Micro RNAs are small, non-coding strands of RNA that bind via homology to the mRNA transcript, disrupting protein synthesis in the ribosomes. Recent research has shown that these micro RNA’s are major players in gene regulation at the post-transcriptional level, for example, Forman et. al. demonstrated that the binding of miRNA to target mRNA, and subsequent requirement of inhibiting proteins interact with the ribosome, hindering elongation. Furthermore, miRNAs are implicated in diseases like cancer. By binding to the mRNA transcript, these miRNAs are able, through destabilization the mRNA and ribosomal interference, to inhibit protein synthesis. The precise mode of action by which miRNAs silence gene expression is still not entirely known; however, because of their gene silencing ability, miRNAs have been identified as potential therapeutics and biomarkers for diseases. Because of their incredible potential, it is imperative that more research be done to further understand how they work, and under what conditions they are most effective. For example, research has shown that at over expressed levels, miRNAs that inhibit oncogenes can reduce cancerous cellular activity. Additionally, it is theoretically possible to take advantage of post-transcriptional gene regulation and use it in cancer treatment. However, when engineering a potential miRNA drug, we are met with certain difficulties that stem from a general lack of knowledge about how the physical miRNA actually works.
Micro RNA’s, or miRNA’s are endogenously encoded, non protein-coding RNA segments that imperfectly, yet specifically, bind to the mRNA transcript (1). They are approximately 23 nucleotides in length, and, due to their short length and imperfect binding, miRNAs have many potential binding sites within the mRNA transcript. This brings up the main controversy regarding miRNA activity, namely, where the miRNA actually binds to transcript (1). The mature mRNA (figure 1) is composed of three major parts, the 3’ un-translated region (3’ UTR), the open reading frame (ORF), which is where the protein coding sequence is found, and the 5’ un-translated region (5’ UTR). The miRNA-binding site, located within the transcript, is a critical piece of the miRNA puzzle that we still do not completely know. Additionally, because the mechanism for translational inhibition hinges on the spatial location of the miRNA-binding site relative to the protein-coding region of the mRNA transcript, the miRNA target site has major implications regarding how the silencing of genes, promoted by miRNA binding happens. Knowing where the miRNA specifically binds to the transcript, and which stage of protein synthesis is hindered due to the miRNA recruitment of the protein silencing complex is a key step in potential miRNA drug engineering. A few of those possible-silencing mechanisms for gene silencing will reviewed here.
Binding at the 3’ UTR and 5’ UTR
It is commonly thought that all miRNA segments bind to the 3’ UTR of the mRNA transcript. Its not an unreasonable assumption to make, as the proximity of the 3’ UTR to the poly-A tail and the termination codon make this site less structured and aid in the miRNAs ability to recognize the homologous target site. During translation initiation, a protein complex of eukaryotic initiation factors (eIFs), brings the 5’ G-cap and the 3’ poly-A tail close together, which allows for the formation of the translational loop (Figure 2). This loop stabilizes the mRNA during translation, and allows for a more efficient translation by the ribosome, the tRNA, and all other translational factors. Disruption of this loop would not only hinder the ability of the ribosome to efficiently translate the mRNA transcript into the final polypeptide product, the ultimate goal of the central dogma, but would also destabilize the physical mRNA.
The binding of miRNA to the 3’ UTR of the transcript would do exactly this, prevent the eIFs from associating the 3’ and the 5’ ends of the mRNA, and thereby prevent the translational loop from forming. If there is no loop formation, translation will not occur, and as a result, protein synthesis of that specific gene will be, phenotypically, unexpressed.
Forman et.al came up with an algorithm to identify evolutionarily conserved sequence motifs. They searched for these motifs, specifically, within coding regions of the genome. They adjusted the search to look for short binding sequences characteristic of miRNA binding sites (~8bp long). They identified four miRNA target sites that scored exceptionally high in sequence motif matching, one of which was the miRNA let-7 target site (2).
Forman, in 2008, was aware of the common thought that miRNA binds to the 3’UTR. This 3’ theory, combined with his data that indicated that there was highly conserved miRNA binding site in a coding sequence, led Forman to investigate whether target sites found outside of the 3’ UTR are responsive to miRNA binding. Since the binding site for let-7 was so highly conserved, they transfected human cells with let-7 and monitored the downstream effects in transcript levels. What they found changed the way we thought about how miRNA works. They observed regulation from miRNA target sites outside of the 3’UTR and down regulation of Dicer (the protein regulated by let-7). This indicated that functional miRNA binding sites can and do exist outside the 3’ UTR.
To analyze the mechanistic differences between traditional 3’ UTR binding and non 3’ UTR binding, Forman et al used the RNAcofold computer program to theoretically fold miRlet-7 around all of it’s target sites (3’ UTR or otherwise. They found preference for base pairing at nucleotides 13-16 on the mRNA’s 3’ UTR. Furthermore, bases 10-12 showed preference for loop formation around the mRNA. In 5’ UTR miRNA binding-sites, however, there was no obvious preference for loop formation. This points to a potential difference, mechanistically, in the 5’ UTR-bound miRNA silencing of mRNAs, and 3’ UTR-bound miRNA silencing of mRNAs (2).
Dissenting to Forman’s school of thought is Lytle et.al who observed that translational repression of IRES-containing mRNA, when miRNA binding sites were 3’ to the ORF, was comparable to the mRNA that had miRNA binding site 5’ to the ORF (3). They concluded that there is no mechanistic difference in the mode of repression between 3’UTR binding of miRNA and 5’ UTR binding of miRNA. However, that conclusion seems to be premature.
Lytle’s experiment was conducted using electroporation to transfect HeLa cells (the cell line used their experiment) with mRNA that contained miRNA-binding sites. This transfection increases the permeability of the cell membrane through the use of an electric field, allowing substrates like DNA and RNA to enter the cell by. The expriemt was repeated using a different transfection technique, cationic lipid transfection. In CLT, negatively charged substrates, like DNA or RNA, are coordinated with hydrocarbons that have positively charged heads. The hydrocarbon then facilitates the entry of the entire complex into the cell ( 3). The results from the CLT experiment were not consistent with the results from the electroporation experiment.
In the CLT experiment, Lytle et.al. only observed translational repression when the miRNA binding sites were in the 5’UTR. They explained this difference in data as a function of CLT’s extremely high efficiency. If there are a lot of mRNAs, there are a lot of miRNA binding sites. If all of those binding sites are introduced into the cell, there will not be enough endogenous miRNA to effectively bind to, and repress the translation of the mRNA transcript. This answer, however, does not explain why translation repression was observed at the 5’ end. Furthermore, Lytle et.al took care to use the lowest concentration of DNA possible in the CLT experiment. All this evidence indicates that there is some other reason that translational repression was not observed when the miRNA interacted the 3’UTR of the mRNA in the CLT experiment.
One possible explanation hinges on the charged nature of CLT. Using Let-7 as a model, when the miRSC (micro RNA silencing complex- an miRNA-protein complex that binds to the mRNA transcript) binds to the 3’UTR or 5’UTR target site in the mRNA, there are base pairing interactions (bases 13-16). These interactions occur when Watson and Crick base pairs form hydrogen bonds. When the target site is in the 3’UTR, there are additional, non-base specific interactions at bases 10-12 that promote looping. This looping is contingent upon the shape and the charge of the mRNA. If either of those criteria were disrupted, it would not be unreasonable to suggest that the looping described by Forman et.al. would be also be disrupted.
If we extend the data from Forman et.al to other miRNA and their specific target sites (which is not unreasonable, as Forman et.al. shows that miRNA target sites are evolutionarily conserved), it is reasonable to theorize that there is preference towards looping when the miRNA target site is in the 3’UTR. If the positively charged head from the cationic lipid interacted with the 3’UTR of the mRNA transcript after transfection (which it very well might since the mRNA has a negatively charged backbone), this electrostatic interaction, while not affecting the base pairing interaction, might have an effect on the looping of the miRNA. This loop-forming inhibition might be the reason that Lytle et.al. (and others who have done similar experiments with the same results) only saw translational inhibition in CLT cells when miRNA target sites were in the 5’UTR (since the only interaction between the 5’UTR of the mRNA and miRNA is base pairing). If this were true, this would indicate a mechanistic difference between
miRNA-silencing of mRNA when bound to the 3’ UTR and 5’ UTR.
While these two mechanisms differ, they both share a commonality in that they both inhibit protein synthesis in the ribosome by blocking the initiation of translation. Because the mRNA transcript degrades, translation is unable to start. There is, however, another possible mechanism whereby miRNA inhibits gene expression post initiation of translation.
One possible way to resolve the uncertainty of whether 3’ UTR binding of the miRNA is mechanistically and structurally different from 5’ UTR could be to have sample mRNAs, some with known miRNA target sites in the 3’UTR and others in the 5’UTR. We bind miRNA to them, and then run the products on a 2D gel. If the 3’ UTR bound products appear to be less linear, that could indicate the 3’ miRNA loop formation, and a mechanistic difference between miRNA binding at the 3’ and 5’ UTRs.
Binding to the ORF
Another novel possibility outlined by Forman et.al. is miRNA-ORF interactions. When identifying miRNA target sites, they discovered certain highly conserved sequences, characteristic of miRNA binding sites within the coding sequence of the mRNA transcript. While miRNA-binding at the 3’ and 5’ UTR most likely disrupt the mRNA loop within the ribosome, miRNA interacting with the ORF would have no effect on mRNA loop formation. The most likely mode of translational suppression, given a miRNA-ORF interaction, is ribosomal drop-off (Figure 4). This happens when the ribosome is unable to interact with the coding sequence of the mRNA, and dissociates from the transcript before translation has been completed.
While miRNA binding to the ORF may have a complete silencing effect, regarding gene expression of proteins, partial silencing may also be observed. This happens when enough of the amino acid chain is synthesized for some functionality to happen, even though elongation is prematurely terminated. If, however, not enough of the mRNA was translated into protein, the partial amino acid sequence would be degraded after the ribosome detached, resulting in complete silencing of the target gene. This, once again is dependent on the location of the miRNA-binding site within the ORF.
This shows yet another possible mechanism for how miRNA might exact it’s post transcriptional control over gene expression.
Competition of miRNA Binding Sites
The precise location and mechanism of miRNA (relative to the mRNA transcript), while important, does not fully illustrate the post-translational gene regulation effect of miRNA. Considering the fact that typical mRNAs are expressed at 10-00 copies per cell, and each mRNA has approximately 1-2 binding sites (4), and many miRNAs can specifically bind to many miRNA binding sites (due to their small size and imperfect base pairing nature), we are left with the realization that within a single cell there are roughly 10-200 miRNA binding sites per gene per cell. When you consider that the human genome is roughly 25,000 genes in size (5), that translates to anywhere between 250,000 and 5 million miRNA binding sites in a given cell for all miRNA! This rough estimate is obviously contingent upon details such as cell type, site specificity of the miRNA-binding site, and number of mRNA transcripts with the specific site. If the miRNA in question is rare and eccentric in sequence, and regulates a gene that is associated with a small number of mRNAs, there will be few genomic binding sites for that specific miRNA. On the other hand, if the miRNA in question has a sequence similar to other miRNAs, and regulates a gene that requires a high number of mRNAs, there would be many more potential genomic miRNA-binding sites for that specific substrate. The reason the number of miRNA binding sites is important to post transcriptional gene regulation is because of binding competition between open miRNA binding sites, and available miRNA.
When thinking about regulation, there is often the misconception that the regulatory substrate is in excess to the target. This model would allow for easy regulation of any target gene. The fact is, the concentration of miRNA binding sites are at a molar equivalence (if not in excess) to the available miRNA substrate. This means that mRNA can compete to bind miRNA. This can result in smaller, muted effect of the miRNA on gene regulation (at certain miRNA concentrations) (4).
Similarly, there are certain RNAs that have many miRNA binding sites. These binding sites, if located in a non-regulatory region (a site in the mRNA that, when bound with miRNA, does not result in post transcriptional gene regulation) sponge, or soak up, available miRNA, further decreasing the regulatory effect of miRNA on the target gene (figure 5). Furthermore, this sponging phenomenon can promote other gene activity since the sponging activity will mute the gene silencing activity of miRNA.
While there are still many unknowns about the precise mechanism and effects of miRNA in regard to post-transcriptional gene regulation, one thing is clear: abnormal expression of miRNAs is associated with many diseases such as cancer (9)(10)(6). There is some miRNA research is currently being conducted, and one of the questions that researchers are asking is how can we use miRNA as a therapeutic? One possible approach to silencing oncogenes might be to administer a miRNA that would bind to the 5’ UTR of the oncogene transcript (as this is easily and efficiently achievable using CLT, and would avoid the problem of partial silencing sometimes observed in ORF targeting miRNA) since they have been observed to the most effective. It would then be important to identify all mRNAs present in the cancerous tissue in order to determine the concentration of potential miRNA binding sites that might compete with the prime oncogene target.
Armed with that information, it would be possible to calculate the appropriate concentration of the miRNA that should be administered to the cancerous tissue in order to effectively silence the oncogene (11).
Similarly, if a tumor suppressor gene were under-expressed (resulting in cancer) the inverse therapy would also be possible (12). An RNA sponge could be administered to absorb free miRNA that otherwise would bind to and silence the tumor suppressor gene. With this added competition for the miRNA, the tumor suppressor gene would be able to be expressed at a less regulated level (effectively higher levels), and fight the cancerous tissue by regulating the cell cycle. These are some ways that diseases could be treated in the future. What we need now is for more research to be conducted in this brand new field of genetics.
- Cipolla, G.-A. (2014). A non-canonical landscape of the microRNA system.Frontiers in Genetics, 5, 337. doi:10.3389/fgene.2014.00337
- Forman, J. J., Legesse-Miller, A., & Coller, H. A. (2008). A search for conserved sequences in coding regions reveals that the let-7microRNA targets Dicer within its coding sequence. Proceedings of the National Academy of Sciences of the United States of America, 105(39), 14879–14884. doi:10.1073/pnas.0803230105
- Lytle, J. R., Yario, T. A., & Steitz, J. A. (2007). Target mRNAs are repressed as efficiently by microRNA-binding sites in the 5′ UTR as in the 3′ UTR.Proceedings of the National Academy of Sciences of the United States of America, 104(23), 9667–9672. doi:10.1073/pnas.0703820104
- Jens, M., Raiewsky, N. (2015). Competition Between Target Sites of Regulators Shapes Post-Transcriptional Gene Regulators. Nature Reviews Genetics, 16(2): 113-126.
- Collins, F.-S., Lander, E.-S., Rogers, J., Waterston, R.-H. (2004). Finishing the euchromatic sequence of the human genome. Nature,431: 931-945
- Dang, Z., Xu, W.-H., Lu, P., Wu, N., Liu, J., Ruan, B., Zhou, L., Song, W.-J., Dou, K.-F. (2014). MicroRNA-135a Inhibits Cell Proliferation by Targeting Bmi1 in Pancreatic Ductal Adenocarcinoma. International Journal of Biological Sciences, 10(8), 733–745.
- Friend, K., Campbell, Z. T., Cooke, A., Conner-Kroll, P., Wickens, M. P., Kimble, J. (2012). A conserved PUF–Ago–eEF1A complex attenuates translation elongation. Nature Structural & Molecular Biology, 19: 176-183.
- (2008). MRNA Structure.svg. Wikimedia Commons,
- Lin, J., Li, J., Huang, B., Liu, J., Chen, X., Chen, X.-M., Xu, Y.-M., Huang, L.-F., Wang, X.-Z. (2015). Exosomes: Novel Biomarkers for Clinical Diagnosis. The Scientific World Journal, 2015, 657086. doi:10.1155/2015/657086
- Ebrahimi, F., Gopalan, V., Smith, R.-A., Lam, A.-K.-Y. (2013). miR-126 in human cancers: Clinical roles and current perspectives. Experimental and Molecular Pathology, 96(1), 98-107.
- Gu J, Zhang Z, Wang Y, Gu C (2015). MicroRNA-410 functions as a tumor suppressor by targeting angiotensin II type 1 receptor in pancreatic cancer. IUBMB Life 67(1): 42-53.
- Zeng, T., Peng, L., Chao, C., Fu, B., Wang, G., Wang, Y., & Zhu, X. (2014). miR-451 inhibits invasion and proliferation of bladder cancer by regulating EMT. International Journal of Clinical and Experimental Pathology, 7(11), 7653–7662.