Inhibition of miR-21 by 3′/5′-Serinyl-Capped 2′- O -Methyl RNA Interspersed with 2′- O -(2-Amino-3-Methoxypropyl) Uridine Units

Introduction

MiRNAs are the most important class of regulatory small noncoding RNAs (∼19–24 mers) that posttranscriptionally modulate gene expression [1]. They are transcribed as primary hairpin precursors and sequentially processed by Drosha and Dicer, to yield a mature miRNA duplex. After getting incorporated into the mRISC complex, miRNA binds predominantly to the 3′ UTR (untranslated region) of the target mRNA resulting in translational repression and/or decay of mRNA based on their complementarity with their target mRNAs. It is estimated that more than 1,000 miRNAs are present in humans and 60% of the protein coding genes are regulated by miRNAs [2]. miRNAs have various functions in the multitude of biological processes like developmental timing, cellular differentiation, metabolism, cell proliferation, and apoptosis [3]. Because they modulate such fundamental physiological processes, any alteration in their expression or function can lead to pathological states such as, neurological disorders, diabetes, cardiovascular diseases, viral infections, and cancer [4]. The detection and/or modulation of deregulated miRNAs are thus of therapeutic interest, attracting chemical biologists.

Antisense technologies that use synthetic reverse complements have been extensively used to suppress or knockdown the expression of target RNAs. Antisense inhibition of miRNA has been an important tool to uncover miRNA functions. Precise functions of individual miRNAs have been delineated by loss of function studies using various anti-miRNA oligonucleotides (AMOs), which either degrade or sequester the miRNAs making them unavailable to bind to their target mRNAs leading to depression of their function. Furthermore, AMOs have been used as probes for detection [5–7], target validation, functional characterization, and therapeutic intervention of miRNAs [8,9]. AMOs are designed to bind to complementary miRNA through Watson–Crick hybridization and thereby repress the activity. Despite this seemingly straightforward and simple approach of sequence-specific targeting of miRNA, unmodified AMOs with phosphodiester backbone present several shortcomings such as poor cellular uptake, rapid degradation by nucleases in serum, low-target hybridization efficiency, and nonspecific targeting. The challenge still remains to bestow AMOs with adequate pharmacodynamic and pharmokinetic “drug-like” properties for therapeutic strategies. This can be successfully brought about by incorporating various chemical modifications in the nucleobase, internucleotide linkages, or in sugars to increase target binding affinity, specificity, potency, and biostability of AMOs [10]. These modifications include phosphorothioates (PS), 2′-O-methoxyethyl (MOE) [11], 2′-O-methyl (2′-OMe) [12,13], 2′-fluoro [14], peptide nucleic acids (PNAs) [15,16], and locked nucleic acids (LNAs) [17,18]. Although these modifications confer high duplex stability while binding with the target RNA, there is a need to further improve their enzymatic stability such that they do not get degraded by endogenous nucleases. Earlier reports suggested that 2′-O-methyl RNA oligonucleotides can provide efficient and simple way to block small RNA function in vivo [12,19]. Recently, terminally modified 14 mer 2′-O-methyl RNA has been shown to inhibit miRNA function with improved mismatch discrimination [20]. The modifications include trimethoxystilbene residue and three 2′-fluoro-2′-deoxynucleotides at the 5′ and 3′ terminal, respectively [20]. 2′-O-methyl RNA is a naturally occurring modification that confers high binding affinity (enhanced melting temperature, Tm) with the target RNA making stable duplexes [21]. In addition to having high affinity, an ideal AMO should also be resistant to both endo- and exonucleases present in the serum. 2′-O-methyl RNA, owing to its chemistry, impedes the ability of endonucleases to degrade single stranded oligonucleotides, but has the limitation of being sensitive to serum exonucleases [22]. A recent report used a non-nucleotide modifer, N,N-diethyl-4-(4-nitronaphthalen-1-ylazo)-phenylamine (“ZEN”), which improves the efficacy when placed at or near the ends of 2′-O-methyl RNA [23]. In our attempt to enhance nuclease resistance with conserved binding affinity of 2′-O-methyl RNA oligonucleotides, recently we described a novel chemical modification comprising methoxy- and amino-groups, that is, 2′-O-(R-2-amino-3-methoxypropyl) (R-AMP) substitutions [24]. Furthermore, we reported the enhanced nuclease resistance of these oligonucleotides by capping the 3′ and 5′ ends with serinol units. The presence of interspersed U^R-AMP units in 2′-O-methyl RNA oligonucleotide capped with serinyl units (S) at 3′/5′ ends was found to be superior to the unmodified 2′-O-methyl RNA, 2′-O-methyl RNA capped at 3′/5′ ends by serinyl units, as well as LNA-OMe mixmers, when used in a cell-based splice correction assay [25]. These promising results prompted us to investigate the efficacy of these modifications (Supplementary Fig. S1; Supplementary Data are available online at www.liebertpub.com/nat) in 2′-O-methyl RNA oligomers as inhibitors of miRNA in cell culture–based system. In this study, we report our modified oligonucleotides as potent miRNA inhibitors. The chemical synthetic routes to the modified uridine phosphoramidite monomer for introducing (U^R-AMP) [24] and serinyl-end cap [25] were simple and affordable (Supplementary Figs. S2 and S3).

These monomers were used in the automated DNA synthesizer along with 2′-O-methyl monomers to synthesize modified 2′-O-methyl RNA oligonucleotides (Fig. 1 and Supplementary Table S1).

OPEN IN VIEWER

FIG. 1.

The sequence and modifications of AMOs against miR-21. AMO-1 and AMO-2 are uniformly 2′-O-methyl modified RNA reverse complementary to miR-21. AMO-2 harbors three U^R-AMP monomers positioned at 2, 8, and 14 and two non-nucleotidic Serinyl caps at both termini. AMOs, anti-miRNA oligonucleotides.

Materials and Methods

Results and Discussions

MiR-21 is an oncogenic miRNA that is significantly overexpressed in many tumors and is functionally associated with cellular proliferation, survival, and migration [30]. The MCF-7 cell line was taken as our experimental system as it expresses high endogenous levels of miR-21 [31]. For achieving functional readout of miR-21 inhibition, we used a dual luciferase screen. The reporter plasmid (psiCHECK-2 Vector; Promega) expresses the renilla gene fused with the miR-21 target sites and a firefly gene for normalization of signal accounting for variability in transfection (internal control). We compared the antisense potency of 2′-O-methyl RNA (AMO-1) and the serinyl capped and U^R-AMP interspersed 2′-O-methyl RNA (AMO-2) for their ability to suppress miR-21 levels and hence upregulate complementary target levels. We compared the potency of each AMO with respect to chemistry matched 2′-O-methyl scramble RNA to avoid variability caused by transfection itself. Furthermore, we also compared the AMOs to another 2′-O-methyl RNA with full PS-modified backbone, a first generation antisense oligonucleotide known to have nuclease resistance and high binding affinity toward target RNA [22]. The cells were transfected with reporter construct with varying concentrations of the AMOs (10–50 nM) and assayed for renilla and firefly luciferase intensity 24 h post-transfection. Transfection of the oligonucleotides resulted in antagonism of miR-21, as shown by specific increase in the levels of the relative target luciferase intensity (Fig. 2).

OPEN IN VIEWER

FIG. 2.

Comparison of the functional potency of antisense treatments against miR-21. Cells were transfected with different concentrations (10–50 nM) of scrambled control, PS modified, AMO-1, and AMO-2 for 24 h. The cell lysates were subsequently analyzed for luciferase signal of target compared to scrambled control. An increase in luciferase signal is observed for different oligonucleotides till 50 nM. AMO-2 shows a maximal increase in luciferase luminescence at 50 nM compared to AMO-1 and PS modified. Data shown are the mean ± SE of three independent experiments. *P < 0.01 and **P < 0.001 indicate significant statistical difference between the oligonucleotide transfected and scrambled control. All wells were normalized with the internal firefly control. PS, phosphorothioates; SE, standard error.

As evident from Fig. 2, an increase in firefly luciferase signal was observed upon transfection with AMOs at all concentrations. AMO-2 showed ∼6-fold increment of luciferase intensity at 50 nM concentrations with respect to 2′-O-methyl RNA scrambled control. PS-modified 2′-O-methyl RNA also showed an efficient derepression of miR-21. Among the 2′-O-methyl RNAs, AMO-2 was found to be the most potent in inhibiting miR-21 levels in all the tested concentrations. These AMOs were found to be functionally potent at 50 nM, thus affirming this concentration to be used for further experiments.

To assess the efficiency of AMO-1 and AMO-2 to inhibit miR-21, we checked the levels of mature miRNA by qRT-PCR 24 h post-transfection with 50 nM and compared the outcome with that of chemistry-matched scrambled control (Fig. 3). The ability to downregulate miR-21 was supreme for AMO-2 with significant decrease of mature miR-21 levels up to ∼95%. AMO-1 and PS-modified AMO showed ∼70% knockdown post-transfection for 24 h compared to scrambled control. AMO-2 was found to efficaciously repress miR-21 levels to a greater extent than AMO-1, which might be due to the fine tuning between high duplex stability and the stability to nucleases conferred by the interspersed U^R-AMP and the capping by serinyl units. The AMO-mediated repression was also measured after 48 h (Supplementary Fig. S4) showing ∼71% reduction when transfected with PS-modified AMO, ∼85% by AMO-1, and ∼92% by AMO-2, implicating sustained inhibition of miR-21 levels.

OPEN IN VIEWER

FIG. 3.

Reduction of miR-21 levels by various antisense oligonucleotides. AMOs designed against miR-21 were transfected at 50 nM concentration, and mature miRNA levels were measured by qRT-PCR after 24 h. Approximately, 95% inhibition of miR-21 is caused by AMO-2 compared to PS modified and AMO-1, which cause ∼70% downregulation. Data shown are the mean ± SE of three independent experiments with *P < 0.01 and **P < 0.0001 (Student's t-test). qRT-PCR, quantitative real-time reverse transcriptase–polymerase chain reaction.

Translational repression is considered as a major mechanism for miRNA regulation of target gene expression. Tumor suppressor, PTEN, is a principle target of miR-21 [32]. PDCD4 is an important validated target of miR-21, which is downregulated by miR-21 in many cancers [33]. Overexpression of miR-21 has an inverse correlation with its target PDCD4 and PTEN protein levels. Thus, the extent of suppression of miR-21 targets, PDCD4 and PTEN, was examined by western blot (Fig. 4 and Supplementary Fig. S5) to evaluate the antisense potency of the AMOs after 24 h of treatment. It was shown that the PDCD4 and PTEN levels were increased after transfection with modified AMOs compared to 2′-O-methyl scrambled control. The target levels were upregulated to a greater extent when treated with AMO-2 (with a P value <0.05) compared to scrambled control than AMO-1. However, the effects of derepression of targets were typically modest consistent with findings of recent high-throughput proteomic approaches [34,35]. Surprisingly, PS-modified RNA did not have an effect on the endogenous protein levels post 24 h.

OPEN IN VIEWER

FIG. 4.

(a) Western blot detection of PDCD4 levels in MCF-7 cell line in response to transfected AMOs. (b) Fold change analysis of the western blot image shows an inverse correlation of miR-21 target PDCD4 after transfection of AMOs after 24 h. The most potent AMO was AMO-2 as indicated by highest increase in PDCD4 levels upon treatment. GAPDH was used as the loading control. Data shown are the mean ± SE of three independent experiments. PDCD4, programmed cell death 4.

Tumor cell invasion is vital step for cancer cells to metastasize, and miR-21 is known to promote migration and proliferation in cancer cells [36]. We sought to envisage the extent to which knockdown of miR-21 by various AMOs would arrest the capacity of cancer cells to metastasize. For this, we took two different cell lines differing in their properties to metastasize, MCF-7 and MDA-MB-231. MDA-MB-231 has proved to be a highly metastatic model to study migration properties [37] and expresses high levels of miR-21 [36]. To establish that knockdown of miR-21 antagonizes its activity, we transfected each of the chemically modified RNAs and performed a scratch assay (Fig. 5 and Supplementary Figs. S6 and S7). After 24 h, we observed that the cells treated with AMOs had delayed migration compared to the untransfected cells and scramble transfected cells. Knockdown of miR-21 suppressed migration capacity by ∼19.4-fold in PS-modified AMO, ∼20-fold of AMO-1, and ∼33.4-fold in case of AMO-2 in MCF-7 cells compared to scrambled control. In MDA-MB-231, the suppression was ∼17.1-fold in PS modified, ∼16.2-fold in AMO-1, and ∼21.2-fold in case of AMO-2 compared to scrambled control. The difference between scrambled control versus PS modified and AMO-1 had a P value <0.05, whereas the difference between scramble treated and AM0-2 had a P value <0.001.

OPEN IN VIEWER

FIG. 5.

In vitro scratch assay in (a) MCF-7 cell line and (b) MDA-MB-231 cell line at 0 and 24 h after scratch. Knockdown of miR-21 by PS/AMO-1/AMO-2 suppresses the migration capacity to different extents compared to the scrambled control. Scratch repair in case of AMO-2 at 50 nM is maximally repressed compared to scrambled control after 24 h. Data shown are the mean ± SE of three independent experiments. *P < 0.05 and **P < 0.001 (Student's t-test).

The challenge of cytotoxicity has always remained an important point of consideration for chemically modified oligonucleotides for their use for therapeutic intervention. To examine this, we performed concentration-dependent cell viability assay from a range of 10 to 100 nM. We observed that there was no measurable cytotoxicity for AMO-1, AMO-2 even at 100 nM rendering them as suitable candidates for the development of potential therapeutics in future (Supplementary Fig. S8). However, PS-modified AMO showed significant toxicity up to ∼40% at 100 nM. This observation was consistent with previous studies that implicated cytotoxicity as one of the potential limitations of PS backbone [38,39].

Furthermore, we assessed the relative intracellular stability of these chemically modified AMOs by qPCR method. A comprehensive analysis of detection and quantification of modified siRNA using real-time polymerase reaction has been shown previously. Although the modified oligonucleotides are detected at lower efficiency, it does not limit the utility of the assay [40]. To determine whether reverse transcriptase is able to use modified oligonucleotide as template for cDNA synthesis, we polyadenylated all the different oligonucleotides, followed by oligo (dT) hybridization and reverse transcription to cDNA. We then compared the reverse transcription profiles obtained from in vitro synthesis of cDNA for each AMO. The standard curves were plotted for each AMO with Ct (threshold cycle) values versus concentrations of cDNA synthesized (Supplementary Fig S9a). The qPCR efficiency was determined from the slope of the log plot of concentration versus Ct value. The calculation for conversion of slope to efficiency was done using the formula: \documentclass{aastex}\usepackage{amsbsy}\usepackage{amsfonts}\usepackage{amssymb}\usepackage{bm}\usepackage{mathrsfs}\usepackage{pifont}\usepackage{stmaryrd}\usepackage{textcomp}\usepackage{portland, xspace}\usepackage{amsmath, amsxtra}\pagestyle{empty}\DeclareMathSizes{10}{9}{7}{6}\begin{document} \begin{align*} \% \ { \rm{ efficiency }} = { \rm{ }} ( { \rm{1}}{0^{ - { \rm{1}} / { \rm{slope}}}} ) \end{align*} \end{document}

By looking at the standard curve and slopes for each AMO, we found that AMO-1 and PS modifications were detected efficiently compared to AMO-2, which was detected although with poor yields. The serinyl capping present in AMO-2 is a non-nucleotide modification, but contains a free OH group at the 3′ end, which can be utilized for polyadenylation. Thus, we find amplification after qRT-PCR for AMO-2 with lesser yields compared to uncapped AMO-1 and PS-modified oligonucleotides (as confirmed by the Ct values), but this did not limit the utility of the assay to detect their intracellular levels. We also monitored the exponential amplification for the oligonucleotides following conversion to cDNA with antisense-specific forward primer and oligo (dT)-specific reverse primer by gel electrophoresis. As a negative control, we included an unmodified RNA without poly (A) tailing subjected to cDNA synthesis using same conditions. The PCR products were observed as a single band (Supplementary Fig. S9b). Once the assay was established, we transfected the oligonucleotides at 50 nM and measured their relative levels at 0, 12, and 24 h by qRT-PCR (Supplementary Fig. S9c). Post 24 h, AMO-2 was found to be most stable (∼68%) compared to AMO-1 (∼22%) and PS modified (∼38%) owing to the U^R-AMP-interspersed serinol capping. However, the transfection of AMOs did not alter the miR-21-3p (star strand) levels after 24 h (Supplementary Fig. S9d).

In summary, we have successfully designed and evaluated the effectiveness and antisense potency of uniformly substituted 2′-O-methyl RNA (AMO-1) and U^R-AMP-interspersed serinol capped 2′-O-methyl RNA (AMO-2) against miR-21. We demonstrate that the inhibitory efficacy of full 2′-O-methyl RNA is enhanced by serinyl capping at 5′/3′ ends and 2′-O-(R-2-amino-3-methoxypropyl) (2′-R-AMP) modification of uridine at three distinct positions. These modifications render the AMO-2 resistant to nucleases present in the cell thereby enhancing its intracellular stability while preserving the duplex stability, making it a more effective inhibitor of miR-21. There can be multiple mechanisms of miRNA inhibition, in response to transfection of AMOs [10]. The AMOs primarily act by sterically blocking the mature miRNA from binding to its target mRNA [41]. The other mechanisms include recruiting and triggering RNase H-dependent degradation of miRNA [42], or sequestering miRNA in P-bodies followed by its decay [14,43,44]. In comparing different chemistries of PNA-K3, LNA/DNA, or LNA/2′-OMe mixmer, Torres et al., observed that only 2′-O-methyl AMO was able to cause degradation of miR-122 in Huh cells, whereas high affinity K-PNA-K3, LNA/DNA, or LNA/2′-OMe mixmer AMOs did not lead to degradation of the target miRNA when bound in miRISC [14,43,44].

Conclusions

miRNAs are small noncoding RNAs that accomplish crucial roles in major biological processes. The development of anti-miR oligonucleotides as tools for deciphering miRNA functions or as probes for miRNA detection or as therapeutic modules necessitates the identification of optimal chemical modifications to endow better nuclease resistance, as well as enhancing potency. This novel 2′-R-AMP modification in uniformly 2′-O-methyl RNA with serinyl end caps is a simple and novel strategy that fine tunes the balance between binding affinity and nuclease resistance leading to highly potent inhibitor of miR-21, with low toxicity.