Identification of mixed di-cation forms of G-quadruplex in solution

Abstract

Multinuclear NMR study has demonstrated that G-quadruplex adopted by d(G₃T₄G₄) exhibits two cation binding sites between three of its G-quartets. Titration of tighter binding K⁺ ions into the solution of d(G₃T₄G₄)₂ folded in the presence of ${}^{15}\text{NH}_4^{+}$ ions uncovered a mixed mono-K⁺-mono-${}^{15}\text{NH}_4^{+}$ form that represents intermediate in the conversion of $\text{di-}{}^{15}\text{NH}_4^{+}$ into di-K⁺ form. Analogously, ${}^{15}\text{NH}_4^{+}$ ions were found to replace Na⁺ ions inside d(G₃T₄G₄)₂ quadruplex. The preference of ${}^{15}\text{NH}_4^{+}$ over Na⁺ ions for the two binding sites is considerably smaller than the preference of K⁺ over ${}^{15}\text{NH}_4^{+}$ ions. The two cation binding sites within the G-quadruplex core differ to such a degree that ${}^{15}\text{NH}_4^{+}$ ions bound to the site, which is closer to the edge-type loop, are always replaced first during titration by K⁺ ions. The second binding site is not taken up by K⁺ ion until K⁺ ion already resides at the first binding site. Quantitative analysis of concentrations of the three di-cation forms, which are in slow exchange on the NMR time scale, at 12 K⁺ ion concentrations afforded equilibrium binding constants. K⁺ ion binding to sites U and L within d(G₃T₄G₄)₂ is more favorable with respect to ${}^{15}\text{NH}_4^{+}$ ions by Gibbs free energies of approximately −24 and −18 kJ mol⁻¹ which includes differences in cation dehydration energies, respectively.

INTRODUCTION

Quadruplexes are stable structures adopted by DNA guanine-rich sequences that can be found in telomeres, immunoglobulin switch regions and gene promoter regions; these have also been implicated in association with human diseases, as therapeutic targets in drug design and in potential technical applications as nanomolecular devices (1–14). The basic structural motif of G-quadruplex is a G-quartet, which consists of four guanine bases held together in a coplanar arrangement by eight hydrogen bonds. G-quadruplexes differ in stoichiometry and orientation of strands. The major requirement for G-quadruplex formation is the presence of cations, which reduce repulsions of guanine carbonyl oxygen atoms and also promote stacking of G-quartets.

A fairly wide variety of cations is capable of inducing G-quadruplex formation (1,15). DNA oligonucleotides with double G-rich repeats like d(G₄T₄G₄), d(G₃T₄G₃) and d(G₃T₄G₄) form dimeric fold-back G-quadruplex structures in the presence of Na⁺, K⁺ or ${\text{NH}}_4^{+}$ ions (16–18). In contrast, d(G₄T₄G₃) adopts a single G-quadruplex structure only in the presence of Na⁺ ions, whereas multiple structures form in the presence of K⁺ or ${\text{NH}}_4^{+}$ ions (19). In the presence of Na⁺ ions, oligonucleotide consisting of four repeats of the human telomeric motif, d(TTAGGG), has been shown to fold into an intramolecular G-quadruplex composed of three G-quartets connected by one diagonal and two lateral TTA loops (20). In contrast, the crystal structure in the presence of K⁺ ions has demonstrated a propeller-type folding topology (21). Recent study by circular dichroism has shown that vertebrate telomere repeat, d(TTAGGG), sequences favor the formation of parallel-stranded structures with propeller-type topologies under conditions of high potassium and low sodium concentration as found in nuclei (22). Changeable cation concentration or gradients in cation concentration combined with the presence of different cations can lead to even more complicated situations where prediction of prevailing G-quadruplex species is difficult. Differences in binding properties of K⁺ and Na⁺ ions have long been noted and have been discussed in terms of metal ion switches inducing conformational changes from one type of G-quadruplex to another (23–25).

Solution and solid-state NMR (16–19,26–32), X-ray crystallography (33–35), computer simulations (36–39) as well as other spectroscopic methods have been applied to localize cations and evaluate cation-dependent structural changes (15,40–43). Further studies are however needed to understand subtle interrelationship between thermodynamic preference of cations for available binding sites, structural details that are cation specific and stability as well as folding pathways of G-quadruplexes. G-quadruplexes represent a good model system due to their strong coordination of metal ions. Recently, ¹⁵N-labeled ammonium ion has been used as a non-metallic substitute in combination with 2D NMR spectroscopy to demonstrate that cations are localized between two neighboring G-quartets in solution (44,45).

In our previous work, we have shown that d(G₃T₄G₄)₂ quadruplex, which consists of three G-quartet planes, two overhanging guanine residues and diagonal as well as edge-type loops (Figure 1) shows no variation with the nature of monovalent cation although the NMR spectra exhibit specific (de)shielding effects (18,46). The present solution NMR study demonstrates that two ${}^{15}\text{NH}_4^{+}$ ions are localized between three G-quartet planes of the dimeric d(G₃T₄G₄)₂ structure (Figure 1). Monovalent cations in general stabilize G-quadruplexes in the following order: K⁺ > ${}^{15}\text{NH}_4^{+}$ > Na⁺. The introduction of tighter binding cation into solution (e.g. K⁺) results in the replacement of weaker binding cation (e.g. ${}^{15}\text{NH}_4^{+}$ or Na⁺). Progressive addition of tighter binding K⁺ ions was therefore expected to result in a replacement of ${}^{15}\text{NH}_4^{+}$ ions at both binding sites of d(G₃T₄G₄)₂. The prediction as to the preference of K⁺ over ${}^{15}\text{NH}_4^{+}$ ions for one of the two binding sites could not be made. One of the possibilities involved simultaneous replacement of both bound ${}^{15}\text{NH}_4^{+}$ ions by K⁺. Other possibility could be that K⁺ preferentially binds over ${}^{15}\text{NH}_4^{+}$ ions at one of the two binding sites. Titration experiments uncovered mixed mono-K⁺–mono-${}^{15}\text{NH}_4^{+}$ form of d(G₃T₄G₄)₂ which represents intermediate in the conversion of $\text{di-}{}^{15}\text{NH}_4^{+}$ into di-K⁺ form. We herein show that K⁺ for ammonium ion replacement in d(G₃T₄G₄)₂ is a sequential process. Spectral differences amongst di-cation forms of d(G₃T₄G₄)₂ enabled their identification by NMR and evaluation of preference of the two cations for both binding sites in a quantitative way.

Figure 1.

(a) Schematic representation of topology of d(G₃T₄G₄)₂ G-quadruplex and its two cation binding sites labeled as U and L. The guanine bases are shown as rectangles, where filled rectangles represent syn nucleobases. (b) Stereo view of high resolution NMR structure (pdb = 1U64) of d(G₃T₄G₄)₂. Note a 180° rotation to the left with respect to (a). Strands G1–G11 and G12–G22 are colored blue and red, respectively.

MATERIALS AND METHODS

Sample preparation

DNA oligonucleotide d(G₃T₄G₄) was synthesized on an Expedite 8909 synthesizer using phosphoramidite chemistry following the manufacturer's protocol and deprotected with concentrated aqueous ammonia. DNA was purified on 1.0 m Sephadex G15 column. Fractions containing only full-length oligonucleotide were pooled, lyophilized, redissolved in 1 ml H₂O and extensively dialyzed against 10 mM LiCl. The DNA was then lyophilized and subsequently redissolved in 0.3 ml of 90% H₂O/10% ²H₂O. LiOH or HCl were added to adjust pH of the sample to 5.0. Oligonucleotide concentration was 3.5 mM in strand (1.75 mM in G-quadruplex) for ${}^{15}\text{NH}_4^{+}-{\text{K}}^{+}$ experiments and 3.4 mM in strand for ${\text{Na}}^{+}-{}^{15}\text{NH}_4^{+}$ experiments. Aqueous solutions of KCl, NaCl or ¹⁵NH₄Cl were titrated into the samples.

NMR spectroscopy

NMR data were collected on a Varian Unity Inova 600 MHz NMR spectrometer at 25°C. The standard and ¹⁵N-filtered ¹H spectra were acquired using ¹⁵N decoupling. Both 2D NOESY (τ_m of 80 and 300 ms) and ROESY (τ_m of 140 ms) spectra were acquired in 10% ²H₂O at 25°C using WATERGATE solvent suppression, 4096 complex points in the t₂ dimension, 400 increments in the t₁ dimension and a spectral width of 9.9 kHz. Spectra were ¹⁵N decoupled in both F1 and F2 dimensions.

RESULTS

Identification of mixed ${\text{K}}^{+}\text{-}{}^{15}\text{NH}_4^{+}$ form of d(G₃T₄G₄)₂

d(G₃T₄G₄) has been folded into a G-quadruplex structure by titration of ¹⁵NH₄Cl up to a 20 mM concentration that resulted in well-resolved imino (and other) resonances in ¹H NMR spectrum (Figure 2a). NMR assignment and 3D structure determination have been described earlier (18,46). We herein examined the effect of gradual titration of K⁺ ions into the solution of d(G₃T₄G₄)₂ folded in the presence of 20 mM ${}^{15}\text{NH}_4^{+}$ ions. New set of resonances appeared corresponding to a mixed ${\text{K}}^{+}\text{-}{}^{15}\text{NH}_4^{+}$ form of G-quadruplex which was a major species (55%) when K⁺ ion concentration reached 2 mM (Figure 2b). The minor peaks in Figure 2b were assigned to $\text{di-}{}^{15}\text{NH}_4^{+}$ and di-K⁺ forms (22.5% each). It is interesting to note that the percentage of mixed ${\text{K}}^{+}\text{-}{}^{15}\text{NH}_4^{+}$ form of d(G₃T₄G₄)₂ in solution is dependent solely on the ratio between ${}^{15}\text{NH}_4^{+}$ and K⁺ ion concentrations and not on their absolute concentrations. Further increase in the relative concentration of K⁺ ions resulted in the increased relative amount of di-K⁺ form, which became greatly predominant at 8.5 mM KCl and 20 mM ¹⁵NH₄Cl (Figure 2c). The increase in K⁺ ion concentration up to 20 mM did not result in further changes in ¹H NMR spectra, which were equivalent to the spectrum of d(G₃T₄G₄)₂ with only K⁺ ions present in solution (46). Although chemical shifts of imino protons for $\text{di-}{}^{15}\text{NH}_4^{+}$, di-K⁺ and mixed ${\text{K}}^{+}\text{-}{}^{15}\text{NH}_4^{+}$ forms differ, their NOESY cross-peak patterns in the aromatic–anomeric and other spectral regions clearly confirm the formation of G-quadruplex structure with the same general fold as presented in Figure 1 (46).

Figure 2.

Standard and ¹⁵N-filtered 1D ¹H NMR spectra of d(G₃T₄G₄)₂ at 25°C and pH 5.0 in 10% ²H₂O in the presence of 20 mM ¹⁵NH₄Cl and 0 (a and d), 2 (b and e) and 8.5 mM (c) KCl. Assignments of individual imino resonances are indicated (a–c). Labels U (δ7.25 p.p.m.), L (δ7.45 p.p.m.), L^m (δ7.41 p.p.m.) and B (δ7.11 p.p.m.) indicate ${}^{15}\text{NH}_4^{+}$ ions at distinct binding sites within the architecture of d(G₃T₄G₄)₂ and in bulk solution, respectively. Downfield shoulders of the main peaks in (d) and (e) are due to isotopomers like ¹⁵NH₃D⁺. The oligonucleotide concentration was 3.5 mM in strand.

We have also performed ¹⁵N-filtered ¹H NMR experiments which in the presence of ${}^{15}\text{NH}_4^{+}$ ions alone give rise to two resonances at δ7.25 and 7.45 p.p.m. corresponding to ${}^{15}\text{NH}_4^{+}$ ions bound to U and L binding sites within the architecture of d(G₃T₄G₄)₂, respectively (Figures 1a and 2d). The largest peak at δ7.11 p.p.m. corresponds to ${}^{15}\text{NH}_4^{+}$ ions that are found in bulk solution (Figure 2d and e). A decrease in intensity of U and L resonances was observed in the course of titration with K⁺ ions, which was in agreement with decrease in the amount of $\text{di-}{}^{15}\text{NH}_4^{+}$ form. At the same time, a new resonance started to appear at δ7.41 p.p.m. corresponding to ${}^{15}\text{NH}_4^{+}$ ion bound in a mixed ${\text{K}}^{+}\text{-}{}^{15}\text{NH}_4^{+}$ form (L^m, Figure 2e). All three di-cation forms can be identified in equilibrium at given concentrations of KCl and ¹⁵NH₄Cl and their relative ratios do not change even after several months in solution. Further increase in the concentration of K⁺ ions, however, resulted in disappearance of resonance at δ7.41 p.p.m., which is in agreement with the decreasing amount of mixed ${\text{K}}^{+}\text{-}{}^{15}\text{NH}_4^{+}$ form and simultaneous increasing concentration of di-K⁺ form of d(G₃T₄G₄)₂. The mixed ${\text{K}}^{+}\text{-}{}^{15}\text{NH}_4^{+}$ form has also been identified during reverse titration of ${}^{15}\text{NH}_4^{+}$ ions into the solution of G-quadruplex, which was initially folded in the presence of K⁺ ions alone.

ROESY spectra shown in Figure 3 were used to assign and localize ammonium ions within d(G₃T₄G₄)₂ G-quadruplex. ${}^{15}\text{NH}_4^{+}$ ions resonating at both δ7.25 (U) and 7.45 p.p.m. (L) show cross-peaks with imino protons of G1, G21, G9 and G13 residues (Figure 3a) that constitute the middle G-quartet (Figure 1a). In addition, ${}^{15}\text{NH}_4^{+}$ ions at binding site U show correlations with imino protons of G19, G22, G10 and G14 residues (Figure 3a), which places them between the planes of the middle and the outer G-quartet that is adjacent to the edge-type loop consisting of T15–T18 (Figure 1a). The ${}^{15}\text{NH}_4^{+}$ ions resonating at δ7.45 p.p.m., on the other hand, show additional cross-peaks with imino protons belonging to residues G2, G20, G8 and G12 (Figure 3a), and are therefore localized between the middle G-quartet and the outer G-quartet which is spanned by the diagonal loop (Figure 1a). ROESY spectrum shown in Figure 3b was used to localize ${}^{15}\text{NH}_4^{+}$ ions at δ7.41 p.p.m. in the mixed ${\text{K}}^{+}\text{-}{}^{15}\text{NH}_4^{+}$ form of d(G₃T₄G₄)₂. Eight ROE cross-peaks with imino protons of the two neighboring G-quartets (Figure 3b) offer unequivocal evidence that ${}^{15}\text{NH}_4^{+}$ ions in the mixed ${\text{K}}^{+}\text{-}{}^{15}\text{NH}_4^{+}$ form are residing between the planes of the middle G-quartet and the outer G-quartet that is spanned by the diagonal loop (i.e. at binding site L in Figure 1a).

Figure 3.

Part of ROESY spectra (τ_m =140 ms) of d(G₃T₄G₄)₂ at 25°C in the presence of 20 mM ¹⁵NH₄Cl (a) showing cross-peaks between bound ${}^{15}\text{NH}_4^{+}$ and nearby imino protons and in the presence of 20 mM ¹⁵NH₄Cl and 2 mM KCl (b). Cross-peak designated with asterisk in (a) corresponds to G2(NH) and G12(H8) correlation.

Although exchange between di-K⁺ and mixed ${\text{K}}^{+}\text{-}{}^{15}\text{NH}_4^{+}$ forms is slow on the NMR time scale respective chemical shift differences are large enough to identify and quantify both species only in the imino region of ¹H NMR spectrum. ROESY experiments that would enable us to follow exchange between di-cation forms based on imino resonances do not allow separation of such exchange process from the exchange of protons between hydrogen-bonded sites. Anyhow, no cross-peaks have been observed above noise level in the imino region of ROESY spectra supposedly due to the opposite signs of correlation signals originating from dipolar interactions of spatially close imino protons and those from exchange processes. Chemical shift differences in other spectral regions were too small and thus cross-peaks were too close to the diagonal to allow the study of exchange of K⁺ cations between the di-cation forms of d(G₃T₄G₄)₂.

Determination of binding constants for two distinct binding sites

Two cation binding sites within d(G₃T₄G₄)₂ G-quadruplex differ to such a degree that ${}^{15}\text{NH}_4^{+}$ ions bound to site U, which is closer to the edge-type loop, are always replaced first during titration by K⁺ ions (Figures 1 and 4). Furthermore, binding site L is not taken up by K⁺ ion until K⁺ ion already resides at binding site U.

Figure 4.

Cation displacement pathway amongst $\text{di-}{}^{15}\text{NH}_4^{+}$ (1), mixed ${\text{K}}^{+}\text{-}{}^{15}\text{NH}_4^{+}$ (2) and di-K⁺ (3) forms of d(G₃T₄G₄)₂. Three G-quartets shown schematically define two cation binding sites. The outer G-quartets are spanned by edge-type and diagonal loops. Cations have to be dehydrated to enter ion cavity of G-quadruplex.

We have then examined cationic preferences within G-quadruplex core in a quantitative way. The conversion from $\text{di-}{}^{15}\text{NH}_4^{+}$ through mixed ${\text{K}}^{+}\text{-}{}^{15}\text{NH}_4^{+}$ to di-K⁺ form (Figure 4) can be expressed by the two equilibrium constants, K₁ and K₂;

$$K_1=\frac{\left[2\right]\times \left[{\text{NH}}_4^{+}\right]}{\left[1\right]\times \left[{\text{K}}^{+}\right]},$$ 1

$$K_2=\frac{\left[3\right]\times \left[{\text{NH}}_4^{+}\right]}{\left[2\right]\times \left[{\text{K}}^{+}\right]},$$ 2

where 1, 2 and 3 correspond to the three di-cation forms defined in Figure 4. Integration of respective resonances in ¹H NMR spectra enabled us to determine the concentrations of all three di-cation forms of d(G₃T₄G₄)₂ at 12 K⁺ ion concentrations, while the concentration of ${}^{15}\text{NH}_4^{+}$ ions was constant at 20 mM (Figure 5). The concentrations of free K⁺ and ${\text{NH}}_4^{+}$ ions in Equations 1 and 2 were calculated from the total concentration of inorganic salts and the amount of cations bound inside the three forms of d(G₃T₄G₄)₂ assuming that both of its binding sites are completely occupied by cations. Individual equilibrium constants K₁ and K₂ were calculated by the least-squares fitting procedure of concentrations of $\text{di-}{}^{15}\text{NH}_4^{+}$ (1), mixed ${\text{K}}^{+}\text{-}{}^{15}\text{NH}_4^{+}$ (2) and di-K⁺ (3) forms to Equations 1 and 2. The K₁ and K₂ values that best fitted the experimental data are 234 (±20) and 29 (±5), respectively (Figure 5). First K⁺ ion residing at binding site U stabilizes d(G₃T₄G₄)₂ G-quadruplex by apparent Gibbs free energy of −13.5 (±0.2) kJ mol⁻¹ at 25°C with respect to $\text{di-}{}^{15}\text{NH}_4^{+}$ form. Additional stabilization by the second K⁺ ion, which binds to binding site L is −8.3 (±0.4) kJ mol⁻¹. It is noteworthy that dehydration of K⁺ ions with respect to ${}^{15}\text{NH}_4^{+}$ ions is energetically more demanding process. The difference in the Gibbs free energies of their (de)hydration is 10 kJ mol⁻¹ (47). The binding of K⁺ ions to binding sites U and L of d(G₃T₄G₄)₂ are thus more favorable with respect to ${}^{15}\text{NH}_4^{+}$ ions by Gibbs free energies of approximately −24 and −18 kJ mol⁻¹, which include differences in cation dehydration energies, respectively.

Figure 5.

Relative concentrations of the three di-cation forms of d(G₃T₄G₄)₂ as a function of total concentration of K⁺ ions. Stars, circles and triangles represent experimental relative concentrations of $\text{di-}{}^{15}\text{NH}_4^{+}$ (1), mixed ${\text{K}}^{+}\text{-}{}^{15}\text{NH}_4^{+}$ (2) and di-K⁺ (3) forms, respectively. Dotted curves represent the best fits of the experimental data to Equations 1 and 2. The maximum individual discrepancy between experimental and calculated fractions of 1–3 was 5 unit%, whereas root-mean-square deviation was 2 unit%.

Mixed ${}^{15}\text{NH}_4^{+}{\text{-Na}}^{+}$ form of d(G₃T₄G₄)₂

We furthermore examined the effect of addition of ${}^{15}\text{NH}_4^{+}$ ions into the solution of d(G₃T₄G₄)₂ folded in the presence of 10 mM NaCl. At 3 mM ${}^{15}\text{NH}_4^{+}$ and 10 mM Na⁺ ion concentrations, a single resonance was observed in ¹⁵N-filtered ¹H spectrum at δ7.35 p.p.m., which demonstrated that ${}^{15}\text{NH}_4^{+}$ ions selectively bind and replace Na⁺ ions at one of the two binding sites (see Figure S1, Supplementary Material). At 7.5 mM concentration of ${}^{15}\text{NH}_4^{+}$ ions, a mixture of $\text{di-}{}^{15}\text{NH}_4^{+}$ (δ7.25 and 7.45 p.p.m.), mixed ${}^{15}\text{NH}_4^{+}{\text{-Na}}^{+}$ (δ7.35 p.p.m.) and di-Na⁺ forms were present in solution. Increase in ${}^{15}\text{NH}_4^{+}$ ion concentrations up to 40 mM resulted in the complete replacement of Na⁺ ions inside d(G₃T₄G₄)₂ and $\text{di-}{}^{15}\text{NH}_4^{+}$ became the only species in solution. The resonances of mixed ${}^{15}\text{NH}_4^{+}{\text{-Na}}^{+}$ form were not resolved from their pure di-cation analogues in ¹H NMR spectrum, which prevented localization of ammonium ions in the mixed form and determination of individual cation binding preferences. Additionally, signals in ¹H NMR spectra became broader upon titration with ammonium chloride due to intermediate exchange on the NMR time scale between the three di-cation forms. This largely affected the quality of the spectra which did not allow assignment of resonances of individual forms and thus their quantitative evaluation at all concentrations of added ${}^{15}\text{NH}_4^{+}$ ions that would enable us to determine equilibrium constants. We have nevertheless attempted to analyze concentrations of di-cation forms in the limited range from 3 to 12 mM (∼30% of $\text{di-}{}^{15}\text{NH}_4^{+}$, 25% of ${}^{15}\text{NH}_4^{+}{\text{-Na}}^{+}$ and 45% of di-Na⁺ forms) total concentration of ${}^{15}\text{NH}_4^{+}$ ions. The limited number of six data points gave estimates of equilibrium constants K₁ and K₂ of 0.5 and 1.3, respectively. Note that these values should be considered as very rough estimates, as in the ${}^{15}\text{NH}_4^{+}$ ion concentration range used, only approximately half of the di-Na⁺ forms have been transformed into ammonium forms, which greatly affects the fitting procedure.

DISCUSSION

Multinuclear NMR data have demonstrated that G-quadruplex adopted by d(G₃T₄G₄) exhibits two cation binding sites U and L. The cation binding affinities and electrostatic potentials of these two binding sites are different as evident by their preferential binding of K⁺, Na⁺ and ${}^{15}\text{NH}_4^{+}$ ions. The nature and preferential affinity of individual binding site for a given monovalent cation is influenced by different orientation of the loops and syn/anti alternation of the glycosidic torsion angles within individual G-quartets. The binding site U is characterized by syn-anti-anti-anti orientation of guanines in both G19-G22-G10-G14 and G1-G21-G9-G13 quartets (Figure 1a). In contrast, cation at binding site L is sandwiched between G1-G21-G9-G13 and G12-G8-G20-G2 quartets with syn-anti-anti-anti and syn-syn-anti-anti orientations, respectively, which affects directionality of hydrogen bonds and thus head-to-head versus head-to-tail nature of stacking interactions of guanine residues of the two G-quartets (Figure 1a). The characteristics of binding site U are defined by the outer G-quartet, which is spanned by the edge-type loop that consists of four thymine residues (T15–T18). Neighboring guanine residue G11 is stacked on the same G-quartet and in this way influences cation interactions at binding site U. In contrast, the outer G-quartet of binding site L is spanned by five residue diagonal loop consisting of four thymines T4–T7 and G3 (Figure 1a).

In general, monovalent cations stabilize G-quadruplexes in the following order: K⁺ > ${}^{15}\text{NH}_4^{+}$ > Na⁺. Introduction of tighter binding cations results in replacement of cations with weaker stabilization energies. Several scenarios were expected when KCl was titrated into the solution of d(G₃T₄G₄)₂ folded in the presence of ¹⁵NH₄Cl. First, ${}^{15}\text{NH}_4^{+}$ ions at both sites could be replaced simultaneously. Second, K⁺ ions could take preference over ${}^{15}\text{NH}_4^{+}$ ions at one of the two binding sites and subsequently at the remaining binding site. We however found that increase in K⁺ ion concentration first leads to replacement of ${}^{15}\text{NH}_4^{+}$ ion at binding site U of d(G₃T₄G₄)₂ which is afterwards converted into di-K⁺ form. The slow exchange on the NMR time scale enabled us to demonstrate the existence of mixed ${\text{K}}^{+}\text{-}{}^{15}\text{NH}_4^{+}$ form. In this form, the binding site U is occupied by K⁺ ion, whereas binding site L is still occupied by ${}^{15}\text{NH}_4^{+}$ ion. The other two minor forms in solution were di-K⁺ and $\text{di-}{}^{15}\text{NH}_4^{+}$ with the former becoming the only form as K⁺ ion concentration was increased further. Previous observations of related gradual transition from sodium to potassium forms of G-quadruplexes of d(G₃T₄G₃)₂ (16) and d(G₄T₄G₄)₂ (17) were interpreted by the assumed mixed Na⁺-K⁺ forms. No separate resonances could be observed due to fast exchange between cation forms on the NMR time scale (16,17). Existence of mixed cation forms is supported by the recent crystallographic study on d(TGGGGT)₄ quadruplex which is stabilized by both Tl⁺ and Na⁺ ions in a single G-quadruplex (48). Quantitative analysis showed that binding of K⁺ ion to sites U and L within d(G₃T₄G₄)₂ is more favorable with respect to ${}^{15}\text{NH}_4^{+}$ ions by Gibbs free energies of approximately −24 and −18 kJ mol⁻¹ which includes differences in the cation dehydration energies, respectively. The comparison of equilibrium constants K₁ and K₂ for K⁺ binding to d(G₃T₄G₄)₂ suggests that cation binding is a process of negative cooperativity. It is however well established that it is very difficult to distinguish between negative cooperativity and binding to non-identical binding sites. The nature of the two cation binding sites of d(G₃T₄G₄)₂ quadruplex is determined by different orientations of the two loops and stacking interactions of non-identical G-quartets. It can be expected that consecutive binding or replacement of cations between pairs of G-quartets is a general feature of G-quadruplexes with alternating orientations of nucleotides within G-quartets and/or strand directionalities. On the other hand, the analysis of fraction saturation parameter (i.e. {[2]+[3]}/{[1]+[2]+[3]{) as a function of concentration of free K⁺ ions according to the Hill equation afforded a slope of 1.1. The calculated value of Hill coefficient >1, but less than the number of binding sites implies positive cooperativity of cation binding to d(G₃T₄G₄)₂.

Folding pathways of G-quadruplexes depend significantly on the oligonucleotide sequence and the nature of cations in solution. Folding process can proceed in a stepwise fashion where formation of a G-quartet is in the presence of alkali cations followed by stacking of neighboring G-quartet. Further stacking of additional G-quartets on such dimeric species leads to final G-quadruplex structure. Alternative mechanism results in a folded structure with all cation binding sites being occupied simultaneously. Demonstrations of both scenarios can be found in the literature. The thrombin binding aptamer, d(GGTTGGTGTGGTTGG), for example, forms G-quadruplex structures in the presence of K⁺ ions in a stepwise fashion (49). A chair-type G-quadruplex structure is formed at one molar equivalent of K⁺ ions per oligonucleotide. The binding of the second K⁺ ion primarily alters the structure of the loop while G-quadruplex remains in a chair-type structure (49). Tetrahymena telomere sequence d(T₂G₄) forms G-quadruplex monomer at low K⁺ ion concentration (50 mM) and dimer of co-axial G-quaduplexes at 300 mM K⁺ ion concentration (50). The presence of major and minor conformers of d(TAGGAGGT)₄ in 100 mM KCl solution was interpreted by two and three bound K⁺ ions per G-quadruplex, respectively (51). Our NMR experiments have unequivocally shown that titration of K⁺, ${}^{15}\text{NH}_4^{+}$ or Na⁺ ions into aqueous solution of d(G₃T₄G₄) leads to immediate formation of asymmetric bimolecular G-quadruplex structure that consists of three G-quartet planes, two overhanging guanine residues and diagonal as well as an edge-type loops where both binding sites are occupied (18,46). In the process of titrations of unfolded d(G₃T₄G₄) with ammonium chloride, two NMR resonances corresponding to ${}^{15}\text{NH}_4^{+}$ ions bound at binding sites U and L appear simultaneously. There is no experimental evidence that folding of d(G₃T₄G₄) proceeds in a stepwise fashion.

CONCLUSION

Some G-rich oligonucleotides fold into different G-quadruplex structures in the presence of Na⁺, K⁺ or other cations. Situation is even more complex in physiological systems where G-rich nucleotides get exposed to mixed ionic environments. Localization of cations interacting with macromolecules in solution is still elusive. G-quadruplexes represent a good model system due to their strong coordination of cations by four guanine carbonyl oxygen atoms per G-quartet. d(G₃T₄G₄)₂ is a unique system where K⁺ and ${}^{15}\text{NH}_4^{+}$ di-cation forms are in slow exchange on the NMR time scale. Titration of tighter binding K⁺ ions into the solution of G-quadruplex folded in the presence of ${}^{15}\text{NH}_4^{+}$ ions demonstrated formation of one of the two possible mixed ${\text{K}}^{+}\text{-}{}^{15}\text{NH}_4^{+}$ forms of d(G₃T₄G₄)₂ G-quadruplex. Two cation binding sites within d(G₃T₄G₄)₂ differ to such a degree that ${}^{15}\text{NH}_4^{+}$ ions bound to the site which is closer to the edge-type loop are always replaced first during titration by K⁺ ions. The second binding site is not taken up by K⁺ ion until K⁺ ion already resides at the upper binding site. Quantitative analysis afforded equilibrium binding constants of 234 and 29. K⁺ ion binding to sites U and L within d(G₃T₄G₄)₂ is more favorable with respect to ${}^{15}\text{NH}_4^{+}$ ions by Gibbs free energies of approximately −24 and −18 kJ mol⁻¹, which includes differences in cation dehydration energies, respectively. Taken altogether, mixed ${\text{K}}^{+}\text{-}{}^{15}\text{NH}_4^{+}$ form of d(G₃T₄G₄)₂ presented here is to the best of our knowledge the first G-quadruplex structure identified in solution with two different cations bound inside G-quadruplex core. Understanding preference of different cations for binding sites presented here, thermodynamics of cation binding and dynamics of their replacement uncovers intriguing questions on polymorphism of nucleic acids and G-quadruplexes in particular.

SUPPLEMENTARY MATERIAL

Supplementary Material is available at NAR Online.