RNA G‑Quadruplex Reprogramming with Guanine-Rich Antisense Oligonucleotides Inhibits Monoamine Oxidase B’s Translation

Abstract

The human transcriptome contains secondary RNA structures like RNA G-quadruplexes (rG4s) which regulate biological processes such as translation by ribosome stalling. Canonical rG4s, which are stabilized by both Hoogsteen hydrogen bonds and potassium ions, are known to hinder translation in the 5′ untranslated region (5′UTR) of mRNAs. In neurodegenerative diseases, including Parkinson’s disease (PD), rG4s have been shown to influence protein synthesis. However, the impact of rG4s in nonmutated therapeutic targets like monoamine oxidase B (MAOB), an enzyme involved in dopamine metabolism, remains unexplored. In this study, an rG4 located in the MAOB mRNA’s 5′UTR was identified, and ways to either stabilize or reprogram this rG4 were explored. The translation inhibitory role of the rG4 was demonstrated both in vitro and in cellulo and was shown to be further accentuated in the presence of the PhenDC3 ligand. As an alternative to ligands, which cannot specifically stabilize only one G4, the MOAB rG4 was reprogrammed with G-rich antisense oligonucleotides (G-ASOs) from a two-quartets to three-quartets G4. The G-ASOs, either unmodified DNA or 2′OMe, were shown to both induce a new rG4 folding through intermolecular interactions and to specifically reduce the translation of MAOB both in vitro and in cellulo. These findings propose a targeted approach with which to modulate rG4 structures for therapeutics, suggesting that rG4 folding, when stabilized by G-ASOs, could regulate protein synthesis and even potentially alleviate PD symptoms by reducing MAOB activity. This approach opens new avenues as it could be used to reduce the expression of many therapeutic protein targets.

Introduction

Secondary structures are widely found in nucleic acids and are known to be involved in various biological processes such as replication, transcription and translation. For example, protein stalling occurs when structures exhibit high stability. , Among these structures, G-quadruplexes (G4s) are characterized by the superposition of guanine tetrads that are stabilized by Hoogsteen’s base pairing and the presence of monovalent cations, usually potassium (Figure A,B). G4s are generally considered to be canonical when they consist of the following motif: 5′-GGG-N_1–7-GGG-N_1–7-GGG-N_1–7-GGG-3′, where N represents the presence of any nucleotide in the loops. They can occur in DNA, as well as in RNA. In DNA, G4s are widely known for their implication in telomere maintenance and their regulation in promotors. In RNA, RNA G4s (rG4s) are generally more stable. They are found in various RNAs, such as mRNA, lncRNA and miRNA. In mRNA, depending on their location, rG4s can have different effects on translation, transcript degradation, polyadenylation and splicing. We and others have shown that this structure, when located in the 5′ untranslated region (5′UTR), primarily reduces RNA translation by inhibiting the progression of the 43S preinitiation complex. , A few rG4s located in the 5′ UTR have also been reported to increase translation if they are located within an internal ribosomal entry site. ,

1.

Schematic representation of a guanine tetrad, a G-quadruplex and a G-ASO. (A) Guanine tetrad. The circled M⁺ is a monovalent cation. The dashed green lines are hydrogen bonds. (B) Canonical G-quadruplex. The Gs represent the nucleotides implicated in the G4 structure, while the Ns represent random nucleotides involved in the loops. The K⁺s represent the potassium cations stacked between G-tetrads. (C) G-ASO targeting the MAOB mRNA. The G-ASO is divided into three color-coded regions: the green nucleotides represent the duplex region that forms Watson–Crick base pairs, the black nucleotides represent the guanines of the G-ASO that participate in the final stable rG4 structure, the blue nucleotides represent the G-ASO’s loop, the red circle indicates the G-ASO’s linker and N denotes a random nucleotide.

rG4s are pathogenic drivers in several neurodegenerative diseases, including Alzheimer’s disease, Amyotrophic Lateral Sclerosis, Frontal-temporal Dementia and Parkinson’s disease (PD). In Parkinson’s disease, we and others have identified the presence of new rG4s in multiple disease related transcripts. − More specifically, in the genes related to PD found in the Online Mendelian Inheritance in Man (OMIM) database, the G4s’ folding in the 5′ untranslated regions (5′UTRs) of LRRK2, PRKN, SNCA and VPS35 have been shown to repress translation in cellulo. , Most of the genes in the OMIM database have PD gene-phenotype relationships. However, rG4s in nonmutated or nonoverexpressed therapeutic targets still need to be investigated. The enzyme monoamine oxidase B (MAOB) is implicated in the catabolic processing of dopamine (Figure S1). This enzyme plays a crucial role in PD as this disease is characterized by the death of dopaminergic neurons, thereby decreasing dopamine levels. Thus, the enzyme’s activity further reduces the available pool of dopamine. The oxidation reaction is also responsible for the release of and the accumulation of peroxide, both of which contribute to the neuron’s death. Currently, only molecules targeting the enzyme’s active site have been approved by the US Food and Drug Administration (FDA) to target MAOB activity.

The recent approvals of a few RNA therapies by the FDA have led to the rise of research interest in these types of therapies by researchers in the RNA field. One way in which to modulate the folding of rG4s is by using small interacting molecules called G4 ligands. , Among the large library of G4 ligands, many can stabilize rG4s, and a few have been shown to destabilize them, thus they can modulate the rG4-dependent translation either positively or negatively. The primary limitation of small ligands such as PDS and PhenDC3 arises from their lack of specificity, as they target multiple G4 structures, leading to widespread cellular deregulation. , An alternative strategy to G4 ligands, called G-ASO, is to induce a unique rG4 folding using antisense oligonucleotides possessing 3′ or 5′ stretches of unpaired guanines (Figure C). According to this approach, a new rG4 can arise from an intermolecular rG4 folded via the unpaired guanines in the G-ASO and stretches of guanines located in the transcript. This RNA/G-ASO complex has been shown to reduce the translation of eIF-4E, eGFR and mutated MSH2, and the progression of the HIV reverse transcriptase. −

Herein, a new rG4 located in the 5′UTR of the MAOB mRNA was identified and investigated. This rG4 was found to impair translation both in vitro and in cellulo. Moreover, translational inhibition was further enhanced by the presence of the G4 ligand PhenDC3 and by that of two distinct G-ASOs. By comparing both strategies, G-ASOs were shown to be more effective and to specifically repress MAOB expression. This research highlights a new way to synthetically modulate MAOB levels using a G4 interactor. It also emphasizes the effectiveness and specificity of G-ASOs in promoting the folding of an rG4 at a defined position in the transcriptome.

Experimental Section

Bioinformatic Analyses

For the bioinformatic analyses, the MAOB transcript sequence was found on the NCBI RefSeq server. The presence of the G-quadruplex in the full-length MAOB transcript (NM_000898.5) was predicted using the G4RNA Screener tool. The basic parameters, a 60-nucleotides (nts) window and a 10-nts shift, were used. Consecutive windows with positive scores were merged, and the average score was designated as the mean of all windows. The structure predictions of both the 5′UTR and the antisense oligonucleotides (ASOs) were determined using the RNAfold software and are represented with the VARNA software. , Plasmid representation was made using Snapgene (Dotmatics).

In Vitro Transcription

To verify the formation of the rG4 in vitro, the T7 promoter sequence was inserted upstream of the 55- or 98-nts sense strand, and the corresponding double-stranded DNA sequence was prepared using two complementary oligonucleotides (Table ). The association of the oligonucleotides was performed in a solution containing 2 μM of the sense and antisense primers, 2 mM MgSO₄, 200 μM dNTP, 10 mM Tris-HCl pH 8.85, 25 mM KCl, 5 mM (NH₄)₂SO₄, 5% DMSO and pfu DNA polymerase that had been prepared “in house” (2 μL in a 100 μL total reaction volume). The solution was then incubated in a thermocycler at 95 °C for 2 min, followed by 10 cycles of 95 °C for 1 min, 60 °C for 1 min, and 72 °C for 1 min, and a final extension at 72 °C for 5 min. The DNA was then ethanol precipitated, and the resulting pellet dissolved in water.

1. Oligonucleotides Used for the Transcription of MAOB.

name	sequences
MAOB 55-nts sens WT	5′-TAATACGACTCACTATAGGGGGCAGCGCGCAGCAGGCCGGCGGGCAGGCGGGCGGGCTGG-3′
MAOB 55-nts antisense WT	5′-CCTGCCTGCCAGCCAGCCCGCCCGCCTGCCCGCCGGCCTGCTGCGCGCTGCCCCCTATAG-3′
MAOB 98-nts sens WT	5′-TAATACGACTCACTATAGGCTCGCCGAGGCGCTGGTGCACGGGGGCAGCGCGCAGCAGGCCGGCGGGCAGGCGGGCGGG-3′
MAOB 98-nts antisense WT	5′-TTCTGGGCCTCGATCCCAGTCCTGCCTGCCAGCCAGCCCGCCCGCC TGCCCGCCGGCCTGCTGCGCGCTGCCCCCGTGC-3′
MAOB 55-nts sens GA	5′-TAATACGACTCACTATAGGGAGCAGCGCGCAGCAGACCGACGAGCAGACGAGCGAGCTGA-3′
MAOB 55-nts antisense GA	5′-TCTGTCTGTCAGTCAGCTCGCTCGTCTGCTCGTCGGTCTGCTGCGCGCTGCTCCCTATAG-3′
MAOB 98-nts sens GA	5′-TAATACGACTCACTATAGGCTCGCCGAGACGCTGATGCACGAGAGCAGCGCGCAGCAGACCGACGAGCAGACGAGCGAG-3′
MAOB 98-nts antisense GA	5′-TTCTGGGTCTCGATCTCAGTTCTGTCTGTCAGTCAGCTCGCTCGTCTGCTCGTCGGTCTGCTGCGCGCTGCTCTCGTGC-3′

For transcription, a solution containing 2 μM of the double-stranded DNA, 6.25 mM rNTP, 80 mM HEPES-KOH pH 7.5, 24 mM MgCl₂, 2 mM spermidine, 40 mM DTT, 20 μg/mL pyrophosphatase (Roche) and a laboratory-prepared T7 RNA polymerase (2 μL in a 100 μL total reaction volume) was incubated at 37 °C for 3 h. The reaction was then stopped by the addition of 0.3 U/mL of DNase RQ1 (Promega) and then incubating the solution at 37 °C for 15 min. An equal volume of phenol:chloroform (1:1) was then added, and the solution was mixed and centrifuged at 17,000g for 5 min. The aqueous phase was then removed, the RNA ethanol precipitated, and the resulting pellet dissolved in water. Two volumes of loading solution (10 mM EDTA and 95% formamide) were added, and the final solution was electrophoresed on a 5% polyacrylamide (19:1)/8 M urea denaturing gel. The gel slices containing the RNAs with the correct sizes were then cut out of the gel and the RNA was eluted, ethanol precipitated and last dissolved in water. The RNA concentrations were determined by spectrometry at 260 nm using a NanoDrop Lite spectrophotometer (Thermo Fisher Scientific).

Fluorescence Assays

To detect the formation of any potential rG4, either the MAOB 55- or 98-nts RNAs were diluted to a final concentration of either 2 or 4 μM in a solution containing 20 mM Li-Cacodylate pH 7.5 and 100 mM of either KCl or LiCl. The resulting solutions were then heated at 70 °C for 5 min and then cooled at a rate of 1 °C/min to 20 °C. For the fluorescence experiments with N-methylmesoporphyrin IX (NMM), either 2.5 mol equivalents (with 2 μM RNA) or 1.25 mol equivalents (with 4 μM RNA) of NMM was added, and the solution was then excited at 399 nm at 700 V and the emission from NMM was measured between 550 and 650 nM on a Hitachi F-2500 fluorescence spectrophotometer. For Thioflavin T (ThT), 0.5 equiv of ThT was added (with 2 μM RNA), and the solution was then excited at 448 nm at 700 V and the emission from ThT was monitored between 450 and 600 nM on the same instrument. For experiments performed in the presence of ASOs, 1 equiv of the ASO was added to the solution before it was heated to 70 °C.

Radiolabeling and In-Line Probing

Before the radiolabeling of both the MAOB 55- and 98-nts RNAs, the transcribed RNAs were dephosphorylated using Antarctic phosphatase (according to the manufacturer’s instructions, New England Biolabs). The resulting RNAs were then phosphorylated with [γ³²P]-ATP (PerkinElmer) using T4 polynucleotide kinase (according to the manufacturer’s instructions, New England Biolabs). The radiolabeled RNAs were then electrophoresed and purified on denaturing polyacrylamide gels as described previously.

For in-line probing, 50,000 counts per minute (CPM) of the RNA solutions were dissolved in a solution containing 20 mM Li-Cacodylate pH 7.5 and 100 mM of either KCl or LiCl. The resulting solutions were then heated to 70 °C for 5 min and then cooled at a rate of 1 °C/min to 20 °C. The solutions were then adjusted to final concentrations of 20 mM Li-Cacodylate pH 7.5, 100 mM of either KCl or LiCl and 20 mM MgCl₂ before incubating for 40 h at room temperature. The RNAs were then ethanol precipitated, and the resulting pellets dissolved in 20 μL of loading solution (10 mM EDTA and 95% formamide). For the in-line reactions performed in the presence of the ASOs, 100 equiv of ASO were added prior to the RNAs being heated to 70 °C.

To perform the “NaOH” size marker, a solution containing 50,000 CPM of RNA and 0.3 N NaOH was incubated for 1 min at room temperature before being neutralized with 0.3 volumes of 1 M Tris–HCl pH 7.5. The RNA was then precipitated with ethanol and the resulting pellet dissolved in 20 μL of loading solution (10 mM EDTA and 95% formamide).

To produce the size marker “T1”, a solution containing 50,000 CPM of RNA, 0.06 U/μL of RNase T1 (Roche), 10 mM MgCl₂, 100 mM LiCl and 20 mM Tris–HCl pH 7.5 was incubated for 2 min at 37 °C before being neutralized with 2 volumes of loading solution (10 mM EDTA and 95% formamide).

An equal number of CPMs for each reaction and condition (5,000 CPM) was then loaded onto a 10% polyacrylamide (19:1)/8 M urea denaturing gel. After electrophoresis, the gel was dried and exposed to a phosphor screen for 16 h. The results were visualized on a Typhoon imaging system (GE Healthcare), and the quantification of the bands was carried out using the SAFA software. The hydrolyzation ratio was calculated by dividing the intensity of the KCl band by that obtained in the presence of LiCl.

Circular Dichroism and Thermal Denaturation

To determine the secondary structure of the RNA sequences by circular dichroism (CD), the 55-nts RNAs were dissolved at a concentration of 4 μM in a solution containing 20 mM Li-Cacodylate pH 7.5 and 100 mM of either KCl or LiCl. The samples were then heated to 70 °C for 5 min and allowed to cool at room temperature for 1 h. The experiments were performed with a Jasco J-810 spectropolarimeter using a quartz cuvette with a 1 mm optical path. CD spectra were recorded between 350 and 210 nM at a rate of 50 nm per minute, a response time of 2 s, a height of 0.1 nm and a bandwidth of 1 nm. CD experiments were performed at room temperature, except for thermal denaturation where the temperature ranged from 20 to 90 °C at a rate of 1 °C/min. In the case of thermal denaturation, the CD signal was monitored at 263 nm.

In Vitro Translation

In order to perform the in vitro translation assays, either the WT or the GA 5′UTR of MAOB (Biobasics) containing a 5′ SpeI sequence and a 3′ SalI sequence was inserted into the SpeI and SalI sites of the PsiCheck 2.1 plasmid as previously described (Figure S2). The sequence of the MAOB’s 5′UTR in the RNA is presented Table . The PsiCheck 2.1 plasmid containing the MAOB 5′UTR was digested with the enzyme XhoI (NEB) prior to transcription. The RNA containing the MAOB’s 5′UTR fused to the Renilla luciferase sequence was then diluted to 12 nM in a solution containing 20 mM Li-Cacodylate pH 7.5, 100 mM KCl and 1.2 μM of ASO (when necessary). The resulting solution was then heated at 70 °C for 5 min and then cooled at a rate of 1 °C/min to 20 °C. Subsequently, for 9.75 μL of the heated solution, 0.375 μL of a solution containing an amino acid mixture minus leucine (1 mM; Promega), 0.375 μL of a solution containing an amino acid mixture minus methionine (1 mM; Promega), 8.75 μL of rabbit reticulocyte extract nuclease treated (Promega) and 0.75 μL of RNase inhibitor (homemade) were added to the solution before being incubated at 30 °C for 90 min. The reactions were stopped by the addition of 3 μL of RNase at a concentration of 830 μg/mL and incubating for 5 min at room temperature. The reactions were then placed on ice for 10 min. Luciferase signals were evaluated with the Dual-Luciferase Reporter Assay System and a Glomax 20/20 luminometer (Promega), using 75 μL of each reagent for each 10 μL of in vitro translation reaction.

2. RNA Sequences Used in This Study .

^aThe red nucleotides represent the G-to-A mutations present in the mutants, the underlined regions represent the MAOB 55-nts construct and the bold nucleotides represent the consecutive guanines.

For experiments performed in the presence of 100 equiv of PhenDC3, the PhenDC3 ligand was added to the solution after the cooling to 20 °C. The solution was then incubated for 5 min at 30 °C prior to the addition of the reaction elements required for the in vitro translation.

Luciferase Assays

To detect the effect of G-quadruplexes on translation, SH-SY5Y cells were seeded to be at a 70% confluence level at the time of transfection. The cells were cotransfected in a 24-well plate with 0.5 μL of Lipofectamine 2000, 50 ng of a PsiCheck 2.1 construct that contained the MAOB’s 5′UTR (Table ) located 5′ of the Renilla luciferase gene and 500 ng of pUC19 (according to the manufacturer’s Lipofectamine 2000 protocol, ThermoFisher Scientific). After a 24 h incubation at 37 °C, the luciferase signal was assessed with the Dual-Luciferase Reporter Assay System kit and a Glomax 20/20 luminometer (according to the manufacturer’s protocol, Promega).

Cotransfection of MAOB and the ASOs

To detect the effects of G-ASOs on MAOB’s translation, HEK293T cells were seeded to be at a 70% confluence level at the time of transfection. The cells were transfected in a six-well plate with 5 μL Lipofectamine 2000, 200 nM ASO and 750 ng of a PCDNA3 construct containing the full-length MAOB coding sequence and both of its UTRs (Biobasics). After 24 h of incubation at 37 °C, the proteins were extracted with lysis buffer as described previously. Proteins were then assayed using the DC Protein Assay (BioRad), electrophoresed on SDS-PAGE 10% gels and wet transferred to a nitrocellulose membrane. Blocking and washing were then performed as described previously. Anti-MAOB (ab133270; Abcam), antiactin (A5441; Sigma-Aldrich), antimouse 680 (A-21057; ThermoFisher Scientific) and antigoat 800 (926–32213; LI-COR) antibodies were used at dilutions of 1:5000, 1:10,000, 1:10,000, and 1:10,000, respectively. The signal was revealed by fluorescence on an Odyssey (LI-COR).

Results and Discussion

Investigation of Potential G-Quadruplexes in MAOB

In order to investigate the potential G-quadruplexes (pG4) found in the MAOB mRNA as possible targets for translational inhibition, its mature transcript (NM_000898.5) was screened using the G4 predictor G4RNA Screener. G4RNA Screener uses three scores, cGcC and G4 Hunter (G4H), which are sequence-based, and G4 Neuronal Network (G4NN), which has been trained with RNA G4s. ,, Only one region of the MAOB mRNA was found to be positive for all three scores (e.g., cGcC > 4.5, G4H > 0.9 and G4NN > 0.5; Figure A,B). This region spans nucleotides (nts) 21 to 91 that are located within the 151-nts of the 5′UTR. Using QGRS Mapper with a maximum loop length of 7 nucleotides, 155 unique rG4 sequence combinations were found, each of which was formed by two quartets of guanines. Thus, the potential rG4s found are likely noncanonical and composed of a mix of rG4s. Other potential rG4s with one or two positive scores were also found (Figure A and Table S1). However, these were not investigated further since they were in either the coding sequence (CDS) or in the 3′UTR, which makes them less attractive for targeting translation.

2.

Prediction of potential rG4s located in the MAOB mRNA using the G4RNA Screener tool. (A) G4NN, G4H and cGcC scores obtained throughout MAOB’s transcript. The major transcript’s regions (5′UTR, CDS and 3′UTR) are delimitated. The asterisks highlight the potential G4 of interest. (B) Characteristics and sequence of the potential rG4 identified in the 5′UTR of the MAOB mRNA. The consecutive guanines are highlighted in green.

In order to confirm the likelihood of a G-quadruplex forming in the region spanning positions 21 to 90 of the 5′-UTR of the MAOB mRNA, the structure of the full 5′UTR was predicted using the RNAfold software. For this prediction, the entire 5′UTR sequence was used to remain faithful to the global structural context of the mRNA. Initially, only Watson–Crick interactions were predicted by RNAfold using the standard parameters and revealed the presence of a four-way junction structure. These four duplex regions (called stem-loop 1, stem-loop 2, stem-loop 3 and duplex 4) converge toward a single point and have a cumulative minimal free energy of – 69.8 kcal/mol (Figure S3A). In a second step, the possibility of forming G4 structures was then added to the RNAfold prediction parameters, along with Watson–Crick interactions. Here, the algorithm predicted a similar four-way junction with a potential two quartet G-quadruplex located at the end of stem-loop 3 with an energy of – 72.5 kcal/mol (Figure S3B). This G4 predicted by RNAfold is situated in the window predicted by G4RNA Screener. Apart from small modifications (stem-loop 3 and the beginning of stem-loop 2), G-quadruplex formation did not induce a global restructuring of the 5′UTR. The formation of this rG4 modestly stabilizes the structure by 2.7 kcal/mol, resulting in a 4% increase in stability for the 5′UTR. However, when stem-loop 3 is analyzed independently, the algorithm predicts a greater stabilization due to the rG4, specifically from −16.6 to −20.0 kcal/mol, reflecting a significant stabilization of 20.5%. This stem-loop stabilization could be an important factor in halting ribosomal progression. However, the G4’s folding, and its impact on translation, must first be confirmed both in vitro and in cellulo.

Confirmation of the 5′UTR MAOB mRNA’s rG4’s Folding In Vitro

To validate the rG4’s prediction, its folding was evaluated using two different techniques. It is known that the biological context surrounding an rG4 can impact both its folding and its stability. Since smaller RNA constructions are generally recommended for analyzing biophysics data, two different RNA constructions were synthesized. The first was an RNA construct of 55-nts in length spanning from the stretch of five consecutive guanines located in the 70-nts pG4 windows to the last guanine doublet (Figure B and Table ). The second was a construct of 98-nts in length that was composed of the full 55-nts of the first construct with 23 and 20-nts added in 5′ and 3′, respectively (Table ). Both constructs correspond to the wild-type (WT) sequence. G-to-A mutants (G/A) of these constructs, which are unable to form an rG4, were also designed and were used as negative controls.

First, the structure was studied by fluorescence spectroscopy using both N-Methylmesoporphyrin IX (NMM) and Thioflavin T (ThT), two well-characterized fluorescent G4 ligands. , When incubated with either the 55- or the 98-nts WT constructs, both fluorescent probes give a higher signal in the presence of potassium as compared to that observed in the presence of lithium (Figure A,B). More precisely, the signal for the 55-nts RNA went from 250 to 1290 au with NMM, and from 404 to 2130 au with ThT, which represents a 5.2- and a 5.3-fold change favoring the potassium condition, respectively. A similar trend was observed for the 98-nts RNA, where the signal went from 160 to 1180 au with NMM and from 860 to 1230 au with ThT, which represents a 7.4- and a 1.4-fold change favoring the potassium condition, respectively. Except for the weak 1.4-fold change of the 98-nts RNA with ThT, this change due to the salt employed has already been described as being indicative of the presence of an rG4 structure. A signal reduction was observed for both constructs with the different G/A mutants as compared to what was observed with the WT, confirming the abolishment of the rG4s when the guanines are mutated (Figure S4A and B). Although the difference between positive and negative rG4s is clear with the NMM, this effect is less distinctive with both ThT and the WT 98-nts RNA. This could be attributed to ThT’s nonspecific binding to RNA, or to the presence of other structures. Globally, both ligands support the formation of the rG4.

3.

Confirmation of the 5′UTR MAOB mRNA’s G-quadruplex folding in vitro. (A) Fluorescence spectra of 2 μM of MAOB mRNA incubated with 5 μM of NMM. Both WT RNA constructs were tested with 100 mM of either KCl or LiCl. (B) Fluorescence spectra of 2 μM of MAOB mRNA incubated with 1 μM of ThT. Both WT RNA constructs were tested with 100 mM of either KCl or LiCl. (C) In-line probing autoradiograms of both the 55- (left) and 98-nts (right) RNAs WT and G/A mutant constructs. The reactions were analyzed on 10% polyacrylamide gels. The “OH” lane represents the NaOH reaction, while the T1 lane is the T1 ribonuclease reaction. The guanines and conditions are identified for each gel. The guanine stretches are indicated in bold and are underlined in the sequence to facilitate their visualization.

Second, the structure was studied using in-line probing. In-line probing is particularly useful with G-quadruplexes since it can identify which nucleotides participate in the loops and which ones in the guanine core. Since the MAOB mRNA sequence contains multiple stretches of guanines, this technique should be useful in identifying those that are engaged in the rG4. Contrary to either fluorescence spectroscopy or to biophysical techniques like circular dichroism and NMR, in-line probing uses only low nanomolar concentrations, thus reducing the probability of the formation of intermolecular G-quadruplexes. To study the MAOB mRNA’s G-quadruplexes, the RNA was incubated for 40 h at room temperature with a high concentration of magnesium (e.g., 20 mM) in a buffer containing 100 mM of either KCl or LiCl. rG4s are characterized by an increased reactivity of the nucleotides located in the loops for the potassium buffer condition, resulting in an increased level of cleavage. For both RNA constructs, an increased amount of cleavage can be observed on the autoradiograms for some of the nucleotides located between the guanine stretches (Figure C). More precisely, C₅₉, C₆₃ and C₆₇ were cleaved more frequently for the 55-nts RNA, and C₆₃, C₆₇, C₇₁, T₇₂, C₇₅, T₇₆, C₇₉, A₈₀, C₈₃ and A₈₄ more frequently for the 98-nts RNA. In order to support these results, the KCl/LiCl cleavage ratios were also calculated using the SAFA software (Figure S4B). The ratios display a trend like that observed on the autoradiograms, although with some minor differences. These findings suggest the presence of a single noncanonical rG4 structure for the 55-nts transcript, and of an equilibrium of multiple noncanonical rG4s for the 98-nts transcript. They highlight both the importance and the effect of flanking sequences on rG4 formation. The finding of noncanonical rG4s here is also of importance, since they are generally less stable and are more likely to be further stabilized by different techniques.

Bioinformatic predictions always need to be confirmed either in vitro or in cellulo. Here, the window predicted by the G4RNA Screener indeed folds an rG4 (Figure ). However, when compared to the RNAfold prediction where G₅₃-G₅₄, G₅₆-G₅₇, G₆₁-G₆₂ and G₆₄-G₆₅ were predicted to participate in the rG4 (Figure S3), the in-line probing results identified G₅₆-G₅₇ (or G₅₈), G₆₁-G₆₂, G₆₄-G₆₅ (or G₆₆) and G₆₈-G₆₉ (or G₇₀) for the 55-nts RNA and an equilibrium between G₆₁-G₆₂, G₆₄-G₆₅ (or G₆₆), G₆₈-G₆₉ (or G₇₀), G₇₃-G₇₄, G₇₇-G₇₈, G₈₁-G₈₂ and G₈₅-G₈₆ for the 98-nts construct (Figure S4B). The exact guanines participating in the formation of a two quartet rG4 G-quadruplex are harder to identify when stretches of three guanines are also present. These results demonstrate the importance of confirming in silico results.

Effect of the MAOB mRNA’s 5′UTR’s G-Quadruplex on Translation

To elucidate the effect of the MAOB mRNA’s 5′UTR’s G-quadruplex on translation, a cell-free translation system was used. Cell-free translation systems such as rabbit reticulocyte lysate (RRL) are particularly convenient for rG4 studies because the direct impact of the rG4 on translation can be investigated. , To proceed, the 5′UTR of either the MAOB WT mRNA, or that of its G/A mutant, was cloned upstream of a Renilla luciferase gene located in a PsiCheck2 plasmid (Table and Figure S2). The construction was then digested, and an in vitro transcription was performed to obtain an MAOB 5′UTR-Renilla luciferase fusion transcript. The rG4 effect was evaluated in RRL after RNA folding. The luminescence results showed that the WT RNA was about 5-fold less effective for translation as compared to the G/A mutant (Figure A), indicating that the presence of the rG4 negatively impairs translation in vitro. Other examples in the literature have indicated that these 5′UTR rG4s can repress translation. Thus, this outcome was therefore not altogether unexpected.

4.

Translational impact of the MAOB mRNA’s 5′UTR’s G-quadruplex. (A) In vitro translation in RRL of RNA containing either the native or the mutant MAOB’s 5′UTR fused to Renilla luciferase. Statistical significance was tested using a t test. (B) Left axis: Normalized RLUC/FLUC ratios of luciferase activities performed in SH-SY5Y cells. The PsiCheck 2.1 plasmid, either empty or containing either the WT or the G/A mutant 5′UTR of MAOB fused with Renilla luciferase, was used. Firefly luciferase was used as an internal control. Right axis: Digital PCR quantification of RNA ratios between Renilla and Firefly luciferase for the different constructs. Statistical significance was tested using two-way ANOVA with Tukey’s multiple comparison test. (***) P < 0.001 and (****) P < 0.0001.

As the in vitro assays were performed at a lower temperature (30 °C), originated from a different species and resulted from uncapped RNA translation, the effects of the G-quadruplexes can differ from those observed in cells. Therefore, the Psicheck2 plasmids (either empty, WT or G/A mutant MAOB 5′UTR) were transfected into SH-SY5Y cells, a neuroblastoma cell line often used for MAOB studies (Figure S2). , The luminescence results were normalized to the Firefly luciferase control. A day after transfection, the in vitro observations were reproduced in cells, although at smaller intensities. Indeed, the WT construction was about 1.5 times less effective for translation (about 30%) as compared to both the G/A mutant and the empty plasmids (Figure B). Moreover, RT-ddPCR results confirmed that this effect was not due to different transcriptional levels between the different constructs (Figure B). These findings suggest that the folding of the rG4 negatively impacts the translation.

Stabilization of the MAOB mRNA’s rG4 with Ligands

The previously identified MAOB mRNA’s G-quadruplex was shown to repress translation. Given that the MAOB sequence lacks the nucleotides required for a canonical G-quadruplex, and the fact that noncanonical G-quadruplexes are generally less stable, it was hypothesized that the addition of a G4 ligand could further reduce translation. In order to test this hypothesis, the interaction between the MAOB mRNA’s G-quadruplex and PhenDC3, a well-characterized G4 ligand, was examined using circular dichroism (CD). PhenDC3 was selected as it has previously been shown that this ligand is one of the best for inducing the formation of two quartet rG4s. These assays were performed using the 55-nts transcripts described previously to reduce interference from other structures which could bury the rG4 signal. While this 55-nts sequence is shorter than the full-length 98-nt one, a mixture of conformations was still anticipated due to the presence of more nucleotides than the minimum that is required for rG4 formation. The CD spectrum obtained in the presence of PhenDC3 displayed an increase in the peak at 263 nm which is a hallmark of parallel rG4 formation. This increase was found to be absent in the G/A mutant’s spectrum (Figures A and S5A). Additionally, a shift from a negative peak at 210 nm to a positive one supported the conversion from A-form RNA to a G-quadruplex structure, although this transition was also unexpectedly observed with the G/A mutant. Subsequently, RNA stability was assessed by thermal denaturation while monitoring the CD signal at 263 nm. The addition of 1, 2, or 5 mol equiv of PhenDC3 raised the melting temperature by 0 °C, 3 and 12 °C for the WT RNA, respectively (Figure B). This stabilization effect was not observed in the G/A mutant, indicating that PhenDC3 can stabilize the MAOB mRNA’s 5′UTR’s rG4 (Figure S5B).

5.

Stabilization of the MAOB mRNA’s G-quadruplex with PhenDC3. (A) Circular dichroism spectra of 4 μM of the 55-nts RNA incubated with 0, 1, 2, or 5 equiv of PhenDC3. (B) Normalized melting curves of 4 μM of the 55-nts RNA incubated with 0, 1, 2, or 5 equiv of PhenDC3. (C) In vitro translation in RRL of RNA containing either the native or mutant 5′UTR of the MAOB mRNA fused to Renilla luciferase and 750 nM of PhenDC3. NT represents the no ligand condition. NT conditions have been normalized to 1. Statistical significance was tested using two-way ANOVA with Sidak’s multiple comparison test. (***) P < 0.001 and (****) P < 0.0001.

Next, whether PhenDC3 could also induce translational repression was assessed. Since the use of G4 ligands in cells can induce unspecific deregulation, the repression was evaluated in vitro. RRL assays were conducted, as previously mentioned, in the presence of PhenDC3 which was added to the reaction following RNA folding. Compared to the untreated control, the addition of 750 nM of PhenDC3 resulted in a 60% reduction in translation of the WT construct as compared to what was observed in its absence, while the G/A mutant exhibited only a 30% decrease (Figure C). Thus, PhenDC3 significantly impaired translation in the presence of the MAOB mRNA’s 5′UTR’s rG4. The partial reduction in translation observed with the G/A mutant may be due to the folding of other rG4 structures within the RNA. Taken together, the real effect on the desired G-quadruplex is likely to be around 30%, which corresponds to the WT effect minus the G/A mutant effect. This finding supports the ability of PhenDC3 to both stabilize the two quartet rG4 of the MAOB mRNA and to decrease translation, even though it displays a lack of specificity.

Specific Reprogramming of the MAOB mRNA’s 5′UTR’s rG4

To specifically induce a stronger inhibition of translation, the MAOB mRNA’s 5′UTR’s rG4 was targeted with G-ASOs. Briefly, G-ASOs are antisense oligonucleotides that possess either 5′ or 3′ unpaired guanine stretches capable of folding an intermolecular G-quadruplex with stretches of guanines located in the mRNA. − As the G4 sequence of the mRNA is comprised of a GGG-C-GGG motif, this candidate becomes particularly interesting for this strategy (Figure A). Contrary to previous work where the function of the MSH2 5′UTR G4 was restored with G-ASO following a guanine-to-adenine substitution in the G4, the strategy here was to reprogram the MOAB rG4 from a two-quartets to a three-quartets G4. Thus, multiple DNA G-ASOs with different loop sizes (1 or 3-nts), linker sizes (0, 1, 3 or 5-nts) and orientations (two stretches of guanines in 5′ or 3′) were designed and tested using RRL assays (Figure A). Two controls were also tested, the first one being 20 random nucleotides, and the second being the duplex region only. G-ASOs with guanines located in the 5′ position were particularly inefficient in repressing translation. This effect is probably due to the displacement of the duplex region before contact between the small ribosomal subunit and the G-quadruplex. Conversely, for those with guanines located in the 3′ position, the candidates with a loop of one nucleotide were found to be more efficient in reducing translation. G4s with longer loops are known to have their stabilities negatively affected by G-ASOs. Varying the linker size between 1 and 5 nucleotides did not affect the translational efficiency. The duplex only, D_Duplex, was not a good candidate with which to repress translation, suggesting that the addition of stretches of guanines is responsible for the observed phenotype. The candidate D_3 was therefore selected for further investigation.

6.

Translational effect of DNA G-ASOs. (A) Visual representation of the G-ASOs used in (B). The bold guanines are those targeted on the MAOB mRNA. The green nucleotides are complementary to the MAOB mRNA, the red nucleotides form the link between the duplex region (in green) and the guanine series, and the blue nucleotides form the G4 loop on the G-ASO. N20 is the DNA sequence 5′-NNNNNNNNNNNNNNNNNNNN-3′, where N can be A, C, G or T. (B) Incubation of the MAOB 5′UTR Renilla luciferase fusion RNA in RRL extracts in the presence of DNA ASOs. The final luciferase signal from each condition was normalized to the signal from the N20 DNA. Statistical significance was tested using one-way ANOVA with Dunnett’s multiple comparison test. (****) P < 0.0001.

Naked DNA oligonucleotides are usually not used in cells due to their susceptibility to DNase cleavage. Therefore, D_3 was fully synthesized with more resistant 2′-O-methyl (2′OMe) nucleotides. 2′OMe_3, the D_3 converted to 2′OMe, showed greater inhibition of translation in the RRL system than did D_3, even though the duplex alone, 2′OMe_Duplex, also demonstrated a reduction in the translation (Figure S6). This latter result was expected because of the high energy of the RNA-2′OMe interaction. Next, it is known that the tetramolecular G4 with the 2′OMe UG₄U oligonucleotide forms faster than DNA but slower than RNA. However, since the 2′OMe modification in the G-ASO hybrid context is poorly understood, the ability of 2′OMe to fold a two-strand intermolecular G-quadruplex was evaluated by both NMM fluorescence assays and in-line probing. Fluorescent assays showed a 3X increase in the signal, which supports the idea that the original G-quadruplex has been reprogrammed into a new G-quadruplex (Figure S7A). This effect was not observed with the duplex alone, nor with a GAG mutant (2′OMe_GAG), where the two stretches of three guanines have been substituted by GAG (Figure S7A). For in-line probing, the experiment was performed as previously described, but with different ASO constructions. Because C₆₇ is the only nucleotide located in a loop in the radioactively labeled MAOB mRNA, it was expected that the nucleotide C₆₇ would become more accessible for hydrolysis. Indeed, a large increase in intensity was observed for this nucleotide, further supporting the reprogramming (Figure S7B). The band intensities for the rest of the gel remained unchanged.

Application of the G-ASO Tool in Cells

The results of the in vitro experiments prompted the study of the effect of the G-ASO 2′OMe_3 on MAOB translation in cells. As a proof-of-concept for G4 reprogramming applicability, the G-ASOs were cotransfected into HEK293T cells with a PCDNA3 plasmid coding for the entire MAOB transcript (NM_000898.5; Figure A). HEK293T cells were used as no basal MAOB protein level was detected in this cell line (data not shown). Hence, only newly synthesized MAOB protein would be evaluated in these experiments. A day after transfection, the proteins were extracted and quantified by Western Blot. The cotransfection of 200 nM of the G-ASO 2′OMe_3 showed a reduction of 34% in the MAOB protein level (Figure B). This diminution was significant when compared to that observed for 2′OMe_Scr, but not when compared to either the Duplex or the GAG mutant control. Thus, 2′OMe_3 seems to impact MAOB translation, but only weakly in cellulo as compared to what was observed in the in vitro studies.

7.

Cellular effect of G-ASOs on MAOB protein levels. (A) Visual representation of the G-ASOs used in (B). The bold guanines are those targeted on the MAOB mRNA. The green nucleotides are complementary to the MAOB mRNA, the red nucleotides form the link between the duplex region and the guanine series, and the blue nucleotides form the G4 loop on the G-ASO. (B) Quantification of the MAOB protein levels by Western blot of the cotransfection of antisense oligonucleotides with a PCDNA3MAOB construct in HEK293T cells. The data were normalized to actin and then to 2′OMe_Scr. Statistical significance was tested using one-way ANOVA with Tukey’s multiple comparison test. (*) P < 0.05, (**) P < 0.01 and (***) P < 0.001.

General ASO recommendations advise the use of ASOs with an intramolecular energy higher than – 1.1 kcal/mol. For 2′OMe_3, the predicted energy by RNAfold of the RNA G-ASO is about – 7 kcal/mol. The real intramolecular energy is probably stronger since 2′OMe modified oligonucleotides are used instead of RNA. We are unaware of any RNA structure predictor accounting for 2′OMe modifications. One of the challenges with 2′OMe_3 could be the location of the binding region selected for the duplex. Since the region selected is guanine-rich, the cytosine of the duplex has a high probability of interacting with the added stretches of guanines. To minimize the intramolecular energy of the G-ASO, all of the predicted energies of the G-ASOs with a duplex at the same position but with a loop between 1 and 3 nucleotides and a linker between 1 and 5 nucleotides were screened with RNAfold. This led to the identification of 19 449 ASO possibilities with intramolecular energies lower than – 7 kcal/mol (Figures S8A and B). Among them, a G-ASO with a lower energy (LE) of – 3.5 kcal/mol was selected (referred to as 2′OMe_LE; Figure A). This new promising G-ASO was tested and was found to reduce translation in the RRL extract at a level like that of 2′OMe_3 (Figure S9A). 2′OMe_LE has a similar fluorescence to that of 2′OMe_3 when incubated with NMM, supporting a similar formation of an rG4 (Figure S9B). However, the accessibility of C₆₇, as evaluated by in-line probing, was lesser with 2′OMe_LE than with 2′OMe_3, potentially because of the longer loop, but it was still higher than that of the RNA alone (e.g., the MAOB mRNA’s 5′UTR‘sG-quadruplex; Figure S9C). No other variations in band intensity were observed. The 2′OMe_LE and MAOB plasmid cotransfection showed a 50% decrease in the MAOB protein level. This time, the reduction was significantly lower than the scramble, Duplex and GAG mutant control, supporting the impact of the 2′OMe_LE G-ASO. The effect was also 20% stronger than that of the 2′OMe_3, although not significantly, which supports the idea that the intramolecular energy is an important parameter for G-ASO design. In summary, these results show that 2′OMe_LE is a G-ASO capable of significantly repressing MAOB translation and could further be used to reduce MAOB protein levels therapeutically.

Here, a liposomal method was used to transfect cells with G-ASOs. For a broader application of this method, the key challenges associated with G-ASO delivery stay the same as the ASO ones. As MAOB needs to be downregulated in the brain, this challenge grows. The ASOs now need to cross the brain-blood barrier. That said, various methods are still available, including passive diffusion, intracerebroventricular injection, encapsulation in lipid nanoparticles or exosomes and conjugation to cell-penetrating peptides. Upon confirmation that the modification does not affect G4 folding, new types of modifications, like phosphorothioates, could also improve the G-ASO uptake. Other challenges related to toxicity also remain, such as evaluating if the G-ASO induces nonspecific targeting or if the chemistry selected induces an immune response.

Conclusions

This study demonstrates that the MAOB transcript contains a noncanonical two quartet rG4 that is located within its 151-nts 5′UTR. Through both in vitro and in cellulo experiments, it was confirmed that this rG4 structure represses translation by about 80% in vitro and 30% in cellulo. The inhibitory effect is further enhanced by 60% in vitro in the presence of the rG4-stabilizing ligands PhenDC3, which showed nonspecific stabilization, and by 98% with 2′OMe_LE, which showed specific stabilization. In cells, the development of the new G-ASO 2′OMe_LE led to a 50% decrease in translation, offering an alternative strategy with which to correct the nonspecificity problems of G4 ligands. This work highlights the specific translational inhibition achieved by targeting this noncanonical rG4 with G-ASOs, which selectively reduces the MAOB protein level in a cellular model. Further work will be necessary to evaluate if 2′OMe_LE can indeed reduce the endogenous MAOB protein levels in neuronal cells and in vivo and thus contribute to the decreasing of the oxidative stress and the increasing of the dopamine pool.

These findings support the therapeutic potential of G-ASOs for modulating gene expression through G-quadruplex folding. Importantly, this study opens avenues for developing RNA-targeting therapies with greater specificity, especially in cases where global G4 ligand treatments may lead to unintended off-target effects. This technique could also be applied to other challenges, like targeting rG4s in viruses to reduce their replication level. Further studies could refine G-ASO design to optimize both binding specificity and stability, thus advancing G4-based therapeutic strategies, especially for conditions like Parkinson’s disease where MAOB activity is a key therapeutic target.

Glossary

Abbreviations

2′OMe

2′-O-methyl

5′UTR

5′ untranslated region

CD

circular dichroism

FDA

Food and Drug Administration

G4H

G4 Hunter

G4NN

G4 neuronal network

G-ASO

G-rich antisense oligonucleotides

G/A

G-to-A mutant

NMM

N-methylmesoporphyrin IX

PD

Parkinson’s disease

rG4

RNA G-quadruplex

ThT

Thioflavin T

WT

wildtype

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acsbiomedchemau.5c00004.

• G4RNA Screener detailed analysis, MAOB-dependent dopamine degradation pathway, Plasmid representation, RNAfold prediction, In-line probing of controls and G-ASOs, NMM and ThT fluorescent assays of controls and G-ASO, Circular dichroism of controls, In vitro translation of 2′OMe and prediction of stable G-ASO (PDF)

This project was supported by grants from the Natural Science and Engineering Research Council of Canada (NSERC; RGPIN-2023-04178 to J.-P.P.). M.A.T. received student fellowships from the Fonds de Recherche Québec Nature et Technologie (FRQNT) and the Canadian Institutes of Health Research (CIHR). J.P.P. holds the Research Chair of the Université de Sherbrooke in RNA Structure and Genomics and is a member of the Centre de Recherche du CHUS. The funders had no role in study design, data collection and analysis, the decision to publish, nor in the preparation of the manuscript.

The authors declare no competing financial interest.