Thermodynamics and kinetics for base-pair opening in the P1 duplex of the Tetrahymena group I ribozyme.

The thermodynamics and kinetics for base-pair opening of the P1 duplex of the Tetrahymena group I ribozyme were studied by NMR hydrogen exchange experiments. The apparent equilibrium constants for base pair opening were measured for most of the imino protons in the P1 duplex using the base catalysts NH3, HPO4(2-) or TRIS. These equilibrium constants were also measured for several modified P1 duplexes, and the C-2.G23 base pair was the most stable base pair in all the duplexes. The conserved U-1*G22 base pair is required for activity of the ribozyme and the data here show that this wobble base pair destabilizes neighboring base pairs on only one side of the wobble. A 2'-OMe modification on the U-3 residue stabilized its own base pair but had little effect on the neighboring base pairs. Three base pairs, U-1*G22, C-2*G23 and A2*U21 showed unusual equilibrium constants for opening and possible implications of the opening thermodynamics of these base pairs on the undocking rates of the P1 helix with catalytic core are discussed.

INTRODUCTION

The Tetrahymena group I intron has been a powerful model system for studying conformational changes in RNA and for understanding the catalytic potential of RNA (1–3). The Tetrahymena L-21 ScaI ribozyme is derived from the self-splicing group I intron and catalyzes a transesterification reaction as part of the self-splicing reaction (1,4). The internal guide sequence (IGS; 3′-GGAGGG—ribozyme) in the ribozyme forms the P1 duplex with a substrate, for example 5′-CCCUCUpA. The ribozyme also binds exogenous guanosine which reacts with the substrate to yield GpA and CCCUCU (1). Cleavage rates have been measured on a large number of modified substrates or ribozymes, from which it was proposed that the catalytic reaction is mediated by three divalent metal ions at the active site of the ribozyme (2–12). Additional kinetics studies showed that the P1 duplex docks into the catalytic core by a two-step reaction: the first step involves formation of the P1 duplex, where the substrate binds to the IGS of the ribozyme and subsequent formation of a closed complex, where the P1 duplex docks into the catalytic core of ribozyme (1,9,13). Various modifications of the substrate or the IGS have been shown to affect the cleavage kinetics and/or the thermodynamics and kinetics for docking of the P1 duplex into the catalytic core (1,3,7). For example, a DNA substrate binds the ribozyme with 10 000-fold lower affinity than an RNA substrate and is cleaved 10 000 times slower than an RNA substrate (6,14,15). Biochemical studies showed that the 2′-OH groups at positions 22, 25 and –3 on the P1 duplex are important for stabilizing tertiary interactions with the catalytic core of ribozyme (1,12). A dU modification at position −1 of the substrate also reduced the cleavage rate (3) but had no effect on the docking/undocking rates (7). Replacing the conserved U-1·G22 wobble pair with a Watson–Crick C-1·G22 pair leads to a large (∼80-fold) increase in the undocking rate and 2′-OMe modifications at positions −1 or –3 also increased the undocking rates, 20- or 140-fold, respectively (7).

X-ray crystal structures of the group I intron from several species have been reported (16–18) and these structures have been very useful in the interpretation of previous biochemical and biophysical studies (19,20). However, even with these X-ray structures it is not clear whether the changes in cleavage activity observed for a specific modification are due to changes in conformation, changes in dynamics or flexibility of the RNA or a redistribution of the populations of various conformational states. NMR hydrogen exchange experiments are used here to probe the conformational dynamics of the isolated P1 duplex for the Tetrahymena ribozyme with the goal of better understanding specific properties of the P1 duplex that are required for catalytic activity.

NMR hydrogen exchange experiments provide information on the thermodynamics and kinetics for base-pair opening and therefore represent a probe of the dynamic motions of the base pairs. These experiments have been used to probe base-pair opening in various DNAs (21–25), RNAs (25,26) and DNA–protein complexes (27,28). Analysis of the hydrogen exchange of imino protons generally employs a two-state (open/closed) model for the base pair, where hydrogen exchange only occurs from the open state (29–31). The opening/closing rate constants and/or equilibrium constant for base-pair opening can often be determined by measuring the exchange as a function of the concentration of external catalyst. Hydrogen exchange data can also be used to probe how intramolecular or intermolecular interactions stabilize nucleic acid duplexes. For example, NMR imino proton exchange studies of a DNA duplex bound to the enzyme DNA glycosylase showed that the enzyme substantially increases the equilibrium constant for opening an A·T base pair without changing the rate constant for base-pair opening (28).

The imino proton exchange rates were measured here for a model P1 RNA duplex formed by an 11-nt IGS, 5′-GGUUUGGAGGG-3′ and a 13-nt wild-type (wt) substrate, 5′-GGCCCUCUAAACC-3′ (Figure 1A). The equilibrium constants for opening of the two most stable base pairs in the P1 duplex, C-2·G23 and C-4·G25, were determined from measurements of imino proton exchange rates as a function of the concentration of the base catalyst ammonia at 35°C. The less effective base catalysts, or TRIS base, were also used to estimate the equilibrium constants for opening of base pairs for the more rapidly exchanging imino protons in the P1 duplex. These data show that the U-1·G22 wobble in the P1 duplex has a large effect on the equilibrium constants for opening of this base pair, and also asymmetrically affects the equilibrium constants for opening of neighboring base pairs. To further understand the correlation between the base-pair opening and changes in biochemical properties and function of the P1 duplex in the Tetrahymena ribozyme, the equilibrium constants for base-pair opening were measured in a number of previously biochemically characterized modified P1 duplexes (see Figure 1A). Comparison of the wt and modified P1 duplexes indicates that some differences in base-pair opening properties are correlated with changes in undocking rates of the P1 duplex into the catalytic core of the ribozyme.

Figure 1.

(A) The sequences of the wt P1, (-1C)-P1, (-3m)-P1 and (-5U)-P1 duplexes. The numbering of the P1 duplexes follows that of the Tetrahymena group I intron. The modifications are indicated by italic and bold characters. The top strand is the IGS in the ribozyme and bottom strand is the substrate. The P1 helix can be divided into an A·U-rich region (denoted region I) and a G·C-rich region (denoted region II) around the central G·U base pair (B) 1D imino proton spectra of the wt P1, (-1C)-P1, (-3m)-P1 and (-5U)-P1 duplexes dissolved in NMR buffer at 35°C. The resonance assignments for the imino protons were made from analysis of 2D NOESY spectra.

MATERIALS AND METHODS

Sample preparation

The sequence of the wt and three modified P1 duplexes employed here are given in Figure 1A. The substrate strands for the wt and the (-5U)-P1 duplexes were prepared by in vitro transcription as previously described (32). The other oligomers, the IGS-11mer and the substrates for the (-1C)- and (-3m)-P1 duplexes were purchased from Dharmacon, Inc. (Lafayette, CO) or the W.M. Keck facility at Yale University (New Haven, CT). The oligomers were desalted using NAP 5 columns (Amersham Biosciences, NJ), and ∼0.5–0.9 mM RNA duplex samples were formed by combining molar equivalents of IGS and the appropriate substrate in 90% H2O–10% D2O NMR buffer (10 mM sodium phosphate, 100 mM NaCl, 10 mM MgCl2, 0.1 mM EDTA, pH 6.6). The duplexes were then dialyzed to achieve specific buffer conditions using Microsep 1K concentrators (PALL Life Sciences, MI). The pH of each sample was measured directly by using a micro-combination electrode (Fisher Scientific), except for samples containing TRIS buffer. To account for the change in the pKa of TRIS with temperature, the pH of the solution was directly measured at room temperature (23.4°C) and then the pH of the NMR sample at 35°C was calculated using the relation between pKa of TRIS and temperature, ΔpKa = −0.033 × ΔT (33).

Analysis of imino proton exchange

The formalism of catalyzed proton exchange has been previously described in detail (30,31), and is briefly presented here. The rate constant for imino proton exchange, ktr, induced by a base (proton acceptor) in an isolated nucleoside is: where ki is the rate constant for imino proton transfer, kcoll is the collision rate constant, [B] is the base catalyst concentration and ΔpKa is the pKa difference between the imino proton and the base.

Imino proton exchange from a base pair was analyzed by the two-step process, base-pair opening followed by proton transfer to a base such as NH3 or TRIS. The base catalyst contribution to the rate constant for imino proton exchange, kex,B, is given by where kop and kcl are the rate constants for opening and closing of the base pair, respectively, and ktr is the rate constant for proton exchange by base catalyst in the open state [see Equation (1)], α is called the accessibility factor (30), and Kop (= kop/kcl) is the equilibrium constant for base-pair opening [note that this is referred to as Kd by Guéron and coworkers (30,31)].

In the absence of added catalysts, exchange is due to a concerted transfer involving water molecules bridging the imino proton and the nitrogen of the complementary base, which acts as an intrinsic (or internal) catalyst (30). The rate constant for imino proton exchange in the absence of added catalyst, kACC, is where is the rate constant for imino proton transfer by intrinsic base catalyst, and this rate constant is in the range of 106 s−1 (30). When kcl is larger than , Equation (3) becomes:

NMR spectroscopy

All NMR experiments were performed on Varian Inova 500 MHz spectrometer using a HCN triple-resonance probe equipped with z-axis pulsed-field gradients. 1D data were processed with the program FELIX (Accelrys), 2D data were processed with the program NMRPIPE (34) and analyzed by the program Sparky (35). The imino protons in the various P1 duplexes are assigned using 120 and 200 ms Watergate-NOESY or jump-return-echo-NOESY spectra acquired at 15 and 35°C. The apparent longitudinal relaxation rate constants (R1a = 1/T1a) of the imino protons were determined by semi-selective inversion recovery 1D NMR experiments, where a semi-selective 180° inversion pulse was applied to imino proton region (9–15.5 p.p.m.) before the jump-return-echo water suppression pulse (29). The apparent relaxation rate constant of water (R1w) was determined by a selective inversion recovery experiment, using a DANTE sequence for selective water inversion and application of a 1-ms gradient (4 G/cm) after the DANTE pulse and a weak gradient (0.02 G/cm) during the relaxation delay to avoid radiation damping of water signals (29). The probe was detuned for this experiment to help reduce the large water signal. The R1a and R1w were determined by curve fitting of the inversion recovery data to the appropriate single exponential function. The R1w of water containing 90% H2O–10% D2O 10 mM sodium phosphate (pH 6.6) in NMR buffer was 0.274 s−1 at 35°C at 500 MHz. The hydrogen exchange rates of the imino protons were measured by a water magnetization transfer experiment at 35°C, where a selective 180° pulse for water was applied, followed by a variable delay, and then a 3–9–19 acquisition pulse was used to suppress the water signal (29,36). During the delay times between selective water inversion and acquisition pulses, a weak gradient (0.02 G/cm) was applied to prevent the radiation damping of the water signal. The intensities of each imino proton were measured with 20 different delay times ranging from 5 to 100 ms (see Figure 2). The exchange rates were determined by fitting the relative peak intensities, I(t)/I0, of the imino protons to: where R1a and R1w were previously measured by inversion recovery experiments, kex is the rate constant for exchange of the imino proton, and I0 and I(t) are the peak intensities of the imino proton at times zero and t, respectively (29).

Figure 2.

(A) 1D spectra of the water magnetization transfer experiments showing the imino protons of the wt P1 duplex in NMR buffer at 35°C. The control spectrum with no selective water inversion is shown at the bottom with resonance assignments. The delay times between the selective water inversion and acquisition pulses are indicated on the left side of the spectra. The inter-acquisition delay was set to 15 s which allows for complete relaxation of the water magnetization. (B) The rate constants for exchange of the imino protons for the P1 duplex at 35°C determined by fitting to Equation (5), and the error bars represents fitting errors.

Determination of the apparent equilibrium constant for base-pair opening

The apparent relaxation rate constant for an imino proton is the summation of the R1 rate constant, which is the inverse of T1 spin-lattice relaxation time of this proton, and the rate constant for imino proton exchange. It is assumed that the R1 values of the imino protons are not affected by the addition of the base catalyst such as ammonia, TRIS or . Thus, the apparent relaxation rate constant, R1a, for an imino proton is given by: where R1 is the relaxation rate constant of the imino proton and kACC and kex,B are the contribution to imino proton exchange of the intrinsic base and the added base catalyst, respectively. Substituting Equation (2) into Equation (6) yields the following: and curve fitting R1a of the imino protons as a function of the base catalyst concentration with Equation (7) (using Sigmaplot) gives the apparent equilibrium constant for base-pair opening (αKop) and rate constant for opening (kop); where the ki for each base catalyst was calculated using Equation (1). In these fits, all the R1a values were weighted by their inverse variances. Under certain conditions where kcl (= kop/Kop) is much larger than αki[B], Equation (7) simplifies to:

RESULTS

Resonance assignment of imino protons in the P1 duplexes

The 1D imino proton spectra of the wt and modified P1 duplexes at 35°C are shown in Figure 1B, and assignments were made by the analysis of NOESY spectra at 35°C. The three other modified P1 duplexes were assigned by comparison of their NOESY spectra with the wt P1 duplex. All imino proton resonances except G18 and U21 imino protons are well resolved in the 1D spectra (Figure 1B). 1D imino proton spectra of the wt P1 duplex were collected as a function of temperature and the resonances for all the non-terminal base pairs are still observed up to 50°C (data not shown).

Relaxation and hydrogen exchange experiments for the imino protons in the wt P1 duplex

The apparent longitudinal relaxation rate constants (R1a) of the imino protons of the P1 duplex were determined by semi-selective inversion recovery experiments on the imino proton resonances. At 15°C, the R1a data for the imino proton resonances did not fit well to a single exponential and all showed unusual double exponential relaxation (see Supplementary Data Figure 6S, left column). This behavior may result from partial aggregation of the P1 duplex at low temperature. However, inversion recovery relaxation data for all the imino protons fit well to single exponentials at 35°C, (see Supplementary Data Figure 6S, right column), thus the exchange properties of the imino protons for all the P1 duplexes were determined from experiments at 35°C.

Hydrogen exchange rates were determined from water magnetization transfer experiments on the imino protons for the wt P1 duplex at 35°C as described in Materials and Methods section. Some imino protons show large differences in peak intensities as a function of delay after water inversion (Figure 2). For example, rapidly exchanging imino protons such as U19 and U20 show negative peaks at short delay times (50 ms in Figure 2A), whereas the G23 resonance, which is the slowest exchanging imino proton, remains basically unchanged up to 100 ms. Figure 2B shows the kex of the imino protons of the P1 duplex determined by the curve fitting to Equation (5) (see Materials and Methods). The G and U imino protons in the A·U-rich region (region I in Figure 1A) have kex from 9 to 40 s−1, whereas, except for the terminal base pairs, the imino protons in the G·C-rich region (region II) all have kex < 6 s−1. The U21 imino proton next to the central G·U pair has the largest rate constant for exchange of any non-terminal base pair (kex of 40.0 ± 4.8 s−1), and the G22 and U-1 imino protons of the central G·U wobble pair have kex values of 12.7 ± 0.8 and 15.5 ± 0.7 s−1, respectively. As seen in Figure 2B, the central G·U wobble pair has a larger effect on kex for base pairs in region I than region II. The most striking result is that the G23 imino proton is the slowest exchanging proton in the P1 duplex, even though the C-2·G23 base pair is flanked on one side by a relatively rapidly exchanging U-1·G22 base pair and on the other by only a moderately exchanging U-3·A24 base pair.

Base-pair opening stability and kinetics for the P1 duplexes with NH3 or base catalysts

The apparent equilibrium constants for base-pair opening in the wt P1 duplex were determined from ammonia-catalyzed imino proton exchange measurements. The effects of [NH3] on R1a of the imino protons of the P1 helix were measured by inversion recovery experiments and results for the G23 and G25 imino protons are shown in Figure 3. From these data, the base-pair lifetimes (τ0 = 1/kop) (C-2·G23: 31 ± 9 ms; C-4·G25: 9 ± 7 ms) and apparent equilibrium constants for opening (αKop) (C-2·G23: 0.30 ± 0.06 × 10−6; C-4·G25: 0.36 ± 0.06 × 10−6) were determined by curve fitting using Equation (7). The R1a data for the G25 imino proton show an almost linear correlation with [NH3] in Figure 3, indicating that kcl is significantly larger than αki[NH3] [see Equation (7)]. These conditions lead to the large error in τ0 for C-4·G25. Thus, a more conservative interpretation of the data for the C-4·G25 base pair is that τ0 ≤ 9 ms. The data for C-2·G23 can be used to calculate an apparent lifetime for base-pair opening, ατopen, of 9 ± 3 ns (using ατopen = τ0αKop). Except for G23 and G25, the other imino protons showed substantial line broadening for [NH3] >30 mM (data not shown); thus, the lifetimes and equilibrium constants for opening of these base pairs could not be determined from the NH3 data.

Figure 3.

Ammonia-catalyzed exchange of the P1 duplex. Apparent spin-lattice R1 relaxation rate constants (R1a) of the G23 and G25 imino protons as a function of the ammonia concentration in NMR buffer at 35°C and pH 8.0. The NH3 concentration was calculated by , where [NH3]total is the total concentration of ammonia added. The solid lines are the best fits to single exponential, and the error bars represent curve fitting errors during the determination of R1a from inversion recovery data.

is a much weaker catalyst than ammonia and was also used to determine Kop for the base pairs. The R1a values of the imino protons of the P1 duplex dissolved in the 10 mM sodium phosphate NMR buffer were measured at three different pHs, leading to different concentrations for the base catalyst . The phosphate tribasic anion precipitates with Mg 2+ (Ksp of Mg3(PO4)2 is ∼10−24) (37); thus, only three data points were collected here due to precipitation at higher concentrations of the phosphate buffer or higher pH with 10 mM phosphate buffer. As predicted from Equation (8), Figure 4A shows that R1a increases linearly as a function of the concentration. However, the limited number of concentrations, combined with the fact that is a much weaker catalyst than ammonia, allowed only a qualitative interpretation of these data. As seen in Figure 4, base pairs C-5·G26 and U-3·A24 show the lowest dependence of R1a on (and therefore smaller values for αKop), C5·G18, A4·U19 and A3·U20 show intermediate αKop values and both imino protons in the U-1·G22 base pair have the largest αKop values. The R1a for the U21 imino proton could not be determined because its resonance overlaps with the G18 resonance (see Figure 1B)

Figure 4.

-catalyzed exchange experiments on the P1 duplex. The R1a of the imino protons of the P1 duplex as a function of the concentration in NMR buffer at 35°C for G18 (filled circle), U19 (open square), U20 (open triangle), U-1 (open circle), G22 (filled square), U-3 (open hexagonal) and G26 (filled triangle). The concentration of was calculated from the pKa for (7.21), the pH of the sample and the total phosphate concentration. The pH and [Pi]total for each condition are shown on top of the figure. The lines are fits of these data to Equation (8) where the R1a data points were weighted by the inverse of their variance. The fits for U-1 and G22 imino protons are indicated by dashed lines and the other imino protons are indicated by solid lines. The error bars represent curve fitting errors during the determination of R1a from inversion recovery data.

Base-pair stability and kinetics for the wt and modified P1 duplexes using TRIS as a base catalyst

TRIS base proved the most useful catalyst for these hydrogen exchange studies of the P1 duplexes because it was possible to measure the αKop values for all non-terminal base pairs in the four P1 duplexes from analysis of the R1a data using Equation (8) (Figure 5 and Table 1). These data were used to compare the effects of the modifications on base-pair stabilities. In the (-1C)-P1 duplex, where the central U-1·G22 base pair is changed to a C-1·G22 Watson–Crick base pair, exchange of the G22 imino proton shows much smaller dependence on TRIS concentration compared to the wt P1 duplex (Figure 5). This leads to a 200-fold smaller αKop for the C-1·G22 base pair than the U-1·G22 base pair in the wt P1 duplex (Table 1), demonstrating that the G·U base pair is much less stable than a non-terminal Watson–Crick G·C base pairs. This difference in the base-pair stability between G·C and G·U base pairs also affects the stabilities of neighboring base pairs. The A2·U21 and A3·U20 base pairs in the (-1C)-P1 duplex have ∼9- and 2.5-fold smaller αKop values than those of the wt P1 duplex, respectively. However, there are no significant differences in the αKop values between the two duplexes for the C-2·G23 and U-3·A24 base pairs on the other side of the G·U base pair. These results indicate that the change from G·U to G·C base pair in the RNA duplex leads to stabilization of this base pair as well as stabilization of base pairs on only one side of the modification.

Figure 5.

TRIS-catalyzed hydrogen exchange data on the wt and modified P1 duplexes. The R1a of the imino protons of the wt P1 (circle), (-1C)-P1 (square), (-3m)-P1 (triangle) and (-5U)-P1 (diamond) duplexes are shown as a function of the concentration of TRIS base at 35°C at pH 7.67. The concentration of the TRIS base was calculated using a pKa = 7.87 at 35°C and knowing the pH and total concentration of TRIS. The lines are the best fits of these data to Equation (8) where the R1a data points were weighted by the inverse of their variance. The imino protons are labeled in the upper right corner of each figure. The error bars represent curve fitting errors during the determination of R1a from inversion recovery data.

Table 1.

Apparent base-pair dissociation constants (αKop) of the wt and mutant P1 duplexes determined by the TRIS-catalyzed NMR exchange experiments at 35°C

Base pair	Imino		P1 duplexes
			wt	−1 C	−3 m	−5 U
Y-5·G26b	G26	αKop (× 10−6)	31 ± 1	40 ± 1c	30 ± 2	n.d.d
C-4·G25	G25	αKop (× 10−6)	<3.5 (0.36)	<3.5	<3.5	58 ± 2
U-3·A24	U-3	αKop (× 10−6)	17 ± 1	21 ± 2	5.0 ± 0.4	35 ± 2
C-2·G23	G23	αKop (× 10−6)	<1.5 (0.30)	<1.5	<1.5	<1.5
Y-1·G22b	U-1	αKop (× 10−6)	300 ± 12	n.a.e	630 ± 32	280 ± 6
	G22	αKop (× 10−6)	275 ± 16	1.2 ± 0.3	530 ± 34	315 ± 9
A2·U21	U21	αKop (× 10−6)	740 ± 210f	83 ± 3	n.d.f	n.d.f
A3·U20	U20	αKop (× 10−6)	160 ± 12	63 ± 3	190 ± 8	160 ± 7
A4·U19	U19	αKop (× 10−6)	220 ± 20	170 ± 17	220 ± 15	180 ± 12
C5·G18	G18	αKop (× 10−6)	270 ± 13c	380 ± 18	340 ± 11c	290 ± 7c

aParameters used in the calculation: kcoll = 1.5 × 109 s−1, pKa(G-NH1) = 9.24, pKa(U-NH3) = 9.20, pKa(TRIS, 35°C) = 7.87; Sample conditions: 10 mM MgCl2, 100 mM NaCl, 0.1 mM EDTA, [TRIS]total = 1–20 mM, 35°C. The pH of all samples and buffer were adjusted to 8.00 at 23.4°C and then calibrated to 7.67 at 35°C using the equation, pH(35°C) = pH(23.4°C)−0.033 × (35−23.4). The errors for these αKop values were determined from the linear curve fitting using Equation (8). The αKop values determined using ammonia as the catalyst are shown in parenthesis.

bY indicates C or U.

cThese resonances are partially overlapped with another resonance and this overlap may lead to a systematic error in αKop. Thus these values may actually have larger errors than what are given, which only represent the errors derived from the fits.

dNot determined. The U-5·G26 base pair caused severe line broadening of G26 imino resonance.

eThere is no resonance in the (-1C)-P1 duplex.

fThe U21 imino resonances in the P1 duplexes, except the (-1C)-P1 duplex, showed line broadening and overlap with the G18 imino resonances. Thus, these values have larger errors or could not be determined.

Hydrogen exchange experiments were also performed for a duplex with a 2′-OMe modification at residue U-3 [termed here as (-3m)-P1]. The exchange of the U-3 imino protons showed a weaker dependence on TRIS compared to the wt P1 duplex, (Figure 5) leading to a 3-fold smaller calculated αKop for the U-3·A24 base pair in the modified duplex (Table 1). There is no significant effect of this modification on the stabilities of the neighboring C-2·G23 and C-4·G25 base pairs (Figure 5 and Table 1). The 2′-OMe modification at the U-3 position caused only a small effect on the exchange of U-1 and G22 imino protons with TRIS buffer (Figure 5), where the αKop of the U-1·G22 base pair is ∼2-fold larger for the (-3m)-P1 than the wt P1 duplex (Table 1).

The (-5U)-P1 duplex has a wobble pair at U-5·G26 and both the U-5 and G26 imino protons exchanged rapidly with water at 35°C so their resonances were not observed. This effect of the U-5·G26 base pair in the (-5U)-P1 duplex propagated to the C-4·G25 and U-3·A24 base pairs (Figure 5) but had little or no effect on the C-2·G23 and U-1·G22 base pairs (Figure 5 and Table 1).

DISCUSSION

NMR hydrogen exchange experiments have been used to probe the thermodynamics and kinetics for base-pair opening in a variety of nucleic acid systems (30). Most hydrogen exchange studies on nucleic acids are performed at or below room temperature in order to slow the observed exchange rates. However, the P1 duplex studied here showed evidence of aggregation at lower temperatures, so the hydrogen exchange kinetics was measured at 35°C. Ammonia is the most commonly used base catalyst in hydrogen exchange studies of nucleic acids (23,30); however, the high catalytic efficiency of the ammonia (pKa = 9.30) led to rapid exchange for most of the imino protons at 35°C. Thus, the base-pair opening lifetimes and αKop values could only be determined for the two most stable base pairs C-2·G23 and C-4·G25 using ammonia as the catalyst. Two other weaker base catalysts, TRIS base (pKa35°C = 7.88) and (pKa = 7.21) were also employed here to determine αKop values for other base pairs of the P1 duplex (33). For exchange data with both TRIS base and , the rate constant for base-pair closing, kcl, is much larger than αki[B], so the measured R1a is a simple linear function of the base catalyst [Equation (8) and Figures 4 and 5]. The imino protons of G23 and G25 showed only a small change in R1a as a function of (data not shown) which is consistent with small αKop values.

Hydrogen exchange experiments can also provide data on opening and closing rates of base pairs (30). The base-pair life times (τ0 = 1/kop) at 35°C using ammonia as the catalyst could only be determined for the C-2·G23 and C-4·G25 base pairs in the P1 helix and were 31 and 9 ms, respectively. The open lifetimes for all other base pairs were too fast to measure (<1 ms) under the conditions employed here.

The exchange rates of the imino protons in nucleic acids are affected by both intrinsic and external base catalysts (30). In Figures 4 and 5, the slope of the linear correlation between R1a of the imino proton and the concentration of the added catalyst is αkiKop and the y-intercept is under experimental condition, where kcl is much larger than [see Equations (4) and (8)]. Thus, the imino proton of a less stable base pair with larger Kop will have both a larger slope and larger y-intercept in these plots. This pattern is followed for most of the imino protons, but not the two imino protons of the U-1·G22 wobble pair (see Figure 4 and Table 1). For example, although αKop of the U-1·G22 wobble pair is larger than those of the A3·U20 and A4·U19 base pairs (Table 1), the imino protons of U-1 and G22 have the larger slopes but much smaller values of y-intercepts than these two A·U base pairs (Figure 4). This phenomenon is consistent with a different rate of internal catalysis for a U·G base pair compared to a standard Watson–Crick base pair. In a U·G wobble base pair, the hydrogen bond acceptors of U and G imino protons are carbonyl oxygens which are less basic and therefore poorer base catalysts than the A-N1 or the C-N3 in standard base pairs (25,30,38).

The P1 duplex of Tetrahymena group I intron can be divided into three parts, the central U·G base pair, the AU-rich (region I; see Figure 1A), and the GC-rich (region II). The P1 duplex in the Tetrahymena L-21 ribozyme contains only the central U·G base pair and base pairs in region II, thus base pairing in region I is not required for activity (4). The αKop data in Table 1 show that the central U-1·G22 base pair has a ∼230-fold higher population of open states relative to having a C·G base pair at this position. Previous studies for opening of G·U and G·T base pairs in RNA and DNA duplexes showed that these wobble pairs have 10–100 times longer lifetimes in the open state than other Watson–Crick base pairs (25). These results are consistent with the thermodynamic data here and suggest that a large part of the destabilizing effect of the U-1·G22 is due to a longer lifetime for the open state.

The A2·U21 base pair is the least stable non-terminal Watson–Crick base pair in the P1 duplex. This base pair is not required for activity and even when this base pair forms, it has a very large αKop. These results suggest that having a unstable base pair next to the central U-1·G22 pair (on the region I side) is important for the high catalytic activity. Previous studies have indicated that flexibility of the U-1·G22 base pair is important for the activity of the Tetrahymena ribozyme. For example, it was initially proposed that the U-1·G22 base pair interacts with the conserved A-rich J4/5-J5/4 symmetric loop, by a hydrogen bond between the NH2 of G22 and the 2′-OH group of A207 at J5/4 (39). This tertiary interaction was subsequently observed in the crystal structures of the Azoarcus sp. and Twort group I introns, where the dT·G or U·G base pairs in the P1 duplex interacts with the 2′-OH of the A87 or A83, respectively (which are the same positions as A207 in the Tetrahymena ribozyme) (16,17). Formation of the tertiary interaction between G22 and A207 likely leads to destabilization of the U-1·G22 wobble. This is supported by the X-ray structure of Twort group I intron, which has a long hydrogen bonding distance (3.52 Å) between U-1-O2 and G8-N4 of the U-1·G8 wobble pair (equivalent to U-1·G22 wobble pair of Tetrahymena ribozyme). Changing the U-1·G22 wobble pair to a C-1·G22 base pair in the P1 duplex stabilizes this base pair to opening by over a factor of 200, and also stabilizes the A2·U21 base pair in region I by a factor of 9 (Table 1), but has little affect on stability of base pairs in region II. The results here show that the U-1·G22 and A·U base pairs in region I have high αKop in the isolated P1 helix, suggesting that the dynamics of the isolated P1 duplex are tuned to help stabilize the active conformation in the docked state.

The results here also show that the destabilizing effect of the central U-1·G22 base pair is propagated asymmetrically to only one side of the duplex. The A2·U21 base pair on one side of the central U-1·G22 is the least stable (non-terminal) base pair but the C-2·G23 on the other side is the most stable base pair in the P1 duplex and is not affected by large αKop for U-1·G22 (Table 1). Structural studies on RNA duplexes show asymmetric changes in the helical twist around U·G wobble base pairs (40–42). These RNA duplexes are unwound between the U·G wobble pair and the base pair 3′ to the U, but over-wound on the other side (40–42). Such helical unwinding on one side of the U·G wobble pair may lead to the large αKop for A2·U21. The results here are also consistent with previous data on a DNA duplex where modification of a A·T base pair to a G·T wobble destabilized the G·C base pair on the 3′-side of the T more than the A·T base pair on the other side (25). However in this duplex there was still a relatively large effect on both sides of the wobble, with ∼80-fold and 16-fold change in αKop for base pairs on the 5′-side and 3′-side of the T, respectively. As seen in Table 1, there is a >30% change in αKop for any of the base pairs on the 3′-side of U-1, but a ∼9-fold change for the A2·U21 base pair on the 5′ side when comparing the wt to the (-1C)-P1 duplex. These difference could arise from comparing a DNA duplex to a RNA duplex, from difference sequences for the base pairs neighboring the G-U(T) wobble, or the different temperatures at which αKop was measured. The (-5U)-P1 duplex studied here changes a G·C base pair in region II to a G·U base pair and the hydrogen exchange results showed enhanced opening for the G·U as well as the neighboring and next nearest neighboring base pairs on the 3′ side of the U (Table 1). The stability of the base pair on the other side of the G·U pair could not be studied because the imino proton of this terminal base-pair exchanged with water too rapidly to be observed. The U-1·G22 wobble pair appears to be important for ribozyme activity where the enhanced opening and lower stability of this base pair aids the formation of tertiary interaction with A207, and the asymmetry around the U-1·G22 base pair can help stabilize base pairs in region II. The data are also consistent with a model that a reduced undocking rate of the P1 duplex, arising from formation of the tertiary interaction between G22 and A207 (7,39), is thermodynamically linked to the instability of the U-1·G22 base pair at the cleavage site.

The hydrogen exchange studies here showed that 2′-OMe modification at U-3 leads to stabilization of the U-3·A24 base pair with ∼4-fold smaller αKop. This modification also causes a small 2-fold destabilization of the U-1·G22 two base pairs away, but had no measurable effect on the directly neighboring C-2·G23 and C-4·G25 base pairs. Previous studies of unmodified and fully modified, r(CGCGCG)2, showed that 2′-O-methylation stabilized the duplex to thermal denaturation but had little effect on the NMR solution structure (43). The X-ray crystal and NMR solution structural studies of this duplex showed that the 2′-OMe group points towards the minor group (43–45) and is very close to the H5′ of the 3′-neighboring residue (43). The X-ray structure of the fully modified, r(CGCGCG)2 duplex shows that the 2′-OMe group affects the minor groove hydration pattern (44,45), thus stabilization of the U-3·A24 base pair may arise from protection of the 2′-OMe group from water. Previous studies showed that a 2′-OMe modification at U-3 increased the undocking rate of the P1 duplex by ∼140-fold (7). A 2′-deoxy modification at the same position also led to a ∼30-fold increase in the undocking rate (7), even though the 2′-deoxy substitution at this position leads to lower thermal stability for the P1 duplex (ΔΔG° ≈0.87 kcal/mol) (12). Both helical stabilization by a 2′-OMe modification and helical destabilization by a 2′-deoxy modification at U-3 increased the kundock of the P1 duplex, indicating there is no correlation between the thermal stability of U-3·A24 base pair and docking of the P1 duplex. In the crystal structures of the Twort group I ribozyme-product complex (PDB ID: 1Y0Q) and Azoarcus sp. group I intron in complex with both exons (PDB ID: 1U6B), the 2′-OH group of residue C-3 in P1 helix (equivalent to U-3 of Tetrahymena ribozyme) is within hydrogen bonding distance of the N1 of G182 (2.97 Å) or A168 (2.67 Å), respectively. Although this position has no sequence homology in the group I intron (U300 in the Tetrahymena ribozyme, G182 in the Twort group I ribozyme and A168 in the Azoarcus sp. group I intron), the long-range hydrogen bond of the –3 position may still be important for determining the undocking rates. This hypothesis may explain why a 2′-H or 2′-OMe modification at U-3 strongly interferes with the tertiary interaction between the P1 duplex and catalytic core (7).

CONCLUSION

The thermodynamics and kinetics for base-pair opening in the P1 duplex of the Tetrahymena group I ribozyme were studied by NMR imino proton exchange experiments. C-2·G23 and C-4·G25 were the most stable base pairs in the P1 helix where the kinetics and thermodynamics for opening were determined at 35°C from analysis of exchange experiments using ammonia as a catalyst. The equilibrium constants for base-pair opening for the other more rapidly exchanging imino protons were determined using the weaker base catalyst, TRIS. These studies also revealed a high degree of asymmetry in the thermodynamics for opening of base pairs around the central U-1·G22 wobble pair, where base pairs on the 3′-side of the U were substantially destabilized whereas there was little effect on base pairs on the other side of the wobble. This asymmetry may result from the known asymmetry of the helical twist around G·U wobble pairs (40–42), which may lead to the a general phenomenon that replacement of a C·G with a U·G also leads to asymmetric destabilization of base pairs towards the 3′-side of U. The hydrogen exchange NMR studies here revealed an unusual pattern of base-pair stability for residues A2·U21, U-1·G22 and C-2·G23 in the P1 duplex and these results are consistent with the model that flexibility in the P1 duplex is important for forming stabilizing tertiary interactions with the catalytic core of the Tetrahymena ribozyme.

SUPPLEMENTARY DATA

Supplementary Data are available at NAR Online.