MicroRNAs direct rapid deadenylation of mRNA

Results

Down-regulation of mRNA by miR-125b has been observed both in P19 mouse embryonal carcinoma cells, where the increased production of this miRNA upon differentiation into neurons contributes to a marked decline in the concentration of lin-28 mRNA, and in 293T human embryonic kidney cells, where the synthesis of miR-125b from a transfected gene significantly reduces the cellular abundance of luciferase reporter mRNAs bearing multiple copies of either miRE1 or miRE2 in the 3′ UTR (15). To determine whether miR-125b decreases the concentration of mRNAs bearing these elements by expediting mRNA degradation, we examined the effect of lin-28 miRE1 on the decay of a β-globin reporter mRNA (BG) expressed in 293T cells under the control of a transiently inducible c-fos promoter. This well established promoter-reporter system for studying mRNA decay made it possible to monitor the nearly synchronous degradation of mRNA molecules that were similar in age (17). The presence of two copies of miRE1 in the 3′ UTR (BG+2E1) markedly accelerated the decay of the reporter mRNA in cells that had been engineered to produce miR-125b at a concentration comparable with that in differentiating P19 cells (Fig. 1A and B; see also Fig. 6, which is published as supporting information on the PNAS web site). Inserting additional copies of miRE1 resulted in even faster decay (data not shown). No such effect was observed in cells that lacked miR-125b (Fig. 1B; see also Fig. 6).

Further investigation showed that the increased rate of mRNA decay did not result from endonucleolytic cleavage within the lin-28 element. Luciferase reporter mRNAs that bore either miRE1 or a synthetic element (element P) perfectly complementary to miR-125b were extracted from 293T cells containing miR-125b and analyzed by ligation to a synthetic oligoribonucleotide and RT-PCR with primers related to the oligoribonucleotide or complementary to a 3′ UTR segment downstream of the regulatory element. This procedure made it possible to detect any 3′-terminal degradation intermediates that might result from mRNA cleavage within the target element, as the 5′ end of such intermediates could be joined to the oligoribonucleotide by T4 RNA ligase (18). Although miR-125b directed cleavage within the RNA element to which it was perfectly complementary, no such cleavage could be detected in or near the imperfectly complementary miRE1 element, which nonetheless mediated significant reductions in luciferase mRNA and protein levels (Fig. 1C and other data not shown; see also Fig. 7, which is published as supporting information on the PNAS web site). We conclude that miR-125b accelerates the decay of mRNA containing imperfectly complementary elements by a mechanism that is distinct from the site-directed endonucleolytic process characteristic of RNA silencing mediated by perfectly complementary siRNAs (1).

Closer examination of the decay of BG+2E1 mRNA revealed that, in the presence of miR-125b, it underwent noticeable shortening within 3 h after its transient synthesis (Fig. 1B). Site-specific cleavage in vitro by RNase H before gel electrophoresis showed that this decrease in length was due to trimming of the 3′ poly(A) tail (Fig. 1D), most likely by an exonuclease. By measuring poly(A) length as a function of time after inducing transient transcription in cells containing miR-125b, we found that the presence of miRE1 caused extensive deadenylation and that decay of the transcribed portion of BG+2E1 mRNA did not begin until the shortest poly(A) tails had reached a length of 20–30 nt or less. No such effect of miR-125b was observed for the poly(A) tail of an otherwise identical reporter mRNA (BG) lacking miRE1; nor did miRE1 expedite deadenylation in the absence of miR-125b. That BG+2E1 mRNA undergoing miR-125b-mediated deadenylation still retained its 5′ cap was evident from its resistance to 5′-exonucleolytic digestion in vitro, which was lost upon treatment with tobacco acid pyrophosphatase to remove the cap (Fig. 1E). These findings indicate that miR-125b can destabilize mRNA by interacting with imperfectly complementary 3′ UTR elements and hastening deadenylation as an initial step in accelerated decay.

To identify other transcripts down-regulated by miR-125b, we examined its effect on mRNA levels in P19 cells, which produce this miRNA when induced to differentiate into neurons (15, 19). Cytoplasmic RNA extracted from undifferentiated P19 cells that had or had not been transfected with chemically synthesized miR-125b (Fig. 6) was used to probe a mouse genome microarray. With a high degree of confidence (P < 0.003), 22 mRNAs were found to decrease in abundance by at least a factor of 1.4 within 24 h after exposure to miR-125b. As expected, these included lin-28 mRNA, whose concentration fell by a factor of 1.7, a change comparable with the 2-fold increase in Lin-28 protein levels previously observed when the function of miR-125b was inhibited in differentiating P19 cells (15). Most of the affected mRNAs contained one or more 3′ UTR elements with the potential to interact productively with miR-125b, whereas such elements were comparatively rare in P19 mRNAs judged not to be affected (Fig. 2A; see also Table 1, which is published as supporting information on the PNAS web site).

Besides lin-28, the P19 messages down-regulated by miR-125b included Ajuba mRNA, whose 3′ UTR contains a likely miR-125b-responsive element in both mice and humans (A1; Fig. 2B). Ajuba is a LIM-domain protein that plays an important role in controlling the entry of vertebrate cells into mitosis and meiosis (20, 21). In P19 cells induced to differentiate, the concentration of this protein initially rises and then declines (22), suggesting a delayed-onset repression mechanism to which miR-125b may contribute as it accumulates over time. Like lin-28 miRE1, the Ajuba A1 element was found to direct rapid deadenylation and subsequent decay of a reporter mRNA in transfected 293T cells that produced miR-125b (Fig. 8, which is published as supporting information on the PNAS web site). mRNA destabilization involving fast poly(A) removal was likewise observed for an element derived from MKK7 (MAPK kinase 7) mRNA, another message down-regulated by miR-125b in P19 cells (Fig. 2B; see also Fig. 9, which is published as supporting information on the PNAS web site). These findings indicate that miR-125b can expedite the deadenylation and decay of many different mRNAs bearing a variety of elements to which it is imperfectly complementary.

A second miRNA, let-7, also helps to down-regulate lin-28 gene expression in differentiating P19 cells by interacting with a distinct 3′ UTR element (L7; Fig. 3A) (15). In combination, L7 and let-7 proved to be as effective as miRE1 and miR-125b at repressing protein synthesis (15), as judged from the 89% decrease in expression of an L7-containing luciferase reporter (Luc+6L7) in the presence of let-7a. Furthermore, down-regulation of protein synthesis was accompanied by a smaller yet substantial decline in mRNA concentration (Fig. 3B). This reduction in mRNA was a consequence of accelerated deadenylation and decay, as evidenced by the ability of one copy of L7 to cause rapid poly(A) shortening and subsequent degradation of BG+L7 mRNA in 293T cells transfected with a gene encoding let-7a (Fig. 3C; see also Fig. 6). That these effects were not cell-type-specific and could be caused by an endogenous miRNA was demonstrated by showing a similar influence of L7 on mRNA deadenylation and decay in HeLa cells (Fig. 3D), which produce let-7 naturally (23).

To demonstrate that the rapid deadenylation and decay mediated by miRNAs is not a consequence of the reduced frequency with which ribosomes transit the coding region, translation of BG+2E1 mRNA was abolished by inserting a large (40-nt) stem-loop structure within its 5′ UTR to create BG+2E1+hp. This stem-loop had previously been shown to prevent assembly of 80S ribosomes when present at the same location within another mRNA containing an identical 5′ UTR and coding region (24), and its inhibitory effect on translation of BG+2E1+hp mRNA was confirmed by in-frame fusion to a luciferase coding region, which showed it to reduce protein synthesis by >99.5%. Blocking translation initiation in this manner caused only a slight increase in the slow rate of poly(A) shortening previously observed for BG+2E1 in the absence of miR-125b, and this change was dwarfed by the much larger increase in the rate of deadenylation and decay caused by miR-125b (Fig. 4). Thus, impaired translation per se is not sufficient to accelerate deadenylation markedly, nor is translation required for miR-125b to increase the rate of poly(A) removal.

Conversely, a poly(A) tail is not required for miR-125b to repress translation. This was shown by comparing the effect of miR-125b on luciferase reporter mRNAs bearing miRE1 and either a 3′ poly(A) tail (Luc+6E1) or a 3′-terminal stem-loop derived from a histone mRNA (Luc+6E1.HSL). As expected, miR-125b diminished the abundance of only the polyadenylated reporter mRNA (Fig. 5A and B). Interestingly, assays of luciferase protein synthesis showed that, whereas the overall degree of repression was greater for the polyadenylated reporter, the inhibitory effect of miR-125b on translation efficiency (protein synthesis per mRNA molecule) was the same for both mRNAs (Fig. 5C). This finding suggested that the regulatory effects of translational repression and accelerated decay were additive, such that decreases in translation efficiency and mRNA concentration contributed equally to down-regulation of the polyadenylated reporter, whereas repression of the reporter bearing a histone stem-loop was achieved entirely at the level of translation. We conclude that, whereas the mRNA decay mechanism triggered by the interaction of miR-125b with miRE1 involves accelerated deadenylation, the mechanism of translational repression mediated by the same interaction does not require a poly(A) tail.

Discussion

The discovery that both miR-125b and let-7 can expedite poly(A) removal as an initial step in the rapid degradation of mRNAs bearing imperfectly complementary elements suggests that accelerated deadenylation and decay are likely to be frequent consequences of the interaction of mammalian miRNAs with their targets. By analogy to known pathways for mRNA degradation (25), poly(A) tail loss would be expected to facilitate decay of the transcribed portion of targeted mRNAs by exposing either their 5′ ends to cap removal and subsequent degradation by the 5′ exonuclease Xrn1 or their 3′ ends to exonucleolytic degradation by the exosome. The former pathway would help to explain the detection of 3′-terminal decay intermediates resulting from let-7-mediated mRNA degradation in C. elegans (14).

mRNA decay triggered by rapid deadenylation appears to be an important aspect of the mechanism by which miR-125b and let-7 down-regulate gene expression. Thus, replacing an mRNA poly(A) tail with a 3′ histone stem-loop not only eliminates the effect of miR-125b on message abundance but also diminishes its influence on protein synthesis by an amount equal to its effect on the concentration of an otherwise identical mRNA bearing a poly(A) tail. This finding suggests that, by reducing both the translation efficiency and the concentration of targeted mRNAs that are polyadenylated, miRNAs can have a greater overall impact on gene expression than if down-regulation were limited to translational repression. Moreover, unlike translation inhibition, the regulatory consequences of mRNA degradation are irreversible. By extrapolation, accelerated deadenylation may also contribute to the off-target effects of siRNAs, which can likewise reduce the abundance of mRNAs to which they are imperfectly complementary (26).

Although it is possible that miRNA-mediated deadenylation and translational repression may both result from a single precipitating event, our findings imply that these two regulatory mechanisms can operate independently. Translation is not required for miRNAs to hasten poly(A) shortening, nor is a poly(A) tail needed for them to down-regulate translation efficiency. The latter observation is of particular significance in view of recent evidence that miRNAs inhibit translation initiation (5), as poly(A)-binding protein (PABP) facilitates translation initiation by interacting with a protein component of the 5′ cap binding complex (27, 28). Thus, miRNA-mediated poly(A) tail loss might have been expected to impair translation efficiency were the deadenylated messages not rapidly degraded. The ability of miR-125b to inhibit translation of a reporter that ends in a histone stem-loop instead of a poly(A) tail suggests that translational repression by miRNAs occurs by a mechanism that is unrelated to PABP loss or at a step in translation initiation where PABP and the protein that binds histone stem-loops function analogously (28, 29).

Materials and Methods

Supplementary Material