Posttranscriptional Regulation of miRNAs Harboring Conserved Terminal Loops

hnRNP A1 Acts to Remodel the Stem Structure of pri-miR-18a Facilitating Drosha-Mediated Processing

In order to define potential structural features that make pri-miR-18a processing dependent on hnRNP A1, we took advantage of a highly related pri-miRNA sequence, pri-miR-18b, which is part of the homologous primary cluster miR106a∼18b∼20b located on chromosome X (Tanzer and Stadler, 2004) that was shown to be processed independently of hnRNP A1 (Guil and Caceres, 2007). In the case of pri-miR-18b, we detected an increase in the intensity of Pb(II)-lead ion cleavages between G54 and G63 that mapped to the UC bulge (Figure S3 and Figure 2B). Thus, the cleavage pattern of pri-miR-18b resembles the structural rearrangements seen in the stem of pri-miR-18a in the presence of added recombinant hnRNP A1 protein, that is, the relaxation of residues between U56 and U60 that are involved in strong Watson-Crick pairing (Figure 1A and Figure S3). To establish the significance of this observation, we introduced two nucleotide mutations on pri-miR-18a and pri-miR-18b in order to force structural changes in the stems so that the conformation of the mutant pri-miR-18a resembles the one from pri-miR-18b and vice versa (Figure 2B). Footprint analysis confirmed that the designed mutations altered only locally the conformation of the stem loops without perturbing the rest of the structure (Figure S4). We then performed in vitro processing assays on these new substrates in the context of the pri-miR-17-18a-19a minicluster. As previously reported, miR-18a was the only pri-miRNA whose processing was reduced when incubated in extracts from cells depleted of hnRNP A1, since both the neighboring 17 and 19a pri-miRNAs were insensitive to the reduced levels of hnRNP A1 (Figure 2C, lane 3) (see also (Guil and Caceres, 2007). Upon depletion of hnRNP A1 achieved with an unrelated set of siRNAs, we observed a similar result, and the accumulation of the remaining RNA containing unprocessed pre-miR-18a becomes more evident (Figure S5B, asterisk). Strikingly, introduction of a bulge in the pri-miR-18a stem (UC → GU) made its processing more efficient and completely independent of the presence of hnRNP A1 (Figure 2C). This experiment clearly shows that a pri-miR-18a with a local structural change in its stem resembling pri-miR-18b (or pri-miR-18a upon addition of hnRNP A1 as shown on Figure 1C) is now processed independently of hnRNP A1. By contrast, the mutation in pri-miR-18b that eliminates the UC bulge (GU → UC) did not make this substrate dependent of hnRNP A1 for its processing but rather abrogated its processing in HeLa cell extracts (Figure 2D). From these experiments, we conclude that the requirement for hnRNP A1 as an auxiliary factor in the processing of pri-miR-18a reflects the ability of this factor to bind and alter the local conformation of the stem in the vicinity of Drosha cleavage sites. These results revealed that subtle structural features in pri-miRNAs that are innate or arise from chaperone activities of RNA binding proteins might facilitate their processing by Drosha.

Open in a separate window

Figure 2

A Fine Structural Alteration in the Pri-miR-18a Stem Allows for hnRNP A1-Independent Drosha Cleavage

(A) Sequences of the highly related pri-miR-18a and pri-miR-18b.

(B) Schematic of the secondary structures of wild-type and stem mutants of pri-miR-18a and pri-miR-18b. Mutated nucleotides are bolded.

(C) In vitro processing of pri-miR-18a and the mutant pri-miR-18a UC > GU substrate. Both radiolabeled primary RNA sequences (50 × 10³ c.p.m.) were incubated in the presence of either control HeLa extracts (lanes 2 and 5) or hnRNP A1-depleted extracts (lanes 3 and 6). Lanes 1 and 4 show negative controls with no extract added. Products were analyzed on an 8% polycrylamide gel. The asterisk indicates the pre-miR-18a product. M, RNA size marker.

(D) In vitro processing of pri-miR-18b and the mutant pri-miR-18b GU > UC substrate. Both radiolabeled primary RNA sequences (50 × 10³ c.p.m.) were incubated in control HeLa extracts (lanes 2 and 4). Lanes 1 and 3 show negative controls with no extract added. Products were analyzed on an 8% polycrylamide gel. The asterisk indicates the pre-miR-18b product. M, RNA size marker.

Conserved Terminal Loops in Pri-miRNAs

Since the footprint analysis showed that hnRNP A1 has a strong binding site on the pri-miR-18a terminal loop, we next focused on the significance of this interaction. It is assumed that the terminal loops of pri-miRNAs are relatively functionally unimportant (Han et al., 2006). This notion is in agreement with the poor phylogenetic conservation that most pri-miRNAs display along their terminal loop regions, which contrasts with the high level of conservation in the mature miRNA sequences (Berezikov et al., 2005). We have noticed that the terminal loop of pri-miR-18a is atypically well conserved across vertebrate species (Figure 3A, compare with pri-miR-27a, which displays no special conservation in the loop, Figure 3B). Phylogenetic analysis of human pri-miRNAs sequences revealed that ∼14% (74 out of 533) of the miRNAs analyzed had similar high conservation pattern across the whole pri-miRNA sequence (Figure 3C and Figure S6). It is conceivable that the conservation of the terminal loops indicates the need to preserve a platform for the binding of auxiliary factors required for an efficient processing of these pri-miRNAs.

Open in a separate window

Figure 3

Multiple Sequence Alignments for the Human Genome with 16 Other Vertebrate Genomes Show Very High Level of Conservation in the Terminal Loop Regions of Some Human pri-miRNAs

(A) Conservation across pri-miR-18a corresponds to a persistently high phastCons conservation profile (the blue track labeled “Conservation”) across stem and terminal loop region predicted by miRBase and EvoFold.

(B) Conservation across pri-miR-27a includes a clear dip in phastCons conservation profile (the blue track labeled “Conservation”) corresponding to the terminal loop region predicted as above.

(C) A list of 74 pri-miRNAs with unusually highly conserved terminal loops is shown.

Oligonucleotides Complementary to Conserved Terminal Loops, LooptomiRs, Block pri-miRNA Processing

In this regard and to evaluate the role of conserved terminal loops in pri-miRNA biogenesis, we performed in vitro processing assays in HeLa extracts with the addition of 2′-O-methyl-modified oligonucleotides complementary to sequences in the terminal loops, which we termed LooptomiRs (for Loop Targeting Oligonucleotide anti miRNAs). We found that the Drosha-mediated processing of pri-miR-16-1 and pri-miR-27a, which bear nonconserved terminal loops, was not affected by the addition of complementary looptomiRs (Figures 4A and 4B). This observation was in agreement with a previous study showing that pri-miR-16-1 did not require the terminal loop for its efficient processing (Han et al., 2006). By contrast, looptomiRs designed against a number of conserved terminal loop sequences, including those of pri-miR-18a, pri-miR-101-1, pri-let-7a-1, pri-miR-379, and pri-miR-31, were able to specifically block the processing of their corresponding pri-miRNAs (Figures 4A, 4C, and ​and4D4D and Figures S7A and S7B). We believe that this is due to the block exerted by looptomiRs on sequences within the terminal loops that bind auxiliary factors required for the efficient processing of these targeted miRNAs. Furthermore, a looptomiR targeting miR-18a selectively abolishes the processing of pri-miR-18a in the context of the miR-17-18a-19a minicluster (Figure S7C). It is possible that looptomiRs could be used to block access to negative factors that negatively regulated pri-miRNA processing. It is also conceivable that some looptomiRs could potentially change the secondary structure of complementary pri-miRNAs, thus affecting the cleavage by Drosha. However, at least in the case of pri-miR-18a bound to its looptomiR, we did not detect any significant changes in the RNA structure of the region surrounding the Drosha cleavage site (Figure S8).

Open in a separate window

Figure 4

LooptomiRs Designed against Conserved Terminal Loops Block the Processing of a Subset of Pri-miRNAs

(A) Primary RNA transcripts corresponding to pri-miR-16-1 (lanes 1–5) and pri-miR-18a (lanes 6–10) (50 × 10³ c.p.m.) were processed in HeLa extracts in the presence of specific (lanes 3 and 4 and 8 and 9 for pri-miR-16-1 and pri-miR-18a, respectively) or control (lane 5 for miR-16-1 and lane 10 for miR-18a, respectively) looptomiRs (+, 50 μM; ++, 100 μM). Lanes 1 and 6 show negative controls with no extract added. M, RNA size marker.

(B) Processing of pri-miR-27a is not affected by the addition of a specific looptomiR. The RNA substrate (50 × 10³ c.p.m.) was processed in vitro in the presence of the corresponding blocking looptomiR (lane 3) or a control looptomiR (lane 4).

(C and D) In vitro processing of pri-miRNAs 101-1 and let-7a-1 is sensitive to the presence of specific looptomiRs. Radiolabeled pri-miRNA 101-1 and let-7a-1, (50 × 10³ c.p.m.) were processed in HeLa extracts with the addition of a specific looptomiR (lane 3) or with a control looptomiR specific for miR-18a (lane 4). As a control, the reaction was also carried out in the absence of any looptomiR (lanes 2). Lanes 1 represent control reaction without extract. M, RNA size marker.

The above results allowed us to conclude that the conserved terminal loop region of pri-miR-18a is essential for its efficient processing, most likely due to the direct binding of the auxiliary factor hnRNP A1. Moreover, the effect of looptomiRs on the processing of pri-miRNAs with conserved loops suggest a more general role of RNA binding proteins as auxiliary factors in the microRNA processing pathway. In order to find out potential additional auxiliary factors regulating processing of pri-miRNAs harboring conserved terminal loops, we combined RNA chromatography with mass spectrometry. Using this approach, we found that hnRNP A1, which was originally defined as an auxiliary factor for miR-18a, also binds to the apical regions of pri-let-7a-1 and pri-miR-101-1 stem loop (Figure 5 and Figure S9). The specificity of this interaction was demonstrated by the lack of hnRNP A1 binding to the loops of mir-16-1, miR-19a, miR-21, and miR-379 (Figure 5C and Figure S9). Notably, the terminal loop of let-7a-1 has a perfect hnRNP A1 consensus-binding site (UAGGGA/U) (Figure 5D). This loop was also found to bind hnRNP L, whereas PTB was found to bind to the pri-miR-101-1 stem loop. Experiments are currently under way to elucidate the role of these factors in the processing of their corresponding pri-miRNAs.

Open in a separate window

Figure 5

Truncated Stem Loops of Pri-miRNAs with Conserved Terminal Loops Bind Common and Distinct RNA-Binding Proteins

(A) RNA chromatography combined with Mass Spectrometry of selected pri-miRNAs harboring conserved loops was performed in HeLa nuclear extracts. Samples resulting from RNA chromatography were resolved on SDS gel and visualized using GelcodeBlue (Pierce). Bands 1 and 3 correspond to hnRNP A1, band 2 corresponds to hnRNP L, and band 4 corresponds to hnRNP I (PTB).

(B) A close-up of a region of the gel shown in (A). Expectation values from the mass spectrometry analysis are shown.

(C) Western blot analysis of RNA chromatography with truncated stem loops of corresponding pri-miRNAs.

(D) Predicted structures of pri-let-7a-1 and pri-miR-101-1 with highlighted putative protein binding sites are shown.

Pri-miR-18a Complex Formation

In order to have a closer look at the nature of the interaction between pri-miR-18a and hnRNP A1, we performed electrophoretic gel mobility shift (EMSA) experiments. First, we assayed increasing amounts of radiolabeled pri-miR-18a (the same transcript as described in Figure 4) programmed with a constant amount of hnRNP A1 recombinant protein (200 nM). The molar ratio of pri-miR-18a relative to hnRNP A1 ranged from 0.25 to 4, such that RNA binding will saturate the protein. A monomeric complex, migrating in between the substrate and its minor structural conformer, had a binding stoichiometry close to one (Figure S10A). A larger multimeric complex migrating above the minor conformer was formed in the higher RNA:protein molar ratios, indicating that pri-miR-18a is able to bind more than one molecule of hnRNP A1. Importantly, addition of a specific looptomiR targeting pri-miR-18a abrogated the formation of the monomeric miR-18a/hnRNP A1 complex without substantially affecting the multimeric complex (Figure S10B, lanes 5–8). We have shown that addition of looptomiRs specifically blocks in vitro miRNA processing (Figure 4 and Figure S7). Taken together, this would suggest that displacing hnRNP A1 from the terminal loop of pri-miR-18a, without affecting its binding to other regions of this pri-miRNA, is sufficient to block its activity in miRNA processing. Moreover, addition of the looptomiR caused a small shift in the migration of both main pri-miR-18a substrate and its minor conformer. This result points to an efficient binding of the looptomiR to the transcript and also suggests small changes of the global architecture of pri-miR-18a stem-loop structure in the presence of the specific looptomiR, which is in agreement with the structural analysis of such complex (Figure S8).

RNA Probing and Footprint Analysis

In vitro transcription of RNAs subjected to structure probing was carried out as previously described (Sobczak et al., 2003). Transcripts were 5′ end-labeled with T4 polynucleotide kinase (Roche) and [γ-³²P] ATP (4,500 Ci/mmol; Amersham Pharmacia Biotech). Transcripts were purified by electrophoresis in a 10% denaturing polyacrylamide gel. Labeled and unlabeled RNAs were visualized by autoradiography and StainsAll staining, respectively. Prior to structure probing transcription, reactions were subjected to a denaturation and renaturation procedure in a buffer containing 2 mM MgCl₂, 80 mM NaCl, and 20 mM Tris–HCl (pH 7.2) by heating the sample at 80°C for 1 min and then slowly cooling to 37°C. Limited RNA digestion was initiated by mixing 5 μl of the RNA sample (100 000 c.p.m. or 100 pmol for labeled and unlabeled RNAs, respectively) with 5 μl of a probe solution containing lead ions-Pb(II), nuclease S1, or ribonucleases T1 or V1 at concentrations specified in the Figure Legends. The reactions were performed at 37°C for 10 min. For the footprint analysis, recombinant hnRNP A1 protein was added in specified concentrations 5 min prior to the reaction. All reactions were carried out with 5′ ³²P-labeled RNAs and were stopped by adding an equal volume of stop solution (7.5 M urea and 20 mM EDTA with dyes) and sample freezing. Unlabeled RNAs, after incubation with the probes, were subjected to high-salt precipitation procedure. To determine the cleavage sites, the products of the 5′ ³²P-labeled RNA fragmentation reaction along with the products of alkaline hydrolysis and limited T1 nuclease digestion of the same RNA molecule were separated on 10% polyacrylamide gels containing 7.5 M urea, 90 mM Tris-borate buffer, and 2 mM EDTA. The alkaline hydrolysis ladder was generated by incubation of labeled RNA in formamide containing 0.5 mM MgCl₂ at 100°C for 10 min. Partial T1 ribonuclease digestion of RNAs was performed under semidenaturing conditions (10 mM sodium citrate (pH 5.0); 3.5 M urea) with 0.2 U/μl of the enzyme and incubation at 55°C for 15 min. Electrophoresis was performed at 1500 V and was followed by autoradiography at −80°C with an intensifying screen or exposed to the PhosphorImager screen. To determine the cleavages sites obtained for unlabeled RNAs, a reverse transcription-based reaction was used as previously described (Michlewski and Krzyzosiak, 2004). Briefly, 100 ng of RNA was reverse transcribed with the use of 2 pmol 5′ ³²P-labeled pri-miR-18a rev primer and SuperScript II reverse transcriptase. The products of the reverse transcription were separated as described above, along with the sequence ladder generated from the pri-17-18a-19a using Promega fentomol sequencing kit.

Pri-miRNA Substrates and In Vitro Processing Assays

RNA substrates for in vitro processing assays were prepared from the DNA templates by standard in vitro transcription with T7 RNA polymerase in the presence of [α-³²P]GTP. Preparation of total or hnRNP A1-depleted HeLa extracts was described previously (Guil and Caceres, 2007). Assays were done in 30 μl reaction mixtures containing 50% (v/v) total or depleted HeLa extract, 0.5 mM ATP, 20 mM creatine phosphate, 3.2 mM MgCl₂, and 20,000 c.p.m. (∼10 fmol) of each pri-miRNA. Reactions were incubated at 30°C for 30 min, then subjected to phenol-chloroform extraction, precipitation, and 8% (w/v) denaturing gel electrophoresis. For terminal loop blocking with looptomiRs, specific or control 2′-O-methyl oligonucleotides were purchased from Sigma-Aldrich and added to the pri-miRNA at a final concentration of 50 μM 5 min prior to the incubation with extract to allow annealing.

The looptomiRs sequences used were as follows: anti miR-16-1 loop (5′-oToAoAoToToToToAoGoAoAoToCoToToA-3′), anti miR-18a loop (5′-oAoToGoCoToAoAoToCoToAoCoToToCoA-3′), anti miR-27a loop (5-oGoAoCoToToGoGoToGoToGoGoAoCoCoC-3′), anti let-7a-1 loop (5′-oGoGoToGoGoGoToGoToGoAoCoCoCoToA-3′), anti pri-miR-101-1 (5′-oCoCoToToToAoGoAoAoToAoGoAoCoAoG-3′), anti pri-miR-31 (5′- oGoGoToToCoCoCoAoGoToToCoAoAoCoA-3′), anti pri-miR-379 (5′-oGoToCoAoGoAoAoAoToCoAoToAoAoCoG-3′). An Ambion's Decade RNA Size Marker was used.

Purification of Recombinant Protein

Recombinant hnRNP A1 protein was obtained as previously described (Cazalla et al., 2005). In brief, T7 tagged hnRNP A1 protein was expressed in the 293T cells transfected with pCG-hnRNP A1 vector (Caceres et al., 1997). Forty-eight hours after transfection, cells were scrapped and sonicated in lysis buffer. Recombinant hnRNP A1 protein was purified using anti-T7 affinity chromatography.

We are grateful to Javier Martinez (IMBA, Vienna) and Nick Hastie (MRC HGU, Edinburgh) for discussions and critical reading of the manuscript. This work was supported by the MRC, a project grant from the Wellcome Trust, and Eurasnet (European Alternative splicing Network-FP6). S.G. is the recipient of a Ramon y Cajal contract from the MICINN (Spanish Government).