Cooperative interplay of let-7 mimic and HuR with MYC RNA

Abstract

Both RNA-binding proteins (RBP) and miRNA play important roles in the regulation of mRNA expression, often acting together to regulate a target mRNA. In some cases the RBP and miRNA have been reported to act competitively, but in other instances they function cooperatively. Here, we investigated HuR function as an enhancer of let-7-mediated translational repression of c-Myc despite the separation of their binding sites. Using an in vitro system, we determined that a let-7 mimic, consisting of single-stranded (ss)DNA complementary to the let-7 binding site, enhanced the affinity of HuR for a 122-nt MYC RNA encompassing both binding sites. This finding supports the biophysical principle of cooperative binding by an RBP and miRNA purely through interactions at distal mRNA binding sites.

Introduction

It is well established that RNA-binding proteins (RBPs) critically regulate gene expression at every stage, including pre-mRNA splicing, mRNA localization, translational efficiency and mRNA degradation.¹ It is also well known that miRNAs, in association with the RNA-induced silencing complex (RISC), represent a fundamental eukaryotic mechanism of reducing mRNA stability and/or translation.² Less is known about the complex interplay between these 2 families of regulatory factors, yet it is clear that their joint influence on shared mRNAs contributes to a robust, dynamic, and versatile regulation of gene expression at the post-transcriptional level.³

When both miRNAs and RBPs associate with a given mRNA, binding by the RBP can impact upon the interaction of the miRNA, and vice versa. For example, one of the best known RBPs, HuR, selectively binds AU-rich elements (AREs) and protects mRNA from degradation.⁴ One way HuR engenders protection is by competing with decay-promoting RBPs (for example, AUF1 or TTP) that would otherwise recruit the degradation machinery.⁵ Another mechanism is by interacting with miRNA target sites, thus preventing the binding of the miRNA and the subsequent RISC-mediated cleavage. This regulatory paradigm was documented for miR-122 targeting of the cationic amino acid transporter 1 (CAT1) mRNA and for miR-195 targeting Stim1 mRNA.^6,7 In one additional mechanism of HuR-mediated stabilization, COX2 mRNA degradation by miR-16 was repressed via the direct binding of HuR to miR-16.⁸ However, in a less expected regulatory paradigm, HuR reduced c-Myc expression by enhancing the binding of let-7 at an adjacent site on the MYC mRNA 3′-untranslated region (3′-UTR), leading to the recruitment of the let-7/RISC complex.⁹ In this case the interplay between miRNA and HuR caused an effect opposite from usual HuR function. Other examples of such interplay have emerged in recent years.^10-12

A mechanism based on functional competition between an RBP and a miRNA is easily understood if it is supported by physical competition. The binding of 2 types of molecules with the same RNA sequence specificity will clearly result in steric hindrance at their mutual mRNA binding site with a biological outcome that will depend on the interaction that prevails. But the mechanism of HuR recruitment of miRNA/RISC at an adjacent site, as reported by Kim et al.,⁹ is less obvious. The recruitment of let-7 to MYC mRNA by HuR was found not to occur through direct protein-protein interactions between the HuR and the RISC complex. Rather, it was proposed that modulation of the surrounding mRNA structure by HuR led to the unmasking of the let-7 binding site. According to this model, binding at one RNA target sequence enhanced binding at a distal sequence.

In the current study, we set out to test this hypothesis in vitro. Our biophysical assay system examined the binding of HuR to the MYC RNA region that contains both the let-7 and HuR binding sites. We used surface plasmon resonance (SPR), a very sensitive method for the detection of HuR binding to the MYC RNA, in the presence and absence of a let-7 mimic (let-7M) consisting of ssDNA complementary to the let-7 binding site. The assay system did not include RISC components, ruling out the possibility of effects due to HuR interactions with RISC consitituents. Instead, we tested whether the MYC mRNA was modified by let-7M binding to effectively “unmask” the adjacent site that is the target of HuR. While this is the inverse scenario from the cellular experiment, it tests the principle that MYC RNA modulation at one site facilitates binding at the other purely through biophysical interactions with the RNA. We observed cooperativity of miRNA and RBP binding at adjacent sites, lending strong biophysical support to the notion of interplay among miRNAs and RBPs occuring purely through distal mRNA interactions.

Results

In Kim et al.,⁹ it was established that nt 1991–2146 in the 3′-UTR of MYC mRNA (referred to as the B sequence) contains both the let-7 target sequence (nt 2015–2035) and an HuR interaction site. The study demonstrated that HuR binding was able to recruit let-7 in an RNA-dependent manner, and not through protein-protein interactions with the RISC complex.

From consideration of nt 2011–2132 of the ‘B’ sequence (Fig. 1A), it is clear that the let-7 target sequence is at least 34 nt upstream of where HuR binds. The HuR binding site was first predicted to occur at nt 2092–2120 based on the detection of binding of HuB (initially named Hel-N1) to MYC RNA.¹³ However, this study did not specifically identify the HuR binding site and did not explore the possibility that regions upstream of the identified 29-nt sequence could also be target sites for HuR. Clearer identification of the mRNA region bound by HuR, as determined using PAR-CLIP, was found to span nucleotides 2071–2097, overlapping the 29-nt sequence identified for HuB, but including an upstream U-rich region of the sequence.¹⁴ Interestingly, using the MFOLD prediction software, base-pairing for this sequence could exist such that both the let-7 and the HuR target sequences exist within base-paired structures that are quite separate from each other (Fig. 1B). In this case, a structural rearrangement of the RNA would be required for binding by either the let-7 or the HuR.

Figure 1.

Sequence of MYC 3′-UTR (nt 2011–2132) used in the current study. (A) The let-7 binding site is colored orange and the reported HuR binding sites are colored in purple and blue (with the region of overlap colored in darker purple). (B) shows an MFOLD prediction of base-pairing for the same sequence, excepting the 3′ nucleotides used for basepairing to the SPR chip, with let-7 and HuR binding sites highlighted using the same color scheme.

We set out to investigate whether the synergistic binding of HuR and let-7 could occur purely through a direct effect on the RNA. To test this possibility, we used surface plasmon resonance (SPR) as a sensitive means of detecting protein-RNA interactions. We prepared MYC RNA encompassing both the let-7 and HuR target sites (nt 2011–2132) using T7 polymerase and tethered this RNA to the surface of a streptavidin-coated biosensor chip via its 3′-end (through basepairing of the 3′-bases to biotinylated DNA). Full-length HuR was prepared to very high purity, as assessed using SDS-PAGE analysis, so that it could be flowed across the surface of the chip to detect its interactions with the RNA.

The SPR experiment was carried out as follows. Since the Biacore T100 chips contains 4 flow cells, 4 HuR binding experiments could be conducted simultaneously. On the first flow cell, only the biotinylated ssDNA sequence was attached, providing a non-HuR binding control cell (blank). To the second flow cell was attached the MYC RNA. In the third flow cell, the MYC RNA was attached after pre-annealing with a let-7 DNA sequence (“let-7 mimic;” henceforth referred to as let-7M), with perfect complementarity to the let-7 binding site of MYC RNA. On the fourth flow cell, the MYC RNA was attached as for the second flow cell and the let-7M was subsequently flowed across the surface at room temperature to form a complex with the already chip-bound MYC RNA. All of the remaining biotinylated ssDNA on the surface of the chip was then blocked using a ssDNA sequence complementary to the biotinylated ssDNA (Fig. 2A).

Figure 2.

Experimental design and SPR sensorgrams of HuR binding to MYC RNA in the presence or absence of annealed let-7M. (A) Schematic showing the arrangement of RNA bound to the surface of the biosensor chip in each flow cell across which HuR was flowed. (B) Top panel: sensorgram for HuR binding to MYC RNA immobilized on the chip surface. HuR (0–200 nM) was injected from 0 to 180 s otherwise buffer was flowing. The sensorgrams were referenced using a blank cell with only biotinylated DNA captured. Middle panel: sensorgram for HuR binding to MYC RNA with let-7M annealed prior to immobilization of MYC RNA on the chip surface. HuR (0–200 nM) was injected from 0 to 180 s otherwise buffer was flowing. Labels of low concentrations are omitted for clarity. The sensorgrams were referenced using a blank cell with only biotinylated DNA captured. Bottom panel: sensorgram for HuR binding to MYC RNA with let-7M annealed on the chip after MYC RNA immobilisation. HuR (0–200 nM) was injected from 0 to 180 s otherwise buffer was flowing. Labels of low concentrations are omitted for clarity. The sensorgrams were referenced using a blank cell with only biotinylated DNA captured.

Pure full-length HuR was then allowed to flow across the chip surfaces at a series of concentrations. As can be seen in Fig. 2B (where signal from the blank run has been subtracted), the sensorgrams of HuR show that very little binding was seen in flow cell 2 (a very low maximal response of 3 RU). Clear binding was observed, in contrast, for HuR flowed across flow cell 3 where the let-7M sequence was annealed to the MYC RNA (maximal RU ˜22). Together, these data indicate that the presence of the annealed let-7M sequence has availed a HuR binding site, allowing for increased HuR binding to take place. In a variation of this experiment, we tested whether the let-7M sequence could form a complex with the target MYC RNA and assist HuR binding without the assistance of heat and pre-annealing. Strong binding was observed in flow cell 4, demonstrating that the let-7M is able to access the MYC RNA target site effectively at room temperature and facilitate HuR binding.

Binding curves calculated for these experiments are shown in Figure 3. These are only estimates, as not all of the binding experiments reached equilibrium during the course of the experiment. The estimated K_D for HuR binding to MYC RNA preannealed with the let-7M was K_D = 18 nM, and that for HuR after the let-7M was allowed to form a complex with the MYC RNA at room temperature was K_D = 36 nM, suggesting a slightly more efficient interaction of the let-7M sequence when pre-annealed with the MYC RNA.

Figure 3.

Binding curves for HuR interaction with MYC RNA in the presence or absence annealed let-7M using SPR. HuR binding to MYC RNA shown as closed circles, HuR binding to MYC RNA with let-7M annealed prior to immobilization of MYC RNA shown as open circles, and HuR binding to MYC RNA with let-7M annealed prior to immobilization of MYC RNA shown as triangles. Error bars indicate standard deviation of triplicate measurements. Fits to the single-site binding model are shown as solid lines.

Discussion

The interplay between a miRNA and an RBP interacting with a shared target mRNA is fundamental for the post-transcriptional regulation of gene expression. While it is conceptually easy to see that miRNA and RBPs targetted to the same site could act competitively, it is less clear that miRNA interactions with RNA could affect protein binding at a distal site or vice versa.

In the current study, we characterized the effect of a let-7 mimic (let-7M: a ssDNA sequence complementary to the let-7 binding site) on HuR interactions with a 122-nt MYC 3′-UTR RNA encompassing both let-7 and HuR binding sites in vitro. Binding of let-7M to the let-7 binding site served as a miRNA mimic, but did not recruit RISC, as RISC components were absent from the assay. In this system, the HuR binding site was at least 34 nt downstream of the let-7 target sequence. Our study showed that in the absence of let-7M, HuR bound very poorly to the RNA, but when let-7M was bound to the target RNA, HuR bound with an approximated affinity in the nanomolar range. This finding demonstrates that binding interactions with RNA at one site can have a positive impact on another binding interaction at a distal site.

Consideration of the HuR binding sites within the predicted structure of the MYC mRNA (MFOLD) shows their predicted proximity to the let-7 binding site (Fig. 1B). Much of the RNA segment is predicted to contain double-stranded regions, including the sites of let-7 and HuR binding. Since both of these molecules bind to ssRNA, let-7 or HuR binding can only occur when the RNA transiently unfolds, exposing single-stranded regions for binding. The apo-RNA likely exists in a predominantly folded state with regions of double stranded secondary structure and intramolecular interactions between these regions. The binding of let-7M would disengage the let-7 binding region from intramolecular interactions and shift the equilibrium toward the unfolded form (Fig. 4). In this way the HuR would gain better access to its ssRNA binding site, as evidenced by a higher affinity interaction. By the same mechanism, the binding of HuR would be expected to enhance access to the let-7 binding site, explaining the observation made by Kim et al.,⁹ of enhanced let-7 mediated translational repression of MYC in the presence of HuR.

Figure 4.

Schematic representation of MYC RNA (nt 2011–2132) in equilibrium between folded and unfolded forms. Interactions with either let-7M or HuR would stabilize the unfolded form, shifting the equilibrium to the right and thereby increasing accessibility for the second binding event.

While competitive binding between RBPs and miRNA to mRNA have been more frequently reported, cooperative binding between RBPs and miRNA has also been observed.¹⁰ Examples other than that of let-7 /HuR binding to MYC mRNA, include the cooperative binding of HuR and poly-C-binding protein at adjacent sites in the 3′-UTR of androgen receptor mRNA and the cooperation between miR-19 and HuR in the regulation of Ras homolog B (RhoB) mRNA.^15,16 Studies of the effect of AUF1 binding to RNA have also led to the suggestion that AUF1 regulates miRNA binding at adjacent sites.^17,18 In each case, the cooperativity may be explained purely through a shift in equilibrium of mRNA structure to the less folded state by one RNA-binding molecule resulting in enhanced access to the second binding site. No protein-protein interactions or interactions with a further component such as the RISC complex need underlie this phenomenon, as shown in vivo by Kim et al.,⁹ and in vitro in the current study.

Materials and Methods

HuR Purification

Full-length HuR cDNA cloned into pTYB111 containing an intein-tag with an incorporated chitin-binding domain, was kindly provided by Gerald Wilson, University of Maryland. The HuR fusion protein was expressed in BL21 (DE3) E. coli cells in SOB media in the presence of 50 μg/mL ampicillin at 37°C until reaching an OD₆₀₀ of 0.6–0.8. Expression was induced with the addition of 1 mM IPTG and cells were incubated a further 5 h at 25°C. Cells were harvested by centrifugation at 3300 g for 20 min. Cells were resuspended in 25 mM sodium phosphate, 500 mM NaCl, 1 mM EDTA, pH 7.0 (HuR column buffer) with the addition of 20 μM phenylmethanesulfonyl fluoride, and lysed by French press. Lysates were cleared by centrifugation at 20000 g for 20 min, and bound to a chitin affinity column equilibrated with HuR column buffer, followed by wash with 2 M NaCl then with HuR column buffer. Self-induced cleavage was performed with the addition of HuR column buffer containing 50 mM DTT to the resin and incubation with mixing at 4 °C for 40 h. HuR was eluted with the further addition of HuR column buffer containing 50 mM DTT and was finally purified by size exclusion chromatography on a HiLoad superdex 200 16/60 column (GE lifescience). Protein purity was confirmed by Coomassie blue staining after SDS-PAGE analysis.

MYC RNA production

Linearized DNA MYC template with T7 promoter site was generated using plasmid psiCHECK2 with MYC (1891–2146) insert (accession number NM_002467) (10 pg/μL) with primers 5′TAATACGACTCACTATAGGGTTTAGCCATAATGTAAACTG3′and 5′TTAAAAACAATTCTTAAATACAAATCTG3′ (80 nM each) with the addition of 0.2 mM dNTP, 4 mM MgCl₂, KOD DNA polymerase (Novagen) (0.016 mg/μL) in KOD reaction buffer 1 (Novagen) in a reaction volume of 25 μL. The reaction mixture was subjected to pre-denaturing for 2 min at 94°C followed by 30 cycles of denaturing at 94°C for 20 sec, annealing at 60°C for 20 sec and extension at 72°C for 35 sec. The linear DNA template was purified on a 1.5% Agarose/TAE gel and extracted using Wizard® SV gel and PCR clean-up system (Promega). MYC RNA (nt 2011–2132; henceforth referred to as MYC RNA) was produced from the Linearized DNA MYC template using T7 RiboMAX™ Express Large Scale RNA Production system (Promega) following manufacturer's instructions with typical yields of 1 mg/mL.

Surface Plasmon Resonance detection of HuR binding

Biotin-TCATTGTGTAAATCTTAAAA (Bi-DNA) at 1.5 µM was incubated with 7.5 µM MYC RNA in the presence or absence of 7.5 µM of a “let-7 mimic” (let-7M: DNA complementary to the MYC let-7 binding site: GAGGCAGTTTACATTATGGCTA) at 95 °C for 5 min then allowed to cool to room temperature to anneal. Samples were then diluted 1/1000 in 10 mM HEPES, 150 mM NaCl, pH 7.4 (HBS) for immobilisation. On a BIAcore T100, Bi-DNA alone was injected at 30 µL/min for 234 s over flow cell 1 of a BIAcore SA series S sensor chip (GE Life Science) resulting in immobilization to a level of 51 RU. Bi-DNA preannealed with MYC RNA was injected at 30 μL/min for 234 s over flow cell 2, resulting in immobilization to a level of 110 RU. Bi-DNA preannealed with MYC RNA preannealed wth the let-7M was injected at 30 μL/min for 234 s over flow cell 3, resulting in immobilization to a level of 100 RU. Bi-DNA preannealed with MYC RNA was injected at 30 μL/min for 234 s over flow cell 4, resulting in immobilization to a level of 82 RU. Un-annealed single stranded Bi-DNA on all 4 flow cells was then blocked with the injection of 7.5 µM cDNA (TTTTAAGATTTACACAATGA) at 10 µL/min for 400 sec, resulting in the addition of 74 RU, 20 RU, 18 RU and 17 RU on flow cell 1, 2, 3 and 4 respectively. Finally, let-7M (7.5 µM) was hybridized to the MYC RNA on flow cell 4 by injection for 400 sec at 10 µL/min, resulting in the addition of 12 RU.

HuR binding experiments were then performed using 10 mM HEPES, 150 mM NaCl, 0.025% Surfactant P20, 1 mg/mL acetylated BSA, pH 7.4 as the running buffer. All HuR samples (0–200 nM) were diluted in running buffer and injected at 30 µL/min for 180 sec with 300 sec dissociation time through all 4 flow cells. Regeneration was performed with 2 60 sec injections of 2 M NaCl at 30 µL/min after each sample. All experiments were performed in triplicate.

Funding

The current study was funded by an Australian Research Council (ARC) Discovery Grant awarded to MCJW, JAW, and MG, and also through a National Health and Medical Research Council (NHMRC) Senior Research Fellowship awarded to MCJW. JY and MG were supported by the National Institute on Aging-Intramural Research Program of the National Institutes of Health.

Disclosure of Potential Conflicts of Interest

No potential conflict of interest was disclosed.