PMCCPMCCPMCC

Search tips
Search criteria 

Advanced

 
Logo of rnabLink to Publisher's site
 
RNA Biol. 2012 October 1; 9(10): 1275–1287.
PMCID: PMC3583858

Dose-dependent differential mRNA target selection and regulation by let-7a-7f and miR-17-92 cluster microRNAs

Abstract

MicroRNAs (miRNAs) are important players of post-transcriptional gene regulation. Individual miRNAs can target multiple mRNAs and a single mRNA can be targeted by many miRNAs. We hypothesized that miRNAs select and regulate their targets based on their own expression levels, those of their target mRNAs and triggered feedback loops. We studied the effects of varying concentrations of let-7a-7f and the miR-17-92 cluster plasmids on the reporter genes carrying either DICER- or cMYC -3′UTR in Huh-7 cells. We showed that let-7 significantly downregulated expression of DICER 3′UTR reporter at lower concentrations, but selectively downregulated expression of a cMYC 3′UTR reporter at higher dose. This miRNA dose-dependent target selection was also confirmed in other target genes, including CCND1, CDKN1 and E2F1. After overexpressing let-7a-7f or the miR-17-92 clusters at wide-ranging doses, the target genes displayed a nonlinear correlation to the transfected miRNA. Further, by comparing the expression levels of let-7a and miR-17-5p, along with their selected target genes in 3 different cell lines, we found that the knockdown dose of each miRNA was directly related to their baseline expression level, that of the target gene and feedback loops. These findings were supported by gene modulation studies using endogenous levels of miR-29, -1 and -206 and a luciferase reporter system in multiple cell lines. Finally, we determined that the miR-17-92 cluster affected cell viability in a dose-dependent manner. In conclusion, we have shown that miRNAs potentially select their targets in a dose-dependent and nonlinear fashion that affects biological function; and this represents a novel mechanism by which miRNAs orchestrate the finely tuned balance of cell function.

Keywords: microRNAs, let-7a-7f, miR-17-92, miRNA target selection, dose-dependent, oncogene, tumor suppressor gene

Introduction

It is now well established that microRNAs (miRNAs) play a major role in post-transcriptional gene regulation. These small noncoding RNAs are initially transcribed by polymerase II1 and modified by DROSHA and DGCR8 complexes to form ~70 bp pre-miRNAs in the nucleus.2-4 The pre-miRs are exported into the cytoplasm by exportin 5 and then processed by Dicer and other factors5-7 to form duplex miRNAs, which are then incorporated into miRNA-induced silencing complex (miRISC). After additional processing, the mature ~18–22 nucleotide long miRNAs together with Argonaute 2 (Ago2) and other miRISC factors bind to complementary sites (preferably in the 3′UTRs) on mRNA transcripts and induce either translational pausing or degradation.6 miRNAs are involved in a myriad of normal cellular functions, including stem cell differentiation,8,9 organ development10 and tumorigenesis;11,12 and have become an invaluable tool for studying gene regulation.

A number of miRNA target prediction algorithms have been developed, such as Targetscan,13,14 miRanda,15 miRbase Targets16 and PicTar.17 These programs predict that a single miRNA can have several hundred potential targets, and a single mRNA 3′UTR can contain binding sites (miRNA response elements) for a number of miRNAs.18 For example, Targetscan predicted that let-7a can target 85 genes, while with miRanda and Microcosma, the numbers are 5436 and 1046, respectively. Many of these predicted targets, including DICER,19 CCND1,20 cMYC,21 and more than 30 other genes have been experimentally validated.22

We recently reported that the genes involved in miRNA biogenesis such as DICER, DROSHA, DGCR8, TRBP and PACT, are themselves target genes of certain miRNAs.19 A single miRNA can regulate these proteins simultaneously, e.g., miR-17-92 cluster can regulate DICER, DROSHA and PACT, while let-7a-7f can target DICER and TRBP. It has been suggested that the efficiency of miRNA suppression of mRNA function is, in part based on threshold limits for the target genes.23 However, it is still not clear as to how this is influenced by differential target selection of the miRNAs. Further, it has been reported that a single miRNA can exert seemingly opposite effects under different conditions; e.g., miR-17-92 cluster functions either as an oncogene,24 or a tumor suppressor depending on its target gene selection;25 and miR-29 can act as a tumor suppressor or oncogene in different subtypes of chronic lymphocytic leukemia (CLL).26 The involvement of competitive endogenous (ce) RNAs has recently added another layer of complexity to the role that miRNAs play, and the manner in which the rheostat can be adjusted.27

Most miRNAs are organized as clusters within the genome and regulated as a single unit; and this is no exception for either the let-7a-7f or miR-17-92 clusters. In this study, we demonstrated that a dual function of miRNAs can result from distinct target selection, which is modulated by expression levels of the miRNAs, their target mRNAs and feedback pathways. Further, we showed that the biological effects of overexpression of the let-7a-7f and miR-17-92 clusters were dose-dependent; and a narrow window of optimal dose for each miRNA cluster was required for maximal suppression of each target mRNA. In addition, the optimal dose was, in part dependent on the expression level(s) of the target mRNA(s), as well as a feedback mechanism between miRNAs and their own biogenesis genes. These mechanisms will allow miRNAs to fine-tune post-transcriptional gene regulation by allowing their expression levels to modulate target mRNA selection under a variety of cellular conditions.

Results

Dose-dependent target selection by miRNAs

To study the dose effect of miRNAs on suppression of target mRNAs, we used a DICER 3′UTR luciferase reporter system. Because many miRNAs are organized as clusters and the expression of the miRNAs in a cluster are regulated simultaneously, we studied let-7a-7f and miR-17-92 clusters to represent the miRNAs function in vivo. The constructs were co-transfected in Huh-7 cells with different amounts of plasmid for miRNA let-7a-7f ranging from 0.00003 µg to 0.3 µg/well. We found that let-7a-7f downregulated luciferase activity of DICER 3′UTR luciferase reporter at low dose (0.0003 µg/well); and no further change was observed with higher doses. Interestingly, when we increased the plasmid to 0.3 µg/well of let-7a-7f, there was an unexpected increase in reporter activity (Fig. 1A).23 Similar experiments were also performed using a cMYC-3′UTR reporter construct. The effective dose range to target cMYC by let-7a-7f was between 0.03 and 0.3 µg/well (Fig. 1B), which is 100 ~1000 fold higher compared with the effective dose for DICER. Similar target selection was also seen when the miRNA cluster, miR-17-92, was studied for its effects on DICER 3′UTR, in which downregulation required transfection doses as high as 0.3 µg/well (Fig. 1C).

figure rna-9-1275-g1
Figure 1. miRNAs select their targets at different overexpression levels. (A) Responses of DICER 3′UTR luciferase reporter activity to different doses of let-7a-7f. Huh-7 cells were co-transfected with 0.1 µg DICER-3′UTR ...

Confirmation of the dose dependent miRNA target selection at mRNA and protein levels

To confirm our findings with the 3′UTR reporter assays, we also tested mRNA and protein levels of the different target genes after overexpressing let-7a-7f and miR-17-92 at the full range of doses. mRNA levels of DICER as measured by qRT-PCR were decreased with let-7a-7f at low dose (0.0003 µg/well) (Fig. 2A), while higher doses (0.03 µg/well) were required for downregulation of cMYC (Fig. 2B). At the protein level, we observed a similar dose-dependent target selection. DICER expression was significantly reduced at a low dose of let-7a-7f plasmid (0.0003 µg/well), which had no effect on cMYC expression. At higher concentration (0.03 µg/well), let-7a-7f had no effect on DICER but significantly downregulated cMYC (Fig. 2D and E). CCND1 is also a let-7a-7f target gene, and the optimal dose of let-7a-7f to repress CCND1 expression was 0.003 µg/well. Interestingly, we found that CCND1 expression could be suppressed at two concentrations with a 100-fold difference (0.00003 µg/well and 0.003 µg/well); and yet, no effect was observed with the intermediate dose (0.0003 µg/well) (Fig. 2C). It is possible that the downregulation of DICER, a miRNA biogenesis gene, at the intermediate concentration decreased the efficacy of miRNA synthesis. Expression levels of miR-17-92 target genes, such as DICER, CCND1, E2F1 and CDKN1 were also studied after overexpression of miR-17-92 cluster (Fig. 3). Again, we observed that the different target genes required varying doses of miR-17-92 cluster. E2F1 and CDKN1 seemed to share similar dose responses. The results suggested that the individual miRNA cluster had target preferences based on a narrow range of effective dose.

figure rna-9-1275-g2
Figure 2. The endogenous target genes were downregulated after overexpression of let-7a-7f with 5 different doses in Huh-7 cells. The suppression of mRNA levels for (A) DICER, (B) cMYC and (C) CCND1 in response to overexpression of different doses ...
figure rna-9-1275-g3
Figure 3. Downregulation of target genes after overexpression of miR-17-92 cluster in Huh-7 cells. The mRNA levels of (A) DICER, (B) CCND1, (C) E2F1 and (D) CDKN1 were examined after transfection with the 5 different doses of miR-17-92 cluster ...

Nonlinear changes in miRNA caused by negative feedback between miRNA and its biogenesis genes

To correlate the effects of miRNA overexpression to the actual levels of miRNA after transfection, we measured let-7a expression in Huh-7 cells transfected with different doses of the plasmid, and observed a nonlinear change in expression. At 0.00003 µg/well, let-7a expression increased by ~20%, but with increasing doses of let-7a from 0.0003 to 0.003 µg/well, the let-7a levels actually decreased to baseline. The expression levels began to increase again when the dose reached > 0.03 µg/well (Fig. 4A). The dose range of let-7a where the expression level actually decreased coincided with the effective dose range for downregulating DICER. This nonlinear change in let-7a was possibly due to the negative feedback loop between let-7a-7f and miRNA processing genes such as DICER.

figure rna-9-1275-g4
Figure 4. let-7a-7f and miR-17-92 expression levels and cell viability after overexpression of these two miRNA clusters in Huh-7 cells. The expression levels of (A) let-7a-7f and (B) miR-17-92 in Huh-7 cells were determined by qRT-PCR 24 h after ...

The expression levels of miR-17 were also studied in Huh-7 cells transfected with miR-17-92. Again, we observed a similar nonlinear change in expression with increasing doses of miR-17-92. The miR-17 expression levels did not increase when plasmid doses were increased from 0.0003 to 0.003 µg/well (Fig. 4B), which may be related to its regulation of E2F1 and DICER. Both E2F1 and DICER can form negative feedback loops with the miR-17-92 cluster. E2F1 can regulate miR-17-92 expression by activating cMYC, a known transcriptional activator of miR-17-92. Overexpression of miR-17-92 cluster downregulates E2F1 expression which, in turn, can suppress the expression of endogenous miR-17-92. Meanwhile, inhibition of DICER can decrease the generation of both endogenous and exogenous miR-17-92 expression. In this study, we also noticed that the increase in miR-17-92 expression was less than that of let-7a-7f and this might explain why the effective doses of miR-17-92 were generally higher than that of let-7a-7f.

The bivalent/dual biological function of miRNAs is due to the dose-dependent selection of targets

It has been previously reported that miR-17-92 cluster can function as either an oncogene or tumor suppressor gene under varying conditions. We hypothesized that differential target selection was responsible for this dual biological effect. To confirm the bivalent function of miRNAs, we studied Huh-7 cell viability after transfecting different amounts of miRNA overexpression plasmids. For let-7a-7f, cell viability was decreased at high doses (Fig. 4C), and the observed changes were nonlinear with the expression of let-7a-7f (Fig. 4A). This might have been due to a difference in target genes as a function of the different let-7a-7f levels. miR-17-92 cluster has been reported to be an oncogene in colon cancer but a tumor suppressor gene in some subtypes of breast cancer.25 It has been suggested that the dual function of miR-17-92 is tissue-specific. However, we found that the bivalent/dual function of miR-17-92 could be reproduced in the Huh-7 cell line. Overexpression of miR-17-92 cluster at low dose (0.00003 µg/well) increased cell viability, but higher doses of miR-17-92 resulted in decreased cell viability (Fig. 4B and D). It is likely that the seemingly opposite effects of miR-17-92 were caused by selective downregulation of target genes at the different doses.

Dose dependent target selection and bivalent/dual function of miRNAs are common in other cell types

To verify that the target selection and bivalent functions of miRNA were not unique to Huh-7 cells, we performed similar miRNA overexpression studies in 2 other cancer cell lines, HCT116 (colon cancer) and MCF-7 (breast cancer). Again, we found let-7a-7f regulates DICER with a unique dose response in both cell lines (Fig. 5A and B). let-7a-7f and miR-17-92 gave different dose responses on DICER in HCT116 cells (Fig. 5A and C), but similar linear dose responses in MCF-7 cells (Fig. 5B and D). The optimal dose and the dose response curve of let-7a-7f for DICER knock down were dramatically different in each of the three cell lines (Figs. 3 and and55).

figure rna-9-1275-g5
Figure 5. Selection of targets by let-7a-7f and miR-17-92 in HCT116 and MCF-7 cells. Dose response curve of let-7a-7f to DICER in (A) HCT116 and (B) MCF-7 were determined by a luciferase reporter system. Different doses of let-7a-7f were co-transfected ...

We also confirmed that the bivalent/dual functions of miR-17-92 on cell viability were present in HCT116 cells. At lower dose, miR-17-92 decreased cell viability, but this was increased with higher dose (Fig. 6C). In MCF-7 cells, however, miR-17-92 inhibited cell proliferation at all concentrations (Fig. 6D). let-7a-7f, on the other hand, continued to act as a tumor suppressor in both cancer cell lines (Fig. 6A and B).

figure rna-9-1275-g6
Figure 6. Cell viability changes in HCT116 and MCF-7 cells transfected with different doses of let-7a-7f and miR-17-92. Cell viability was determined by Cell Titer-Blue assay in HCT116 and MCF-7 cells transfected with varying amounts of let-7a-7f ...

miRNAs target selection is related to the expression of both miRNAs and their target genes

From earlier reports, it is clear that target selection of miRNAs is not solely dependent on the seed sequences.18 It remains to be clarified what additional factors are required for target selection and also the mechanism(s) by which selection occurs. We speculated that target selection was also affected by the endogenous expression levels of both miRNAs and their target genes, in addition to the feedback regulations between these two RNA species. Thus, we examined the endogenous expression levels of let-7a and miR-17-5p and their experimentally validated targets such as DICER and cMYC in Huh-7, HCT116 and MCF-7 cells. In Huh-7 cells, cMYC expression levels were about 2.5 fold higher than CCND1, and CCND1 about 3.8 fold higher than DICER (Fig. 7A). This is in agreement with our data (Fig. 2A-C) showing that the effective levels of let-7a-7f plasmid to knockdown cMYC were roughly 10-fold higher than that of CCND1, and 100-fold higher than that of DICER. Similarly, the levels of CCND1, E2F1 and CDKN1 were also correlated with effective doses of miR-17-92 plasmid (Figs. 7A and and3B–D).3B–D). These results suggested that the inhibiting dose required for an individual miRNA may be determined by the expression level of its corresponding target gene.

figure rna-9-1275-g7
Figure 7. Endogenous let-7a, miR-17-5p and their target expression levels in Huh-7, HCT116 and MCF-7 cells. (A) The expression levels of several targets of let-7a-7f and miR-17-92 in Huh-7 cells were were studied by qRT-PCR. The endogenous (B ...

Using luciferase reporter assays, we also observed similar expression correlation between let-7a-7f and miR-17-92 and their corresponding target genes in HCT116 and MCF-7 cells (Fig. 5). This observation was further validated by the determination of endogenous expression levels of cMYC, which was 16- and 6-fold higher than that of DICER in these two cell lines (Fig. 7B and C), respectively. Thus, DICER 3′UTR could be inhibited with maximal levels of the miRNA while cMYC was unaffected by any of the 5 doses tested (Fig. 5).

The expression levels of DICER in HCT116 and Huh-7 cells were comparable (Fig. 7C). However, the effective DICER inhibiting dose of let-7a was lower in Huh-7 cells, where the endogenous let-7a expression was 17-fold less than in HCT116 cells (Fig. 7E). This was also supported by the observations that the doses of let-7a-7f plasmid to knockdown DICER was ~100-fold higher in HCT116 than in Huh7 cells (Figs. 1A and and5A).5A). However, expression levels of target mRNAs may be more relevant than the endogenous miRNA levels, as miR-17-5p did not seem to affect DICER levels (Fig. 7F). These observations suggested that endogenous mRNA expression is key to regulation by miRNAs.

To further confirm that target selection of miRNAs is determined by both miRNA and mRNA levels, we examined a set of endogenous miRNAs, which have variable expression levels in different cell lines. We devised a reporter assay to measure the efficacy of endogenous miRNAs in cell lines DLD1, HCT116, Huh-7, JR1, MCF-7 and SW480. Two luciferase reporter constructs were used, which contained either a wild type or mutated miRNA response element region, or seeding site. The difference in reporter activity was designed to detect endogenous miRNA activity (Fig. S1A and B). We initially tested miR-29a, b and c according to the above transfection system. All three miRNAs shared the same seeding sequence on 3′UTRs of DICER and E2F7. With transfection of the luciferase reporter constructs, we observed miR-29 knockdown of both DICER and E2F7 expression in HCT116 and SW480 cells; DICER knockdown in DLD1 cells; and reduced E2F7 in the JR1 cell line. No changes in expression were observed in either the Huh-7 or MCF-7 cell lines (Fig. 8A). Similar results were observed with miR-1 and miR-206 using the psiCHECK-2 reporter system, in which the seeding site of miR-1/206 was inserted into the 3′UTR of Renilla luciferase gene (Fig. S1C). Our results indicated that miR-1/206 can regulate Renilla luciferase activity in HCT116, JR1, MCF-7 and SW480 cells, but not DLD1 and Huh-7 (Fig. 8B). Based on these results, we concluded that target selection is variable in different cell lines, and this may be a result of varying endogenous levels of the miRNAs.

figure rna-9-1275-g8
Figure 8. Association between miRNA target selection and endogenous expression levels of miRNAs and mRNAs. (A) Repression of DICER and/or E2F7 by miR-29 in different cell lines. The 3′UTR reporter constructs with a mutated miR-29 seeding ...

To confirm our observation, we correlated the expression levels of miRNAs to their knockdown activity. We thus measured the expression levels of endogenous miR-29 and DICER/E2F7 (Figs. 8C, 7C and D). The miR-29 family includes miR-29a, b and c, which share the identical seeding sequence on the 3′UTR of DICER and E2F7. Interestingly, the knockdown efficacy did not correlate with either levels of miRNA or mRNA target genes alone (Fig. S2). Instead, we found an excellent correlation (R2 = 0.77) between downregulation of DICER 3′UTR reporter activity and the ratio of miR-29 (total levels of miR-29a, b and c) to DICER expression levels (Fig. 8E). When the efficacy of the individual miR-29s was analyzed, it appeared that the effect of downregulation was more significant with greater levels of endogenous miRNA (Fig. S2B). A similar pattern was observed between the knockdown efficiency of miR-29 and the miR-29 to E2F7 ratio (Fig. 8F). From above results, it is clear that the target selection of individual miRNAs depends on its expression level relative to that of the target genes.

A mathematical model of differential target regulation by miRNAs

Our experimental results have demonstrated that increasing the concentrations of the miRNA let-7a-7f and the miR-17-92 clusters can have differential effects on the levels of their targets. For example, we showed that as the overexpression plasmid concentration of let-7a-7f was increased, the normalized activity of the DICER 3′UTR reporter diverged significantly from that of the cMYC reporter (Fig. 1A and B). Using concepts of the recently described competitive endogeneous RNAs (ceRNAs),27 we have constructed a mathematical model to demonstrate a possible mechanism for differential target regulation by miRNAs.

We considered the binding of miRNAs to target response elements on mRNAs, and probabilistically model the effects of competition between mRNAs with common response elements. In this model, the probability of miRNA binding to a response element on a particular mRNA type depends on abundance of each mRNA type, the number of response elements on each mRNA, and the miRNA abundance. A single miRNA copy can bind, as a simple example to any of three types of target mRNAs (Fig. 9). In such a system, we describe the dependence of the probabilities of each of these binding events on the number of response elements and the abundance of target mRNA transcripts. In the full model, a large network of miRNAs and target genes may be considered, with the possibility of differential effects of miRNA-binding to distinct response elements on each gene. This model allowed us to quantitatively study how competition for miRNA-binding between mRNAs may contribute to (1) differential effects on target regulation by a common miRNA; and (2) regulation of the activity of mRNA transcripts by other mRNAs. Note that since our modeling framework identifies miRNAs by their target element sets, it can also be utilized to model miRNA clusters that share common targets.

figure rna-9-1275-g9
Figure 9. A proposed mathematical model illustrating a simple example with a single copy of one microRNA and three types of target mRNAs. The parameter ri represents the number of mRNA transcripts of type i (where i = 1, 2 or 3), and the parameter ...

We first demonstrated how increasing the concentration of a single miRNA may differentially affect the levels of target mRNA activity. Differential target regulation is exhibited in two target mRNAs as the miRNA level is increased through four orders of magnitude (Fig. 10A). We considered a situation in which target mRNA 1 harbors three response elements matching the miRNA, while target mRNA 2 harbors only one response element. Transcripts of mRNA 1 are more abundant; thus the protein expression level for target mRNA 1 is higher than that of target 2 when miRNAs are present only at low levels. However, as the miRNA level is increased the repressive effect is significantly stronger on target 1. This is due to the presence of more response elements on this gene, which allows for both greater repression and limits the number of miRNAs available to bind to target 2. Thus, as the miRNA level increases beyond a threshold, target 1 is eventually fully repressed while expression levels of target 2 are still nonzero, exhibiting a differential regulation by the miRNA change. For example, let-7 has two seeding sites on DICER but only one on cMYC, which may explain the observation that let-7 can regulate DICER at a much lower dose compared with cMYC.

figure rna-9-1275-g10
Figure 10. (A) Differential target regulation via competition for miRNA binding. The red and blue lines represent the protein expression levels of targets 1 and 2 as the miRNA concentration is increased. Parameters of the model (as described in ...

We then investigated how overexpression of one mRNA (e.g., by transcriptional activation) can result in the de-repression of other transcripts that contain the same response elements. We again considered the same situation in which target mRNA 1 harbors three response elements matching the miRNA, while target mRNA 2 harbors only one response element (Fig. 10B). However, here the miRNA level is held constant at a high value while the transcript abundance of target mRNA 1 is increased. Initially, at low transcript levels of both mRNAs, both proteins are fully repressed; then, as mRNA target 1 is increased it acts as a “sponge” for the miRNAs, thus enabling some expression of target 2. Eventually, beyond a threshold that is dependent on the miRNA level, we observe expression of target 1 increasing with linearly mRNA 1 abundance. Interestingly, at some mRNA 1 levels we may observe the effect of de-repression in target 2 even while no effect on target 1 is detectable.

This mathematical model was constructed to theoretically demonstrate a mechanism for the observed differential target regulation and also the regulation of mRNA activity by other mRNA transcripts. We have made a number of simplifying assumptions in the model to more easily demonstrate these mechanisms. For example, we considered a single miRNA with several targets in the system to demonstrate ceRNA activity. In addition, we assumed a uniform effect of miRNA binding to response elements on each mRNA transcript’s activity. However in order to accurately reproduce experimental findings, the model would need to incorporate the entire network of inter-related miRNAs, target genes and response elements. In addition, to producing experimentally meaningful results the model would also require data quantifying the effects of each type of miRNA binding to response elements on various mRNA transcripts involved in the network. These effects are not necessarily uniform between different mRNAs and even between different response elements on the same gene. Therefore, significant additional steps must be taken to fully correlate this model with experimental results.

Discussion

It has been reported that at least 60% of gene pool in humans may be regulated by miRNAs.14 Over the last decade, it has been shown that miRNAs can serve as potential biomarkers28-30 in diagnosing diseases and establishing the prognosis for a wide variety of clinical conditions.31 In addition, the recent discovery that miRNAs are present in body fluids has generated enormous interest in their use as potential circulating biomarkers.32 Balancing and maintaining the threshold levels of these miRNAs is a highly complex system uniquely designed for the delicate control of gene expression. It is therefore of great importance to understand the relevance of cellular miRNA levels mechanisms whereby miRNAs selectively bind to and regulate their target genes.

In our studies, we found that the overexpression of individual miRNAs regulates their putative targets in a nonlinear fashion. In fact, our results indicate a narrow range of functional dose for each target gene, beyond which the particular mRNA is no longer a viable target. Selection by miRNAs is not only dependent on the recognition of complementary target sequences but also on the conditions of cellular environments. For example, we found that overexpression of let-7a-7f in Huh-7 cells could also effectively target DICER at certain concentrations. Higher or lower doses of this miRNA, however, were no longer effective for DICER but were effective against cMYC and CCND1. miR-17-92 also showed a similar type of dose-dependent target selection, suggesting that this may be a common characteristic of miRNAs.

Not surprisingly, we found that the shift of the targets significantly changed the biological function of the miRNA. In our studies, miR-17-92 decreased cell viability at low dose expression, but was increased with high dose transfection in Huh-7 cells. Interestingly, the opposite results were observed in HCT116 cells. These results shed light on the previous reports that miR-17-92 can act as either a tumor suppressor gene or oncogene under varying conditions.

Our work has also further characterized the conditions involved in target selection. Based on our findings, it seems that miRNAs select their targets not only by sequence homology, but also by their respective endogenous levels. It is possible that, genes with lower levels of expression are more easily targeted by miRNAs, requiring lower concentrations of miRNA. We also demonstrated that the dynamics of miRNA targeting are also affected by feedback mechanisms between the miRNAs and their target genes. As examples, the target genes of miR-17-92 and let-7a-7f, DICER and cMYC, respectively, directly regulate the biogenesis of these two miRNAs, thus providing a mechanism for fine-tuning the expression of target genes. Further, we confirmed that target selection of miRNAs was determined by both miRNA and mRNA levels. Reporter assays designed to detect endogenous miRNA activity showed that target gene selection was variable in different cell lines, suggesting that target selection of individual miRNAs depends on endogenous miRNA levels and those of the target genes. These observations are in agreement with the recent findings that miRNAs can act both as a switch and as a fine-tuner of gene expression depending on the threshold levels of target.23

We have constructed a mathematical model to investigate competition for miRNA binding as a possible mechanism for differential target regulation by miRNAs. Using this model, we demonstrated how the ceRNA hypothesis27 could explain differential repression between targets as the miRNA level is increased, similar to the phenomena we have observed experimentally. The model also investigates how mRNAs interact with each other through competition for miRNA binding sites. In particular, we studied how overexpression of one mRNA can result in the de-repression of other transcripts that contain the same response elements. Interestingly, we observed that in some ranges of overexpression, de-repression effects of other transcripts might be noticeable while protein levels of the overexpressed mRNA are not yet apparent.

This modeling strategy could potentially be used to quantitatively decipher the complex network of miRNA-mRNA interactions and better understand how disease states differ from normal; and to design therapeutic strategies to manipulate those networks. For example, by knowing the abundance of a network of gene response elements, we can quantitatively study the effects of transcriptional activation of one or a group of mRNAs on the functional activity of other mRNAs in network, via competition for various types of miRNAs. Thus, it may have potential clinical uses in identifying secondary effects of specific targets or identifying alternative drug targets. Additional studies will confirm whether the observed nonlinear dose dependent phenotypes are a universal phenomenon in different tumor and primary cell lines.

Here we have assumed a static mathematical model of miRNA binding; however, this static model can be hybridized with a set of differential equations to incorporate feedback mechanisms in the temporal dynamics of the system. For example, after let-7 miRNAs enter a system, a dynamic negative feedback system may commence wherein let-7 downregulates some targets such as DICER, which in turn reduces the biogenesis of let-7. In addition, basic model assumptions can be relaxed to incorporate differential binding affinity of miRNAs to response elements on different genes, or distinct response elements on the same genes. Given the wide range of target sites for each miRNA, and the multiplicity of miRNAs, it requires further research especially in systems biology to fully understand the role of miRNAs in the diverse biological processes. The mathematical framework we have constructed can be expanded to incorporate the complex network structure and specific relationships between miRNA binding and target gene activity, as these parameters are experimentally explored. Herein, we have utilized this simplified mathematical model to illustrate a theoretical, quantitative framework for this proposed mechanism of differential regulation.

Materials and Methods

Cell culture, transfection and 3′UTR luciferase reporter assay

Human DLD1, Huh-7, HCT116, JR1, MCF-7 and SW480 cells were cultured in either high-glucose DMEM or RPMI 1640 media as described.33,34 DICER-, cMYC- and E2F7–3′UTR constructs were purchased from SwitchGear Genomics. let-7a-7f cluster was cloned into pcDNA3.1 (Invitrogen) as reported.19 The miR-17-92 expression construct was kindly provided by Dr. He Lin (University of California, Berkeley). The E2F7 reporter plasmid with mutated miR-29 seeding sequence, and miR-1/206 reporter constructs have been previously validated.35 DICER 3′UTR-sGG vectors with mutations in the miR-29 binding region were constructed using QuickChange site-directed mutagenesis kit (Stratagene) following the manufacturer’s instructions. The cells were plated at 70% ~90% density 24 h prior to transfection in 48 well plates. Different 3′UTR reporter constructs were co-transfected with miRNA constructs and the SV40-RL internal control plasmid (Promega) using Lipofectamine 2000TM. The miRNAs constructs were transfected with 5 different dosages ranging from 0.00003 µg to 0.3 µg per well. The cells were harvested 24 h after transfection, and the luciferase activity determined by the Dual-Glo Luciferase Assay System (Promega) using a Synergy 2 microplate reader (BioTek).

Quantitative reverse transcriptase-polymerase chain reaction (qRT-PCR)

RNA was extracted using the TRIZOL Reagent (15596–018, Invitrogen) following the manufacturer’s instructions; and processed with TURBO DNA-free kit (AM1907, Ambion). For mRNA studies, RT was done with 1 µg total RNA using High Capacity RNA-to-cDNA kit (4368814, Applied Biosystems); and PCR amplification was performed with Power SYBR Green PCR Master Mix (4367659, Applied Biosystem) in ABI 7300 real-time PCR System. GAPDH was used as an internal control. For miRNA studies, 50 ng total RNA was treated with miScript reverse transcription kit (218161, Qiagen); and PCR amplifications were carried out with miScript SYBR Green PCR kit (218073, Qiagen) using U6 as an internal control. Primer sequences are listed in Table S1. At least three independent qRT-PCR and/or transfections were performed, and the average of replicates was used in the data analysis.

Western blot analysis

miRNA expression constructs and pcDNA3.1 empty vector ranging from 0.00003 µg to 0.3 µg per well were transfected into Huh-7 cells in 48 well plates and the cells harvested after 24 h. For DICER studies, total protein was extracted and western blots were performed as previously described19 with anti-Dicer antibody (sc 136981, Santa Cruz); and GAPDH (G9545, Sigma) as an internal control. For cMYC studies, the nuclear fraction was extracted as previously described19 with anti-cMYC antibody (sc-40, Santa Cruz); and Histone 3 (H3, ab 1791) was used as an internal control. A minimum of three independent transfections was performed.

Cell viability analyses

miRNA expression constructs over the 105 dose range were transfected into Huh-7, HCT116 and MCF-7 cells in 48 well plates and the cells harvested after 24 h. Cell proliferation was determined using Cell Titer 96 Aqueous One Solution Cell Proliferation Assay (G3582, Promega) in Synergy 2 microplate reader.

Statistical analysis

Significance was determined with a two-tailed Student’s t test.

Mathematical modeling

We considered a simple framework to model interactions between the abundance of a single type of miRNA and the protein expression levels in a system of G genes, where each gene may harbor (possibly multiple) response elements of the miRNA. The parameter ri represents the number of mRNA transcripts of type i, and the parameter xi represents the number of response elements on mRNA i, for i between 1 and G. Let pj then represent the protein expression level of gene j in the absence of any miRNA repression; we assumed for simplicity that pj [proportional, variant] rj, i.e., that the native protein expression level was proportional to the number of mRNA transcripts in the system. If a miRNA binds to a response element on an mRNA of type j, it reduces the protein expression level by dj units. We also assumed for simplicity that each miRNA copy could bind to any of the available response elements in the system with equal probability, regardless of their location on a particular mRNA. In considering m miRNAs in the system, we were interested in the expected number of miRNAs that bind to each type of mRNA. Under these basic assumptions, this can be recognized as a classic problem in probability called “sampling balls from urns.” In this situation we can think of each response element in the system as one ball, which is colored according to its mRNA type. There are a total of

An external file that holds a picture, illustration, etc.
Object name is rna-9-1275-e1.jpg

balls in the urn. Then if we sample, without replacement after each draw, m balls from the urn, the distribution of ball colors in this sample will correspond to the distribution of the number of miRNAs that bind to each type of mRNA. We recognized this distribution as the multivariate hypergeometric distribution, which can be written the terms of our problem as follows. Let mi, i = 1…G be a sequence of numbers such that m1 + m2 + … + mG = m. Then, the probability that m1 miRNAs bind to mRNAs of gene 1, m2 miRNAs bind to mRNAs of gene 2, and so on, is given by:

An external file that holds a picture, illustration, etc.
Object name is rna-9-1275-e2.jpg

Where

An external file that holds a picture, illustration, etc.
Object name is rna-9-1275-e3.jpg

represents the binomial coefficient. In terms of protein expression, when An external file that holds a picture, illustration, etc.
Object name is rna-9-1275-e4.jpg miRNAs are introduced, the protein expression levels in the system change from (p1, p2,…,pG) to (p1d1m1, p2d2m2,…,pGdGmG) with probability

An external file that holds a picture, illustration, etc.
Object name is rna-9-1275-e5.jpg.

Using the form of the distribution, we can calculate the expected protein expression change from the event:

An external file that holds a picture, illustration, etc.
Object name is rna-9-1275-e6.jpg

This expression also allows us to calculate the expected effect of changing the level of one mRNA on the activity of the other mRNA types in the system.

Supplementary Material

Additional material

Acknowledgments

We thank members of both the C.J.S. and S.S. laboratories for their technical assistance and comments on the manuscript. This work was supported by National Institutes of Health ARRA Grant R01 DK081865–01 (to C.J.S.); Department of Defense Grant W81XWH-10–1-0556; and University of Minnesota Academic Health Center and Minnesota Medical Foundation (to S.S.).

Glossary

Abbreviations:

miRNA
microRNA
miRISC
miRNA-induced silencing complex
Ago2
Argonaute 2
DGCR8
DiGeorge syndrome critical region gene 8
TRBP
TAR RNA binding protein 2
PACT
interferon-inducible double stranded RNA-dependent protein kinase activator A
ceRNA
competing endogenous RNA
CLL
chronic lymphocytic leukemia

Disclosure of Potential Conflicts of Interest

No potential conflicts of interest were disclosed.

Supplemental Material

Supplemental material may be found here:
www.landesbioscience.com/journals/rnabiology/article/21998

References

1. Cai X, Hagedorn CH, Cullen BR. Human microRNAs are processed from capped, polyadenylated transcripts that can also function as mRNAs. RNA. 2004;10:1957–66. doi: 10.1261/rna.7135204. [PubMed] [Cross Ref]
2. Denli AM, Tops BB, Plasterk RH, Ketting RF, Hannon GJ. Processing of primary microRNAs by the Microprocessor complex. Nature. 2004;432:231–5. doi: 10.1038/nature03049. [PubMed] [Cross Ref]
3. Landthaler M, Yalcin A, Tuschl T. The human DiGeorge syndrome critical region gene 8 and Its D. melanogaster homolog are required for miRNA biogenesis. Curr Biol. 2004;14:2162–7. doi: 10.1016/j.cub.2004.11.001. [PubMed] [Cross Ref]
4. Gregory RI, Yan KP, Amuthan G, Chendrimada T, Doratotaj B, Cooch N, et al. The Microprocessor complex mediates the genesis of microRNAs. Nature. 2004;432:235–40. doi: 10.1038/nature03120. [PubMed] [Cross Ref]
5. Saito K, Ishizuka A, Siomi H, Siomi MC. Processing of pre-microRNAs by the Dicer-1-Loquacious complex in Drosophila cells. PLoS Biol. 2005;3:e235. doi: 10.1371/journal.pbio.0030235. [PMC free article] [PubMed] [Cross Ref]
6. Chendrimada TP, Gregory RI, Kumaraswamy E, Norman J, Cooch N, Nishikura K, et al. TRBP recruits the Dicer complex to Ago2 for microRNA processing and gene silencing. Nature. 2005;436:740–4. doi: 10.1038/nature03868. [PMC free article] [PubMed] [Cross Ref]
7. Yeom KH, Lee Y, Han J, Suh MR, Kim VN. Characterization of DGCR8/Pasha, the essential cofactor for Drosha in primary miRNA processing. Nucleic Acids Res. 2006;34:4622–9. doi: 10.1093/nar/gkl458. [PMC free article] [PubMed] [Cross Ref]
8. Kanellopoulou C, Muljo SA, Kung AL, Ganesan S, Drapkin R, Jenuwein T, et al. Dicer-deficient mouse embryonic stem cells are defective in differentiation and centromeric silencing. Genes Dev. 2005;19:489–501. doi: 10.1101/gad.1248505. [PubMed] [Cross Ref]
9. Murchison EP, Partridge JF, Tam OH, Cheloufi S, Hannon GJ. Characterization of Dicer-deficient murine embryonic stem cells. Proc Natl Acad Sci USA. 2005;102:12135–40. doi: 10.1073/pnas.0505479102. [PubMed] [Cross Ref]
10. Sood P, Krek A, Zavolan M, Macino G, Rajewsky N. Cell-type-specific signatures of microRNAs on target mRNA expression. Proc Natl Acad Sci USA. 2006;103:2746–51. doi: 10.1073/pnas.0511045103. [PubMed] [Cross Ref]
11. Lu J, Getz G, Miska EA, Alvarez-Saavedra E, Lamb J, Peck D, et al. MicroRNA expression profiles classify human cancers. Nature. 2005;435:834–8. doi: 10.1038/nature03702. [PubMed] [Cross Ref]
12. Volinia S, Calin GA, Liu CG, Ambs S, Cimmino A, Petrocca F, et al. A microRNA expression signature of human solid tumors defines cancer gene targets. Proc Natl Acad Sci USA. 2006;103:2257–61. doi: 10.1073/pnas.0510565103. [PubMed] [Cross Ref]
13. Lewis BP, Burge CB, Bartel DP. Conserved seed pairing, often flanked by adenosines, indicates that thousands of human genes are microRNA targets. Cell. 2005;120:15–20. doi: 10.1016/j.cell.2004.12.035. [PubMed] [Cross Ref]
14. Friedman RC, Farh KK, Burge CB, Bartel DP. Most mammalian mRNAs are conserved targets of microRNAs. Genome Res. 2009;19:92–105. doi: 10.1101/gr.082701.108. [PubMed] [Cross Ref]
15. Betel D, Wilson M, Gabow A, Marks DS, Sander C. The microRNA.org resource: targets and expression. Nucleic Acids Res. 2008;36(Database issue):D149–53. doi: 10.1093/nar/gkm995. [PMC free article] [PubMed] [Cross Ref]
16. Griffiths-Jones S, Saini HK, van Dongen S, Enright AJ. miRBase: tools for microRNA genomics. Nucleic Acids Res. 2008;36(Database issue):D154–8. doi: 10.1093/nar/gkm952. [PMC free article] [PubMed] [Cross Ref]
17. Lall S, Grün D, Krek A, Chen K, Wang YL, Dewey CN, et al. A genome-wide map of conserved microRNA targets in C. elegans. Curr Biol. 2006;16:460–71. doi: 10.1016/j.cub.2006.01.050. [PubMed] [Cross Ref]
18. Bartel DP. MicroRNAs: target recognition and regulatory functions. Cell. 2009;136:215–33. doi: 10.1016/j.cell.2009.01.002. [PubMed] [Cross Ref]
19. Shu J, Kren BT, Xia Z, Wong PY, Li L, Hanse EA, et al. Genomewide microRNA down-regulation as a negative feedback mechanism in the early phases of liver regeneration. Hepatology. 2011;54:609–19. doi: 10.1002/hep.24421. [PMC free article] [PubMed] [Cross Ref]
20. Zhao Y, Deng C, Wang J, Xiao J, Gatalica Z, Recker RR, et al. Let-7 family miRNAs regulate estrogen receptor alpha signaling in estrogen receptor positive breast cancer. Breast Cancer Res Treat. 2011;127:69–80. doi: 10.1007/s10549-010-0972-2. [PubMed] [Cross Ref]
21. Sampson VB, Rong NH, Han J, Yang Q, Aris V, Soteropoulos P, et al. MicroRNA let-7a down-regulates MYC and reverts MYC-induced growth in Burkitt lymphoma cells. Cancer Res. 2007;67:9762–70. doi: 10.1158/0008-5472.CAN-07-2462. [PubMed] [Cross Ref]
22. Thomson DW, Bracken CP, Goodall GJ. Experimental strategies for microRNA target identification. Nucleic Acids Res. 2011;39:6845–53. doi: 10.1093/nar/gkr330. [PMC free article] [PubMed] [Cross Ref]
23. Mukherji S, Ebert MS, Zheng GX, Tsang JS, Sharp PA, van Oudenaarden A. MicroRNAs can generate thresholds in target gene expression. Nat Genet. 2011;43:854–9. doi: 10.1038/ng.905. [PMC free article] [PubMed] [Cross Ref]
24. He L, Thomson JM, Hemann MT, Hernando-Monge E, Mu D, Goodson S, et al. A microRNA polycistron as a potential human oncogene. Nature. 2005;435:828–33. doi: 10.1038/nature03552. [PubMed] [Cross Ref]
25. Hossain A, Kuo MT, Saunders GF. Mir-17-5p regulates breast cancer cell proliferation by inhibiting translation of AIB1 mRNA. Mol Cell Biol. 2006;26:8191–201. doi: 10.1128/MCB.00242-06. [PMC free article] [PubMed] [Cross Ref]
26. Pekarsky Y, Croce CM. Is miR-29 an oncogene or tumor suppressor in CLL? Oncotarget. 2010;1:224–7. [PMC free article] [PubMed]
27. Salmena L, Poliseno L, Tay Y, Kats L, Pandolfi PP. A ceRNA hypothesis: the Rosetta Stone of a hidden RNA language? Cell. 2011;146:353–8. doi: 10.1016/j.cell.2011.07.014. [PMC free article] [PubMed] [Cross Ref]
28. Mitchell PS, Parkin RK, Kroh EM, Fritz BR, Wyman SK, Pogosova-Agadjanyan EL, et al. Circulating microRNAs as stable blood-based markers for cancer detection. Proc Natl Acad Sci USA. 2008;105:10513–8. doi: 10.1073/pnas.0804549105. [PubMed] [Cross Ref]
29. Bohanes P, Courvoisier DS, Perneger TV, Morel P, Huber O, Roth AD. Survival predictors for second-line chemotherapy in Caucasian patients with metastatic gastric cancer. Swiss Med Wkly. 2011;141:w13249. [PubMed]
30. Hu Z, Chen X, Zhao Y, Tian T, Jin G, Shu Y, et al. Serum microRNA signatures identified in a genome-wide serum microRNA expression profiling predict survival of non-small-cell lung cancer. J Clin Oncol. 2010;28:1721–6. doi: 10.1200/JCO.2009.24.9342. [PubMed] [Cross Ref]
31. Liu C, Tang DG. MicroRNA regulation of cancer stem cells. Cancer Res. 2011;71:5950–4. doi: 10.1158/0008-5472.CAN-11-1035. [PMC free article] [PubMed] [Cross Ref]
32. Cortez MA, Bueso-Ramos C, Ferdin J, Lopez-Berestein G, Sood AK, Calin GA. MicroRNAs in body fluids--the mix of hormones and biomarkers. Nat Rev Clin Oncol. 2011;8:467–77. doi: 10.1038/nrclinonc.2011.76. [PMC free article] [PubMed] [Cross Ref]
33. Yoo BJ, Selby MJ, Choe J, Suh BS, Choi SH, Joh JS, et al. Transfection of a differentiated human hepatoma cell line (Huh7) with in vitro-transcribed hepatitis C virus (HCV) RNA and establishment of a long-term culture persistently infected with HCV. J Virol. 1995;69:32–8. [PMC free article] [PubMed]
34. Takebayashi Y, Goldwasser F, Urasaki Y, Kohlhagen G, Pommier Y. Ecteinascidin 743 induces protein-linked DNA breaks in human colon carcinoma HCT116 cells and is cytotoxic independently of topoisomerase I expression. Clin Cancer Res. 2001;7:185–91. [PubMed]
35. Li L, Sarver AL, Alamgir S, Subramanian S. Downregulation of microRNAs miR-1, -206 and -29 stabilizes PAX3 and CCND2 expression in rhabdomyosarcoma. Lab Invest. 2012;92:571–83. doi: 10.1038/labinvest.2012.10. [PubMed] [Cross Ref]

Articles from RNA Biology are provided here courtesy of Landes Bioscience