Mg²⁺ modulation of EMCV IRES key activity fragment equilibria and r(G·C) base-pair kinetics

Abstract

NMR magnetization transfer from water and ammonia-catalyzed exchange of the imino proton have been used to probe enhanced thermostability and conformational rearrangements induced by Mg^2 + in two key activity fragments r(CACCUGGCGACAGGUG) and r(GGCCAAAAGCC) of the encephalomyocarditis virus (EMCV) picornavirus internal ribosome entry site (IRES). We have measured some of their r(G·C) base-pair lifetimes and dissociation constants under different MgCl₂ conditions, and we compare them with those of other short RNA duplexes. The RNA fragment r(CACCUGGCGACAGGUG) adopts two topologies, a palindromic duplex with two conformations and a hairpin, whose equilibrium can be monitored: the duplex form is destabilized by Mg^2 + and temperature, a delicate balance wherein the entropic contribution of the free energy helps populate the hairpin state. For both fragments, the opening rates of the r(G·C) pairs are lower in the presence of Mg^2 + and their dissociation constants are smaller or comparable. Analysis of the results suggests that Mg^2 + has a preferential and specific effect on the r(CACCUGGCGACAGGUG) hairpin in the region close to the r(G·C) closing pair of the GCGA tetraloop, and the ion moves diffusively around r(GGCCAAAAGCC), thereby differentiating the GNRA and RAAA hairpin motifs that are both involved in the biological regulation functions of the EMCV IRES.

Introduction

Picornaviruses are the leading cause of infectious morbidity in the world and replicate in the eukaryotic cell cytoplasm with the help of a genome segment called the internal ribosome entry site (IRES) [1–3]. Encephalomyocarditis virus (EMCV) is a member of the cardiovirus genus of the positive-sense picornaviruses. Its IRES is organized in five structural domains, termed 1 to 5 or G to L from the 5^′ to the 3^′ end [4]. Domain 3 contains a hammerhead region (Fig. 1) partly constituted by apical stem loops with two phylogenetically conserved motifs bearing the sequences GNRA and RAAA (N can be A, C, G or U; R is A or G). In the EMCV IRES, the GCGA tetraloop (in mauve, Fig. 1) and the AAAA submotif (in blue, Fig. 1) are essential for the activity of the IRES within mammalian cells [5, 6] and maintain the IRES three-dimensional structural integrity, including its dependence on ions. Specific binding of the GNRA hairpin to the central zone of domain 3 was also observed and may be stabilized by the RAAA hairpin local conformation [7]. ¹H NMR and UV melting experiments [8] have shown that addition of Mg^2 + induces selective changes to the chemical shifts of the imino protons and a dramatic increase in apparent melting temperature of fragments containing the GCGA tetraloop and the AAAA tetranucleotide, suggesting that Mg^2 + may promote folding of the EMCV IRES and enhance its thermodynamic stability. Here, we present an experimental study of Mg^2 + effects on the conformational equilibria of two EMCV IRES fragments, one containing a GNRA hairpin motif and the other an RAAA hairpin motif, and make a quantitative local analysis of the retardation of solvent exchange with r(G·C) base-pairs upon Mg^2 + addition, interpreting it in terms of base-pair kinetics.

Fig. 1.

Schematic diagram showing the EMCV IRES hammerhead region containing the 16-mer (in mauve) and 11-mer (in blue) hairpins. The secondary structure base-pair numbering is based on the nomenclature proposed previously [2]. The four residues spanning 5^′ -[G⁵⁴⁷-A⁵⁵⁰]-3^′ form the GCGA tetraloop. The four residues spanning 5^′ -[A⁵⁶⁸-A⁵⁷¹]-3^′ constitute the AAAA tetranucleotide motif

Base pairing disruption studies constitute a useful characterization and complement to nucleic acid structure determination by NMR, giving a relatively complete picture of the base-pair elementary dynamics and a comprehensive local description of the structure accessibility. Over the past two decades, base-pair kinetics has proven to be a very sensitive probe of nucleic acid structures [9, 10]. For example, d(G·C) base-pair lifetimes are increased by two orders of magnitude in Z DNA as compared to B DNA [11, 12]. To address the question of a possible EMCV IRES conformational rearrangement induced by Mg^2 +, we have undertaken an NMR investigation of the imino proton exchange for three r(G·C) base-pairs in the two key activity fragments r(CACCUGGCGACAGGUG) (in mauve, Fig. 1) and r(GGCCAAAAGCC) (in blue, Fig. 1) under different MgCl₂ conditions. The fragments r(CACCUGGCGACAGGUG) and r(GGCCAAAAGCC) respectively contain the GCGA tetraloop and the AAAA submotif sequences, and are henceforward designated as the 16-mer and the 11-mer, respectively, in the following text.

Materials and methods

Imino proton exchange theory

Imino proton exchange from a base-pair is a two-step process [13] requiring base-pair opening, followed by transfer to a proton acceptor such as NH₃. The proton acceptor contribution to the exchange time is given by:

1

where τ₀ is the base-pair lifetime, K_diss the base-pair dissociation constant, and the proton transfer rate from the open pair. The imino proton transfer rate from the open pair induced by NH₃ is:

2

where k_coll is the collision rate, [NH₃] is the ammonia concentration, and ΔpK is the pK_a difference between the imino proton (pK_G = 9.3, pK_U = 9.2) and ammonia ().

Oligonucleotide synthesis

The fragments r(CACCUGGCGACAGGUG) and r(GGCCAAAAGCC) were synthesized using phenoxyacetyl -cyanoethyl phosphoramidites on a 2-M scale and purified as previously described [10]. The RNA concentration was determined from the UV absorbance using the A ²⁶⁰ values computed according to a nearest neighbor model [14].

Sample preparation

The NMR samples contained 0.2 mM 2,2-dimethyl-2-silapentane-5-sulfonate, whose methyl peak was set to 0 ppm for chemical shift reference. The sample pH was measured at room temperature before and after each experiment and adjusted using 0.1 to 1 M HCl and NaOH solutions.

Polyacrylamide gel electrophoresis

The 16-mer and 11-mer RNA fragments were subjected to electrophoresis at 0°C for 45 min (35 V/cm) in a 15% non-denaturing polyacrylamide gel (acrylamide/bis-acrylamide, 19:1) in 1 × TAE, pH 8.0.

NMR methods

All the NMR experiments were performed on a 500 MHz Bruker Advance spectrometer. The imino protons were assigned by 2D NOESY (nuclear Overhauser enhancement spectroscopy) and spectra collected with mixing times of 90 and 250 ms. The ¹H NMR titration of the 16-mer sample at 0°C, pH 6.0, was carried out by monitoring the integrals of the imino proton resonance peaks. Ammonia-catalyzed exchange was measured at pH 8.9, a value high enough to provide a fair proportion of NH₃. Varying the pH from 6.0 to 8.9 had little effect on the chemical shifts. The ammonia was added to the samples from a stock solution whose concentration was 6.5 M. The ammonia-catalyzed exchange experiments were carried out as previously described [10].

Results and discussion

16-mer and 11-mer topologies

The 16-mer and 11-mer structures were mostly investigated by NOESY NMR. Figure 2 (top) displays the 2D ¹H NOESY spectrum of the 16-mer which shows two families of exchangeable protons at the frequencies of the imino protons (10–15 ppm), demonstrating the presence of two different species. RNA is especially prone to adopting conformations other than the native state. NOESY methods allowed us to disentangle the resonance peaks whose assignments are colored in mauve and orange, corresponding to a monomer (the hairpin in mauve) and a palindromic duplex (orange) of the 16-mer (Fig. 2, bottom). The hairpin loop r(G⁵⁴⁹) imino proton is not visible and is in fast exchange with the solvent on the chemical shift scale. In contrast, the presence of the hairpin loop r(G⁵⁴⁷) imino proton resonance peak at 10.4 ppm and its weaker intensity indicate that it is partially involved in a non-canonical r(G·A) base-pair. As regards the 16-mer duplex, the duplex r(G⁵⁴⁹) imino proton resonance peak at 11.9 ppm shows that the additional Watson–Crick base-pairs r(G⁵⁴⁹·C) is formed. Additionally, the presence of the duplex r(G⁵⁴⁷) imino proton resonance peak at 10.1 ppm and its weaker intensity indicate that the duplex r(G⁵⁴⁷) imino proton is partially involved in a non-canonical r(G·A) base-pair. The hairpin and duplex r(U⁵⁵⁵) and r(G⁵⁵⁶) imino protons have identical chemical shifts (14.1 and 12.7 ppm, respectively).

Fig. 2.

Top: examples of imino proton NOESY cross-peaks of the monomer and the duplex of the 16-mer. The mauve color indicates the peaks of the monomer form, i.e., the 16-mer hairpin. The protons of the duplex form are colored orange. This color coding is conserved throughout the text. The lower inset shows an enlargement containing the 2D region between 11.8 and 12.1 ppm. r(G⁵⁴⁶) and r(G⁵⁴⁹) imino protons of the 16-mer duplex conformer with r(G⁵⁴⁷·A) in its closed state are termed G⁵⁴⁶ and G⁵⁴⁹; r(G⁵⁴⁶) and r(G⁵⁴⁹) imino protons of the 16-mer duplex conformer with r(G⁵⁴⁷·A) in its disrupted state are named G; their respective cross-peaks are labeled with orange stars. Solution conditions: strand concentration 0.5 mM, 15°C, pH 6.0. Bottom: schematic of the equilibrium between the hairpin (mauve) and the duplex (orange) form of the 16-mer. Non-canonical r(G⁵⁴⁷·A) base-pairs are indicated by black stars

Several features reveal an internal motion in the region of the duplex r(G⁵⁴⁷·A) base-pair. Two conformers of the 16-mer duplex are present: the first one has its non-canonical r(G⁵⁴⁷·A) base-pair formed and the second one has its r(G⁵⁴⁷·A) base-pair in a disrupted state. This 16-mer duplex conformational exchange is implied by cross-peaks (labeled with orange stars in the inset of Fig. 2) between the r(G⁵⁴⁶) and r(G⁵⁴⁹) imino protons (G⁵⁴⁶ at around 11.8 ppm and G⁵⁴⁹ at around 11.9 ppm in the inset of Fig. 2) of the duplex conformer with r(G⁵⁴⁷·A) in its closed state and the r(G⁵⁴⁶) and r(G⁵⁴⁹) imino protons (G at around 12.0 ppm and G at around 12.1 ppm in the inset of Fig. 2) of the duplex conformer with r(G⁵⁴⁷·C) in its disrupted state. The broadening of r(G⁵⁴⁶) and of r(G⁵⁴⁹) imino proton resonance peaks (1D spectrum with orange labeling, Fig. 2) of the duplex conformer with r(G⁵⁴⁷·A), in its closed state, is due to exchange with water which is facilitated by the disruption of the adjacent r(G⁵⁴⁷·A) base-pair. The difference between the chemical shifts of G⁵⁴⁶ and of G, or between the chemical shifts of G⁵⁴⁹ and of G, approximately equals 100 Hz (0.2 ppm × 500 Hz ppm^− 1). The frequency of r(G⁵⁴⁷·A) motion between r(G⁵⁴⁷·A) disrupted and closed states is then much less than 100 Hz, the corresponding constant rate being therefore much larger than 10 ms. This result, together with the fact that the 16-mer duplex r(G⁵⁴⁷) imino proton is not visible in the r(G⁵⁴⁷) disrupted state leads to the conclusion that the r(G⁵⁴⁷) imino proton in the r(G⁵⁴⁷) disrupted state is accessible and in fast exchange with the solvent on the chemical shift scale. The r(G⁵⁴⁷) disrupted state is radically different from the r(G⁵⁴⁷) open state described in the imino proton exchange theory (Section 2.1).

We also observed that the 16-mer hairpin/duplex equilibrium is progressively displaced in favor of the duplex form. The 16-mer hairpin lifetime is around 3 months at 5°C, as monitored by real-time 1D ¹H NMR. The 16-mer hairpin is trapped in a metastable state and evolves into the 16-mer duplex as a result of spontaneous random fluctuations. Conversely, the 11-mer sample exhibited only a monomer form (the 11-mer hairpin in blue, Fig. 1) as unraveled by 1D ¹H spectra and 2D ¹H NOESY spectra (not shown).

In order to confirm these observations, two types of experiments were carried out. Firstly, polyacrylamide gel electrophoresis corroborated that indeed the two 16-mer species migrate as a monomer and a duplex, whereas the 11-mer only exists as a monomer. Figure 3 (left) displays a polyacrylamide gel stained with ethidium bromide. In the REF-I and REF-II lanes, RNA linear fragments of known length were loaded: lane REF-I a refers to the band of the RNA duplex [r(CGCGAUCGCG)]₂; lane REF-II b and c correspond respectively to the bands of the RNA duplex [r(GCGCGCGAUCGCGCGC)]₂ and of the RNA monomer r(GCGCGCGAUCGCGCGC). Lane EIH-16 (EIH-16 stands for EMCV IRES Hammerhead 16-mer) shows two bands corresponding to the 16-mer monomer and duplex. Lane EIH-16 Q (EIH-16 Q stands for EMCV IRES Hammerhead 16-mer quenched) displays a unique band from a 16-mer sample heated at 90°C for 3 min and rapidly quenched in an ice-water bath. The 11-mer exists in a monomer form, as shown by lane EIH-11 (EIH-11 stands for EMCV IRES Hammerhead 11-mer).

Fig. 3.

Left: Polyacrylamide gel electrophoresis of the 16-mer and 11-mer. Samples in lanes REF-I and REF-II are references (see text). Samples in lanes EIH-16 and EIH-16 Q contained respectively, the 16-mer (2.5 g) and the quenched 16-mer (1.0 g). Lane EIH-11 displays the 11-mer (1.5 g). Right: NMR titration of the 16-mer duplex form vs. the hairpin concentration at 0°C, pH 6.0. The line of slope = 2 drawn through the data points confirms the formation of a duplex

Secondly, the ¹H NMR titration of the 16-mer sample at 0°C, pH 6.0, which proved that the concentration of the duplex increases as the second power of the monomer concentration (Fig. 3, right), defines the formation of the 16-mer duplex, as one expects for a stoichiometry of 2:

3

where K_dilution is the equilibrium dissociation constant of the duplex in the 16-mer hairpin/ duplex equilibrium at each dilution point. The ¹H NMR titration of the 16-mer was also performed within the temperature range 5–40°C in steps of 5°C, pH 6.0, establishing the formation of the 16-mer duplex as well (not shown). The equilibrium between the two 16-mer forms (Fig. 2, below) is partly explained by the two supplementary r(G⁵⁴⁹·C) Watson–Crick hydrogen bonds in the 16-mer duplex form and the formation of the two non-canonical r(G⁵⁴⁷·A) base-pairs [15]. The 11-mer case does not reveal any advantageous hydrogen bond balance in favor of a duplex form, and the entropic driving force simply leads to the sole formation of a hairpin monomer.

Mg^2 + control of the 16-mer hairpin/duplex equilibrium

The analysis of such energetic properties would arguably suggest that the 16-mer hairpin/ duplex equilibrium could be monitored by distinguishing the electrostatic contribution of both topologies. Figure 4 demonstrates that Mg^2 + favors the 16-mer hairpin formation and provides evidence that 16-mer hairpin enhanced stability arises from Mg^2 +. Also, we did not observe any changes in the chemical shifts and the integrals of the 16-mer imino proton resonance peaks upon addition of NaCl (1, 3, and 6 molar equivalents), of KCl (1, 3, and 6 molar equivalents), of MnCl₂ (3 molar equivalents), or of CoCl₂ (3 molar equivalents). Therefore, a simple increase in the ionic strength did not produce the same effect as the addition of Mg^2 +. This observation seems to indicate the specificity of Mg^2 + for the 16-mer hairpin. The temperature dependence of the Gibbs free energy ΔG is provided by the equilibrium ratio between the two 16-mer species:

4

where RT is the thermal energy. The thermodynamic parameters were derived for the equilibrium between the hairpin and the duplex: ΔH increases from 22 to 24 kJ mol^− 1 as Mg^2 + is added, and ΔS takes an approximately constant value of 74 J mol^− 1 K^− 1. The derivative of the Gibbs free energy ΔG with respect to the number of added Mg^2 + equivalents provides the variation of the chemical potential :

5

where n is the number of added Mg^2 + equivalents, n_{16 − mer} is the number of moles of 16-mers, and T and P are respectively the temperature and the pressure. does not change much with temperature and varies between 330 J mol^− 1 eq^− 1 at 0°C and 510 J mol^− 1 eq^− 1 at 50°C.

Fig. 4.

The temperature dependence of the Gibbs free energy, ΔG, of the equilibrium between the hairpin and duplex forms of the 16-mer, parameterized by Mg^2 + concentration from 0.1 to 3 molar equivalents. For the sake of clarity, the experimental points are only displayed for 0 (gray), 0.1 (blue), and 3 (red) molar equivalents of Mg^2 +. A color gradient from orange to mauve symbolizes displacement of the equilibrium between the hairpin (mauve) and the duplex form (orange) of the 16-mer

Figure 5 schematically presents a plausible energetic path for the RNA 16-mer sequence, incorporating the various data obtained from the evolution of the 16-mer hairpin with temperature and upon Mg^2 + addition. The single-stranded 16-mer random conformation was observed at 90°C; its energy level is therefore placed above those of the 16-mer hairpin and duplex in the diagram. The temperature variation results help in positioning plausible 16-mer transition state energy levels. The 16-mer hairpin/duplex equilibrium Gibbs free-energy differences ΔG and the activation energies and were obtained from the previously mentioned and other ¹H NMR titrations of the 16-mer performed within a range of temperatures 0–50°C, with and without Mg^2 +. ΔG is around 6.6 kJ mol^− 1 and per se, whose average over this temperature range roughly equals 420 J mol^− 1 eq^− 1, has little influence on the 16-mer hairpin/duplex equilibrium, as corroborated by the coexistence of the two species in solution. An input of thermal energy combined with a Mg^2 + addition has a larger activation effect in favor of the 16-mer hairpin and agrees with the explanation that the 16-mer hairpin is trapped in a metastable state: . The 16-mer hairpin lifetime is then around 6 months at 5°C, as monitored by real-time 1D ¹H NMR. This duplex-to-hairpin transition induced by Mg^2 + is surprising, since one might have expected the reverse switch to occur at increasing polyelectrolyte concentrations: given the salt screening of the electrostatic repulsions between the negatively charged phosphate groups, the 16-mer hairpin/duplex equilibrium should have been even more displaced in favor of the duplex form. An explanation for the greater stability of the hairpin may be the following. The favorable stacking energy and the adverse conformational entropy are the two largest factors that determine RNA stability [16]. In most models, the thermodynamic stability of an RNA hairpin is ascribed to a cooperative two-state (all-or-none) process, where ΔG of the tetraloop is assumed to be purely entropic and sequence-independent. The directionality of the GNRA backbone is reversed to reveal the two antiparallel strands of the stem and provokes steric repulsion. In compensation, the GNRA stability comes from intraloop interactions such as non-canonical r(G·A) base pairing and cation-mediated interactions, which cause a sequence dependence and an enthalpic contribution to ΔG of the tetraloop. The 16-mer hairpin r(G⁵⁴⁶) site belongs to the closing base-pair between the loop and the stem (Fig. 1, in mauve), and constitutes a strong candidate for binding a Mg^2 + ion [8, 17]. Its immediate surroundings in the 16-mer duplex form may be perturbed such that the hairpin/duplex equilibrium is displaced into a state with a larger 16-mer hairpin proportion. The chemical shift of the 16-mer hairpin r(G⁵⁴⁷) increases by around 0.3 ppm after incremental addition of 3 molar equivalents of Mg^2 + [8], whereas the chemical shift of the 16-mer duplex r(G⁵⁴⁷) varies by less than 0.01 ppm. The Mg^2 +-induced local reorganization observed for the 16-mer hairpin may not be possible in the 16-mer duplex, where the addition of Mg^2 + may hamper the internal motions of its non-canonical r(G⁵⁴⁷·A) base-pairs, implying an uncompensated loss of conformational entropy. The conformationally restricted 16-mer r(G⁵⁴⁶) region helps populate the hairpin conformer at the expense of the competing duplex.

Fig. 5.

Energy diagram summarizing the Mg^2 + effects on the equilibrium between the hairpin and duplex forms of the 16-mer. The dagger indicates transition states and their corresponding activation energy variations in the presence and absence of Mg^2 +. Variations of the chemical potential due to the addition of Mg^2 + are displayed as a function of temperature

Mg^2 +-parameterized ammonia-catalyzed r(G·C) imino proton exchange

Motivated by the importance of quantifying the Mg^2 +-mediated changes, we focused on how much the Mg^2 + cation modifies the rate constants of the base-pair opening–closing mechanisms and thereby helps stabilize the 16-mer hairpin GCGA tetraloop and the 11-mer simplex AAAA tetranucleotide. In this study, we characterized the 16-mer hairpin base-pairs r(G⁵⁴⁶·C) and r(G⁵⁵³·C), and the 11-mer simplex base-pair r(G⁵⁷²·C) as a first probing of the Mg^2 + local effects on the base-pair opening–closing kinetics. The base-pairs r(G⁵⁴⁶·C) and r(G⁵⁷²·C) are both the closing base-pairs between the loop and the stem of the 16-mer and the 11-mer, and may give information about motional changes at these structural junctions upon Mg^2 + addition. On the other hand, r(G⁵⁵³·C) is an inner-stem base-pair and may help for the analysis with respect to the previously studied RNA duplexes [10]. Using 1D ¹H NMR magnetization transfer from water and ammonia-catalyzed exchange of the imino proton, we measured their base-pair lifetimes and their dissociation constants in the presence and in the absence of Mg^2 +. NH₃ remains the best choice for imino proton exchange studies due to its small size, high pK_a, good solubility, neutral character, and lack of detectable proton on the ¹H NMR spectrum at alkaline pH. The presence of the equilibrium between the two 16-mer species could have complicated the evaluation of the base-pair kinetic parameters. Since we were primarily interested in the base-pair kinetics of the 16-mer hairpin, the 16-mer sample was heated to 90°C for 3 min and rapidly cooled in an ice-water bath before conducting the measurements.

As predicted by the two-state (closed/open) model of the base-pair [13], the plot of vs. the inverse of ammonia concentration is a straight line whose extrapolation to infinite catalyst concentration yields the base-pair lifetime . The apparent dissociation constants K_diss were computed according to Eq. 1 from the ratio of the rates of exchange catalysis measured in the RNA fragments and for the isolated nucleoside representing the base in the open pair situation. The apparent lifetime of the open pair, τ_open, is equal to the product . The ammonia contribution to imino proton exchange in the r(CACCUGG CGACAGGUG) hairpin displayed in Fig. 6 is typical of the exchange measurements performed in the presence or in the absence of Mg^2 +. The linear dependence of vs. the inverse of the ammonia concentration is consistent with an exchange process occurring from a single open state. Without Mg^2 +, the base-pair lifetime of r(G⁵⁵³·C) obtained by extrapolation of the NH₃ exchange contribution to infinite ammonia concentration is 22 ms at 15°C. Upon addition of 1 molar equivalent of Mg^2 +, the r(G⁵⁵³·C) lifetime increases to 41 ms.

Fig. 6.

Ammonia contribution to the exchange time of the r(G⁵⁵³·C) imino proton of the 16-mer hairpin with and without Mg^2 +. The slopes of the plots indicate a slight decrease of base-pair stability, whereas extrapolations to infinite ammonia concentration show an increase of the base-pair lifetimes upon addition of Mg^2 +

r(G·C) base-pair kinetics with and without Mg^2 +

Several differences emerge from comparison of base-pair kinetics and stability in the RNA fragments with and without Mg^2 + (Table 1).

1. Kinetics of the r(G·C) base-pairs without Mg^2 +

   The r(G·C) base-pair lifetimes fall in the same range of values as those of RNA duplexes previously studied [10]. The base-pairs r(G⁵⁴⁶·C) and r(G⁵⁷²·C) are compared to the r(G·C) base-pairs in the second position of the previously studied duplexes [10], in order to get a first understanding of the stabilization provided by a loop with regard to a terminal r(G·C) base-pair influenced by fraying effects. Additionally, the inner-stem base-pair r(G⁵⁵³·C) is compared to r(G·C) base-pairs in the third position of the previously studied duplexes [10], which represent typical inner-stem base-pairs. This approach offers a set of reference values to serve as benchmarks for undertaking the more subtle comparison when Mg^2 + is brought into play later on. More precisely, r(G⁵⁵³·C) behaves like an RNA duplex inner r(G·C) pair but r(G⁵⁴⁶·C) and r(G⁵⁷²·C) show a second position base-pair character. The similarity between the double-stranded stems of the two hairpins and short RNA duplexes is largely explained by their common right-handed helices composed of Watson–Crick base-pairs with an A-form geometry. The 16-mer r(G⁵⁴⁶·C) apparent dissociation constants are much greater than those of control RNA duplexes, whereas the 11-mer r(G⁵⁷²·C) apparent dissociation is just slightly greater. The lifetimes of the open r(G⁵⁴⁶·C), which are about ten times longer than those internal to RNA duplexes, indicate an increased end-fraying propagation.

2. Kinetics of the r(G·C) base-pairs with Mg^2 +

   The 11-mer r(G⁵⁷²·C) base-pair lifetime is comparable to that obtained without Mg^2 +. In contrast, the 16-mer r(G⁵⁵³·C) and r(G⁵⁴⁶·C) base-pair lifetimes are longer upon addition of 3 molar equivalents of Mg^2 +, respectively increasing from 22 to 43 ms and from 4.8 to 11 ms. An Mg^2 + preferential effect for the 16-mer hairpin may explain the variations of r(G⁵⁵³·C) and r(G⁵⁴⁶·C) opening–closing kinetics which strongly suggest a structural effect of this cation. This observation supports the hypothesis of Mg^2 +/GCGA tetraloop binding, which may provide a nucleation site to ensure proper folding of the larger RNA [15, 18, 19]. The 16-mer r(G⁵⁴⁶·C) apparent dissociation constant is much lower in the presence of Mg^2 + and the r(G⁵⁵³·C) apparent dissociation constant remains unaltered, suggesting that Mg^2 + serves to enhance stability of the 16-mer [16]. It is also unlikely that the changes observed upon addition of 3 molar equivalents of Mg^2 + could be due to intermolecular effects, because no indication for such interactions could be found on the 2D NMR spectra. They more probably might be attributed to further structural changes. Their characterization and the distinction between global/local structural effects induced by the addition of Mg^2 + constitute an important problem for the future and may be solved by determining the 16-mer hairpin structures without and with Mg^2 +. Furthermore, Mg^2 + magnetic relaxation dispersion measurements combined with residual dipolar coupling determinations in oriented media may help in completing the dynamical aspects provided by the base-pair kinetics results. The 11-mer r(G⁵⁷²·C) apparent dissociation constant is slightly higher. The lifetimes of the open r(G⁵⁷²·C) argue for it being comparable to those in the absence of Mg^2 +.

Table 1.

Lifetimes, apparent dissociation constants, and apparent open pair lifetimes in r(G·C) base-pairs of the 16-mer and the 11-mer hairpins at 15°C

		[Mg2 + ]	(ms)		τopen (ns)
16-mer simplex	G546·C	0 eq	4.8 (3.4, < < 1)a	21 (2.2, 20)a	101 (7.5, < < 20)a
		1 eq	6.5	7.6	49
		3 eq	11	15	160
	G553·C	0 eq	22 (33, 7)b	1.6 (0.28, 0.57)b	35 (9.2, 4)b
		1 eq	41	4.0	160
		3 eq	43	2.2	95
11-mer simplex	G572·C	0 eq	11 (3.4, < < 1)a	4.9 (2.2, 20)a	54 (7.5, < < 20)a
		1 eq	13	9.5	120
		3 eq	11	11	120

The uncertainties on base-pair lifetimes are estimated as ±30%. The uncertainties on dissociation constants are about ±20%

^aAs a comparison, the parameters measured for base-pair (G²· C) respectively in [r(CGCGAUCGCG)]₂ and [d(CGCGAUCGCG)]₂ are given inside parentheses

^bThe parameters measured for base-pair (C³· G) respectively in [r(CGCGAUCGCG)]₂ and [d(CGCGAU CGCG)]₂ are given inside parentheses

Conclusion

Divalent Mg^2 + cations are efficient and specific in promoting RNA folding and stability. Their distribution around an RNA molecule may be represented as continuous and dynamic. However, the observation of different values of r(G⁵⁵³·C) and r(G⁵⁴⁶·C) base-pair lifetimes with and without Mg^2 + shows that this alteration in exchange kinetics is related to the local presence or absence of Mg^2 +. Moreover, the chemical shift of 16-mer r(G⁵⁴⁷) showed a variation of around 0.3 ppm after incremental addition of molar equivalents of Mg^2 + [8]. Crystallographic studies [20] have already led to an understanding of the stereochemistry of metal ion binding to bases, to nucleosides and to nucleotides, recognizing that the phosphate oxygen, the sugar 2^′ hydroxyl, or the base keto oxygen and ring nitrogens are good ligands for alkaline earth cations, since these atoms carry lone electron pairs. Mg^2 + seems to bind or at least to have a preferential effect on a specific region of the 16-mer hairpin, involving r(G⁵⁵³·C) and r(G⁵⁴⁶·C), and may be classified as a specific ion for the 16-mer hairpin. The strength of the Mg^2 +-induced force appears strong enough to slow down the neighboring base-pair kinetics against random motion. Our NMR approach is local, although the possibility cannot be totally excluded that the observed Mg^2 +-induced changes in exchange rate constants and dissociation constants may originate from long-range effects of Mg^2 + binding/affecting other sites. On the other hand, the addition of Mg^2 + was not accompanied by chemical shift changes of the 11-mer imino protons (less than 0.01 ppm) and the constant value of 11-mer r(G⁵⁷²·C) base-pair lifetime indicates that Mg^2 + seems to move diffusively without altering any specific sites in the 11-mer hairpin. Interest in the role of Mg^2 + in promoting IRES function has been rekindled recently, since it may bring light to bear on the active structure of the picornavirus genome. Recent experiments have shown that RNA folding mechanisms depend critically on the initial conditions [21]. We are now therefore characterizing the other 16-mer and 11-mer base-pairs which seem to open independently of each other, both with and without Mg^2 +, a result which would appear to be in agreement with general views on RNA behavior. Their exchange kinetics at different temperatures is still under analysis but seem to strengthen the interpretation of the preferential and specific effect of Mg^2 + and will be presented elsewhere.