and eukaryotic Pol II are similar in the regions that interact with the transcriptional bubble [43]. Therefore, the long double-stranded structures formed by inverted SINEs used in our study might act in a comparable manner leading to an increase in Pol II pausing [44].