It is known that the guanine-rich strands in proto-oncogene promoters can fold into G-quadruplex structures to regulate gene expression. An intramolecular parallel G-quadruplex has been identified in MET promoter. It acts as a repressor in regulating MET expression. However, the full guanine-rich region in MET promoter forms a hybrid parallel/antiparallel G-quadruplex structure under physiological conditions, which means there are some antiparallel and hybrid parallel/antiparallel G-quadruplex structures in this region. In the present study, our data indicate that g3-5 truncation adopts an intramolecular hybrid parallel/antiparallel G-quadruplex under physiological conditions in vitro. The g3-5 G-quadruplex structure significantly stops polymerization by Klenow fragment in K+ buffer. Furthermore, the results of circular dichroism (CD) spectra and polymerase stop assay directly demonstrate that the G-quadruplex structure in g3-5 fragment can be stabilized by the G-quadruplex ligand TMPyP4 (5,10,15,20-tetra-(N-methyl-4-pyridyl) porphine). But the dual luciferase assay indicates TMPyP4 has no effect on the formation of g3-5 G-quadruplex in HepG2 cells. The findings in the present study will enrich our understanding of the G-quadruplex formation in proto-oncogene promoters and the mechanisms of gene expression regulation.
MET proto-oncogene encodes the hepatocyte growth factor receptor (HGFR or MET), which can be activated by its ligand hepatocyte growth factor (HGF). METs are involved in several diverse physiological events through Ras/MAPK, PI3K/Akt, c-Srcand STAT3/5 signaling pathways . However, abnormal activation of MET can result in cell proliferation, antiapoptosis, angiogenesis, altered cytoskeletal function, metastasis and poor prognosis [2–5]. In many types of human malignancies, including cancers of kidney, liver, stomach, breast and brain, METs are abnormally activated and can promote tumor growth, angiogenesis and cancer metastasis . Hence, MET becomes an attractive target for cancer therapy [6,7].
G-quadruplex structure can be formed in single-stranded guanine-rich nucleic acid fragments, which is a kind of noncanonical secondary structure. It is a four-stranded nucleic acid that contains stacked G-quartets of four guanines connected by Hoogsteen hydrogen bonds [8,9]. In the human genome, these G-quadruplex structures can play different roles. Recently, several studies have revealed that G-quadruplex structures in proto-oncogene promoters are capable of regulating gene transcription and can be used as candidates for cancer therapy [10–17]. The G-quadruplex ligand TMPyP4 binds and stabilizes G-quadruplexes in human promoter sequences, resulting in inhibition of gene transcription [11,12,17–19]. Currently, TMPyP4 and its analogs are under study as potential anticancer agents.
Our previous research has discovered that Pu23WT fragment in MET promoter (-48 to -26 upstream of transcription start site) can adopt an intramolecular parallel G-quadruplex structure, which functions as a transcriptional repressor of MET . Moreover, the cationic porphyrin TMPyP4 can stabilize the Pu23WT G-quadruplex. It has been shown that Pu23WT G-quadruplex could be a candidate target for anticancer treatment. However, CD spectrum in the present study showed that the entire G-rich strand in MET proximal promoter (met, -91 to -26 upstream of transcription start site) folded into a hybrid parallel/antiparallel G-quadruplex structure under physiological conditions (Table 1 and Figure 1), which means there are some potential antiparallel or hybrid parallel/antiparallel G-quadruplex structures in this region. These relationships led us to truncate the met strand into several short fragments (Table 1) to investigate the G-quadruplex structures and their biological effects on MET transcription.
CD spectra of the truncations of met and mutants of g3-5
|Name .||Sequence .|
|Name .||Sequence .|
G-islands were underlined. The guanine (G) was replaced in G-islands by the thymine (T) in bold.
Preparation of oligonucleotides
The DNA sequence of human MET promoter was acquired in GenBank database [GenBank GI: 224589819, NC_000007.13]. The truncated oligonucleotides and their mutants were synthesized by Sangon Biotech Co., Ltd. (Shanghai, China). The potentiality of G-quadruplex formation was evaluated by the online software Quadparser (http://www.quadruplex.org/?view=quadparser). The oligonucleotides were prepared at 5 μM in 20 mM sodium cacodylate with a 140 mM KCl buffer (pH 7.4) unless specified otherwise.
Circular dichroism spectroscopy and ultraviolet absorption spectroscopy
The CD spectra and the ultraviolet (UV) absorbance spectra at 295 nm were acquired according to the procedures described in our previous study . A slow melting–annealing cycle was performed prior to each CD and UV measurements.
The CD spectra were recorded over a wavelength range of 220–350 nm at room temperature for at least three independent scans. They were baseline-corrected by subtracting the signal contributions of the buffer. The 5 μM oligonucleotide samples were prepared with different concentrations of KCl, NaCl, MgCl2, CaCl2, and TMPyP4. When recording the TMPyP4-dependent CD spectra, the samples were prepared with 10 mM KCl.
The UV absorbance spectra at 295 nm were used to analyze the thermal stability of samples. The spectra were recorded between 20 and 80°C with 1°C increment during heating and cooling processes. A slow heating/cooling rate of 0.3°C/min was applied to ensure the thermal transition under an equilibrium condition. The melting and annealing curves were averaged with three independent measurements. The data were normalized by the formula: FF(θ) = (Y − Ymin)/(Ymax− Ymin), where θ is the fraction of folded G-quadruplex at a given temperature, Y is the signal at 295 nm. The Tm or Ta values were calculated according to the normalized profiles (fraction folded profiles) at θ = 0.5 .
g3-5 and its mutants were denatured at 95°C for 10 min in the 140 mM KCl, then cooled down slowly to room temperature and left overnight. All the samples were analyzed in 20% non-denatured polyacrylamide gels at 4°C, which were stained with Stains-all (Sigma-Aldrich, U.S.A.) in darkness. The standards were the 35 nt single-stranded oligonucleotide and the 20 bp double-stranded oligonucleotide. The images were recorded using a Personal Scanner (FangZheng, China).
Polymerase stop assay
The g3-5 strand was linked with either a long (AGCCTAGCTGCAAGTATGTTCACG) or short (ATGTTCACG) tail sequence at the 3′-end to formulate templates for LTg3-5 and Tg3-5 (Table 2). The primer 5′-d(ATCCAGAAGTACGTGAACAT)-3′ contained a complementary sequence to LTg3-5 and Tg3-5 tails . LTg3-5 or Tg3-5 was denatured with primer then annealed to form asymmetric duplex. The two-way-extension of the asymmetric duplex by Klenow fragments was kept for 30 min at room temperature. We then employed the same procedure shown in our previous work . The two-way-extension can be stopped by the G-quadruplex in the template.
|Name .||Sequence .|
|Name .||Sequence .|
The additional tail sequences of g3-5 were shown as italic.
The complementary sequences between template and primer were underlined.
Dual luciferase assay
Recombinant DNA 0.7met-pGL4.10 was the firefly luciferase reporter driven by MET promoter, which contained the full G-rich region within MET promoter . Site-directed mutations in met strand (G to T, Table 3) were performed using the QuikChange® Lightning Site-Directed Mutagenesis Kit (Stratagene, U.S.A.). To assess the effect of TMPyP4 on MET promoter activity, HepG2 cells were separately transfected with 0.7met-pGL4.10 constructor, the mutant g3-1-mut, and the mutant g3-5-mut. The post-transfected HepG2 cells were treated with 50 μM TMPyP4 for 24 h. The cells were protected from light during the period of TMPyP4’s treatment. The activities of firefly luciferase and renilla luciferase were assayed by the Dual Luciferase Reporter Assay System E1910 (Promega, U.S.A.) on Infinite M200 Multi-plate Reader (Tecan, Switzerland). Three independent experiments were done in triplicate.
|Name .||Sequence .|
|Name .||Sequence .|
The guanine (G) was replaced in G-islands by the thymine (T) in bold.
The data were presented as the mean ± SD in at least three separate experiments. Statistical analysis was performed using one-way ANOVA by GraphPad Prism 5.0 software. The statistically significant difference was defined as P<0.05.
Results and discussion
g3-5 forms an intramolecular hybrid antiparallel/parallel G-quadruplex structure in K+ solution
The met strand in MET approximal promoter and its truncated fragments were listed in Table 1. Each truncation was composed of at least four consecutive G-islands. A serial of mutants (G to T) of g3-5 were used to decipher which Gs were potentially involved in the formation of G-quadruplex structure.
We recorded CD spectra of the met strand and its truncated fragments in 140 mM K+ buffer (pH 7.4) (Figure 1a). The CD spectrum of met exhibited a positive peak at 264 nm, a negative peak at 240 nm, and a shoulder peak approximately 290 nm, indicating the formation of a hybrid parallel/antiparallel G-quadruplex structures. The spectrum of met-m (mutant of met) showed a positive peak at 280 nm and a negative peak at 250 nm. There was no characteristic CD spectrum in met-m. Truncation g3-1/Pu23WT, g3-2, and g3-3 exhibited a positive peak approximately 260 nm and a negative peak approximately 240 nm. In terms of the CD spectra, they all adopted parallel G-quadruplex structures. The truncation of g3-4 did not show any characteristic of G-quadruplex spectrums. However, the CD spectrum of g3-5 was uniform with that of met, exhibiting a positive peak at 264 nm, a negative peak at 240 nm, and a shoulder peak approximately 290 nm. Both of them formed a hybrid parallel/antiparallel G-quadruplex structure. Due to the hybrid parallel/antiparallel G-quadruplex in the full G-rich strand of met, there had to contain antiparallel or hybrid parallel/antiparallel G-quadruplex structures. According to CD spectra, only g3-5 was able to adopt a hybrid parallel/antiparallel G-quadruplex structure. Moreover, g3-5 truncation got high score when assessing the ability to fold into G-quadruplex via the online software Quadparser. In addition, UV absorbance spectra of g3-2, g3-3, and g3-4 showed no G-quadruplex structures in these strands (Supplementary Figure S1). Hence, g3-5 and g3-1/Pu23WT made contributions to the formation of the hybrid parallel/antiparallel G-quadruplex in MET promoter.
The g3-5 contained five G-islands, four of them contain three or four guanines. Only one G-island was composed of GG repeat (Table 1). Consequently, there would be several ways to form G-quadruplex structure using different G-islands. In order to decipher the number of stacked G-tetrads and the guanines involved in g3-5 G-quadruplex structure, we compared the CD profiles of g3-5 and its mutants (Figure 1b). The CD spectra were collected in the same conditions. The mutant g3-5-m2 exhibited a very similar CD spectrum with that of g3-5, suggesting a hybrid parallel/antiparallel G-quadruplex structure in K+ solution. The CD spectrum of g3-5-m3 showed a positive peak approximately 260 nm and a negative peak approximately 240 nm, which indicated a parallel G-quadruplex structure. Other mutants could not fold into any G-quadruplex structure according to their CD spectra. Therefore, the Gs in the g3-5-m2 mutant formed the hybrid parallel/antiparallel G-quadruplex structure of g3-5 in K+ buffer.
In the K+ buffer, the absorbance spectra at 295 nm of g3-5 showed a hypochromic effect except a small evaporation artifact between the melting and annealing curves (Figure 2a), which suggested the process was indeed in thermodynamic equilibrium. This reversible phenomenon supported the formation of an intramolecular G-quadruplex. Tm and Ta values of g3-5 in 140 mM KCl were determined by the folded fraction profiles at θ = 0.5 (Figure 2b), which were 47.46 ± 0.27 and 45.64 ± 0.31°C. The Tm value of g3-5 was much lower than that of g3-1/Pu23WT (81.07 ± 0.35°C) in 140 mM KCl .
The melting and annealing curves of g3-5 at 295 nm
The mobilities of g3-5 and its mutants were printed on the non-denatured polyacrylamide gel containing 140 mM KCl at 4°C. The g3-5 strand migrated faster than the 35 nt control strand (lane 3, Figure 3), indicating that g3-5 folded into a compact intramolecular structure, and g3-5-m2 ran faster than g3-5, which also appeared as a single species (lanes 5, Figure 3).
Non-denatured PAGE data of g3-5 and its mutants
In conclusion, the g3-5 strand could fold into an intramolecular hybrid parallel/antiparallel G-quadruplex structure by using the Gs in the g3-5-m2 mutant in K+ buffer. It was assumed that g3-5 and g3-1/Pu23WT G-quadruplexes coexisted in met strand, resulting in a hybrid parallel/antiparallel G-quadruplex, even though the loops in this G-quadruplex structure were unusual .
K+ facilitates the G-quadruplex formation in the g3-5 sequence
Cations played an important role in changing the G-quadruplex topology structure and its stability. The sequence of AG3TTAG3TTAG3TTAG3 adopted a basket-type antiparallel G-quadruplex structure in Na+ solution [24,25]. However, it folded into a hybrid antiparallel/parallel G-quadruplex structure in K+ solution [25–27]. In addition, the native thrombin binding aptamer (TBA) in Li+, Na+, Cs+, Ca2+, and Mg2+ solutions showed the same CD spectra with the K+ condition . No matter what topological structure TBA forms, K+, Na+, Mg2+, and Ca2+ can stabilize the G-quadruplex structure [29–33]. To explore the effects of metal ions on the g3-5 G-quadruplex formation, their CD spectra were recorded in KCl, NaCl, MgCl2, or CaCl2 solution. A similar CD spectra of g3-5 with lower peaks, were found in the Na+ and Mg2+ buffer, as well as the K+ buffer (Figure 4a). However, the G-quadruplex structure of g3-5 disappeared in the Ca2+ buffer (Figure 4a). The CD spectrum of g3-5 in 140 mM KCl solution was stronger than that in 50 mM KCl solution (Figure 4b), implying that 140 mM K+ promoted formation of G-quadruplex structure. In contrast, CD spectra of g3-5 in 200 mM Mg2+ and Na+ showed weaker profiles (Supplementary Figure S2) than that in 50 mM buffer. However, the melting and annealing processes of g3-5 at 295 nm in the buffer of Mg2+ or Na+ exhibited no regular characteristic hypochromic curves (data not shown), which meant no stable G-quadruplex structure. Taken together, the strong elliptic signal of g3-5 CD spectrum in K+ solution supported that K+ favored the formation of the hybrid antiparallel/parallel G-quadruplex structure .
CD spectra of g3-5 in the presence of different cations
G-quadruplex structure can arrest DNA extension [14,31,35–37]. DNA polymerase stop assay was performed to further test the key role of K+ on the G-quadruplex formation and the effect of G-quadruplex structure on DNA extension. We used the long tail template LTg3-5 and the short one Tg3-5, after considering spatial hindrance of the additional tail in the formation of G-quadruplex structure (Table 2). We found both of them could form partial hybrid double-strands with the primer. The hybrid double-strand was then bidirectionally extended by Klenow fragment at room temperature . The extension products were analyzed with 15% polyacrylamide gel. There was no difference between these two templates on the formation of G-quadruplex structure. The extension products of LTg3-5 in the NaCl buffer were similar to those in the control buffer (Figure 5a,c). However, KCl significantly reduced the products of LTg3-5 (Figure 5a,c). It was revealed that K+ could facilitate the G-quadruplex formation of g3-5 to stop DNA extension, but Na+ could not. Moreover, extension products were decreasing with the increasing concentration of K+ (Figure 5b). The results were entirely consistent with our CD and UV spectral data. In our previous study, CD spectra and polymerase stop assay results revealed that K+ was essential for the formation of Pu23WT G-quadruplex structure in MET promoter and the inhibitory effect of DNA polymerization . Therefore, it may be assumed that there existed two mixed types of G-quadruplex structures folded in equilibrium by g3-1/Pu23WT and g3-5 in cells where the concentration of K+ was high enough.
Klenow fragment polymerase stop assay of LTg3-5 and Tg3-5
TMPyP4 attenuates the polymerization products by stabilizing the g3-5 G-quadruplex structure
It has been demonstrated that the cationic porphyrin TMPyP4 can stabilize the G-quadruplex structure [38–41]. In order to assess the effect of TMPyP4, CD spectra, polymerase stop assay, and MET-dependent firefly luciferase activities of g3-5 strand were analyzed. CD spectra of g3-5 hybrid antiparallel/parallel G-quadruplex structure became more dominant when TMPyP4 was added into the buffer (Figure 6a). The stronger elliptic peak arose in the higher concentration of TMPyP4, which indicated a TMPyP4 concentration-dependent G-quadruplex stability of g3-5. Furthermore, the effect of TMPyP4 on Tg3-5 extension was analyzed using DNA polymerase stop assay (Figure 6b). It showed that many extension products of Tg3-5 existed in the absence of TMPyP4. However, the extension products of Tg3-5 were decreasing with the increasing concentration of TMPyP4. The products almost disappeared in the 15 μM TMPyP4 (Figure 6b). We also assessed the G-quadruplex function of g3-5 in cells. 0.7met-pGL4.10 is the MET-dependent firefly luciferase recombinant plasmid, whose insertion fragment is the promoter sequence of the MET gene located at −662 to +47 , where it was named WT. The mutants of WT were then constructed according to Table 3. The transfected HepG2 cells were treated with 50 μM TMPyP4 for 24 h, after which the relative firefly luciferase activities were detected. The relative luciferase activities of WT plasmid in the transfected HepG2 cells decreased significantly to 44.72% due to the stable G-quadruplex structure (Figure 6c). However, TMPyP4 had little effect on the g3-1-mut plasmid. But TMPyP4 still reduced the relative firefly luciferase activity to approximately 65.84% in the g3-5-mut plasmid. Therefore, it was concluded that the full MET promoter retained the G-quadruplex structure, and g3-1/Pu23WT G-quadruplex was the primary regulatory element for MET proto-oncogene expression in the cells. Though the G-quadruplex structure of g3-5 arrested DNA extension in vitro, g3-5 strands in the g3-1-mut plasmid could not respond to the treatment of TMPyP4. It is possible there could exist an interplay between g3-1/Pu23WT and g3-5 G-quadruplex structures due to a steric effect in recombinant plasmid. The Tm value of g3-5 was 47.46 ± 0.27°C in 140 mM KCl, which was much lower than that of g3-1/Pu23WT (81.07 ± 0.35°C) . Hence, the g3-1/Pu23WT G-quadruplex structure was much more stable than g3-5 G-quadruplex. Furthermore, there was one Sp1 binding site located in g3-1/Pu23WT, which bound to Sp1 factor to initiate MET transcription [42,43]. The stable G-quadruplex of g3-1/Pu23WT strongly prevented the binding of Sp1 factor to its binding site, resulting in a decrease in gene transcription. Therefore, the G-quadruplex structure in g3-1/Pu23WT fragment played a crucial role in MET expression.
Inhibitory effect of g3-5 G-quadruplex
Our study revealed that the g3-5 strand adopted an intramolecular hybrid antiparallel/parallel G-quadruplex structure. And the G-quadruplex could stop DNA polymerization reaction in K+ and TMPyP4-containing buffer in vitro, although its effect on gene expression in HepG2 cells was very weak.
In the present study, we found a novel G-quadruplex motif formed in the Human MET promoter region. The G-rich strand g3-5 could fold into an intramolecular hybrid antiparallel/parallel G-quadruplex structure under physiological conditions in vitro. The G-quadruplex structure in g3-5 strongly arrested the DNA extension in the presence of K+ and TMPyP4. However, this G-quadruplex structure could not decrease the luciferase activities in HepG2 cells. All the results showed that there existed a stable G-quadruplex structure in g3-5 in vitro, which could not act as a repressive element when it coexisted with the more stable g3-1/Pu23WT G-quadruplex structure. The findings in the present study will enrich our understanding about the G-quadruplex formation and gene expression regulation in the proto-oncogene promoters.
This project was funded by National Natural Science Foundation of China ; Doctoral Scientific Research Foundation of Liaoning Province ; and China Postdoctoral Science Foundation [2017M611286].
J.Y. wrote the manuscript and performed the experiments. D.Z. critically revised the manuscript. L.D. and S.P. conducted data processing and statistical analysis. F.H. prepared the figures. J.Y. and Y.G. conceived and designed the experiments. All authors read and approved the final manuscript.
The authors declare that there are no competing interests associated with the manuscript.