How MicroRNAs Modulate Gene Expression
By: Carol A. Rouzer, VICB Communications
Published: April 12, 2013
Combined proteomics and transcriptomics analyses reveal the importance of both transcriptional repression and mRNA decay in the control of gene expression by microRNAs.
MicroRNAs (miRNAs) are small (~21 nt) single-stranded RNAs that bind to the messenger RNAs (mRNAs) of protein-coding genes. The interaction of an miRNA with its target mRNA leads to repression of gene expression by two possible mechanisms, promotion of mRNA decay and inhibition of translation. The degree to which each of these mechanisms contributes to miRNA-mediated regulation of gene expression is unclear. Contributing to this uncertainty is the focus in many prior studies on target mRNAs with structures known to promote mRNA decay, and the frequent use of mRNA expression data as the major or only indication of miRNA-dependent effects. Now, VICB members Bing Zhang and Dan Liebler and their laboratories have combined comprehensive proteomics analysis with miRNA and mRNA expression data to obtain a better understanding of the relative roles of mRNA decay and translation repression in miRNA function [Q. Liu et al. (2013) Mol. Cell. Proteomics, published online April 2, DOI:10.1074/mcp.M112.025783].
The project actually began when Rob Slebos and Patrick Halvey in the Liebler lab became interested in the role of mismatch repair defects on cellular proteomes. To address this question, they chose nine colorectal carcinoma cell lines and performed triplicate liquid chromatography-tandem mass spectrometry-based shotgun proteomic analyses of each cell line. The number of peptides identified (ranging from 25,159 to 31,526) led to the identification of from 5,162 to 5,698 proteins for each cell line, thus providing an extensive dataset for the intended studies. During the bioinformatics analysis of the results, Qi Liu and Bing Zhang realized that comparable miRNA and mRNA transcriptomics data were also available for these cell lines, leading them to propose that a combination of the proteomics and transcriptomics data would pave the way for a comprehensive exploration of miRNA function.
Analysis of the proteomics and transcriptomics data provided 5,144 genes with paired mRNA and protein expression data. The Zhang lab used bioinformatics techniques to search the data for relationships between miRNA expression levels and the levels of expression of their target mRNAs and proteins (Figure 1). They reasoned that, for miRNAs that promote mRNA decay, there would be a negative correlation between miRNA and mRNA expression (miRNA-mRNA correlation). For miRNAs that repress mRNA translation, there would be a negative correlation between miRNA expression and the ratio of protein to mRNA expression (miRNA-ratio correlation). Either mechanism, or both working together would lead to a negative correlation between miRNA and protein expression (miRNA-protein correlation).
Figure 1. Identification of miRNA mechanisms of action. Promotion of mRNA decay results in a negative miRNA-mRNA correlation. Translation repression produces a negative miRNA-protein/mRNA ratio correlation. The combined mechanisms lead to a negative miRNA-protein correlation. Reproduced with permission from Q. Liu et al. (2013) Mol. Cell. Proteomics, published online April 2, DOI:10.1074/mcp.M112.025783. Copyright 2013, ASBMB.
Prior work has identified key features of the binding sites on the target mRNA that facilitate miRNA-dependent expression regulation. The “seed region” of the miRNA (positions 2 through 7) is particularly important, and complementarity to this region in the mRNA target site is critical. Extension of complementarity to position 8 and the presence of A across from position 1 also facilitate target site efficacy. There is usually poor complementarity at positions 9 through 12 of the miRNA, often leading to a “bulge” structure. This is followed by a second region of complementarity, particularly at miRNA positions 13 through 16 (Figure 2). An mRNA will usually possess more than one miRNA target site, and these are most often found in the 3′-UTR (untranslated region). A surrounding region of high AU content also facilitates target site efficacy.
Figure 2. Key structural features of an miRNA binding site. Of critical importance is a region that is highly complementary to the seed region (nucleotides 2 through 7, shown in red and blue) of the miRNA. Greater efficacy is obtained if the mRNA bears an A across from miRNA position 1 (shown in purple), and if the mRNA and miRNAs are also complementary at miRNA position 8 (shown in green and orange). An uncomplementary bulge is usually present in the region corresponding to miRNA positions 9 through 12, followed by a second region of complementarity in the 3’ end of the miRNA, especially corresponding to positions 13 through 16 (shown in red and blue).
Most of the studies that had characterized optimal miRNA binding sites as described above used mRNA decay as the measure of target site efficacy. Thus, the Zhang group hypothesized that these characteristics might not be optimal to promote miRNA-mediated translational repression. They tested this hypothesis by investigating the effects of individual target site characteristics on mRNA decay and translation repression using their miRNA-mRNA and miRNA-ratio correlations. First, they investigated the characteristics of the target site in the vicinity of the miRNA seed region, using four types of sites (Figure 3). The 6mer site contains nucleotides complementary to positions 2 through 7 of the miRNA seed region. The 7mer-m8 site is also complementary at miRNA position 8. The 7mer-A1 site is complementary at miRNA positions 2 through 7 and contains an A across from miRNA position 1 (which may or may not be a complementary U). The 8mer site combines the characteristics of the two 7mer sites. The Zhang lab found that interaction of an miRNA with an mRNA bearing at least one 8mer site in its 3′-UTR was most likely to promote mRNA decay. Efficacy for the decay pathway progressively decreased when the 8mer site was substituted with a 7mer-m8, 7mer-A1, and 6mer site, in that order. These results were consistent with prior findings. However, the investigators also determined that the presence of an 8mer site did not facilitate miRNA-mediated translation repression. Furthermore, they discovered that the presence of an 8mer site in the 5′-UTR or ORF (open reading frame) regions of the mRNA did not facilitate mRNA decay, but its presence in the ORF did promote translation repression. Pairing the presence of an 8mer site with good complementarity to the 3′ region of the miRNA also promoted mRNA decay, as reported previously; however, this characteristic had the opposite effect on translation repression. Thus, most of the characteristics previously identified as promoting target site efficacy did so only for mRNA decay while having no effect or a negative effect on translation repression. Only the presence of an 8mer site in a region of high AU content facilitated both mRNA decay and translation repression.
Figure 3. Four types of target sites identified on the basis of mRNA structure in the vicinity of the miRNA seed region. The 6mer site is complementary to miRNA positions 2 through 7. The 7mer-m8 site is also complementary to miRNA position 8. The 7mer-A1 site is complementary to positions 2 through 7 and has an A opposing position 1 (that need not be complementary). The 8mer site combines the characteristics of the two 7mer sites.
The Zhang lab next performed a more detailed examination of 580 interactions involving 60 miRNAs and 423 genes identified as potential targets using the TargetScan, miRanda, and/or MirTarget2 software packages. Using the correlations between expression of these miRNAs with mRNA, protein, or the protein/mRNA ratio, the investigators identified six categories of interactions:
• RD - mRNA decay is the major mechanism, characterized by a negative miRNA-mRNA and miRNA-protein correlation with no miRNA-ratio correlation.
• RD_o - mRNA decay is the major mechanism, but other compensating mechanisms counteract the effects of miRNA on protein expression. This is characterized by a negative miRNA-mRNA correlation, but miRNA-protein exhibits no correlation and miRNA-ratio may actually be positive.
• TR - translation repression is the major mechanism, characterized by a negative miRNA-ratio and miRNA-protein correlation with no miRNA-mRNA correlation.
• TR_o - translation repression is the major mechanism, but other compensating mechanisms counteract the effects of the miRNA on mRNA translation. This is characterized by a negative miRNA-ratio correlation, but miRNA-protein exhibits no correlation and miRNA-mRNA may actually be positive.
• B_s - both mRNA decay and translation repression occur with strong effects. A negative correlation is observed between miRNA and mRNA, protein, and protein/mRNA ratio.
• B_w both mRNA decay and translation repression occur with weak effects. Only miRNA-protein shows a negative correlation
Investigation of the 580 miRNA-gene interactions revealed that 248 (43%) could be characterized as RD, RD_o, or B_s (Figure 4). For these, promotion of mRNA decay was strong enough that miRNA-mediated effects could be identified from measurement of mRNA expression levels alone. However, the remaining interactions, 30% of which were TR or TR_o and 27% of which were B_w, involve a significant contribution from translation repression and would not be detected without proteomics analysis.
Figure 4. Characterization of 580 miRNA-gene interactions on the basis of correlations between miRNA and mRNA, protein, and the protein/mRNA ratio. Reproduced with permission from Q. Liu et al. (2013) Mol. Cell. Proteomics, published online April 2, DOI:10.1074/mcp.M112.025783. Copyright 2013, ASBMB.
More detailed analysis of miR-138 revealed that modulation of 10 out of 16 of its targets occurred by TR or TR_o, and no targets modulated by the RD or RD_o mechanisms were observed (Figure 5). Of the genes controlled by miR-138, 14 are involved in cell migration or tumor metastasis. Consistently, the expression of miR-138 was 8-fold higher in the non-metastatic SW480 cell line as compared to the SW620 line which was derived from a metastatic tumor from the same patient. Despite the large difference in miR-138 expression between these two cell lines, a less than 2-fold difference was observed in the expression of any of the targeted mRNAs. In contrast, expression of 9 of the regulated proteins was over two-fold higher in SW620 cells as compared to SW480 cells, and the difference in expression of 6 of the target proteins was statistically significant. These results are consistent with the conclusion that miR-138-mediated gene regulation occurs primarily by translation repression as opposed to mRNA decay, confirming the importance of the translation repression mechanism in this miRNA, which likely plays an important role in the malignant phenotype.
Figure 4. Mechanisms used by miR-138 to regulate its 16 target genes, and the functional roles of those genes. Reproduced with permission from Q. Liu et al. (2013) Mol. Cell. Proteomics, published online April 2, DOI:10.1074/mcp.M112.025783. Copyright 2013, ASBMB.
Together the data reveal the importance of comprehensive proteomics analysis in the evaluation of miRNA function. Clearly, reliance on transcriptomic data alone risks missing important miRNA-gene interactions. Furthermore, past attempts to identify binding sites of miRNAs to their target mRNAs based on miRNA-mediated effects on mRNA levels alone have likely failed to identify the characteristics of target sites that facilitate miRNA-dependent translation repression. These findings pave the way for future efforts to develop a more complete appreciation of the mechanisms by which miRNAs regulate gene expression.