Locating Counterions in Guanosine Quadruplexes: A Contrast

May 3, 2005 - Paolo Mariani , Francesco Spinozzi , Francesco Federiconi , Heinz ... Melanie Berghaus , Rana Seymen , Loana Arns , Scott A. McCallum ...
0 downloads 0 Views 741KB Size
Download PDF
J. Phys. Chem. B 2005, 109, 11037-11045

11037

Locating Counterions in Guanosine Quadruplexes: A Contrast-Variation Neutron Diffraction Experiment in Condensed Hexagonal Phase Francesco Federiconi,*,† Pamela Ausili,† Giovanna Fragneto,‡ Claudio Ferrero,§ and Paolo Mariani† Dipartimento di Scienze Applicate ai Sistemi Complessi and INFM, UniVersita` Politecnica delle Marche, Via Ranieri 65, I-60131 Ancona, Italy, Institut Laue LangeVin, BP 156, F-38042 Grenoble Cedex 9, France, and European Synchrotron Radiation Facility, BP 220, F-38043 Grenoble Cedex, France ReceiVed: January 11, 2005; In Final Form: March 18, 2005

Guanosine, one of the primary components of nucleic acids, self-associates in water to form G-quadruplexes, four-stranded helicoidal aggregates, made by stacked planar tetramers, consisting of four planar guanine molecules. Essential for the stability of these supramolecular aggregates is the presence of monovalent cations. As G-quadruplexes show a lyotropic polymorphism, neutron diffraction, in combination with the H2O/D2O contrast variation technique, has been applied to study the cation structural features of quadruplexes in hexagonal phase at different hydrations and counterion concentrations. The guanosine 5′-monophosphate, dipotassium salt, was considered and G-quadruplexes in hexagonal phase were prepared in the different experimental conditions (contrast, hydration and KCl solution concentration) by using the osmotic stress technique. The calculated scattering length density distribution maps show that counterions fill the helix inner cavity and that atmospheric cations are bound to a second site, close to the external phosphate groups. The site occupancy turned out to be very high: we found on the inner site 0.87 K ions per tetramer in G-quadruplexes prepared in pure water and 0.97 K ions per tetramer in G-quadruplexes prepared in KCl 0.5 M, while in all cases 6 ions per unit cell were detected to occupy the outer sites, partially neutralizing the two formal negative charges per phosphate group. The very large K ions concentration difference between the binding sites and the bulk solution demonstrates that counterions are structurally involved in the formation and in the stabilization of the helices.

Introduction Among bases constituting nucleic acids, guanine shows the special and unique ability to self-associate in water solution to form four-stranded helical aggregates (the so-called G-quadruplexes).1,2 The basic building blocks are stacked, planar tetramers (G-quartets), consisting of four planar guanine molecules, arranged in a cyclic pattern, bound by double hydrogen bonds in the Hoogsteen manner between N1-O6 and N2-N7 of successive guanines2-5 (see Figure 1). While the noncovalent interactions that enable formation of a G-quadruplex are certainly interconnected, the self-assembling process can be analyzed at different structural organization levels.6 As sketched in Figure 1, self-association starts in diluted aqueous solutions: the first level of organization is the formation of the G-quartets because of the guanine self-complementary hydrogen bonds. In the presence of nucleating cations (such as K+, Na+, Rb+, Cs+, Li+, and NH4+), a dimerization process, in which two G-quartets aggregate one on the top of the other at the van der Waals distance of 3.3 Å, then occurs. The cation, located between the two G-quartets and forming cation-dipole interactions with the O6 ketone groups of eight separate molecules of guanine, stabilizes the hydrogen-bonded quartets and enhances base-stacking interactions, providing a C4symmetric G8‚M+ octamer (where M+ represents the cation). * Corresponding author. Phone: +39 071 2204608. Fax: +39 071 02204605. E-mail: [email protected]. † Universita ` Politecnica delle Marche. ‡ Institut Laue Langevin. § European Synchrotron Radiation Facility.

Indeed, at guanine concentration higher than 0.1 M (about 35 mg/mL), only G-quartets and dimers of G-quartets have been detected in aqueous solutions of guanosine 5′-monophosphate.7 The successive level of organization occurs for increasing guanine concentration: extended columnar aggregates (Gquadruplexes) form by the stacking of discrete cation-bound G8‚M+ octamers. As a direct consequence of the intrinsic chirality of the compound, each G-quartet result rotated with respect to the adjacent ones, so that a helicoidal structure is built. Therefore, despite the absence of the sugar-phosphate backbone, the final structure of the G-quadruplexes resembles that of four-stranded B-DNA. Indeed, G-quadruplexes are formed in vitro by DNA and RNA oligonucleotides containing G-rich sequences, as occurs in chromosomal telomeres, gene promoter regions, recombination sites, RNA packaging sites, and RNA dimerization domains.6,8-10 As shown by their crystal structures, DNA and RNA G-quadruplexes can differ in their chain number and orientation, but in all cases they are stabilized by alkali and alkaline-earth cations and their structure is cationdependent.5,6,11-13 In the crystallographic structures, noticeable is the presence of collinear cations, spaced 3.3 Å apart, arranged into the central channel (see, for example, the structure reported by Philips and co-workers13). The electrostatic repulsion expected between the channel cations is clearly minimized by the G-quartet oxygen atoms and aromatic rings. Moreover, cation binding presumably reduces the repulsion of the four central oxygen atoms in the hydrogen-bonded quartet, enhancing the hydrogen-bond strength, and then stabilizing the G-quartet stacking. Therefore, while there is a debate over the relative

10.1021/jp0501751 CCC: $30.25 © 2005 American Chemical Society Published on Web 05/03/2005

11038 J. Phys. Chem. B, Vol. 109, No. 21, 2005

Federiconi et al.

Figure 1. Schematic representation of hierarchical self-assembly processes of GMP in water: (A) arrangement of four guanosine molecules in a G-quartet, and the monovalent cation occupies the tetramer inner cavity; (B) C4-symmetric G8‚M+ octamer, formed by two subsequent tetramers, and the cation is located between the two G-quartets; (C) four-stranded helix (G-quadruplex) formed by stacking of discrete cation-bound G8‚M+ octamers; (D) cholesteric and (E) hexagonal liquid-crystalline phases (G-quadruplexes are represented as cylinders).

importance of the electrostatic and dispersive forces that drive aromatic-stacking interactions, experiments and calculations both on DNA and RNA G-quadruplexes and on four-stranded helices obtained by stacking of discrete cation-bound G8‚M+ octamers show that cation-dipole interactions are essential for formation of quadruplex structures.6 Depending on concentration and temperature, G-quadruplexes show an additional level of organization, showing a lyotropic liquid crystalline polymorphism.4,14 In particular, GMP oligomers in water form cholesteric and hexagonal phases.4,14-16 Structural results obtained with different counterions suggested that the ability of cations to promote G-quartet structure formation and to stabilize the helices is also reflected in the characteristics of the temperature-concentration-dependent phase diagrams.14,17,18 Moreover, due to the electrostatic nature, ions were also observed to modify long-range inter-helix lateral forces:19,20 osmotic stress measurements indicated that the repulsive force between helices in the hexagonal phase is dependent on the ionic strength of the aqueous compartment, but from the calculated density charge per unit surface it was argued that in G-quadruplexes there are about 90% of phosphate charges balanced by counter-charges.20 In this context, three points are relevant for the present work. First, G-quadruplexes and lyotropic phases form in solutions only in the presence of monovalent cations. The nucleating ion occupies the inner cavity in the G-quadruplexes,6,21-23 but atmospheric cations also bound to a second site, close to the external phosphate groups, where they induce a partial neutralization of the helix negative charges.6,19 Second, the inner site is ion-selective, while the outer site is rather unspecific.5,6 The selectivity of the inner site has been attributed to the better fit of potassium and ammonium ions into a G8-octamer cage because of the small size of their ionic radii (K+, 1.33 Å; NH4+, 1.45 Å). Ross and Hardin, however, concluded that the “optimal size” proposal did not fully explain the K+/Na+ selectivity of the G8-octamer and suggested that electronic factors must be

important.24 Later, Hud and co-workers provided evidence that the hydration energy of the cations was the major determinant of K+/Na+ selectivity.25 While both cations fit between Gquartets, it is easier to dehydrate potassium ions. Calculations support this ion dehydration argument.26 Third, in the past decade there have been many NMR studies and a growing number of X-ray investigations on DNA and RNA G-quadruplexes structures, and the structural features of coordinated cations have been reviewed in many articles.6,9,11,22 Solid-state 23Na and 39K NMR spectroscopy and extended X-ray absorption fine structure (EXAFS) have also been used to directly locate cations in G-quadruplexes.6,27 However, experiments have been performed neither in hydrated systems, as a function of the ionic strength of the solution, nor in the liquid crystalline phases, where the intrinsic disorder of the long-range structure is a serious obstacle toward high-resolution studies. On the basis of these considerations, we decided to use neutron diffraction to analyze the structural features of potassium ions in G-quadruplexes in the hexagonal phase. Neutron scattering studies offer indeed unique opportunities for obtaining information on macromolecular assemblies. The distinguishing feature of neutron scattering is the ability to exploit the difference between the scattering lengths of protons and deuterons, allowing contrast variation techniques, that can provide structural information on individual elements of multicomponent systems.28 As the scattering length of potassium is rather different from the scattering length of the other components, contrast variation neutron diffraction should allow the ion position in the hexagonal cell to be identified. Note that to control the hydration of the G-quadruplexes and the KCl concentration in the aqueous compartment, samples were prepared using the osmotic stress technique.29 Osmotic stress is the controlled removal of water from the system under investigation: leaving the GMP to come in equilibrium with a polymer solution of known osmotic pressure and prepared with different amounts of D2O and H2O and at various KCl

Locating Counterions in Guanosine Quadruplexes

J. Phys. Chem. B, Vol. 109, No. 21, 2005 11039

Figure 2. Neutron diffraction profiles obtained from GMP samples condensed from a PEG 43% w/w solution at different contrast levels and with 0 M KCl (upper frames) and 0.5 M KCl (lower frames). Low-angle diffraction profiles are shown on the left side, while the corresponding highangle diffraction profiles are reported on the right side.

concentrations, it is possible to obtain samples in the hexagonal phase at different inter-helix separation distances suitable for contrast-variation neutron diffraction experiments. Materials and Methods GMP (guanosine 5′-monophosphate) potassium salt was obtained from the GMP free acid form (ICN Biomedicals, 99% purity) by titration with potassium hydroxide solution; KCl and poly(ethylene glycol) (PEG) 15000-20000 MW were of commercial type (Sigma, 99% purity) and were used without further purification. To analyze the GMP quadruplexes in hexagonal phase at low hydration levels, and to be able to control the KCl salt activity also in the low concentration regime, samples were prepared using the osmotic stress method.19,20,29 In particular, GMP was left to come in equilibrium against PEG-salt aqueous solutions of known osmotic pressure in vast excess (5 mg/mL) for approximately 24 h at room temperature. When polymer solutions of at least 20 wt % were used, GMP condensed into a hexagonal phase separate from the solution. In the present case, PEG solutions of 35, 43, and 50% w/w, which correspond to osmotic pressures Π of 2.233, 4.003, and 6.240 × 10-9 dyne/ Å2, respectively, were considered.29 The polymer solutions were at three different concentrations of KCl (namely, 0, 0.1, and 0.5 M) and prepared at variable D2O/H2O ratios: due to hydrogen exchange between PEG and solvent, the final compositions of the investigated aqueous mixtures were 0, 12, 27, 40, 52, and 65% D2O (v/v, %).

The condensed phases were then mounted in vacuum-tight cells with thin mica windows for X-ray diffraction measurements and in 2-mm-thick quartz cells for neutron scattering experiments. X-ray diffraction experiments were performed to test the homogeneity of the condensed phase and to detect structural effects possibly induced by deuterated water. Neutron diffraction experiments were performed to calculate the scattering length density maps and to derive direct structural information on the location of counterions. From the measured unit cell dimension and by comparison with previous results,19 the sample composition was derived. X-ray diffraction experiments were performed using a Philips PW1830 X-ray generator equipped with a Guinier-type focusing camera operating in a vacuum: a bent quartz crystal monochromator was used to select the Cu KR1 radiation (λ ) 1.54 Å). The diffraction patterns were recorded on a stack of two Kodak DEF-392 films. The sample cell temperature was kept constant at 25 °C with an accuracy of 1 °C by using a circulating thermostat. Neutron scattering experiments were performed at the D16 diffractometer at the Institut Laue-Langevin (Grenoble, France), at room temperature by using a neutron wavelength λ of 5 Å. The measure range of the scattering vector Q ) (4π sin θ)/λ (where 2θ is the full scattering angle) was between 0.035 and 1.5 Å-1. In a few cases, the investigated Q-range was extended up to 2 Å-1, so as to detect the high-angle diffraction peak arising from the stacked G-quartets. Peak intensities were measured from scattered intensities, corrected for electronic

11040 J. Phys. Chem. B, Vol. 109, No. 21, 2005

Federiconi et al.

Figure 3. Dependence on osmotic pressure of the 2D hexagonal unit cell as measured at different KCl concentrations. Lines are guides to the eye.

Figure 4. Dependence on KCl concentration of the tetramer stacking distances at different osmotic stresses. Lines are guides to the eye.

Figure 5. Dependence on the solvent scattering density (that is, on the D2O fraction in the solvent) of the Bragg peak intensities for GMP samples condensed from a PEG 43% w/w solution at different contrast levels and with 0 M KCl (upper frame) and 0.5 M KCl (lower frame).

noise and sample holder signals. For data evaluation purposes, all Bragg peaks were fitted by Lorentzian distributions. The fittings were carried out with the software package Igor Pro 4.08 (WaveMetrics, Lake Oswego, Oregon). Thereafter, to calculate the peak intensity Ih,k (h,k are the reflection indices), a Lorentz correction was applied by multiplying each peak area by its corresponding wave vector (Q/2π)2 (for discussion, see Rappolt and co-workers30). Two-dimensional scattering length density profiles were obtained by calculating the two-dimensional Fourier distribution:

F(x,y) )

∑h,k((Fh,kcos(Q(x)h,kx)cos(Q(y)h,ky))

(1)

where Fh,k is the Fourier coefficient of the peak at the position Qh,k obtained by the peak intensity using Fh,k ) (Ih,k/mh,k)1/2, where mh,k is the multiplicity of the h,k reflection. Note that the phase information for each diffraction order is either positive or negative for a center-symmetric scattering length density profile, such as that pertaining to the hexagonal phase. According to the contrast-variation technique,28,31,32 the correct sign of the Fourier coefficients Fh,k can be obtained analyzing the dependence of the diffracted peak intensities on the contrast, defined as the difference of the average scattering length densities of the solvent (which is a linear function of the D2O fraction) and the G-quadruplexes (see below).

Figure 6. Section of the cylindrically symmetric geometrical model used to represent the G-quadruplex. By way of illustration, the G-quartet arrangement is superimposed. Dimensions and scattering length densities, as derived from molecular models and chemical compositions, are: R1 ) 1.5 Å, R2 ) 7 Å, R3 ) 12 Å. In 0% D2O, F1 ) 1.3 × 1010 cm-2, F2 ) 4.3 × 1010 cm-2, F3 ) 1.3 × 1010 cm-2, F0 ) -0.56 × 1010 cm-2. In 65% D2O, F1 ) 1.3 × 1010 cm-2, F2 ) 4.3 × 1010 cm-2, F3 ) 3.4 × 1010 cm-2, F0 ) 3.9 × 1010 cm-2.

peaks with spacing ratios in the order 1:x3:x4:x7:x9:x12 ... which can be indexed considering the two-dimensional hexagonal lattice of p6m symmetry.33 From the Bragg spacings Qh,k, the unit cell dimension, a, (that is, the inter-axial distance between the four-stranded helices) has been obtained from

Results According to our previous works,19,20 at the investigated osmotic pressures, X-ray and neutron diffraction results were consistent with the presence of the hexagonal columnar phase. The neutron diffraction profiles reported in Figure 2 show a low-angle diffraction region characterized by a series of narrow

a)

2 2 4π x(h + k - hk) Qh,k x3

(2)

As a result, the inter-axial distance has been detected to decrease both with increasing PEG concentration (that is, by

Locating Counterions in Guanosine Quadruplexes

J. Phys. Chem. B, Vol. 109, No. 21, 2005 11041

Figure 7. Amplitudes of the Bragg peaks as a function of Q position. Best-fit curves are Fourier transforms of the cylindrically symmetric geometrical models for the G-quadruplexes.

increasing the osmotic pressure) and with increasing the salt molarity. Unlike that, the unit cell does not change when the ratio between deuterated and light water in the solvent changes. In Figure 3, the inter-axial distance measured at different KCl concentrations is reported as a function of the osmotic pressure. Note that the inter-helix distance changes from 32.1 ( 0.1 Å (PEG 35%, KCl 0 M) to 30.7 ( 0.1 Å (PEG 50%, KCl 0.5 M). Experimental evidence of the columnar nature of the GMP condensed phases has been obtained by the analysis of the highangle diffraction region. According to our previous works,4,14 a narrow band is in fact observed at about Q ) (2π/3.34) Å-1 (see Figure 2). This band is related to the nature of the order inside the structure elements, which are actually columns composed of G-quartets stacked perpendicularly to the column axis. From the position of this peak, the distance between the neighboring tetramers is obtained. As shown in Figure 4, this stacking distance seems to be independent of the osmotic pressure and of the amount of deuterated water in the solvent, but it shows a small dependence on the salt concentration. Our analysis mainly concerned the intensities of the Bragg peaks observed in the low-angle neutron diffraction region. As Figure 2 shows, a gradual addition of D2O to the aqueous solution determines a change in the intensity of the diffraction peaks. According to the contrast variation method,31,32 a linear dependence of (Ih,k)1/2 as a function of contrast is expected, so that the direct observation of the sign change for each diffraction peak should be possible. However, due to the presence of PEG in the bulk solution, full contrast-variation conditions remained unachieved (note that the matching point is expected to occur

at about 68% D2O), while the presence of exchangeable protons in the guanosine molecules makes it more difficult to work out the positions of the zeros of the Fourier coefficients (see Figure 5). Therefore, the signs of the Fourier coefficients were derived by fitting via a geometric model of the four-stranded helices the intensity data sets obtained in the same contrast conditions, but under different osmotic stresses. For sake of simplicity, the model was made to be cylindrically symmetric. The unit cell was split up into four regions: inner cavity (where the potassium ions are expected to be located), guanosine quartet region, sugar residue region, and water region, as shown in Figure 6. The corresponding scattering length densities were obtained by adding uniform disks of the appropriate density. The Fourier transform of these disks is given by the continuous function:

F(Q) ) 2(R12(F1 - F2)J1(QR1)/QR1 + R22(F2 - F3)J1(QR2)/QR2 + R32(F3 - F0)J1(QR3)/QR3)/ (R12(F1 - F2) + R22(F2 - F3) + R32(F3 - F0)) (3) where J1 is the first-order Bessel function of the first kind; F1, F2, and F3 are the scattering length densities of the potassium, of the guanosine, and of the sugar residue, respectively; and F0 is the scattering length density of the aqueous solvent. Moreover, R1 is the radius of the central hole, R2 is the outer radius of the guanosine tetrameric shell, and R3 is the outer radius of the shell where the sugar residues are located. The model parameters, calculated in two different solvent conditions, are reported in the caption of Figure 6.

11042 J. Phys. Chem. B, Vol. 109, No. 21, 2005

Federiconi et al.

Figure 8. Scattering length density maps calculated for GMP samples condensed from a PEG 43% w/w, 0 M KCl solution at different contrast levels.

A few examples of the F(Q) functions for different D2O concentrations are shown in Figure 7. It should be noticed that, since these systems are quite soft and disordered, we expect thermal fluctuations and undulations to have important effects on the model structure factors. Some corrections should thus be considered to describe the model parameters: as the model analysis could become too intricate, we preferred to use the derived continuous structure functions to set the signs of the experimental Fourier coefficients (see eq 2) and to directly obtain the relevant structural parameters from the calculated scattering length density maps. The scattering length density maps for the series at 0 M KCl and PEG 43%, calculated from the experimental intensities and

the signs derived from the continuous functions of Figure 7, are shown in Figure 8. The circular G-quadruplex contour can be easily appreciated, but other important features are evident, such as the presence of the hole in the central region of the helix, the approximately constant density corresponding to the tetrameric contour region, and the strong dependence on the solvent composition of the scattering length density in regions far away from the cylinder surfaces. The helix characteristic radii directly derived from the maps and the scattering length densities measured in significant regions of the hexagonal cell are reported in Table 1. In particular, we considered a mean inner radius, Rchannel, defined as the mean distance from the center of the helix to the inner

Locating Counterions in Guanosine Quadruplexes

J. Phys. Chem. B, Vol. 109, No. 21, 2005 11043

TABLE 1: Quadruplex Characteristic Radii and Scattering Length Densities in Significant Regions of the 2D-hexagonal Cell as Measured from the Scattering Length Density Maps Calculated from Experimental Intensities PEG (w/ w%)

KCl (M)

D2O (v/v%)

Rchannel (Å)

Rcore (Å)

Rhelix (Å)

Fsol (1010 cm-2)

Fext (1010 cm-2)

Fchannel (1010 cm-2)

50

0.5

0 12 27 40 52 65

1.9 2.1 2.3 2.5 2.4 2.1

4.5 4.5 5.0 5.0 5.0 5.5

10.1 10.1 10.0 10.2 10.0 10.1

-0.29 0.47 1.40 2.19 2.79 3.50

0.62 1.45 1.86 2.53 3.07 3.63

2.19 1.99 2.06 1.92 1.83 2.14

50

0

0 12 27 40 52 65

2.5 2.5 2.4 2.7 2.3 2.2

5.0 5.0 5.0 5.0 5.0 5.5

10.2 10.1 10.1 9.8 10.2 10.0

-0.09 0.47 1.19 1.97 2.77 3.47

0.74 1.28 1.73 2.31 2.99 3.65

1.32 1.41 1.62 1.83 1.92 2.20

43

0.5

0 12 27 40 52 65

2.2 2.2 2.3 2.4 2.2 2.1

5.0 5.0 5.0 5.0 5.0 5.5

10.0 10.0 10.2 10.2 10.2 10.1

-0.24 0.43 1.45 2.01 2.94 3.60

0.86 1.33 1.92 2.13 3.02 3.76

1.82 1.87 1.86 1.99 2.21 2.20

43

0

0 12 27 40 52 65

2.4 2.4 2.6 2.7 2.3 2.3

5.0 5.0 5.0 5.0 5.0 5.5

10.1 10.2 9.9 10.2 9.8 10.1

-0.28 0.43 1.18 2.00 2.83 3.55

0.84 1.15 1.68 2.25 3.00 3.75

1.51 1.61 1.38 1.73 2.01 2.19

35

0.5

0 12 27 40 52 65

2.4 2.3 2.5 2.3 2.2 2.2

5.0 5.0 5.0 5.0 5.0 5.5

10.5 10.3 10.1 10.2 10.0 10.2

-0.28 0.46 1.77 2.27 2.79 3.55

0.79 1.45 2.37 2.61 2.94 3.72

1.91 1.77 1.82 1.98 2.09 2.10

35

0

0 12 27 40 52 65

2.2 2.5 2.6 2.7 2.2 2.2

4.5 5 5 5 5 6

10.3 10.2 10.0 10.2 10.0 10.1

-0.30 0.52 1.37 2.19 2.89 3.50

0.87 1.09 2.18 2.52 2.95 3.63

1.56 1.63 1.48 1.88 2.09 2.34

inflection point of the scattering length density; a mean helix core radius, Rcore, defined as the mean distance from the center of the helix to the maximum of the scattering length density map; a mean helix external radius, Rhelix, defined as the mean distance from the center of the helix to the external inflection point (roughly speaking, the helix core radius corresponds to the region where the guanine residues are located, while the helix external radius defines the position where the sugar residues are situated). On the other side, scattering length density values were measured in points corresponding to the minimum in the central region of the helix, Fchannel, to the external inflection point, Fext, and in a region far away from the quadruplex surface, Fsol (that is, in the aqueous compartment of the cell). Discussion To acquire some meaningful information on counterion structural features, GMP quadruplexes in hexagonal phase were investigated by neutron diffraction in different contrast conditions. In each of the experimental conditions, at least 5 Bragg peaks were detected (see Figure 2), and their positions and intensities were measured. From peak positions, the inter-axial distance between the helices was obtained: as shown in Figure 3, at the same osmotic pressure, changes in ionic strength induce small variation in the hexagonal unit cell, confirming the role of ions in reducing the inter-helix electrostatic repulsive forces. From peak intensities, scattering length density maps were

calculated (see Figure 8). In particular, the signs of the Fourier coefficients were derived fitting intensity data sets with a cylindrically symmetric model for the four-stranded helices (see Figures 6 and 7). Key features of the scattering length density maps (some of which are shown in Figure 8) are the circular helix contour, the presence of a hole in the central region of the helix, the fairly constant density in correspondence to the G-quartet region and the strong dependence of the scattering length density on solvent composition both in the quadruplex contour region and in the region far away from the cylinder surfaces. A series of relevant structural parameters, directly derived from the maps, are reported in Table 1. Noticeable is the uniformity of the characteristic helix radii: Rchannel, Rcore, and Rhelix appear in fact independent of contrast, salt concentration, and osmotic pressure, indicating that the structure of the G-quadruplexes does not depend on the experimental conditions. On the contrary, scattering length densities show large differences in the different experimental conditions: in particular, since the structure is conserved, and the variations cannot be explained only considering D2O/H2O exchange, the values measured for different KCl concentrations on a point corresponding to the minimum of the central helix region (Fchannel) and far away from the helix surfaces (Fsol) suggest a different level of occupancy by counterions. To allow a quantitative analysis, scattering length density values were averaged among samples with identical ionic

11044 J. Phys. Chem. B, Vol. 109, No. 21, 2005 strength: these values are reported as a function of the D2O fraction in the solvent in Figure 9. As shown in the upper frame, scattering length densities measured far away from the helix surfaces (that is, in the aqueous regions of the cell) show a linear dependence on the D2O fraction in the solvent. Moreover, the values observed in samples prepared in the presence and in the absence of KCl are rather similar. It should be noted that scattering length densities expected for pure H2O/D2O mixtures are quite different (scattering length densities vary linearly from -0.56 to 3.9 ×1010 cm-2 when the heavy water volume fraction changes from 0 to 65%, respectively), while changes determined by 0.5 M KCl are well below the experimental errors. We suggest that the observed values are related to the “crowded” condition of the aqueous compartment (that is, on the presence of sugar and phosphate residues protruding out from the fourstranded helix surfaces) and on a locally rather high concentration of counterions, condensed around the phosphate charges. Indeed, the observed scattering length densities were fitted using a linear combination of sugar, phosphate, potassium, and water model scattering densities and considering D2O/H2O exchange. Best-fit results show that the volume concentration of sugar and phosphate residues in the aqueous compartment is quite low (around 5%), while the volume concentration of potassium ions was found to be around 12%. This concentration corresponds to 6 (( 1) ions per unit cell. Considering that there are 4 negatively charged phosphates per unit cell (for a nominal charge of -8), this number indicates that about 75% of helical charges are balanced by counter-charges, in full agreement with previous results.19,20 Noticeable is the fact that this result is independent of the nominal salt concentration of the bulk phase. In the central frame of Figure 9, the average scattering length density measured at the inflection point on the external surface of the G-quadruplex (Fext) is reported as a function of the D2O fraction in the solvent. As for Fsol, the scattering length densities continuously change versus the D2O fraction, while no changes with the KCl concentration of the solution are observed (very slight differences are within the experimental errors). In this case, however, best fit results were obtained using a linear combination model of sugar and water model scattering densities only, and accounting for D2O/H2O exchange. The concentration of the sugar residues was determined to be around 0.7 v/v, confirming that sugar residues, protruding out from the tetramer helix surface, are located in a highly hydrated region, close to the helix; moreover, counterions are almost excluded from this region. Finally, the last plot of Figure 9 (lower frame) shows that Fchannel, the average scattering length density measured in the central hole, differs significantly in samples prepared with and without KCl. In particular, in the absence of salt, scattering length densities are rather low and show a strong dependence on the D2O composition of the aqueous mixture; in KCl 0.5 M, the scattering length densities are larger and only slightly dependent on the solvent D2O fraction. Such a behavior can be explained considering that the inner hole is filled by a few water molecules in addition to potassium ions and that the amount of K ions inside the tetramers is related to the salt concentration of the bulk solution. Accordingly, data points have been fitted using a linear combination of scattering length densities from bare K ions and water molecules, considering D2O/H2O exchange. Fitting results indicate a potassium hole occupancy of 0.87 ( 0.02 v/v in the absence of salt, which increases to 0.97 ( 0.02 v/v in KCl 0.5 M. Therefore, the analysis of the scattering length densities as a function of contrast allowed the counterion distribution inside

Federiconi et al.

Figure 9. Dependence on the D2O fraction in the solvent of the averaged scattering length densities measured in relevant positions of the cell. Error bars refer to the standard deviation of the average. Upper frame: averaged scattering length density measured far away from the four-stranded helix surfaces. Data have been fitted using Fsol ) cV,sug[Fsug,H(1 - fD2O) + Fsug,D fD2O] + cV,KFK + (1 - cV,sug - cV,K)[FD2O fD2O + FH2O(1 - fD2O)]. Best-fitting parameters are cV,sug ) 0.05 ( 0.03 and cV,K ) 0.13 ( 0.02 at 0 M KCl, cV,sug ) 0.05 ( 0.02, and cV,K ) 0.11 ( 0.02 at 0.5 M KCl. Middle frame: averaged scattering length density measured at the inflection point on the external surface of the helix. Data have been fitted using Fext ) cV,sug[Fsug,H(1 - fD2O) + Fsug,DfD2O] + (1 - cV,sug)[FD2O fD2O + FH2O(1 - fD2O)]. Best-fitting parameters are cV,sug ) 0.69 ( 0.03 at 0 M KCl and cV,sug ) 0.73 ( 0.03 at 0.5 M KCl. Lower frame: averaged scattering length density measured in the inner channel of the helix. Data have been fitted using Fchannel ) {cV,KFKVK + (1 - cV,K)[FD2OVD2OfD2O + FH2OVH2O(1 - fD2O)]}/{cV,KVK + (1 - cV,K)[VD2OfD2O + VH2O(1 - fD2O)]}. Best-fitting parameters are cV,K ) 0.87 ( 0.02 at 0 M KCl and cV,K ) 0.97 ( 0.02 at 0.5 M KCl. Used symbols: cV,sug, concentration of the guanosine sugar residue; cV,K, concentration of potassium ions; Fsug,H ) 1.3 × 1010 cm-2, scattering length density of the guanosine sugar residue, totally hydrogenated; Fsug,D ) 4.4 × 1010 cm-2, scattering length density of the guanosine sugar residue, with all labile hydrogens exchanged; FK ) 1.3 × 1010 cm-2, scattering length density of potassium; FH2O ) -5.6 × 109 cm-2, scattering length density of light water; FD2O ) 6.3 × 1010 cm-2, scattering length density of heavy water; fD2O, volume fraction of heavy water in the solvent; VK ) 2.2 × 10-23 cm3, molecular volume of potassium ion; VH2O ) 2.99 × 10-23 cm3, molecular volume of light water; VD2O ) 3.02 × 10-23 cm3, molecular volume of heavy water.

Locating Counterions in Guanosine Quadruplexes the cell to be enlightened, confirming that G-quadruplexes have two different sites for the binding of metal ions, the inner cavity and the outer sites, close to the external phosphate groups. Even without excess counterions, the occupancy of these sites is very high: we found in the inner site 0.87 K ions per tetramer in G-quadruplexes prepared in pure water and 0.97 K ions per tetramer in G-quadruplexes prepared in KCl 0.5 M, while in both cases 6 ions per unit cell have been detected to preferentially occupy the outer sites, partially neutralizing the two formal negative charges per phosphate group brought together in the self-assembled structures. The K ion concentrations at the binding sites and in the bulk are therefore largely different, demonstrating that counterions are structurally involved in the formation and in the stabilization of the helices. It should be noticed that a very few water molecules are present in the central channel of the quadruplexes (the water hole occupancy is 0.13 ( 0.02 v/v and 0.03 ( 0.02 v/v when GMP samples are prepared in pure water and in KCl 0.5 M, respectively): tetramers specifically need alkali metal cations in order to form, but some water molecules, which have the suitable size to fit in the central cavity, seem to be incorporated in the structure during the stacking process of discrete cationbound G8‚M+ octamers. The amount of water in the inner site depends on the KCl concentration of the bathing solution. These defects do not prevent the formation of stable helices, but they could be the cause of their different lengths, as detected by X-ray diffraction and dynamic light scattering and 31P NMR spectroscopy analyses.14,17,18,34 Indeed, Mariani and co-workers14,18 found that KCl in excess induces a stabilization of the hightemperature hexagonal phase in dGMP and the occurrence of direct isotropic-to-hexagonal (with increasing concentration) and cholesteric-to-hexagonal (with increasing temperature) phase transitions: both were ascribed to the formation of longer quadruplexes. Moreover, stronger aggregation has been also related to the different mechanical behavior of GMP helices at high pressure.18,35 In particular, the analysis of the compression work in terms of stacking elasticity allowed the longitudinal quadruplex elastic constant to be derived:35 in the presence of 30% water w/w, the elastic constant changes from 3.85 to 4.64 and 4.69 × 105 dyne cm-1 as KCl concentration changes from 0 to 1 and 2 M, respectively, indicating that excess potassium increases the helix rigidity. Therefore, the present work confirms that G-quadruplexes bind alkali metal cations, but indicates that counterion occupancy of the tetramer inner cavity in solution depends on solvent conditions. As occupancy affects some relevant biological properties, such as aggregate stability and rigidity, these data could be useful for an appropriate understanding of the role of nonduplex structures in the packing of DNA in cells and viruses

J. Phys. Chem. B, Vol. 109, No. 21, 2005 11045 as well as of the biological function of DNA and RNA quadruplexes. References and Notes (1) Gellert, M.; Lipsett, M. N.; Davies, D. R. Proc. Natl. Acad. Sci. U.S.A. 1962, 48, 2013. (2) Saenger, W. Principles of Nucleic Acid Structure; SpringerVerlag: Heidelberg, Germany, 1984. (3) Fisk, C. L.; Becker, E. D.; Miles, H. T.; Pinnavaia, T. J. J. Am. Chem. Soc. 1982, 104, 3307. (4) Mariani, P.; Mazabard, C.; Garbesi, A.; Spada, G. P. J. Am. Chem. Soc. 1989, 111, 6369. (5) Sen, D.; Gilbert, W. Curr. Opin. Struct. Biol. 1991, 1, 435. (6) Davis, J. T. Angew. Chem., Int. Ed. 2004, 43, 668. (7) Detellier, C.; Laszlo, P. J. Am. Chem. Soc. 1980, 102, 1135. (8) Siddiqui-Jain, A.; Grand, C. L.; Bearss, D. J.; Hurley, L. H. Proc. Natl. Acad. Sci. U.S.A. 2002, 99, 11593. (9) Simonsson, T. Biol. Chem. 2001, 382, 621. (10) Smirnov, I.; Shafer, R. H. Biopolymers 2000-2001, 56(3), 209. (11) Gilbert, D. E.; Feigon, J. Curr. Opin. Struct. Biol. 1999, 305. (12) Leonard, G. A.; Zhang, S.; Peterson, M. R.; Harrop, S. J.; Helliwell, J. R.; Cruse, W.; Langlois d’Estantiot, B.; Kennard, O.; Brown, T.; Hunter, W. N. Structure 1995, 3, 335. (13) Phillips, K.; Dauter, Z.; Murchie, A. I. H.; Lilley, D. M. J.; Luisi, B. J. Mol. Biol. 1997, 273, 171. (14) Gottarelli, G.; Spada, G. P.; Mariani P. In Crystallography of Supramolecular Compounds; Tsouscaris, G., Atwood, J. L., Lipkowski, J., Eds.; Kluwer Academic Publishers: Dordrecht, Olanda, 1996; pp 307330. (15) Gottarelli, G.; Proni, G.; Spada, G. P.; Bonazzi, S.; Garbesi, A.; Ciuchi, F.; Mariani, P. Biopolymers 1997, 42, 561. (16) Proni, G.; Gottarelli, G.; Mariani, P.; Spada, G. P. Chem. Eur. J. 2000, 6, 3249. (17) Spindler, L.; Drevensek-Olenik, I.; Copic, M.; Mariani, P. Mol. Cryst. Liq. Cryst. 2003, 395(317), 147. (18) Federiconi, F. Ph.D. Thesis, Polytechnic University of Marche, Ancona, Italy, 2004. (19) Mariani, P.; Saturni, L. Biophys. J. 1996, 70, 2867. (20) Mariani, P.; Ciuchi, F.; Saturni, L. Biophys. J. 1998, 74, 430. (21) Su¨hnel, J. Biopolymers 2001, 56, 32. (22) Keniry, M. Biopolymers 2001, 56(3), 123. (23) Hardin, C. G.; Perry, A. G.; White, K. Biopolymers 2001, 56, 123. (24) Ross, W. S.; Hardin, C. C. J. Am. Chem. Soc. 1994, 116, 6070. (25) Hud, N. V.; Smith, F. W.; Anet, F. A. L.; Feigon, J. Biochemistry 1996, 35, 15383. (26) Gu, J. D.; Leszczynski, J. J. Phys. Chem. A 2002, 106, 529. (27) Wong, A.; Wu, G. J. Am. Chem. Soc. 2003, 125, 13895. (28) Byron, O.; Gilbert, R. J. C. Curr. Opin. Biotechnol. 2000, 11, 72. (29) Parsegian, V. A.; Rand, R. P.; Fuller, N. L.; Rau, D. C. Methods Enzymol. 1986, 127, 400. (30) Rappolt, M.; Hickel, A.; Bringezu, F.; Lohner, K. Biophys. J. 2003, 84(5), 3111. (31) Worcester, D. L.; Franks, N. P. J. Mol. Biol. 1976, 100, 359. (32) Franks, N. P.; Lieb, W. R. J. Mol. Biol. 1979, 133, 469. (33) Luzzati, V. X-ray diffraction studies of lipid-water systems. In Biological Membranes, Physical Facts and Function; Chapman, D., Ed.; Academic Press: New York, 1968; Vol. 1, pp 71-123. (34) Spindler, L.; Drevensek Olenik, I.; Copic, M.; Romih, R.; Cerar, J.; Skerjanc, J.; Mariani, P. Eur. Phys. J. E 2002, 7, 95. (35) Ausili, P.; Pisani, M.; Finet, S.; Amenitsch, H.; Ferrero, C.; Mariani, P. J. Phys. Chem. B 2004, 108, 1783.