Translational repression of cyclin D3 by a stable G-quadruplex in its 5′ UTR

Abstract

cyclin D3 (CCND3) is one of the three D-type cyclins that regulate the G1/S phase transition of the cell cycle. Expression of CCND3 is observed in nearly all proliferating cells; however, the presence of high levels of CCND3 has been linked to a poor prognosis for several types of cancer. Therefore, further mechanistic studies on the regulation of CCND3 expression are urgently needed to provide therapeutic implications. In this study, we report that a conserved RNA G-quadruplex-forming sequence (hereafter CRQ), located in the 5′ UTR of mammalian CCND3 mRNA, is able to fold into an extremely stable, intramolecular, parallel G-quadruplex in vitro. The CRQ G-quadruplex dramatically reduces the activity of a reporter gene in human cell lines, but it has little impact on its mRNA level, indicating a translational repression. Moreover, the CRQ sequence in its natural context inhibits translation of CCND3. Disruption of the G-quadruplex structure by G/U-mutation or deletion results in an elevated expression of CCND3 and an increased phosphorylation of Rb, a downstream target of CCND3, which promotes progression of cells through the G1 phase. Our results add to the growing understanding of the regulation of CCND3 expression and provide a potential therapeutic target for cancer treatment.

Introduction

The progression of cells through the cell cycle is driven by the phosphorylating activities of cyclin-dependent kinase (CDK) enzymes, which are regulated by cyclins, a family of proteins implicated in the induction and control of mitosis.¹ Different cyclins are required for the progression through different phases of the cell cycle. During the G1 phase, cells make critical decisions about their fate, including the optional commitment to cell division. The first step in the mitotic cycle is the transition from the G1 to the S phase when the D-type cyclins (including cyclins D1, D2 and D3) are required.² By binding to and activating CDK4 and CDK6, the D-type cyclins phosphorylate the retinoblastoma tumor suppressor protein (Rb), leading to the release of the E2F transcription factors and the subsequent expression of E2F-regulated genes that are critical for S phase progression.^3,4 The three D-type cyclins show high homology despite being encoded by separate genes, and the human cyclin D3 (CCND3) gene was originally identified by cross-hybridizing cyclin D1 probes with a human cDNA library.⁵ CCND3 is expressed in nearly all proliferating cells and shows the most broad expression pattern of the three D-type cyclins.⁶

Emerging evidence suggests that CCND3 regulates the G0/G1 to S phase transition in a variety of cell types, and is required for the development of normal cells as well as cells originate from human malignancies.^6-23 In multiple myeloma and mature B-cell malignancies, overexpression of CCND3 was observed.^24,25 Moreover, high levels of CCND3 were suggested to be associated with a poor prognosis in patients with superficial melanoma or diffuse large B-cell lymphoma.^26,27 Given its role in cell fate decisions, the expression of CCND3 is tightly regulated. Transcription of CCND3 has been shown to be activated by transcription factors such as Sp1, E2A, E1AF, and E2F1, and repressed by HDAC5 and Meis1.^28-32 The stability of the CCND3 protein is regulated either in a Thr-283-dependent or -independent manner.^33-35 In addition, the 3′ untranslated region (UTR) is involved in the regulation of CCND3 expression. RNA-binding proteins (e.g., HuR) have been shown to bind to the 3′ UTR of CCND3 mRNA and regulate its stability and translation.^36,37 Besides, as reported by our group and two other groups, three microRNAs (including miR-195, miRNA-16, and miR-138) inhibit the translation of CCND3 by basepairing with its 3′ UTR.^38-40 Despite these findings, an understanding of the mechanisms underlying the regulation of CCND3 expression is still far from complete, especially on the question of whether the 5′ UTR is involved in the regulation.

Recently, it was shown that a new type of cis-acting element, RNA G-quadruplex, exists in the 5′ UTRs of several genes and represses their translation by blocking the formation or scanning of the preinitiation complex.⁴¹ G-quadruplexes (also known as G4) are non-canonical, four-stranded nucleic acid structures that build on a series of π-stacked G-quartet planes, each of which consists of four guanine bases that interact via Hoogsteen hydrogen bonds. The quadruplex secondary structure can be further stabilized by monovalent metal ions, especially potassium.^42-44 Potential DNA G-quadruplex sequences have been identified in G-rich eukaryotic telomeres and non-telomeric genomic DNA regions, such as nuclease-hypersensitive promoter regions, whereas research on the RNA G-quadruplex is relatively limited. Recently, two research groups have conducted bioinformatic analyses and estimated that potential RNA G-quadruplexes (PG4) were enriched in human 5′ UTRs, with a 4- to 5-fold higher density than that of the human genome.^41,45 Experimental evidence also supports the existence of RNA G-quadruplexes in the 5′ UTRs of several mRNAs, including FGF-2, VEGF, NRAS, Zic-1, ESR-1, MT3-MMP, Bcl-2, TRF2, ADAM10, KRAS, and EBAG9.^45-55 However, not all of the predicted PG4 sequences were found to fold into G-quadruplex structures in vitro.⁴⁵ Therefore, there is an urgent need to identify bona fide 5′ UTR G-quadruplex sequences, especially for functionally important genes, and to determine the roles of these G-quadruplex structures in diverse cellular processes, for example, in the regulation of cell-cycle progression.

In this study, we identified a G-quadruplex structure (CRQ) in the 5′ UTR of the CCND3 gene and highlighted its roles in the translational repression of the CCND3 protein and the subsequent inhibition of the G1-S phase transition of the cell cycle. Various methods, including circular dichroism (CD), thermal denaturation and RNase T1 footprinting, were adopted to verify the formation of the CRQ G-quadruplex in vitro. In addition, dual-luciferase reporter assays, western blot analyses and quantitative RT-PCRs were performed to determine the translation repression ability of the CRQ G-quadruplex in cellulo, and flow cytometry and MTT assays were performed to study the effect of this G-quadruplex structure on the cell cycle distribution and cell proliferation. Our study represents the first report on the regulation of CCND3 by an RNA G-quadruplex structure, and more importantly, on the functional analysis of an RNA G-quadruplex structure regarding cell cycle control.

Results

Potential G-quadruplex-forming element within the 5′ UTR of the human CCND3 mRNA

To search for elements that regulate the translation of human CCND3, we constructed CCND3-expression vectors with or without a 5′ UTR immediately in front of the coding sequence (CDS), which we referred to as p5U-wt-D3 or pno5U-D3, respectively. The vectors were then transfected into HepG2, a human hepatoma cell line, in which the expression levels of CCND3 have been shown to be very low,⁶ and the expression of CCND3 was determined. A 60% to 85% decrease in CCND3 protein levels was observed when the 5′ UTR was present, as shown in Figure 1A, which revealed the presence of inhibitory elements inside the 5′ UTR. We thus decided to further characterize the sequence of the 5′ UTR of CCND3. The G content of the CCND3 5′ UTR (165-nucleotides-long) is 37.6%, significantly higher than the average of the four nucleotides, suggesting the existence of G-rich RNA motifs. More detailed scanning using QGRS Mapper and Quadfinder^56,57 revealed a potential RNA G-quadruplex-forming motif (hereafter referred to as CRQ, Fig. 1B) which was located 30 nucleotides downstream of the 5′ cap and 116 nucleotides upstream of the translation start site. The motif is highly conserved among different mammal species in its sequence and the flanking sequences, as well as in its position with respect to the translation start site (Table 1; Fig. S1), indicating that it may be functionally important.

Figure 1. The 5′ UTR of CCND3 contains a potential G-quadruplex inhibitory element. (A) The entire 5′ UTR of CCND3 inhibits the expression of CCND3. HepG2 cells were transfected with pcDNA3.0 (3.0) expression vectors with (+) or without (-) the 5′ UTR in front of the CDS of CCND3 and were analyzed by western blot 48 h later. (B) Schematic representation of the human CCND3 gene. The sequence of the 5′ UTR is shown, with the CRQ motif underlined and the guanines that are predicted to be involved in the formation of the potential G-quadruplex highlighted in red.

Table 1. Conservation of the CRQ motif in the 5′ UTR of CCND3.

Species	Latin name	Sequence a	Position b
Human	Homo sapiens	GGGCGGCGGGCGGGCUGGGG	-116
Chimpanzee	Pan troglodytes	GGGCGGCGGGCGGGCUGGGG	-116
Gorilla	Gorilla gorilla	GGGCGGCGGGCGGGCUGGGG	-116
Orangutan	Pongo pygmaeus	GGGCGGCGGG - - - CGGGCGGGCUGGGG	-116
Gibbon	Nomascus leucogenys	GGGCGGCGGGCGGGCUGGGG	-116
Rhesus	Macaca mulatta	GGGCGGCGGGCGGGCUGGGG	-116
Marmoset	Callithrix jacchus	GGGCGGCGGGCGGGUUGGGG	-122
Mouse	Mus musculus	GGGUGGCGGGCGGGCUGGGG	-125
Opossum	Monodelphis domestica	GGGAAGCCGGG - - - CGGGCGGGCUGGGG	-136
Consensus		GGGCGGCGGGCGGGCUGGGG

^a Dashes represent gaps in the alignment. Nucleotides in bold are runs of guanines able to form G-quartets. ^bThe position of the last G in the potential G-quadruplex motif is given relative to the translation start site.

The CRQ sequence forms a thermodynamically stable RNA G-quadruplex in vitro

CD spectroscopy is a classical technique commonly used to investigate the formation of quadruplex structures. An RNA G-quadruplex structure is compelled to adopt a parallel four-stranded helical structure because of the nature of its sugar,⁵⁸ which results in a CD spectrum that is characterized by a positive peak at approximately 264 nm and a trough at approximately 240 nm. To examine whether the CRQ motif is able to form a G-quadruplex structure, we performed CD spectroscopy using RNA oligonucleotides of the CRQ (CRQ-wt) and the corresponding G/U-mutant (CRQ-mut). The CRQ-wt oligonucleotide showed a characteristic CD signature for a parallel RNA quadruplex at physiological pH, with a positive peak at 263 nm and a negative peak at 239 nm (Fig. 2A). The addition of KCl led to a significant transition to a higher positive peak at 263 nm and a negative peak at 241 nm, the amplitude of which was dependent on the concentration of K⁺ (Fig. 2A; Table S1). On the contrary, the CRQ-mut showed no corresponding transition, regardless of the concentration of KCl present in the buffer (Fig. 2A; Table S1). The effect of different ionic conditions on the CD spectrum of CRQ-wt was also determined. As shown in Figure 2B, 100 mM KCl or NaCl resulted in a significant transition, but LiCl did not, consistent with a previous finding that potassium and sodium ions could stabilize the formation of G-quadruplex structures, while lithium ions could not.⁵⁹

Figure 2. Biophysical analysis confirms the formation of the CRQ G-quadruplex in vitro. (A) CD spectra of the wild-type (CRQ-wt) or G/U-mutated (CRQ-mut) CRQ RNA oligonucleotide in the presence of various concentrations of KCl. (B) CD spectra of the CRQ-wt RNA oligonucleotide in the presence of different types of monovalent cations. (C) Thermal UV melting curves of the CRQ-wt RNA oligonucleotide in the absence (NS) or presence of 1 mM KCl.

To further confirm that the CRQ oligonucleotide does indeed fold into a stable G-quadruplex structure, we performed thermal denaturation experiments by monitoring changes in the CD intensity of CRQ-wt at 263 nm. The T_m value of CRQ-wt in the absence of added salt was 51.71 ± 0.67°C as determined by “fraction folded” (θ) vs. temperature plots (Fig. 2C). In the presence of 1 mM KCl, the T_m value significantly increased by more than 20°C (73.27 ± 0.70°C). The annealing and melting curves were virtually identical, indicating that the molecules were at thermodynamic equilibrium. The T_m values of CRQ-wt in the presence of various monovalent cations showed the expected trend for a G-quadruplex, with K⁺ > Na⁺ > Li⁺ (Table 2). Specifically, in the presence of 100 mM KCl, the folded structure could not be fully unfolded even at 95°C, which is comparable with other G-quadruplex molecules and suggestive of a very stable structure.^{45,46,49,51,52} Notably, the addition of LiCl also increased the T_m value of CRQ-wt, although less significantly than did NaCl or KCl, which has been reported and can be explained by the non-specific counterion effect of the cation, which reduces the repulsion of the negative charge of the phosphate backbone.⁴⁵

Table 2. T_m values and thermodynamic parameters of CRQ RNA under various cations.

Cation	Tm (°C)	∆H° (kJ∙mol−1)	∆S° (kJ∙mol−1∙K−1)	∆G°37°C (kJ∙mol−1)
NS a	51.71 ± 0.67	-118.75 ± 5.57	-0.37 ± 0.02	-5.98 ± 0.61
100 mM Li+	63.07 ± 0.91	-139.01 ± 1.19	-0.42 ± 0.01	-11.46 ± 0.10
100 mM Na+	75.69 ± 2.10	-160.17 ± 1.85	-0.46 ± 0.00	-18.28 ± 1.08
100 mM K+	> 83	N/A b	N/A b	N/A b
1mM K+	73.27 ± 0.70	-183.76 ± 5.76	-0.53 ± 0.02	-20.08 ± 0.30

^a NS means no added salt. ^bThe thermodynamic parameters of CRQ RNA in the presence of 100 mM K⁺ is unable to be calculated since it cannot be fully unfolded even at 95°C.

We also calculated the thermodynamic parameters from the annealing and melting curves of CRQ-wt. In the presence of 1 mM KCl, the Gibbs free energy ∆G° at 37°C was -20.08 ± 0.30 kJ∙mol⁻¹, with a more than 3-fold decrease compared with the value in the absence of added salt (Table 2), suggesting the formation of a stable G-quadruplex structure at 37°C. Moreover, the calculated ∆H° and ∆S° values were also comparable to published data of other G-quadruplexes.^49,52,60

These thermal denaturation experiments were all performed with CRQ-wt RNA oligonucleotides at a concentration of 5 μM. When we conducted these experiments using CRQ-wt oligonucleotides over a 10-fold concentration, ranging from 1 μM to 10 μM, the T_m value in the presence of 1 mM KCl showed no apparent change (Table 3), suggesting that CRQ-wt forms an intramolecular quadruplex. Taken together, these results provide extensive evidence that the CRQ sequence located in the 5′UTR of CCND3 adopts an extremely stable, intramolecular, parallel G-quadruplex structure under physiological conditions (i.e., 100 mM K⁺).

Table 3. T_m values of CRQ RNA at various concentrations.

RNA concentration (μM)	1	5	10
Tm (°C)	73.23 ± 1.26	73.27 ± 0.70	74.45 ± 2.90

The CRQ G-quadruplex consists of three G-quartets and is resistant to RNase T1 cleavage

In addition, we conducted RNase T1 footprinting experiments to better determine the specific features of the CRQ G-quadruplex, such as the positions and the nucleotides of the loops and the guanosines involved in G-quartet formation, as previously described.^49,54,55 Because of the specificity of RNase T1 for guanine residues in single-stranded RNA, guanosines in the loop or the unstructured regions are more susceptible to attack by RNase T1 than those that directly participate in quartet formation. Thus, a G-quadruplex-forming RNA oligonucleotide will have different patterns of RNase T1 cleavage under conditions that vary in their ability to stabilize the G-quadruplex structure.^49,54,55 The addition of 100 mM Na⁺ led to a slight change in the cleavage pattern of CRQ-wt compared with Li⁺, whereas the same concentration of K⁺ resulted in a distinct pattern (Fig. 3A). First, the amount of CRQ-wt RNA that had not been cleaved was lowest in the presence of Li⁺, followed in order by Na⁺ and K⁺, suggesting that the presence of K⁺ stabilized the formation of secondary structures of the RNAs and thus protected them from RNase T1 cleavage. Second, all of the guanosines that are expected to directly participate in the formation of quartets were largely protected by K⁺ (Fig. 3A, in bold), except for G17, which may not be preferred for G-quartet formation. Third, the guanosines G5 and G6, which were expected to locate at the single-stranded loop region, instead, became more susceptible to RNase T1 cleavage. In contrast, footprinting on CRQ-mut showed almost the exact same pattern, regardless of the species of monovalent cation added, with all of the guanosines underwent cleavage (Fig. 3A), suggesting the inability to form a secondary structure. Therefore, on the basis of the RNase T1 footprinting pattern, we propose that the CRQ RNA has a strong tendency to fold into an intramolecular G-quadruplex structure, which consists of three G-quartets formed by four runs of guanines, specifically G1-G3, G8-G10, G12-G14 and G18-G20, while guanines G5, G6 and G17 are most likely located in the loop regions (Fig. 3B).

Figure 3. RNase T1 footprinting experiments showing protection of G-quartet-forming guanines from RNase T1 cleavage. (A) Alkaline hydrolysis ladders (L) were loaded and run on a polyacrylamide gel with RNase T1-treated RNA samples. The sequences for the CRQ-wt and CRQ-mut are indicated on the right side of their respective figures. For CRQ-wt, the guanines involved in the formation of G-quartets are in bold, while the other guanines that are not protected by KCl from RNase T1-mediated cleavage (G5, G6, G17) are indicated with closed arrow heads. (B) Proposed secondary structure of the CRQ G-quadruplex.

The CRQ G-quadruplex inhibits translation of a reporter gene in human cell lines

G-quadruplexes within 5′ UTRs have been shown to repress translation of their downstream genes in vitro and in cellulo.^45-53 To examine whether the CRQ G-quadruplex inhibits translation, we performed dual-luciferase reporter experiments. The reporter plasmids were constructed by insertion of the wild-type or G/U-mutated CRQ sequence exactly in front of the CDS of renilla luciferase and transfected into HEK293T and HepG2 cells. Forty-eight hours post-transfection, the cells were lysed, and the ratios of renilla luciferase activity normalized to firefly luciferase activity were determined. Insertion of a wild-type CRQ motif upstream of the renilla luciferase resulted in a more than 50% reduction in its expression in both cell lines compared with the vehicle plasmid (p < 0.01), whereas insertion of the G/U-mutated form had no effect on the renilla lucifease expression (Fig. 4A and B). To test whether the inhibition of the reporter gene expression was caused by regulation at the transcriptional or the post-transcriptional level, we performed qPCR with mRNA extracted from the transfected cells. After being normalized to the firefly luciferase mRNA, the renilla luciferase mRNA level showed no change in HEK293T cells transfected with either plasmid, while in HepG2 cells it decreased by approximately 15% when transfected with the wild-type plasmid (Fig. 4C and D). However, this slight decrease in mRNA levels could not explain the more than 50% reduction in protein levels as previously reported.⁵² These results show that the CRQ motif has a common inhibitory effect on the expression of the reporter gene in cellulo, which is achieved, not at the transcriptional level, but at the translational level.

Figure 4. The CRQ G-quadruplex inhibits the translation of a Renilla luciferase reporter gene. (A and B) Dual luciferase reporter assays were performed 48 h after HEK293T (A) and HepG2 (B) cells were transfected with the indicated reporter plasmids. Renilla luciferase activity (Rluc) was normalized to Firefly luciferase activity (Fluc), and the value for cells transfected with the empty psiCHECK-2 plasmid (vector) was set to 100%. **, p < 0.01. (C and D) The mRNA levels of Rluc and Fluc in transfected HEK293T (C) and HepG2 (D) cells were determined by quantitative RT-PCR. The ratio of Rluc / Fluc in cells transfected with CRQ-mut-psiCHECK-2 (CRQ-mut) plasmids was set to 100%. *, p < 0.05.

The CRQ G-quadruplex in the context of its full-length 5′ UTR inhibits the translation of CCND3

It has been suggested that the natural context of the G-quadruplex either affects the formation of the secondary structure or influences its function.^45,61 This urged us to investigate the role of CRQ in its natural context in the regulation of CCND3 expression. HepG2 cells transfected with CCND3-expression vectors with wild-type 5′ UTR (p5U-wt-D3) expressed significant lower levels of CCND3 protein, as compared with those transfected with vectors (p5U-mut-D3) harboring G/U mutations within the CRQ motif (Fig. 5A). Notably, the differential expression of CCND3 protein levels between the p5U-wt-D3 and the p5U-mut-D3 vectors was unlikely to be the result of transcriptional regulation, as evidenced by the almost equal expression of CCND3 mRNA from these two vectors (Fig. 5B). The inconsistence of protein and mRNA expression between the p5U-wt-D3 and the p5U-mut-D3 vectors was also observed when they were transfected into HEK293T cells (Fig. S2). Taken together, these results suggest that the CRQ G-quadruplex sequence in the context of its full-length 5′ UTR is responsible for the translational repression of CCND3 in cellulo.

Figure 5. CRQ G-quadruplex affects the cell cycle distribution and proliferation in HepG2 cells. (A) HepG2 cells were transfected with the indicated expression plasmids. Forty-eight hours later, the cells were harvested, and western blot analysis was performed. (B) Quantitative RT-PCR showing the relative mRNA levels of CCND3. The CCND3 mRNA level was normalized to that of GAPDH, and the value for cells transfected with the empty pcDNA3.0 plasmids (vector) was set to 1. (C) Forty-eight hours after HepG2 cells were transfected with the indicated expression vectors, the cells were stained, and DNA content was determined using flow cytometry. Top, representative histograms showing the cell cycle of transfected HepG2 cells. Bottom, percentages of cells in each phase of the cell cycle calculated from three independent experiments. (D) Twenty-four hours post-transfection, the cells were replated, and MTT assays were performed to evaluate cell proliferation.

Disruption of CRQ G-quadruplex formation promotes the G1/S cell cycle transition and cell proliferation

Because of the role of CCND3 in promoting cell cycle progression through the G1/S transition, we evaluated the potential of the CRQ motif in cell cycle and proliferation control. As shown in Figure 5A, HepG2 cells transfected with each of the three CCND3-expression vectors showed higher levels of p-Rb compared with those with the vehicle vector. Mutation or deletion of the CRQ sequence in the expression vector increased the levels of p-Rb, consistent with the expression change of CCND3 (Fig. 5A).

Since phosphorylation of Rb is critical for the G1/S cell cycle phase transition and cell proliferation, we next performed flow cytometry and MTT assays to determine the impact of the CRQ motif on the cell cycle distribution and cell proliferation of HepG2 cells, respectively. Overexpression of CCND3 under the control of the wild-type 5′ UTR resulted in a slight decrease in the number of cells in the G1 phase and a slight increase in the number of cells in the S phase (Fig. 5C). This change in cell cycle distribution was more significant when the CRQ motif within the 5′ UTR was mutated, with more cells having progressed through the G1 phase (Fig. 5C). The results were consistently observed not only in HepG2 cells, but also in HEK293T cells, regardless of whether the CDK6 was co-expressed or not (data not shown), demonstrating that the CRQ motif has an inhibitory effect on cell cycle progression. In addition, HepG2 cells transfected with the p5U-mut-D3 expression vector proliferate significantly faster than those with the vehicle vector or even the p5U-wt-D3 vector, especially 4 to 5 d after seeding (Fig. 5D), indicating that disruption of the CRQ motif in CCND3 mRNA promotes HepG2 cell proliferation.

Discussion

Emerging research has supported the role of CCND3 in the development of normal and cancer cells in addition to its basic role in cell cycle regulation and proliferation.^6-23 However, study of the regulation of CCND3 in cells is yet to be fully defined. In this study, we report that a cis-acting element, located between nucleotides 31 and 50 of the human CCND3 5′ UTR, forms a stable G-quadruplex structure and represses translation of a reporter gene and the CCND3 gene in human cell lines, and we show for the first time that this RNA G-quadruplex sequence is involved in the regulation of the cell cycle and proliferation. Our results uncover a novel mechanism regarding the regulation of the CCND3 gene.

Under physiological conditions, the CRQ motif folds into an extremely stable, parallel, intramolecular G-quadruplex in vitro. First, the parallel four-stranded structure of CRQ was demonstrated by CD spectroscopy from which the characteristic peaks at approximately 264 nm and 240 nm were observed (Fig. 2A and B), which is consistent with previously reported RNA G-quadruplexes,^45-53,62 as well as by RNase T1 footprinting in which the involvement of three G-quartet planes in the formation of this secondary structure was demonstrated (Fig. 3A). Second, the CRQ quadruplex displayed extreme stability when it was chelated by monovalent metal ions, such as K⁺ or Na⁺, as reflected by the low ∆G° at 37°C and the high T_m value (Table 2). Specially, in the presence of 1 mM KCl, the T_m value of CRQ is 73.27 ± 0.70°C, comparable to or even higher than published data of other RNA G-quadruplexes.^46,49,51,52 Resistance to RNase T1-mediated cleavage in the presence of KCl also indicated the high stability of this RNA G-quadruplex. Third, we calculated the T_m values of CRQ at various RNA concentrations and showed that the secondary structure was intramolecular, which was also supported by the cleavage pattern of CRQ by RNase T1. When the G-to-U mutations were introduced into the CRQ sequence, however, the characteristics for an RNA G-quadruplex were completely lost, similar to that of the TRF2 G-quadruplex.⁵¹

Accumulated evidence from bacteria to human cells supports the fact that stable G-quadruplex structures located in the 5′ UTR of mRNAs inhibit translation.^63-65 Specifically, a systematic study performed by Halder and coworkers revealed a universal translation-suppressing role of synthetic and natural G-quadruplex motifs introduced into 5′ UTRs in mammalian cell lines.⁶⁴ However, there are also reports that 5′ UTR RNA G-quadruplexes promote the initiation of translation.^54,55 In our study, we found that the CRQ motif from the 5′ UTR of CCND3 caused a more than 2-fold decrease of the renilla luciferase reporter gene in two human cell lines. Notably, the mRNA levels of the renilla luciferase were almost unchanged, suggesting that the CRQ sequence, when introduced into 5′ UTR, reduces gene expression by a translation-inhibitory mechanism similar to most of the known 5′ UTR G-quadruplexes.^45-52

Recent reports have emphasized the importance of studying 5′ UTR G-quadruplex-forming sequences in their natural context, including elements such as the flanking sequences and the positions of the 5′ UTR G-quadruplex-forming sequences relative to their 5′ caps.^45,61 Specifically, Kumari and coworkers showed that the NRAS G-quadruplex motif inhibited in vitro translation only when it was located close to the 5′ cap (within approximately 50 to 100 nt).⁶¹ The CRQ motif in the 5′ UTR of CCND3 is very close to the 5′ cap, with a distance of only 30 nt. To mimic the endogenous CCND3 mRNA, we cloned the CRQ motif-containing full-length 5′ UTR into the CCND3-expression vector and evaluated the effect on CCND3 expression. The full-length wild-type 5′ UTR resulted in a remarkable decrease of CCND3 expression compared with the mutated counterpart, whereas the mRNA levels of CCND3 were not influenced, further demonstrating that the CRQ motif inhibits translation in the context of its natural full-length 5′ UTR. Interestingly, the decrease of CCND3 expression by the insertion of the entire 5′ UTR in the expression plasmid was almost completely restored by the mutation of the CRQ motif (Fig. 5A), indicating that the CRQ G-quadruplex is the most important inhibitory element in the CCND3 5′ UTR that is responsible for its translation-repressing ability.

Although significant progress has been made in the identification of RNA G-quadruplexes and the exploration of their roles in the regulation of gene expression, little is known about the cellular effects of RNA G-quadruplexes. While a previous report suggested that the reduction in the Aurora A protein level during a G-quadruplex stabilizer-induced cell cycle arrest may result from the stabilization of a potential RNA G-quadruplex in its 5′ UTR, no direct evidence was provided to support the formation of this RNA G-quadruplex and the role of this sequence in cell cycle regulation.⁶⁶ We demonstrated in this study that the CRQ RNA G-quadruplex within the 5′ UTR of CCND3 plays a role in the G1/S phase transition of the cell cycle. Mutations that disrupt the formation of the G-quadruplex structure led to the production of more CCND3 protein from the expression vectors, which promoted the phosphorylation and inactivation of the Rb tumor suppressor gene and subsequently resulted in accelerated cell cycle progression and cell proliferation. Our results thus provide direct evidence for the cellular function of an RNA G-quadruplex. Because deregulation of the cell cycle and inappropriate cell proliferation could ultimately lead to cancer,⁶⁷ our findings may also have implications in the prevention and treatment of cancer.

As mentioned above, the CCND3 protein is critical in normal cellular processes and the development of cancer. Cellular levels of CCND3 can be regulated by different stimulus through different mechanisms. For example, deprivation of L-Arginine impaired cyclin D3 mRNA stability in activated T cells through arresting the protein synthesis of an RNA binding protein HuR,³⁷ whereas all-trans retinoic acid enhanced cyclin D3 expression in human T lymphocytes by stimulating IL-2-mediated signaling.⁶⁸ Our fingding that a novel cis-acting element within the 5′ UTR of the CCND3 gene represses its translation thus provides a new way for regulating CCND3 expression. However, the understanding of the formation or unfolding of this G-quadruplex structure in vivo requires further investigation. Until recently, several RNA G-quadruplex-binding proteins have been reported, including FMRP, Pat1, Stm1, XRN1p, RHAU and DHX9, which bind to and stabilize or unwind the G-quadruplex structures.^69-74 It is possible that under different cellular context and conditions, different G-quadruplex-binding proteins bind to the CRQ G-quadruplex structure and hence regulate the expression of CCND3. In addition, as G-quadruplex structures are becoming attractive targets for cancer therapeutics,⁶⁵ the existence of an RNA G-quadruplex in the 5′ UTR of CCND3 thus offers a site for the binding of small molecules, which provides a new opportunity for the modulation of CCND3 expression and the control of cell cycle progression and cancer development.

Materials and Methods

Oligonucleotides

HPLC-purified RNA oligonucleotides (Thermo Fisher Scientific) were dissolved in RNase-free water, and the concentrations were determined by measuring the absorbance at 260 nm using an ND-1000 spectrophotometer (Nanodrop Technologies). The RNA sequences were as follows:

CCND3-wt: GGGCGGCGGGCGGGCUGGGG

CCND3-mut: GAGCGGCGAGCGAGCUGAAG

Spectroscopy

For the spectroscopy measurements, RNA samples at a concentration of 5 μM were prepared in buffers containing 10 mM TRIS-HCl (pH 7.4) either in the absence or presence of a monovalent salt. The samples were annealed by heating at 90°C for 10 min and slowly cooling to 4°C at a controlled rate of 0.01°C/sec.

CD experiments were performed on a J-810 spectropolarimeter equipped with a Jasco Peltier temperature controller. Briefly, the sample in a 200 μl volume was transferred to a quartz cuvette with an optical path length of 1 mm, placed in the spectropolarimeter and allowed to equilibrate at 20°C for 5 min. In each experiment, a CD spectrum ranging from wavelengths of 220 to 320 nm was scanned five times at 50 nm/min with a 2 sec response time, 0.5 nm pitch and 1.71 nm bandwidth.

Thermal UV melting curves were obtained by heating the samples from 20°C to 90°C and melting back at a controlled rate of 1°C/min, while monitoring the 263 nm CD peak every 0.2 min. The melting temperature (T_m) values and other thermodynamic parameters were calculated using the Van’t Hoff method.⁷⁵

RNase T1 footprinting

RNA oligonucleotides were 5′-end-labeled with ³²P, as previously described,⁷⁶ and pre-folded, as described above. The RNase T1 digestion reaction was performed by incubating 30,000 c.p.m. RNA samples and 3 μg yeast tRNA with 1 unit of RNase T1 (TaKaRa) at 37°C for 5 min. Alkaline hydrolysis ladders were prepared by incubating 60,000 c.p.m. 5′-end-labeled RNA and 3 μg yeast tRNA in 15 μl alkaline hydrolysis buffer (50 mM sodium carbonate, pH 9.2) at 95°C for 10 min. The reactions were quenched by placing the samples on ice immediately and adding formamide dye buffer (95% formamide, 10 mM EDTA, 0.025% bromophenol blue and 0.025% xylene cyanol). An equal volume of each sample was loaded on a 20% polyacrylamide gel containing 8 mol/L urea, run at 45 V/cm at room temperature for 3 h, and then dried and exposed to a phosphorimager screen. The signals were visualized by a Typhoon 8600 imager (GE Healthcare).

Plasmid construction

To construct plasmids for the dual-luciferase assays, a DNA duplex carrying either the CRQ-wt or CRQ-mut sequence was inserted at the NheI restriction site within the psiCHECK-2 vector (Promega). The DNA duplexes were annealed using the following synthetic oligonucleotides:

CRQ-wt sense: CTAGCGGGCGGCGGGCGGGCTGGGGA

CRQ-wt antisense: CTAGTCCCCAGCCCGCCCGCCGCCCG

CRQ-mut sense: CTAGCGTGCGGCGTGCGTGCTGTTGA

CRQ-mut antisense: CTAGTCAACAGCACGCACGCCGCACG

For the construction of the expression plasmids, CCND3 cDNA clones with or without the entire 5′UTR (namely p5U-wt-D3 or pno5U-D3, respectively) were subcloned into the pcDNA3.0 vector between the HindIII and EcoRI restriction sites. A p5U-mut-D3 plasmid with the mutated variant of the G-quadruplex motif was constructed by replacing the sequence between the HindIII and BamHI restriction sites in the p5U-wt-D3 vector with the G/U mutations-containing DNA duplex as described above. The CDK6 cDNA clone was also subcloned into the pcDNA3.0 vector using the same restriction sites as in CCND3. All these expression plasmids had 6x His-tags fused to the C-terminus of the target genes. For all of the plasmids, the insertion of the correct sequences was verified by sequencing.

Cell culture and transfection

The human embryonic kidney cell line HEK293T and human hepatoma cell line HepG2 were obtained from the cell bank of the Chinese Academy of Sciences (Shanghai) and cultured in DMEM media supplemented with 10% FBS, 1% penicillin/streptomycin and 2mM glutamine at 37°C in the presence of 5% CO₂. Transfections were performed using Lipofectamine 2000 (Life Technologies) according to the manufacturer’s instructions.

Dual-luciferase reporter assays

The cells were reverse transfected with psiCHEK-2 plasmids. Forty-eight hours later, cells were lysed and renilla luciferase (Rluc) and firefly luciferase (Fluc) activities were measured with a GloMax 96 Microplate Luminometer (Promega) according to the supplier’s instructions. For each lysate, the value of Rluc was divided by the value of Fluc, and the ratios for CRQ-wt-psiCHECK-2 and CRQ-mut-psiCHECK-2 were normalized to that of the vehicle plasmid. Both the mean value and the standard deviation were calculated from at least three independent experiments.

Quantitative RT-PCR assays

The total cellular RNA was extracted from cells after a 48 h transfection using TRIzol reagent (Life Technologies). First-strand cDNA was synthesized using PrimeScript RT reagent Kit With gDNA Eraser (TaKaRa). Real-time quantitative PCR was conducted using a SYBR Premix Ex Taq Kit (TaKaRa) on an ABI StepOne Detection System (Life Technologies) in the presence of the appropriate set of primers: Rluc (forward: AGAGCGAAGAGGGCGAGAAA; reverse: AGGCGTTGTAGTTGCGGACA); Fluc (forward: CCCTGCTAACGACATTTACAACG; reverse: TCGAAAGACTCTGGCACGAAG); CCND3 (forward: GCTTACTGGATGCTGGAGGTATGTG; reverse: AGTTTTTCGATGGTCAGGGGC); GAPDH (forward: TGACCTGCCGTCTAGAAAAACC; reverse: GCCAAATTCGTTGTCATACCAGG). Negative controls were included for each primer pair, and each PCR reaction was performed in triplicate. The baseline and quantification cycle were automatically determined using StepOne software. Both the mean values and the standard deviations were calculated from at least three independent reverse transcription assays.

Western blot analysis

HEK293T and HepG2 cells were seeded in 48-well plates and transfected CCND3 expression plasmids. Forty-eight hours after transfection, cells were lysed and subjected to western blot analysis as previously described⁷⁶ using anti-CCND3 (#2936), anti-p-RB (#9301) and anti-GAPDH (#2118) antibodies from Cell Signaling Technology and anti-His (010–21861) antibody from Wako. GAPDH was used as a loading control.

Cell cycle analysis

The cells were harvested 48 h after transfection, washed once in PBS, and stained with Krishan’s reagent (0.05 mg/mL propidium iodide, 0.1% Na citrate, 0.02 mg/mL RNase A, 0.3% NP-40) at 37°C for 30 min. Cell fluorescence was measured on a FACSCalibur flow cytometer (Becton Dickinson). Nuclear debris and overlapping nuclei were gated out, and the DNA content was analyzed based on FL2-A vs. FL2-W linear plots using FlowJo software.

Cell proliferation assays

HepG2 cells were counted with an Auto T4 cellometer (Nexelom Bioscience) 24 h after transfection and seeded in 96-well cell culture plates with 1,000 cells/well. MTT assays (Promega) were performed every 24 h from the day of seeding for 5 consecutive days, as instructed by the manufacturer.

Glossary

Abbreviations:

CCND3

cyclin D3

CDK

cyclin-dependent kinase

CRQ

CCND3 RNA G-quadruplex

UTR

untranslated region

G

guanine

CD

circular dichroism

Disclosure of Potential Conflicts of Interest

No potential conflicts of interest were disclosed.