1H and 31P NMR of 1 with double helix B-DNA d(CGTACG)2 and d(AAGAATTCTT)2.
The NMR spectra of both the self-complementary oligonucleotides d(CGTACG)2 (“CG”) and d(AAGAATTCTT)2 (“AATT”) display signals in a region ranging from 12 to-13.5 ppm, which are characteristic of the NH imino protons of CG and AT base pairs. The presence of these signals confirms that both oligomers adopt, in Na+ solution, a double helix conformation. For this reason, they were used as models for CG- and AT-rich sequences, respectively.
The phosphorus spectra of “CG” with 1 showed a low-field shift variation of G2pT3, T3pA4 and C5pG6 signals (Figure. 2a, Table 1). It is known that 31P resonance is a sensitive probe to detect changes in the phosphoribose chain of the oligonucleotides due to the intercalation process; in fact, the chemical shift variation of the 31P resonances reflects a deformation at the level of P-O(5’) and P-O(3’) bonds. As a consequence of the intercalation of a ligand into the oligonucleotide double helix, the alfa= O(3’)-P–O(5’)-C(5’) and zeta = C(3’)-O(3’)–P-O(5’) angles change from a gauche, gauche conformation (-60° and -90°) to a gauche, trans conformation (-60°, +180°). This conformational change is usually associated with a low-field shift up to 1.0–2.5 ppm for 31P resonances.19, 20 In our case the lower Δδ values (-0.2 ppm) found for G2pT3, T3pA4 and C5pG6 suggest that 1 is bound to the oligonucleotide, however through a partial intercalation binding mode, at these three sites, with a possible exchange among them. The addition of 1 to “AATT” induced insignificant chemical shift variation of the phosphate signals in the 31P NMR spectra (Figure 2b and Table 1). This is the proof that an intercalation process did not occur. Nevertheless, the A1pA2, A5pT6 and A4pA5 signals became very broad even at R=0.25. The 1H NMR titration with 1 gave different results for the two oligonucleotides. Broadening of the imino NH signals with no relevant chemical shift variation was observed for “AATT” resonances, whereas only the H1’ anomeric protons belonging to “AATT” tract were slightly perturbed (Δδ = -028/-0.11 ppm) (Table S1). A line broadening was observed for the aromatic protons of A5 and T6, while the H2 proton of A4, which lies in the minor groove, splitted into two signals just at R=0.25. This suggests the formation of both a free and a bound species; for R ≥1.0 the free species disappears (Figure. S1).
The addition of 1 to a solution of “CG” presented a significant up-field chemical shift variation either for imino, aromatic or anomeric protons (Figure S2). The most affected were the protons of the G2 and C5 units, in line with the interaction at these sites. The unique exception was represented by the T3 anomeric proton (Δδ +0.19 ppm) (Table S1). A possible explanation of this deshielding effect is that the intercalation of the ligand at G2pT3 places the piperidine moiety in the minor groove at the level of T3.
2D-NOESY experiments did not show intermolecular Nuclear Overhauser Effect (NOE) contacts between the ligand and both oligonucleotides, probably due to a weak interaction or to a rapid exchange between different binding sites. Therefore, it was not possible to build a model for the complexes with 1.
Table 1. 31P chemical shift assignments of phosphate in the free oligonucleotides and in the complex with 1a,b. Oligonucleotide “CG” corresponds to duplex d(CGTACG)2 and “AATT” is duplex d(AAGAATTCTT)2.
“CG”
|
δ (ppm)
|
Δδc
|
“AATT”
|
δ (ppm)
|
Δδc
|
C1pG2
|
-0.27
|
+0.07
|
A1pA2
|
broad
|
n.d.
|
G2pT3
|
-0.49
|
+0.21
|
A2pG3
|
-0.25
|
+0.07
|
T3pA4
|
-0.13
|
+0.26
|
G3pA4
|
-0.5d
|
0.00
|
A4pC5
|
-0.54
|
-0.07
|
A4pA5
|
broad
|
n.d.
|
C5pG6
|
+0.04
|
+0.24
|
A5pT6
|
broad
|
n.d.
|
-
|
|
|
T6pT7
|
-0.45d
|
+0.05
|
-
|
|
|
T7pC8
|
-0.38
|
+0.07
|
-
|
|
|
C8pT9
|
-0.50d
|
0.00
|
-
|
|
|
T9pT10
|
-0.38
|
0.00
|
a Measured at 15°C in ppm (δ) from external DSS. Solvent H2O-D2O (90:10 v/v), of 0.1 M NaCl and 10 mM sodium phosphate buffer solution, pH = 7.0 bR = 2.0. cΔδ = δbound – δfree. dThe assignment might be interchanged.
Overall, these results show that 1 partially intercalates with a CG-rich sequence, whereas it gives a slight external interaction with an AT-rich sequence. In conclusion, the interaction of 1 with double helix oligonucleotides can be considered not relevant.
However, these findings cannot rule out an interaction with G-quadruplex DNA structures, whose stabilization has been found to activate PARP-1 enzyme.14 Thus, we focused our study on the interaction of selected compound 1 with G-quadruplex structures of telomeres and proto-oncogenes.
Interaction of 1 with telomere d(TTAGGGT)4 quadruplex.
The titration of the oligonucleotide solution with 1 induced significant line broadening of G4 NH signal and less pronounced broadening of the aromatic resonances of the three guanines (Figure 3a). An upfield shift variation was observed for the imino, aromatic and anomeric proton signals of all units, except for T7 (Table S2). Also the resonances of the ligand gave an upfield shift. They were partially overlapped to the signals of the oligonucleotide, however they were identified by a TOCSY experiment at 7.02 and 7.49 ppm (pyrrole moiety), at 7.09 and 7.38 ppm (phenyl and pyridine moieties). NOE contacts were found for A3 H8 and G4 NH with the pyrrole protons. Other NOE interactions involved NH and H8 of G6 with both pyrrole and aromatic protons of the ligand (Table 2) (Figure 3b). These results indicate that 1 binds to the G-quadruplex at two sites: between A3 and G4 and over G6.
The interaction of 1 with the quadruplex d(TTAGGGT)4 was also investigated using the molecular docking technique, followed by optimization via molecular dynamics (MD). The molecule was docked at sites A3-G4 and G6-T7. In both cases the ligand does not adopt a center-symmetrical stacking interaction with the upper and lower tetrads, but it is rather displaced towards two of the four residues (Figure 4).
Table 2. Intermolecular NOE in the complex of 1 with d(TTAGGGGT)4 a
|
A3G4 binding site
|
NOE
|
d (Å)b
|
1
|
(TTAGGGT)4
|
|
H2
|
A3H8
|
5.01
|
H3
|
A3H8
|
3.61
|
H aromatic/Phe
|
A3H8
|
4.37, 4.60
|
H2
|
G4H1
|
4.26
|
|
|
|
|
G6T7 binding site
|
H2
|
G6H1
|
4.97
|
H3
|
G6H1
|
3.93
|
H aromatic/Phe
|
G6H1
|
3.69
|
H aromatic/H3
|
G6H8
|
4.57, 4.79
|
aAcquired at 25°C in H2O-D2O (90:10 v/v), 25 mM K-phosphate buffer, 150 mM KCl, 1 mM EDTA, pH 6.7. bDistances obtained by molecular modelling of the complex.
At the AG intercalation site the 7-azaindole portion of the ligand is inserted between A3 and G4. In this intercalation site the complex is also stabilized by a hydrogen bond between the NH2 group of A3 and the aromatic nitrogen of the pyrrole[2,3-b]pyridine moiety, at a distance of 2.98 Å. The piperidine ring is oriented outward from the quadruplex, with the quaternary nitrogen forming a hydrogen bond with N7 of A3 (2.63 Å), a cation-π interaction with G4, and a ionic interaction with OP2A3 (figure 4B).
At the GT intercalation site, the 7-azaindole moiety of the ligand gives rise to π-π stacking interactions with G6 and T7. In this case, the complex is stabilized by a hydrogen bond between O4 T7 and the CONH2 group of the ligand, at a distance of 2.36 Å. The -CONH2 group itself is locked in position by an intra-molecular hydrogen bond with the pyridine nitrogen (2.24 Å). Again, the piperidine ring points outward from the quadruplex, with the quaternary nitrogen forming a hydrogen bond with N3 of G6 (2.58A) and an ionic bond with OP1T7 (Figure 4C).
The best docked conformations of the complexes at the A3G4 and G6T7 intercalation sites are in good agreement with the reported NOE contacts (Table 2).
Interaction of 1 with G-quadruplex Pu22T14T23 sequence.
Pu22T14T23 gave high quality spectra in K+ solution in comparison with wild-type sequence.21,22 For this reason, it has been chosen as a good model to study ligand-quadruplex interaction. 1H NMR titration experiments were performed by adding increasing amounts of ligand to Pu22T14T23 solution, with ratios R=[ligand]/[DNA] ranging from 0 to 2.5. The NMR data indicated a single G-quadruplex conformation for the complex, with each proton showing a single chemical shift value and NOEs characteristic of the three G-tetrad stacked structure (Table S3 and Table S4). The titration experiment proved a chemical shift perturbation of the imino protons (fig. 5). The largest perturbations were observed for G18 and G22 (3’-end) and for G7, G11, G16 and G20 (5’-end). Less relevant perturbations were observed for the imino protons of G8, G12, G17 and G21, in the central guanine tetrad. No relevant chemical shift variation was observed for the residues located in the loops of the G-quadruplex, such as A15, T14 and T19 (Table S3). This excludes the interaction of 1 with the groove. These findings suggest that 1 stacks on the 3’ and 5’ sites of Pu22T14T23.
NOE contacts were found between the ligand and the imino NH protons of G13, G18 (3’-end), G7, G16 and G20 (5’-end) of the quadruplex (Figure 6). The aromatic protons of the ligand could not be unambiguously assigned because of their overlapping. However, the resonances of H2 and H3 of the pyrrole moiety were identified by TOCSY experiments at 7.49 and 7.04 ppm, respectively. These protons show NOE contacts with the imino NH of G16 and G18 (H3) and with G7 and G20 (H2) (Table 3). Other resonances of the ligand were identified at 7.38 and 7.55 ppm, showing a NOE interaction with NH G13 of the quadruplex. No NOEs between the ligand and the flanking chains were detected.
Table 3. Intermolecular NOEs in the complexes of 1 with Pu22T14T23a and distances from modeling.
3’-binding site NOE
|
d (Å )b
|
1
|
Pu22T14T23
|
|
H4, H5
|
G13H1
|
5.09, 5.59
|
H3
|
G18H1
|
5.50
|
5’-binding site NOE
|
|
H2
|
G7H1
|
5.55
|
H3
|
G16H1
|
3.74
|
H3
|
G5H8
|
4.52
|
H2
|
G20H1
|
4.89
|
aAcquired at 25 °C in H2O-D2O (90:10 v/v), 25 mM K-phosphate buffer, 70 mM KCl, 1 mM EDTA, pH 6.9. bDistances obtained by molecular modelling of the complex.
In order to obtain a three-dimensional model congruent with the NOE contacts we performed a molecular docking experiment, followed by MD calculations (fig.7). At 5'-end, 1 is stabilized by an extensive network of π–π interactions involving the underlying 5'-end G-tetrad, with the pyrrole[2,3-b]pyridine moiety located near the center of the tetrad (Figure. 7(a) and (b)). The above cited aromatic rings interact with the π systems of G5, G11 and G16. They are held in place by the cation-π interaction between the potassium ion and the pyrrole moiety (4.96 Å). Also the central phenyl ring creates π–π interactions with G11 and G16, while the piperidine ring is oriented outside the system, towards G11, T14 and A15, without giving rise to observable interactions. The -CONH2 group is coplanar with the pyridine aromatic system and positioned above G16, without giving noteworthy interactions with Pu22.
At 3'-end, the complex is also stabilized by a dense network of π–π interactions involving all the guanines of the tetrad: specifically, the pyrrole[2,3-b]pyridine moiety interacts with the π systems of G18 and G13 (Figure.7 (c) and (d)). The phenyl group of 1 forms π–π interactions with the G13 unit as well. This is complemented by a further π–π interaction with the G9 aromatic system. The CONH2 group is oriented towards G22 and T23, forming two hydrogen bonds: one with O6G22 (2.84 Å) and the other with O4T23 (3.02 Å). The piperidine ring is oriented outside the system, in the area between G9, G13 and A25, and does not present particular interactions with Pu22. In both 3’-end and 5’-end positions, the piperidine ring is arranged along the main groove of Pu22, with a docking score difference in favour of the complex in 3' (ΔE = 5.77 kcal/mol). The best docked conformations of the complexes at 5' and 3' are in good agreement with the reported NOE contacts (Table 3).
The flanking chains at the two terminals in the complex, even though no NOE interactions with the ligand were detected, showed significant chemical shift variations, especially in the segment T23-A24-A25 at the 3’-end. Many resonances move upfield: T23 Me (Δδ =-0.28 ), T23 H6 and H1’ (Δδ= -0.11 and -0.30); H8 of A24 and A25 (Δδ= -0.38 and -0.32). Other resonances move low-field, the most significant being the anomeric protons H1’of A25 and A24 (Δδ +0.57 and +0.25, respectively), and a slight deshielding of G9 NH (Δδ +0.10). This finding suggests that the three units T23, A24 and A25 do not prevent the binding. The ligand, positioned on the 3’-end tetrad, changes the architecture of the tail of this terminal. Actually, the structure of the free nucleotide Pu22T14T23 shows that A25 folds back to form a base pair with T23, thus protecting the external 3’-end G-tetrad. The entry of the ligand breaks the Hoogsteen-type H-bond between T23 and A25,23 pushing away T23 toward the top of G22, thus experiencing the stacking effect of this guanine. This is highlighted by the up-field shift of T23 protons, being also in line with the model. The A25 unit is no more folded over the G9 aromatic moiety, as in the free nucleotide.23 This justifies the deshielding of G9 NH and of the anomeric H1’ protons of both A25 and A24. The aromatic portion of these units is slightly shielded. This may be explained with the increased flexibility of the tail. At 5’-end the ligand induces small conformational changes as shown by the slightly low-field shift of the T4, G5 and A6 aromatic protons (Δδ = +0.09, -0.15 ppm), which indicates that also this tail is pushed away from the tetrad.
Similar experiments were also performed using ABT-888 as a ligand, in order to find possible interactions with Pu22T14T23 and to compare the results with those above described for 1. The titration of Pu22T14T23 with ABT888 did not show any chemical shift variation (Figure S3). and no NOE contact was found in the NOESY spectra. Consequently, we must conclude that the Pu22T14T23 quadruplex of the c-MYC promoter is not a target for ABT-888.
CD and fluorescence studies.
Both DNA G-quadruplex sequences, d(TTAGGGT)4 and Pu22T14T23, used for NMR investigation, form stable parallel G-quadruplex structures at the experimental conditions used for CD and fluorescence studies (Figure 8). The intramolecular G-quadruplex formed by Pu22T14T23 showed a Tm value around 85 oC (Figure S4), which indicated a high thermal stability of this structure. On the other hand, the human telomeric sequence formed an intermolecular G-quadruplex structure. Therefore, the kinetics of the folding/unfolding process depends so strongly on DNA concentration that determination of the right Tm value would need an extremely small heating rate, due to the presence of hysteresis. Consequently, the midpoint of the transition determined in the conditions used in this work should be named as T1/2,24 being its value 50 oC for d(TTAGGGT)4.
CD-monitored titrations of both G-quadruplexes with the ligand 1 were carried out to determine potential structural and, global changes due to the interaction. Titrations were carried out at 15 oC, where both G-quadruplex structures are the major species. No clear changes were observed that could be related to dramatic structural modifications of the G-quadruplex (Figure 8). Upon addition of the ligand, no variations could be observed in Tm value of the G-quadruplex formed by Pu22T14T23, in agreement with the high stability of this structure.
The fluorescence spectra of ligand 1 in presence of both G-quadruplex sequences are shown in Figure S5. The ligand showed strong fluorescence in potassium phosphate buffer solution without oligonucleotides. The addition of the G-quadruplex structures induced a decrease of the fluorescence signal intensity. From the titration curves, an estimation of the stoichiometry and the binding constants (Kb) relative to the interaction were done with the EQUISPEC program, which is based on the multivariate analysis of the whole spectra measured along the titration. In both cases, the Kb values obtained for the interaction of ligand 1 with Pu22T14T23 and d(TTAGGGT)4 considering a 1:1 stoichiometry are in the order of 106.1 M-1, which suggests a relatively strong interaction between this ligand and both structures. Alternatively, titration curves were studied in PBS buffer obtaining a Kb value for Pu22T14T23 similar to that obtained in K buffer. However, the binding constant for d(TTAGGG)4 is in the order of 105.1 M-1 (Figure S6).