|Home | About | Journals | Submit | Contact Us | Français|
Telomeres protect chromosome ends from being recognized as double-stranded breaks. Telomeric function is ensured by the shelterin complex in which TRF2 protein is an essential player. The G-rich strand of telomere DNA can fold into G-quadruplex (G4) structure. Small molecules stabilizing G4 structures, named G4 ligands, have been shown to alter telomeric functions in human cells. In this study, we show that a guanine-rich RNA sequence located in the 5′-UTR region of the TRF2 mRNA (hereafter 91TRF2G) is capable of forming a stable quadruplex that causes a 2.8-fold decrease in the translation of a reporter gene in human cells, as compared to a mutant 5′-UTR unable to fold into G4. We also demonstrate that several highly selective G4 ligands, the pyridine dicarboxamide derivative 360A and bisquinolinium compounds Phen-DC(3) and Phen-DC(6), are able to bind the 91TRF2G:RNA sequence and to modulate TRF2 protein translation in vitro. Since the naturally occurring 5′-UTR TRF2:RNA G4 element was used here, which is conserved in several vertebrate orthologs, the present data substantiate a potential translational mechanism mediated by a G4 RNA motif for the downregulation of TRF2 expression.
In vitro, single-stranded G-rich nucleic acid sequences can adopt an unusual conformation called a G4-quadruplex (G4) (1). G4 structures result from the successive stacking of G-quartet formed by four guanines located on a plan and interacting by Hoogsteen hydrogen bonds (2). The formation of G4 structures is strongly dependent on monovalent cations such K+ and Na+ and, hence, physiological buffer conditions favor their formation (3). G4 structures can be formed by one or more nucleic acid strands. Numerous physical analyses have revealed that G4 structures are highly polymorphic and can be subgrouped into various families, such as parallel or antiparallel, according to the orientation of the strands (4–6). In vitro, G4 structures are selectively bound and stabilized by small molecules called G4 ligands (7).
The existence of G4s has been demonstrated in vivo in ciliates using specific antibodies (8), whereas only indirect evidence sustained their existence in mammalian cells [for a review see (9,10)]. Bioinformatic analysis of the human genome indicated that it contains as many as 370000 sequences possessing the Potential G-Quadruplex-forming Sequences (PQS) (11,12). As expected, most of these sequences are located in repetitive DNA regions, such as telomeres and rDNA. In addition, a statistically significant enrichment of PQS was found in regulatory regions such as gene promoters (13), splice sequences and UTRs regions (14) raising the possibility that G4 structures could play a role in the regulation of gene expression.
In eukaryotes, chromosomes ends are protected from DNA repair systems by a particular nucleoprotein structure, the telomere (15). In humans, the telomere is composed of thousands of G-rich double-stranded TTAGGG repeats (16) and a 3′ single-stranded G-rich extension called the G-tail or G-overhang (17). The telomeric DNA is bound by a telomere-specific six-protein complex called shelterin (18). Shelterin stabilizes a special DNA structure, the t-loop, in which the G-tail invades the duplex telomeric repeats, forming a D-loop structure (18). The t-loop masks chromosome ends and blocks the activation of the DNA damage response at telomeres (19).
TRF2 protein plays an essential role in the shelterin function (18). TRF2 has been shown to promote and stabilize loop formation (20). In conjunction with its partner RAP1, TRF2 also triggers the inhibition of the the non homologous end joining relying on the DNA-dependent protein kinase at telomeres (21). Thus, overexpression of dominant-negative mutants of TRF2 induces telomere uncapping triggering end-to-end chromosome fusions (22) or stochastic deletions of telomeric DNA through a homologous recombination-mediated mechanism (23). TRF2 is overexpressed in several human tumors, such as liver hepatocarcinomas (24), breast carcinomas (25) and lung carcinomas (26), suggesting that TRF2 may play a role in tumorigenesis.
In this study, we describe the biophysical and functional characterization of G-rich sequences present within the TRF2 mRNA. We show that a G-rich sequence located in the 5′-UTR region of the TRF2 mRNA adopts a stable intramolecular G4 RNA structure in vitro, which is able to repress the expression of a reporter gene in vitro and in cells. Mutation of this sequence impairing quadruplex stabilization leads to an increased expression.
Furthermore, using biophysical analyses, we show that the G-quadruplex RNA motif adopted by the G-rich sequence located within the 5′-UTR of TRF2 mRNA is bound by several highly selective G-quadruplex ligands. In vitro studies show that the stabilization of the G4 RNA motif has a significant effect on the expression of a reporter gene. These data suggest that G4 formation in the 5′-UTR from TRF2 represents a new mechanism to control TRF2 expression.
All oligonucleotides described in Table 1 except +75UTRATGTRF2 and mut+75UTRATGTRF2 (Sigma Aldrich) were purchased from Eurogentec.
Circular dichroism (CD) spectra were recorded on a JASCO-810 spectropolarimeter using 1-cm path length quartz cuvettes in a reaction volume of 580µl, as previously described (27). Oligonucleotides 91TRF2G:RNA and mut91TRF2G:RNA (Table 1) were prepared as a 4µM solution in 10mM lithium cacodylate pH 7.2, 100mM NaCl or KCl buffer and annealed by heating to 90°C for 2min, followed by slow cooling to 20°C. Scans were performed at 20°C over a wavelength range of 235–350nm with a scanning speed of 500nm/min, a response time of 1s, 1nm data pitch and 1nm bandwidth.
Oligonucleotides 91TRF2G:RNA and mut91TRF2G:RNA (Table 1) were synthesized by Eurogentec (Seraing, Belgium) at the 200 nmol scale and used without further purification. Concentrations were estimated using extinction coefficients provided by the manufacturer. Melting assays were performed on a Uvikon 940 spectrophotometer in a 10mM lithium cacodylate pH 7.2 buffer (supplemented with either 0.1M KCl or NaCl, hereafter referred to as potassium and sodium conditions, respectively), as previously described (28–30). Melting experiments were typically performed at a concentration of 4µM per strand. All transitions were reversible, as shown by superimposable heating and cooling profiles at a fixed rate of 0.2°C/min.
Tm of F21T (0.2µM) was recorded alone, in the presence of 1µM of G4 ligands, in the presence of 3µM competitor or in presence of G4 ligands + competitor. The competitors were a double-stranded (ds26:5′-CAATCGGATCGAATTCGATCCGATTG-3′), or 91TRF2G:RNA or Mut91TRF2G:RNA. All experiments were performed in duplicate. The emission of fluorescein is normalized between 0 and 1, and the T1/2 is defined as the temperature for which the normalized emission is 0.5 (31).
Formation of G4 RNA was confirmed by nondenaturing PAGE. Oligonucleotides were directly visualized by UV shadow (see Supplementary Data). Prior to the incubation, the DNA samples were heated at 90°C for 5min and slowly cooled (2h) to room temperature. Samples were incubated at 30µM strand concentration in Tris–HCl 10mM pH 7.5 buffer with 100mM Na+ or K+. Ten percent sucrose was added just before loading. Oligothymidylate markers (dT15, dT21 or dT30) or double-stranded markers (Dx9:5′-d-GCGTATCGG+5′-d-CCGATACGC; Dx12:5′-d-GCGTGACTTCGG +5′-d-CCGAAGTCACGC) were also loaded on the gel. One should note that the migration of the dTn oligonucleotides does not necessarily correspond to single strands (32): these oligonucleotides were chosen here to provide an internal migration standard, not to identify intramolecular or higher order structures.
To construct pWUTRF2 and pMUTRF2 plasmids, a synthetic DNA duplex carrying either the +65UTRATGTRF2 or the mut+65UTRATGTRF2 sequence was inserted at the BamHI–EcoRI sites within the pcDNA3 vector (Invitrogen). The resulting plasmids were utilized to clone an EcoRI–XhoI DNA fragment encoding the green fluorescent protein (GFP) obtained by PCR amplification of the GFP sequence from pEGFP-C1 plasmid (Clontech). The insertion of correct sequences was verified by sequencing.
TRF2ΔB construction result from the insertion of a 5′ deleted form of TRF2 cDNA (23) into the BamHI–EcoR1 sites of the pcDNA3 plasmid (Invitrogen).
In vitro coupled transcription–translation of pWUTRF2 and pMTRF2 plasmids was carried out using the rabbit reticulocyte lysate-based transcription–translation coupled system (TNT® Quick Coupled Transcription/Translation System, Promega) according to the manufacturer’s instructions. Briefly, a 1µg plasmid mixture consisting of 0.2µg of the TRF2ΔB plasmid and either 0.8µg of pWUTRF2 or pMUTRF2 plasmid was mixed to 12.5µl of TNT rabbit reticulocyte system, 1µl of TNT reaction buffer, 0.5µl of amino acid mixture minus methionine, 1µl of [35S]-l-methionine at 1175Ci/mmol (PerkinElmer) and 0.5µl of T7 RNA polymerase in a total reaction volume of 25µl. The reaction mixture was incubated at 30°C for 90min and immediately put at −20°C. To evaluate GFP production, 5µl of each reaction was diluted 5-fold with 1X Laemmli buffer, heated at 70°C for 10min and loaded on a 12% SDS–polyacrylamide gel. After separation, [35S]-methionine incorporation was visualized by a Phosphorimager (Typhoon 9210 Amersham). To study the effect of G4 ligands on in vitro gene expression indicated, concentrations of molecules (in a volume of 2.5µl) were added to the complete reaction mixture before incubation at 30°C.
The 293T cells were growth at 37°C in a humidified atmosphere containing 5% CO2 in Dulbecco’s modified Eagles medium supplemented with 10% heat-inactivated fetal calf serum and the antibiotics penicillin and streptomycin.
Transfection was carried out in six-well plates using 3µg of each plasmid and Lipofectamine 2000 (Invitrogen) transfectant reagent according to the manufacturer’s instructions. Twenty-four hours after transfection, cells were recovered by trypsination and the cellular pellet was divided into two equal portions to proceed to western blot and Quantitative real time PCR (RT–qPCR) analyses.
SDS gel electrophoresis was performed in a 12% polyacrylamide gels with a protein content of 50µg at 150V in 0.025M Tris pH 8.3, 0.192M glycine, 0.1% SDS. Gels were blotted onto Immobilon-P polyvinylidene difluoride membrane (Millipore). The membrane was blocked with 5% dry milk in PBS-T [(1× phosphate buffer saline, 0.1% v/v Tween-20 (Sigma-Aldrich)] for 30min at room temperature and incubated with the anti-GFP primary antibody (1:1000 in 1× phosphate buffer saline, 0.1% v/v Tween-20) for 1h at room temperature. A secondary horseradish peroxidase conjugated antibody (1:10000, 45min at room temperature) was used and detected using the enhanced chemiluminescence kit (ImmunofaxA, Yelen).
Total cellular RNA was isolated using RNeasy Plus Mini Kit (Qiagen). For quantification of mRNA levels, 1µg of total RNA was used in a 20µl reverse transcription reaction using the RevertAid H Minus First Strand cDNA synthesis Kit (Fermentas) in the presence of random hexamer. A control reaction, without reverse transcriptase enzyme, was performed in parallel to ensure the absence of plasmid DNA contamination. Real-time PCR was performed with 25ng of cDNA and 300nM concentration of both sense and antisense GFP primers (sense 5′-CGACCACATGAAGCAG and antisense 5′-TCCTGGACGTAGCCTTCGG) in a final volume of 25µl using the SYBR Green TaqMan Universal PCR master mix (Applied Biosystems). Fluorescence was monitored and analyzed in a GeneAmp 7300 detection system instrument (Applied Biosystems). Analysis of 18S ribosomal RNA and GAPDH RNA (Applied Biosystems TaqMan probes) was performed in parallel to normalize for gene expression.
For RNA quantification of in vitro transcription/translation products, 5µl of this mix was used in 20µl reverse transcription reaction. Real-time PCR was performed with 5µl of reverse transcription products.
Sequence analysis of the 5′-part of the human TRF2 mRNA shows the presence of several G-rich sequences surrounding the AUG start codon (Figure 1A). Bioinformatic analysis of the TRF2 mRNA using QGRS Mapper program (33) shows the existence of several potential G4 structures within the 5′-part of the TRF2 mRNA. Two sequences present high G-score and are favorable for G4 folding: the first one matching the sequence 5′-GGGAGGGCGGGGAGGG-3′ (91TRF2G) is located within the 5′-UTR, 19nt upstream of the AUG start codon and the second one matching the sequence 5′-GGGCCCGGCGGGGGCGCCACGAGCCGGGGCUGGGGGG-3′ (199TRF2G) at 66nt downstream of the translation start site. The DNA sequence alignment of human TRF2 with its orthologs present in six amniota vertebrates shows that the G-rich sequence present in the 5′-UTR of human TRF2 is evolutionarily conserved across species (Figure 1B), indicating a potential regulatory role. Thus we decided to further characterize these sequences, both at the molecular and functional level.
In a first attempt to determine which G-rich TRF2 sequences may adopt a G4 DNA structure in vitro, we performed a rapid screening by electrophoretic mobility shift assays with several DNA oligonucleotides designed from the PQS (91TRF2G:DNA, 131TRF2G:DNA, 195TRF2G:DNA and 199TRF2G:DNA; see Table 1) and purified recombinant human Topoisomerase IIIα (Topo III), in the presence of increasing concentrations of the 360A pyridine dicarboxamide derivative, a potent and highly specific G4 ligand (Supplementary Figure S1 and Supplementary Data).
We conclude from these experiments that 91TRF2G:DNA sequence is the best candidate in the 5′-UTR of the TRF2 mRNA to adopt a G4 DNA structure in vitro, and we decided to further characterize its biophysical and biological properties.
To test whether the 91TRF2G sequence forms a G4 RNA structure, we carried out biophysical analyses on the RNA oligonucleotide 5′-CGGGAGGGCGGGGAGGGC-3′. CD experiments using different ionic conditions (Na+, K+ and Li+) at physiological pH (see ‘Materials and Methods’ section) were performed, and spectra show the characteristic features of parallel G4 structures, with a positive peak at 263nm and a trough at 240nm the amplitude of which is higher in Na+ or K+ conditions than in Li+ (Figure 2A) (34). Under the same conditions, thermal difference spectrum analysis of the 91TRF2G:RNA is characterized by a negative minima at 262 and 295nm, positive maxima at 255 and 270nm, a shoulder at 245nm, which are the particular signatures of the G4 structures (data not shown) (29).
The thermal melting of quadruplex nucleic acids can be characterized by an inverse UV transition at 295nm, and has generally been found to have significant and characteristic cation dependence. The monovalent ion dependence for the stabilization of folded 91TRF2G:RNA as judged by Tm, was in the order K+>Na+>Li+ (Table 2), which is characteristic of G4 nucleic acids (3).
At 100mM KCl, the folded 91TRF2G RNA quadruplex could not be unfolded, even at 95°C, which is indicative of a very stable quadruplex. The 91TRF2G G4 RNA was very stable (Tm=61.58°C) even in the presence of low-salt concentrations (1mM KCl). Studies over a 10-fold strand concentration range (from 2 to 20μM) showed no change in the Tm-value (Table 3), this last result being consistent with the folding of an intramolecular quadruplex.
Formation of G4 structures is dependent on Hoogsteen bonding established between guanines. Thus, single guanine replacements in the 91TRF2G:RNA sequence should prevent the folding of this sequence in a G4 RNA structure. Mut91TRF2G:RNA sequence 5′ CGUGAGUGCGCUGAGGGC 3′ was designed by replacing four key guanine nucleotides present in the 91TRF2G:RNA sequence at positions G3, G7, G11 and G12. As shown in Figure 2B, CD spectrum analysis of Mut91TRF2G sequence, in the presence of Na+ or K+ ionic concentrations, did not show the main features associated to G4 formation. In addition, native gel electrophoresis revealed that Mut91TRF2G:RNA migrates at its expected molecular weight position, as compared to 91TRF2G:RNA whose migration is lowered, as expected for a G4 structure (Supplementary Figure S2).
Altogether, these data indicate that under near-physiological pH and salt conditions, 91TRF2G:RNA folds into a very stable, intramolecular G4 RNA that presents the characteristic features of a parallel G4. Since CD spectra only provide indirect evidence for G4 topology, additional experiments are necessary to confirm this finding.
To evaluate the impact of the G4 structure formed by the 91TRF2G sequence on gene expression, we cloned the 62nt of the 5′-UTR from TRF2, containing either the wild-type 91TRF2G sequence (plasmid pWUTRF2) or its mutated version Mut91TRF2G (plasmid pMUTRF2), upstream of the GFP-coding sequence and immediately downstream of the minimal T7 promoter present on pcDNA3 plasmid (Figure 3A). Using an in vitro coupled transcription–translation assay, based on rabbit reticulocyte lysate system, the GFP expression resulting from these constructions was evaluated by radioactive quantification of [35S]-methionine incorporation and showed a 2.45-fold increase in GFP synthesis for the mutated sequence relative to the wild-type sequence (Figure 3B and C). To further characterize the repressive effect of 91TRF2G sequence on gene expression in vitro, we next proceeded to quantification of RNA molecules produced by the transcription–translation coupled system. As shown in Figure 3D, RT-PCR analysis did not show any significant difference of the in vitro transcription of two constructions. This result indicates that the presence of a G4 sequence within the 5′-UTR of the gene reporter construction does not affect the transcriptional process. Together, these results suggest that the presence of a G4 motif within the 5′-UTR of the TRF2 mRNA inhibits the gene expression at a posttranscriptional level.
In order to investigate the impact of 91TRF2G G4 structure in a cellular context, 293T cells were transfected with plasmids pWUTRF2 or pMUTRF2. Twenty-four hours after transfection, cells were harvested and protein lysates were prepared. Western blot analysis of GFP expression indicated that 293T cells transfected by pMUTRF2 present a 2.8-fold higher GFP protein level (normalized to actin), as compared to the signal obtained using pWUTRF2 (Figure 4A). As RNA quadruplexes in genes promoters have been reported to modulate transcription levels of several genes (35–37), and to exclude any variation due to the transfection efficiency, mRNA levels arising from transcription of pWUTRF2 and pMUTRF2 plasmids were quantified using RT–qPCR assays. Relative quantification of GFP mRNA molecules produced by pWUTRF2 or pWUTRF2 showed no significant difference of the transcriptional expression between the two constructions (Figure 4B). This result together with our in vitro data showing the formation of a highly stable parallel G4 RNA structure under physiological conditions for 91TRF2G:RNA strongly suggest that the repression of GFP expression observed in vivo is due to the ability of 91TRF2G:RNA G4 structure to reduce translation in human cells.
The interaction of the 91TRF2G:RNA sequence with G4 ligands has been investigated by a competition FRET assay using a real-time PCR apparatus and the doubly labelled F21T oligonucleotide which mimics the human telomeric single-strand overhang, as previously described by Mergny and coworkers (31). To this end, melting of F21T is performed in the presence of three different, highly selective, G4 ligands: the 360A molecule (38) and two bisquinolinium compounds Phen-DC(3) and Phen-DC(6) (39) (see structures in Supplementary Figure S3), and various competitors: a 26bp duplex (ds26), the 91TRF2G:RNA sequence and its mutant counterpart, the Mut91TRF2G.
As shown in Table 4, in the presence of sodium or potassium, the three G4 ligands display a good selectivity for the quadruplex over the control duplex, since stabilization of F21T is only moderately affected by a 15M excess of the ds26 DNA duplex. In contrast, when the 91TRF2G:RNA sequence was added, the competition pattern shows a strong decrease in ΔT1/2 for all compounds. In the presence of sodium or potassium, 91TRF2G:RNA sequence provokes a complete loss of the stabilization of the F21T by bisquinolinium compounds. For the pyridine dicarboxamide derivate, the 360A molecule, competition analyses show a differential binding to 91TRF2G:RNA sequence dependent on ionic conditions. While in the presence of sodium, 91TRF2G:RNA sequence induces a complete displacement of 360A from F21T oligonucleotide; in the presence of potassium, the G4 ligand presents a lower affinity for the RNA sequence as compared to F21T sequence although a significant decrease in ΔT1/2 is observed (ΔT1/2 decreases of 14°C). In contrast, the mutant sequence, Mut91TRF2G:RNA, did not compete with the stabilization of F21T by G4 ligands (Table 4). Together, our results demonstrate that the 91TRF2G:RNA sequence is able to bind G4 ligands and that the binding to these molecules is associated with its capacity to adopt a G4 structure.
To further demonstrate that the repressive effect of 5′-UTR 91TRF2G sequence was due to the formation of a G4 structure, we studied the effect of G4 ligands on GFP synthesis in vitro.
As shown in Figure 5B, all of the three G4 ligands used in this assay have a more pronounced effect, concentration dependent, on GFP expression from the wild-type (pWUTRF2) than from the mutant construction (pMUTRF2).
GFP expression from the wild-type construction (pWUTRF2 containing the 91TRF2G motif), is decreased by 15 and 60% using 1 and 10µM 360A, respectively (Figure 5A and B, left). In contrast, up to 10µM 360A causes a very limited reduction of GFP protein synthesis of the mutant construction (~10% at 10µM; Figure 5A and B, left).
The bisquinolinium compound Phen-DC(3) exhibits a more marked inhibition effect. At 3µM concentration, Phen-DC(3) inhibits GFP expression from pWUTRF2 by four-fold, while at the same concentration this molecule did not show any significant effect on pMUTRF2 construction. Although at 10µM concentration, Phen-DC(3) showed a marked effect on GFP expression of from the mutant construction (~60%), GFP expression from the wild-type construction was inhibited by 10-fold (Figure 5B, central).
When compared to the two others molecules, Phen-DC(6) compound shows a more pronounced nonspecific effects. At 3µM, the GFP expression for the mutant reaches 40% inhibition and this value was of 50% at 10µM. Nevertheless, at the same concentrations, the inhibition of GFP synthesis from the wild-type construction reaches 75 and 90%, respectively (Figure 5B, right).
To investigate if the inhibition of the GFP synthesis induced by G4 ligands was due to a posttranscriptional effect, we next quantified the RNA molecules. As shown in Figure 5C, 10µM 360A did not show any significant effect on RNA production from either pWUTRF2 plasmid or its mutant counterpart pMUTRF2 (Figure 5C, left and right). For Phen-DC(3), a limited 10% inhibition effect was observed only on RNA production from pMUTRF2 plasmid. In contrast, Phen-DC(6) shows a significant inhibition effect on RNA production from both pWUTRF2 and pMUTRF2 plasmids. At 10µM, Phen-DC(6) provokes a 25% inhibition of the RNA production for both constructions (Figure 5C) which cannot account for the profound and preferential inhibition of GFP synthesis observed from the pWUTRF2 plasmid under these conditions (Figure 5B, right). Together, these results indicate that the inhibition of the protein production induced by the G4 ligands used in this report is mainly due to their ability to interfere with a posttranscriptional process.
In the past years, several works have reported the presence of stable G4 structures in RNA molecules. For instance RNA G4-forming sequences have been found in the HIV-genomic RNA, hTERT, FMR1, FGF-2, NRAS, Zic-1, IGF-II, ESR1 and MT3-MMP mRNAs (40–48). In this report, we show that a G4-forming sequence, located 19nt upstream of the initiation codon of the TRF2 protein, the 91TRF2G sequence, represses protein synthesis in vitro and is able to reduce the translation rate of a reporter gene in human cells.
Biophysical studies of the 91TRF2G:RNA sequence using CD, UV melting and FRET analysis suggest that the G4 RNA structure is extremely stable and could be formed under physiological conditions. Remarkably, this sequence may adopt the features of a stable G4 (typical G4 CD spectrum and ΔTm >20°C) for K+ concentrations as low as 1mM. Native gel electrophoresis, as well as the independence of the Tm-values over a 10-fold strand concentration range, is consistent with the formation of an intramolecular quadruplex rather than a dimer. The formation of G4 structures is entirely dependent on Hoogsteen hydrogen bonding established between guanine nucleotides, and the guanine modifications introduced in the mut91TRF2G:RNA sequence completely abolish the formation of the RNA quadruplex, as evidenced by CD and UV melting analysis. Interestingly, the 91TRF2G:RNA sequence is characterized by a ‘Type I spectra’ with a trough at 240nm and a positive peak at 263nm, which are the main features of parallel G4 structures (34). Although the structural characterization of this quadruplex remains to be determined, the parallel folding of a RNA quadruplex is highly probable due to the more favored C3′-endo conformation of the ribose ring (49), although a combination of C2′-endo and C3′-endo has also been observed (50).
Stable secondary RNA structures located in the 5′-UTRs of mRNA have been shown to regulate translation (51). For example, in bacteria, a RNA G4 motif located directly upstream of the Shine–Dalgarno sequence showed a very strong inhibition of the translation of a reporter gene (52). In human cells, the RNA G4 motif located in the 5′-UTR region of Zic-1 and MT3-MMP mRNAs have been shown to repress translation (40,46). Furthermore, using in vitro and functional assays, Halder and coworkers (53) have shown that natural and synthetic G4 motifs introduced into the 5′-UTR region of a mRNA reporter system act as universal translational suppressors in mammalian cell lines.
In our study, using a GFP reporter system, we show that the 91TRF2:RNA G4-forming sequence, located in the 5′-UTR region of the TRF2 mRNA, causes a 2.4-fold decrease of the expression of a reporter gene in vitro and a 2.8-fold decrease in the rate of expression in human cells (Figures 3 and and4).4). Interestingly, our results indicate that the 91TRF2G quadruplex motif does not affect the GFP mRNA accumulation both in vitro and in cellulo, as compared to its mutated counterpart (Figures 3 and and4).4). Taken together, these observations show that the G4 motif formed by the 91TRF2G sequence located within the 5′-UTR region of the TRF2 mRNA downregulates gene expression by a translational suppressor mechanism.
In a previous work, Kumari and colleagues (54) demonstrated that translation modulation by an RNA G4-forming sequence is dependent on its position in the 5′-UTR of the message, with the inhibition of translation occurring only when the G4-forming-sequence is rather proximal to the 5′-cap. In our constructions, the G4 motif adopted by the 91TRF2G:RNA sequence is located at 92 bases downstream of the putative transcriptional start site of T7 polymerase or at 125 bases downstream of the putative transcriptional start site of CMV promoter. Since in the normal context, this sequence is located closer to the 5′-end of the TRF2 mRNA (90 bases), we can reasonably suppose that its repressive effect on translation may be at least as efficient as what we observe with the reporter constructions.
The search for specific G4 ligands was initially directed to target DNA telomere repeats and lead to the synthesis of compounds that present marked effects on telomere replication and capping (55). Interestingly, these compounds did not present potent selectivity towards other G4 structures (56), and several reports indicated that these ligands should not be considered as simple telomere ‘binders’ (57) and may act in extratelomeric regions (58,59), or at the mRNA level (60) to modulate gene promoter expression (61) and rDNA biosynthesis (62).
The 360A molecule, a pyridine dicarboxamide derivate, Phen-DC(3) and Phen-DC(6), two bisquinolinium compounds, are three G4 ligands among the most potent ones reported so far (63). In this study, we demonstrated that these G4 ligands bind to the 91TRF2G:RNA sequence and that this binding is dependent on the ability of this sequence to adopt a G4 structure. Thus, as evaluated by competition FRET experiments, these ligands exhibit a clear affinity for the 91TRF2G:RNA sequence, while they are not able to bind the mutant counterpart, which did not show the main features associated to G4 formation (Figure 2).
Using an in vitro coupled transcription–translation system, we demonstrated that the binding of G4 ligands to the 91TRF2G:RNA was correlated with their ability to inhibit the protein synthesis in vitro. Thus, even if at 10µM G4 ligands exhibit an inhibitory effect on the protein synthesis from the control plasmid pMUTF2 (~20% inhibition for 360A and 60% inhibition for bisquinolinium compounds): the inhibition effect on the expression of the GFP protein from the construction containing the G4-forming sequence was ~60% for the 360A and reached 90% for bisquinolinium compounds. In our assay, bisquinolinium compounds are more active than the 360A molecule. While, at 1µM 360A molecule has a limited inhibitory effect on the expression of GFP protein from pWUTRF2 plasmid, Phen-DC(3) and Phen-DC(6) inhibit the protein synthesis by two-fold. Interestingly, competition FRET studies show that, in the presence of potassium, the binding of 360A to the 91TRF2G:RNA was weaker than the binding of bisquinolinium compounds to the same sequence (Table 4). When compared to two others molecules used in this study, Phen-DC(3) exhibits a more selective effect. At 3µM, Phen-DC(3) molecule inhibits GFP expression from pWUTRF2 by four-fold, whereas it does not affect the expression of the GFP protein from the control plasmid. For comparison, the commonly used G4 ligand TMPyP4, show no RNA G4 specificity (64), and in a recent publication Bugaut and coworkers (64) identified a pyridine-2,6-bis-quinolino-dicarboxamide derivate, RR82 that exhibits an two-fold inhibitory effect on NRAS 5′-UTR G4-forming sequence at 10µM concentration.
More interestingly, we demonstrated that G4 ligands used in this report do not affect the RNA production and/or stability, as evaluated by RT-PCR analyses (Figure 5C). Although we cannot exclude that in a natural chromatin context, the folding of the 91TRF2G sequence in a G4 structure could also affect the transcription of the endogenous TRF2 mRNA; all reports so far that demonstrate the impact of G4 DNA structures on transcriptional processes concern G4-forming sequences located in the promoter regions (13).
Taken together, these observations are consistent with G4 ligands inhibiting protein synthesis via interactions that stabilize the 91TRF2G:RNA G4 motif. Furthermore, our study demonstrated the proof-of-concept for G4 ligand-mediated regulation of translation by targeting natural G4 RNA motifs.
Mice overexpressing TRF2 in the skin develop both spontaneous and carcinogenic-induced epithelial tumors (65). Conversely, downregulation of TRF2 was found to reduce the tumorigenic potential of grafted tumors in mice (66). Thus, these studies suggest that the level of TRF2 protein in human cells may play an important role on cell fate and cancer development. Several reports show that TRF2 is overexpressed in human tumors (24–26). Although most of these works have evaluated TRF2 expression at the mRNA level, Nijjar and coworkers (25) have shown that the upregulation of the TRF2 protein in tumor cells, as compared to normal cells results from differences in posttranscriptional regulation, since their mRNA levels remained comparable. According to the data reported here, it may be interesting to check if G4-forming sequences in TRF2 mRNA could be implicated in part of the changes in TRF2 protein level associated with cancer.
Supplementary Data are available at NAR Online.
Institut National Contre le Cancer (TELINCA program) and Ligue Nationale Contre le Cancer (‘équipes labellisées’, to P.C., D.G., B.S., J-.F.R. and A.G., in part). Funding for open access charge: CNRS - Délégation Midi-Pyrénées.
Conflict of interest statement. None declared.
P.C. is a scientist from INSERM, France.