Target mRNAs are repressed as efficiently by microRNA-binding sites in the 5′ UTR as in the 3′ UTR

A Target Reporter mRNA Is Repressed by Endogenous let-7a miRNA in HeLa Cells.

Previously, a region of the C. elegans lin-41 3′ UTR containing two conserved LCSs separated by a 27-nt spacer was observed to be sufficient to down-regulate C. elegans LIN-41, whereas deletion of these LCSs and the intervening spacer interfered with temporal down-regulation (33). We constructed FF luciferase reporters containing two copies of this 85-bp sequence (Fig. 1A, bolded sequence from the C. elegans lin-41 3′ UTR) and thus four copies of the LCS or of its reverse complement, also containing mutations in the seed region of the LCS (white or gray small boxes in Fig. 1B; see Materials and Methods). These constructs were cloned into expression plasmids under the control of either the strong CMV or the weak herpes simplex virus thymidine kinase (HSV-TK) promoter (Fig. 1B) (34, 35). HeLa cells were transiently transfected [using the cationic lipid TransIT-HeLaMONSTER transfection reagent from Mirus (Madison, WI)] with these plasmids and, simultaneously, with a control plasmid expressing Renilla (R) luciferase under the control of the alternate promoter, to prevent promoter competition (Fig. 1B). This system was chosen instead of experiments involving synthetic, exogenous miRNAs because endogenous human let-7a, which is expressed in HeLa cells (refs. 30 and 36 and data not shown), should target these LCSs.

We observed a 3- to 4-fold decrease in FF luciferase activity of LCS-containing mRNAs relative to those with reversed, mutated LCSs, whether under the control of the strong or weak promoter (Fig. 1C). This degree of repression agrees with published data (5, 20, 37). Also, in agreement with recent reports of mRNA degradation associated with miRNA-mediated regulation (2), we observed some (≈35%) degradation of our LCS-containing reporter mRNAs relative to their reversed, mutated LCS controls (determined by RNase protection) (Fig. 1 D and E). After normalization for mRNA levels, translation efficiency was diminished by ≈2-fold (Fig. 1F). Increasing let-7a expression up to 2-fold in these cells, either by expressing the let-7a primary transcript from a plasmid or by transfecting siRNA, did not lead to an increase in translational repression (data not shown).

Reporters Under the Control of the HCV IRES Are also Translationally Repressed by let-7a Binding Sites.

To test whether initiation is the translation step affected by miRNA-mediated repression, the HCV IRES (383 bp including the first 50 bp of coding sequence) was fused to the N terminus of the FF luciferase ORF to bypass most of the steps necessary for translation initiation (31, 32). The HCV IRES initiates translation dependent only on eukaryotic initiation factor 3 and the initiator tRNA–ternary complex. Such IRES-containing reporters therefore allowed us to study the effect of the position of miRNA-binding sites relative to the ORF without interfering with preinitiation scanning from the cap by ribosomes. Forward or reversed LCSs were placed in either the 5′ or 3′ UTRs of HCV IRES-containing reporter mRNAs (Fig. 2A), and HeLa cells were transfected by TransIT-HeLaMONSTER (Mirus) with these DNA plasmids.

When the LCSs were present in the 5′ UTR of an IRES-containing mRNA, either 5′ or 3′ to a hairpin designed to prevent leaky ribosome scanning (38, 39), luciferase activity was repressed (Fig. 2B). As we observed with reporters without an IRES (Fig. 1 D and E), repressed mRNAs are present at slightly lower levels (≈33% down; second to fifth lanes of Fig. 2 C and D). After normalization for RNA levels, repression was 2-fold or more with miRNA-binding sites present in the 5′ UTR of the IRES-containing mRNAs. The simplest interpretation of these results is that miRNPs inhibit protein synthesis at a step downstream of translation initiation.

As a control, we placed the LCSs in the 3′ UTRs of the same IRES-containing constructs and surprisingly observed no effect on luciferase activity or mRNA levels relative to the reversed LCS controls (Fig. 2 B, C, and E). This observation suggests the contradictory conclusion: that translation initiation is the step affected during miRNA-mediated repression because the IRES is able to alleviate repression. We reasoned that perhaps the repression of an IRES-containing mRNA with miRNA-binding sites in the 5′ UTR (Fig. 2E) occurs because a “normal” scanning ribosome is different in its ability to associate with miRNP factors from a ribosome that initiates at an IRES. A normal ribosome would scan the 5′ UTR up to the hairpin preceding the IRES and then halt in close proximity to LCSs, allowing miRNP(s) to be recruited to the 5′ UTR, with resulting repression. In contrast, a ribosome that enters the ORF at an IRES might lack not only translation initiation factors but also the ability to recruit miRNPs, therefore allowing an IRES-containing message with miRNA-binding sites located at a distance in the 3′ UTR to escape repression.

To test whether stalled ribosomes that scan from the 5′ cap are important to the results observed with the IRES-containing reporters, G- and A-capped, polyadenylated mRNAs were transcribed from the same plasmids (described in Figs. 1B and 2A). A-capped mRNAs would presumably not have normal ribosomes on any portion of the 5′ UTR and therefore would not be expected to exhibit the repression observed with endogenous G-capped mRNAs (40). Upon transfection into HeLa cells using Lipofectamine 2000 (Invitrogen, Carlsbad, CA), G-capped IRES-containing FF luciferase transcripts were repressed whether the LCSs were in the 5′ or 3′ UTR (Fig. 3 D–F). Similar results were obtained for the same transcripts with an A-cap (Fig. 3 G–I). Most puzzling, the results obtained with LCSs in the 3′ UTR differed for RNA versus DNA transfections. Whereas a lack of repression was observed by DNA transfection (sixth and seventh lanes in Fig. 2E), nearly 2-fold repression was obtained by RNA transfection (fifth and sixth lanes in Fig. 3F).

Because the Lipofectamine 2000 used to introduce the RNA transcripts into cells (Fig. 3 D–F) is another cationic lipid-mediated transfection reagent different from the TransIT-HeLaMONSTER we used for DNA transfections, the experiments in Fig. 2 were repeated with this new reagent (Fig. 3 A–C). DNA transfections gave the same results regardless of which cationic lipid formulation was used (compare Fig. 2E with Fig. 3C). Specifically LCSs 5′ to the ORF in our IRES-containing reporters promoted repression of translation, whereas LCSs in the 3′ UTR did not exhibit miRNA-mediated repression.

miRNA-Mediated Translational Repression Is Affected by the Transfection Method.

While we were completing these studies, Barreau et al. (41) reported that caution should be used with liposome-mediated transfections of RNA and advised that electroporation should be used instead. We therefore transfected HeLa cells by electroporation with both our DNA and RNA constructs (Fig. 4). Interestingly, our RNA transfection results were reproducible using either cationic lipid-mediated transfection or electroporation, with the LCSs effecting repression in all positions (Figs. 3F and 4F, but note the larger error bars for RNA electroporations). In contrast, the DNA electroporation results for LCSs in the 3′ UTR showed a >2-fold repression (Fig. 4C), in direct contrast to the lack of repression observed by liposome-mediated transfection (Figs. 2E and 3C). These results highlight that caution should be taken when using different transfection methods and that conclusions should not be based on a single method.

Translational Repression of a Reporter mRNA Occurs When Human Ago2 Is Tethered to Either Its 5′ or 3′ UTR.

Finally, we examined the effect of tethering the human Ago2 (hAgo2) protein, a core component of the miRNA-induced silencing complex, to the 5′ UTR rather than the 3′ UTR, where it is known to cause translation repression (42). IRES-containing mRNAs with the LCSs substituted by five phage boxB hairpins in either the 5′ or 3′ UTR (Fig. 5A) were coexpressed with a N-HA-Ago2 fusion protein (42), both from DNA plasmids. Because ribosomes enter the IRES-containing message at an internal start codon, binding of the fusion protein should not interfere with ribosome entry into the FF ORF. A mutant N-HA-hAgo2 protein, in which the first 62 aa of Ago2 containing the PRP domain were deleted (ΔPRP) (Fig. 5A), was used as a control because it was previously shown that tethering this mutant protein does not induce repression (42). After normalization to the R reporter, a control reporter containing the five boxB hairpins in reverse orientation and mRNA levels, an ≈2-fold level of repression was observed (Fig. 5B), comparable to that published previously (43). This repression occurred regardless of whether hAgo2 was tethered to the 5′ or 3′ UTR of the mRNA (Fig. 5B).

DNA Constructs.

FF4LCS and its reverse, mutated control were created by subcloning the FF luciferase and 4LCS (containing a duplicated sequence from the C. elegans lin-41 3′ UTR) into pcDNA3 (Invitrogen), with a CMV promoter, or into phRL-TK (Promega, Madison, WI), with a HSV-TK promoter and R luciferase removed. The mutated 4LCS maintained the same base composition (UUAUACAACCGUUCUACACUCA→UACCCAUUAAGUUCUACACUCA).

pc4LCShpHCV/FF, pchp4LCSHCV/FF, pcHCV/FF4LCS, and their reverse controls were made by subcloning into pcDNA3 4LCS, a hairpin, and an HCV/FF fusion containing the HCV IRES (383 bp) fused to FF luciferase, in the order indicated. The hp/HCV IRES/4LCS DNA constructs had 5′ UTRs ranging from 521 to 723 bp and 3′ UTRs ranging from 157 to 333 bp. pc5BBhpHCV/FF, pchpHCV/FF5BB, and their reverse controls were created by cloning the HCV/FF fusion, five boxB domains, and a hairpin into pcDNA3. These constructs had 5′ UTRs of 521 or 836 bp and 3′ UTRs of 157 or 470 bp. R luciferase was amplified from another vector and introduced into pcDNA3, to create pcR, or replaced R in phRl-TK to produce phR. phR generates an RNA transcript that has a 65-nt 5′ UTR and a 231-nt 3′ UTR.

For complete details of plasmids, see supporting information (SI) Materials and Methods.

RNA Transcription.

Cold, G-capped RNAs were in vitro-transcribed using the mMESSAGE mMACHINE kit (Ambion, Austin, TX) from plasmids described above cleaved with DraIII. A-capped RNAs were transcribed using this same kit, but the 2× NTP/CAP solution was replaced with a 40 mM Tris buffer containing 1.5 mM ATP, 7.5 mM other NTPs, 6 mM A cap-A(5′)ppp(5′)G (NEB, Ipswich, MA), and 15 mM MgCl₂. The RNAs were polyadenylated with the Poly(A) Tailing kit (Ambion), extracted with phenol/chloroform, and precipitated with ammonium acetate and isopropanol. The hp/HCV IRES/4LCS RNA constructs had 5′ UTRs ranging from 481 to 683 nt and 3′ UTRs ranging from 559 to 735 nt (before polyadenylation). The R RNA had a 5′ UTR of 15 nt and a 3′ UTR of 634 nt.

Transfection Protocols.

For TransIT-HeLaMONSTER (Mirus) transfections, HeLa JW36 cells were transfected with DNA plasmids. One of six wells received 1.5 μg of phFF4LCS or its reversed, mutated control along with 25 ng of pcR or 200 ng of pcFF4LCS or its control along with 2 μg of phR. A total of 200 ng of pcHCV/FF construct was transfected with 2 μg of phR. The RNA from three wells and luciferase activity from one well were analyzed.

For Lipofectamine 2000 (Invitrogen) plasmid transfections, 200 ng of each pcFF construct per well with 2 μg of phR was transfected. Four wells were used per construct, with luciferase activity and RNA determined as described above. For RNA transfections, 1 μg of G-capped or A-capped FF RNA was transfected along with 1 μg of R RNA. Two wells of a six-well plate were transfected per construct; one well was harvested for luciferase activity, and one well was harvested for RNA.

For plasmid transfection by electroporation, 3 × 10⁶ cells were added to 7.5 μg of a single pcFF construct and 7.5 μg of phR and immediately transferred to two wells of a six-well plate. One well was harvested for luciferase activity, and one well was harvested for RNA. Electroporation with RNA transcripts was done with 9 μg of G-capped FF RNA and 9 μg of R RNA added to 6.625 × 10⁶ cells. After pulsing, the cells were transferred to three wells of a six-well plate; one well was harvested for luciferase activity, and two wells were harvested for RNA.

Tethering hAgo2.

A total of 100 ng of pc5BBhpHCV/FF, pchpHCV/FF5BB, or their reverse controls were transfected by TransIT-HeLaMONSTER (Mirus) along with 500 ng of pNHAAgo2 or pNHAAgo2ΔPRP and 2 μg of phR into one well of a six-well plate as described above.

Luciferase Assay.

The contents of one well were isolated by using the Dual-Luciferase Reporter Assay according to the manufacturer's directions. A total of 50 μl of the 500 μl of 1× PLB were assayed by a 20/20n luminometer (Turner BioSystems, Savannah, GA) in a 0.5-ml Eppendorf tube using 100 μl of LARII and 100 μl of Stop & Glo Reagent. FF and R luciferase activities were measured.

RNase Protection Assay.

The RNase protection protocol was adapted from one published earlier (45) and is described in detail in SI Materials and Methods and elsewhere (51).