¹H NMR determination of base-pair lifetimes in oligonucleotides containing single base mismatches

Abstract

Proton nuclear magnetic resonance (NMR) spectroscopy is employed to characterize the kinetics of base-pair opening in a series of 9mer duplexes containing different single base mismatches. The imino protons from the different mismatched, as well as fully matched, duplexes are assigned from the imino-imino region in the WATERGATE NOESY spectra. The exchange kinetics of the imino protons are measured from selective longitudinal relaxation times. In the limit of infinite exchange catalyst concentration, the exchange times of the mismatch imino protons extrapolate to much shorter lifetimes than are commonly observed for an isolated GC base pair. Different mismatches exhibit different orders of base-pair lifetimes, e.g. a TT mismatch has a shorter base-pair lifetime than a GG mismatch. The effect of the mismatch was observed up to a distance of two neighboring base pairs. This indicates that disruption in the duplex caused by the mismatch is quite localized. The overall order of base-pair lifetimes in the selected sequence context of the base pair is GC > GG > AA > CC > AT > TT. Interestingly, the fully matched AT base pair has a shorter base-pair lifetime relative to many of the mismatches. Thus, in any given base pair, the exchange lifetime can exhibit a strong dependence on sequence context. These findings may be relevant to the way mismatch recognition is accomplished by proteins and small molecules.

INTRODUCTION

DNA mismatches, or non-complementary base pairs, arise in vivo as a result of the misincorporation of bases during replication (1), heteroduplex formation during homologous recombination (2), mutagenic chemicals (3,4), ionizing radiation (5) and spontaneous deamination (6). These errors are usually detected and eliminated by DNA polymerase and postreplicative mismatch repair system (7–9). How these DNA mismatches are detected by the repair machinery of the cell requires an understanding at the molecular level. Therefore, it is useful to characterize the structure, dynamics and biochemistry of various mismatched base pairs in DNA and to determine how they affect the structure of the double helix in terms of both global and local perspectives.

The structures of several DNA duplexes containing mismatched base pairs have been characterized by X-ray crystallography (10–12) and NMR methods (13–16). In all of these structural studies, the mismatches were shown to have minimal effect on the global conformation of the DNA; the distortions produced are limited to the mismatched site and neighboring base pairs. ¹H NMR studies show that the mismatches GG (17–19), AA (20–22), TT (20–22), CC (23,24), GA (25–28) and GT (29) are well stacked in the helix and the bases remain in an intrahelix orientation. In fully base-paired right-handed B-form DNA duplexes, there are NOEs evident between base protons (H8 or H6) and the 5′-flanking sugar H1′ and H2′2″ protons, allowing an NOE walk from the 5′ to the 3′ end of the oligonucleotide. In the case of these mismatched duplexes, the NOE walk is conserved, again supporting the notion that the mismatches are inserted and stacked well between the flanking base pairs, and the oligonucleotides adopt the classical B form duplex with minimal local disruption. Additionally, ³¹P NMR studies support a B₁ conformation (17,19,20,23).

Much less is known about the dynamic properties of the mismatched bases in the DNA duplex and about any possible role of dynamics in mismatch recognition. The bases in DNA move rapidly within the double helix, undergoing thermally driven structural fluctuations in solution. Since base motions occur within a multidimensional potential well, determined by a combination of base-stacking and base-pairing forces, it is reasonable to expect that the motions of a mismatched base pair in DNA should be different from that of the fully matched pair; this dynamic difference may influence the interactions of mismatched base pairs with repair enzymes. Furthermore, the dynamics of mismatches, as well as fully matched base pairs, may play a pivotal role in modulating charge transport through DNA, which is a topic of considerable current interest (30–38).

The dynamics of mismatched duplexes have been the focus of spectroscopic studies by Millar and co-workers (39). Time-resolved fluorescence anisotropic decay measurements were obtained for a series of oligonucleotides containing intervening AP·X base pairs, where AP is the fluorescent adenine analog 2-aminopurine, and X = A, T, G or C. This technique allowed the detection of base motions in DNA on the picosecond timescale. Motions such as helical twisting, propeller twisting, base tilting and base rolling could potentially alter the emission dipole of AP, thereby contributing to changes in the decay of the fluorescence anisotropy. AP pairs differently with each of the different bases and these differences in its relative pairing ability were reflected in the internal dynamics.

A complementary method of probing base-pair dynamics is through ¹H NMR studies of imino proton exchange rates. Clearly, such studies probe base-pair motions on a much slower timescale. The imino protons of the aromatic heterocyclic base exchange with the solvent protons when the hydrogen bond in the base pair is disrupted (40,41). The chemical proton transfer step from the open state is usually rate limiting, and a proton acceptor, the base catalyst, must be added to the solution in order to accelerate the exchange close to opening-limited conditions. Base-pair lifetimes are then obtained by extrapolation of the exchange times to infinite catalyst concentration, where the dissociation constant of a base pair is estimated by comparing the exchange rates of the imino proton in the base pair and in the mononucleoside. The unknown factor in this comparison is the accessibility of the imino proton in the open base pair, which is related to properties of the open state. Measurements of the NMR relaxation rates of the imino protons as a function of solvent exchange have yielded lifetimes in the range of 1–40 ms for fully paired bases in B-DNA (42). In general, AT base-pair lifetimes have been found to be in the range of 1–5 ms at 15°C, except for AT tracts where lifetimes >100 ms have been observed (43). For GC base pairs, lifetimes ∼10× longer than for AT base pairs have been observed (42), as one might expect, given the presence of an additional hydrogen bond in the GC base pair. However, it is important to note that the sequence composition (43,44), the charged state of the double helix (45) and drug interactions (46) all serve to modulate base-pair dynamics sensitively.

Here, we report a systematic study of the dynamics of single base mismatches within DNA duplexes, through the measurements of imino proton exchange. Using ¹H NMR, we have determined the base-pair lifetimes of different mismatches in a given sequence context and have compared the exchange times with those of fully matched base pairs. We observe strong sequence dependence of the base-pair opening times and that rates are increased at the mismatched site and in the directly neighboring site. These results underscore the importance of DNA sequence and of sequence context in governing base dynamics.

MATERIALS AND METHODS

Oligonucleotide preparation

Oligonucleotides were synthesized using standard phosphoramidite chemistry on an Applied Biosystems 392 DNA synthesizer with a dimethoxy trityl protecting group on the 5′ end (47). Oligonucleotides were purified on a reversed-phase Rainin Dynamax C₁₈ column on a Waters HPLC and then deprotected by incubation in 80% acetic acid for 15 min. After deprotection, the oligonucleotides were purified again by HPLC. Following purification, these oligonucleotides were desalted on a Waters C₁₈ SepPak column and then converted to a sodium salt using CM Sephadex C-25 (Sigma) equilibrated in NaCl. The concentration of the oligonucleotides was determined by UV-visible spectroscopy (Beckman DU 7400) using the extinction coefficients estimated for single-stranded DNA: ε(260 nm, M^–1cm^–1) adenine (A) = 15 400; guanine (G) = 11 500; cytosine (C) = 7400 and thymine (T) = 8700. Single strands were mixed with equimolar amounts of complementary strand and annealed using a Perkin Elmer Cetus Thermal Cycler by gradual cooling from 90°C to ambient temperature in 90 min. The NMR samples were prepared by dissolving the oligonucleotides in a buffer solution (5 mM Na₂HPO₄, 15 mM NaCl, pH 7.0). The concentrations of the samples varied between 0.5 and 1.2 mM duplex.

Melting temperature experiments

The melting temperatures of the oligonucleotides were determined from absorbance versus temperature curves measured at 260 nm on a Beckman DU 7400 UV-visible spectrophotometer. Ten micromolar duplex was used in a buffer of 5 mM Na₂HPO₄, 15 mM NaCl, pH 7.0. The melting profile of the duplexes was obtained by slowly lowering the temperature (0.5°C/min) from 75°C to 10°C and measuring the absorbance at 260 nm at each temperature. The T_m values represent the midpoint of the transition as obtained by fitting the melting profiles with a sigmoidal expression in Origin.

NMR spectroscopy

One- and two-dimensional ¹H NMR spectra were taken in both D₂O and 90/10 H₂O/D₂O at 277 K on a Varian INOVA 600 MHz spectrometer. For the spectra taken in 90/10 H₂O/D₂O, WATERGATE gradient pulse water suppression (48) was used. The conditions of acquiring the spectra were: 600 MHz, 20 p.p.m. sweep width, TPPI, 2048 complex points, hypercomplex mode, 256 t₁ blocks, 64 scans per t₁ block, 1.3 s relaxation delay, 120 ms mixing time. ROESY spectra (49,50) were obtained at variable mixing times of 80, 150 and 300 ms keeping the other parameters the same as for the WATERGATE NOESY experiments with the r.f. strength for the spin-lock field set at 3.5 kHz. Linear prediction was employed in the indirect dimension to increase resolution close to the diagonal. All two-dimensional processing was carried out with VNMR 6.1b software (Varian) on a SUN workstation. Chemical shifts were reported relative to the internal standard, sodium 3-trimethylsilyl-[2,2,3,3-D₄] propionate (TMSP).

Imino proton exchange measurements

Imino proton exchange within the DNA duplex occurs through the opening of the base pair followed by exchange from the open state. The conformational features of this state are not yet fully understood. To account for proton exchange data, it is generally assumed that, in the open state, the imino groups are fully accessible to solvent and the base-pairing hydrogen bonds are broken such that the imino proton becomes available for hydrogen bonding with water or catalyst molecules (40). The chemical proton transfer step from the open state is rate limiting, and a proton acceptor like OH^– must be added to accelerate the exchange close to opening-limited conditions. The base-pair lifetimes are obtained by extrapolation of the exchange times to infinite base catalyst concentration.

k_ex = Σ_n kⁿ_op kⁱ_tr[B]/kⁿ_cl/αⁿ + kⁱ_tr[B]

For a base pair with multiple open states, formed with rates kⁿ_op and closed with rates kⁿ_cl, and provided that Σ_n kⁿ_op« k¹_cl, k²_cl, …., kⁿ_cl, the total imino proton exchange rate k_ex equals the sum of the exchange rates from each mode (40). To simplify, where kⁱ_tr is, as above, the intrinsic imino proton transfer rate from the mononucleoside and αⁿ is a parameter reflecting the different accessibility of the imino proton in the open states and in the mononucleoside and [B] denotes the concentration of the exchange catalyst, OH^– in this case, then for a base pair with a single opening mode (n = 1) and with k_op« k_cl, this can be written as:

τ_ex = τ_op + 1/αK_dk_trⁱ[B]1

where τ_ex and τ_op are the inverse exchange and opening rates, respectively, and K_d = k_op/k_cl is the base-pair dissociation constant. A plot of τ_ex versus 1/[B] yields a straight line where τ_op is obtained from the y-axis intercept and αK_d from the slope. The limit where equation 1 is valid is known as Linderstrøm–Lang kinetics (51).

In this work, the oligonucleotides were titrated using ammonia (4 M ammonia, 1.25 M ammonium chloride, pH 10.0; Ricca Chemical Co.) as the exchange catalyst. The oligonucleotide concentrations of the samples varied between 0.5 and 1.2 mM duplex, and the oligonucleotide sample was buffered with 5 mM Na₂HPO₄, 15 mM NaCl, pH 7.0. The imino proton exchange times (τ_ex) at different catalyst concentrations were obtained from measurements of the inversion recovery times in the presence (T_rec) and absence (T_aac) of exchange catalyst according to the equation

1/τ_ex = 1/T_rec – 1/T_aac2

Except for longitudinal dipolar relaxation, direct exchange to water as well as exchange catalyzed by OH^– ions and the acceptor nitrogen of the opposite base contributes to the recovery rate of the imino protons in the absence of added catalyst 1/T_aac. However, these contributions remain constant when the catalyst is added and will be canceled in equation 2. Consequently, the exchange time τ_ex represents exchange only via the added catalyst. Thus, effectively, through measurements of the longitudinal relaxation times, T₁, for various imino protons as a function of catalyst concentration, we can determine the imino exchange rates within the oligonucleotide duplex.

The experiment was carried out by creating a designed pulse sequence comprised of a 3.2 ms 180° Gaussian pulse (g3) for selective inversion, a variable delay, and a 5 ms 90° Gaussian observe pulse (g4). Right shift and linear prediction of the free induction decay are employed to correct for magnetization evolution during the observed pulse. The carrier frequency was centered on the imino proton region. The spectral width was 7000 Hz and the experiments were carried out at 600 MHz.

RESULTS AND DISCUSSION

Sequence design

The general 9mer oligonucleotide sequence d(GACAXT GTC)₂; where X = G, T, A, C, was employed to determine the base-pair lifetimes for DNA duplexes containing different central single base mismatches (Table 1). A palindromic sequence was utilized that is self-complementary except at the mismatch site (X). This simplifies the ¹H NMR spectra and facilitates the assignment of the protons. Table 1 also shows the melting temperatures for the different duplexes and demonstrates that duplexes containing mismatches are thermodynamically destabilized as compared with the matched pairs.

Table 1. The sequences and the melting temperatures (T_m) of the duplexes containing a single base mismatch (XY) used in the NMR experiments.

Duplexesa	Tm (°C)b
5′-GACAGTGTC-3′	34.9
3′-CTGTGACAG-5′
5′-GACACTGTC-3′	24.5
3′-CTGTCACAG-5′
5′-GACAATGTC-3′	32.1
3′-CTGTAACAG-5′
5′-GACATTGTC-3′	32.4
3′-CTGTTACAG-5′
5′-GACAGTGTC-3′	47.8
3′-CTGTCACAG-5′
5′-GACAATGTC-3′	44.1
3′-CTGTTACAG-5′

^aThese sequences represent the six oligonucleotides employed for the imino exchange experiments. The single base mismatched or matched base pairs are highlighted in bold.

^bShown are the melting temperatures (T_m) of the duplexes determined as described in Materials and Methods. Samples include 10 µM duplex in a buffer of 5 mM Na₂HPO₄, 15 mM NaCl, pH 7.0. The order of the melting temperature varies as GC > AT > GG > TT ∼ AA > CC.

For the mismatched duplexes, only one strand was used, resulting in a duplex with either AA, TT, CC or GG mismatch at the central site (X); to obtain a complementary base pair at site X, however, two different strands had to be used. This led to the possibility of the formation of a mixture of matched and mismatched duplexes as two separate strands were added to form the fully matched GC and AT sequences. Given the melting temperatures, that possibility was minimized by slow cooling during hybridization of the duplexes and repeated annealing of the sample to allow nucleation. In the case of the AT-matched duplex, no resonances corresponding to the imino protons of mismatched base pairs were observed. For the GC matched duplex only a small fraction was observed (see below), confirming the isolation of duplexes with primarily Watson–Crick base pairs. The 9mer duplex was also chosen in part to preserve the sequence used in guanine oxidation studies reported elsewhere (38). The possibility of hairpin formation is negligible with 9mer duplexes. Moreover, non-denaturing agarose gel electrophoresis with the mismatch-containing duplexes indicated no formation of hairpins in these duplexes (data not shown).

Imino proton resonance assignment

The imino proton resonances of the different oligonucleotides were assigned primarily from the imino-imino crosspeaks in the WATERGATE NOESY spectra in 90:10 H₂O/D₂O solution (Table 2). The B-form structure of the DNA duplex constrains the position of imino protons in the duplex to be 3.4 Å above or below one another; in the fully matched duplex, each base pair contains one imino proton (N₃H in thymine and N₁H in guanine). These interactions between the imino protons lead to crosspeaks in the imino-imino region of the two-dimensional NOESY spectra of the DNA duplexes, which can be exploited for the assignment of these protons.

Table 2. The chemical shifts^a (p.p.m.) of the imino proton resonances in the oligonucleotides containing the intervening mismatches (XY).

Base pairsb	X/Y5NHc (p.p.m.)	T6NHd (p.p.m.)	G7NHd (p.p.m.)	T8NHd (p.p.m.)	G9NHd (p.p.m.)
G·G	10.65	13.76	12.70	13.92	12.71
	10.22	13.46	12.47	13.89
T·T	10.72	14.25	12.44	14.03	12.79
	10.41(sm)	13.87(sm)
C·C	–	14.20	12.54	14.04	12.80
		13.87(sm)
A·A	–	13.55	12.46	13.92	12.76
G·C*e	13.54	13.81	12.57	13.97	12.78
A·T*e	13.72	14.01	12.47	13.85	12.79
	13.42	13.96	12.41

123456789

5′-GACAXTGTC-3′     where X, Y = A, T, G, C

3′-CTGTYACAG-5′

^aSamples for NMR studies are prepared as described in Materials and Methods and the proton chemical shifts are relative to TMSP (0.00 p.p.m.). The samples were at a concentration of 0.5–1.2 mM duplex and taken in a buffer solution of 5 mM Na₂HPO₄, 15 mM NaCl, pH 7.0 in 90:10 H₂O/D₂O.

^bDesignation of XY as shown in assembly.

^cThe imino proton attached to the central mismatched base pair is denoted by X/Y5NH. The assignment of the imino protons was accomplished from the imino-imino cross peaks of the WATERGATE NOESY spectra of the oligonucleotides in 90:10 H₂O/ D₂O as described in Results and Discussion.

^dBecause of the symmetry of the duplexes and the fact that each base pair can have only one imino proton (except the mismatched pairs), imino protons at G1, T2, G3 and T4 sites are usually equivalent to that of G9, T8, G7 and T6, respectively. The presence of the two values of the chemical shift reflects the loss of this symmetry.

^eXY* corresponds to the Watson–Crick paired sequence.

Figure 1 shows the exchangeable imino region (9– 15 p.p.m.) of the one-dimensional ¹H NMR spectrum for sequences containing GG, AA, TT, CC mismatches and for the fully matched sequences containing AT and GC base pairs. For all of the oligonucleotides under study, one observes three main sets of imino protons: (i) thymine imino protons in the range of 14.3–13.4 p.p.m., (ii) guanine imino protons mostly in the range 12.8–12.4 p.p.m. and (iii) a set of imino protons from the mismatched sites (GG and TT) in the range 10.2– 10.7 p.p.m.. These shifted resonances at the mismatch site are more exposed and subject to exchange with solvent than those of the stacked base pairs. Nonetheless, most of the imino protons are quite well resolved. As is evident in Table 2, for all of the duplexes, the chemical shifts of the outer imino protons (G9N₁H, T8N₃H and G7N₁H) remain quite close to each other. This similarity in chemical shift is consistent with previous observations that perturbations caused by the mismatches are localized, limited essentially to the flanking bases (10–12).

Figure 1.

One-dimensional ¹H NMR spectra of the exchangeable imino region (9–15 p.p.m.) of the oligonucleotides with intervening GG, TT, AA, CC mismatches and the fully matched AT and GC base pairs at 600 MHz. Samples contained 0.5–1.2 mM oligonucleotides, 5 mM Na₂HPO₄ and 15 mM NaCl buffer at pH 7.0 in 90:10 H₂O/D₂O at 277 K. The peaks are assigned from the respective imino-imino region of the two-dimensional NOESY spectra. Chemical shifts are reported relative to TMSP (0.00 p.p.m.). The chemical shifts of the individual imino resonances in all the duplexes are given in Table 2.

The line-widths of the imino proton(s) for the different duplexes containing mismatches clearly differ depending upon the base composition of the mismatch. TT and CC mismatch-containing duplexes exhibit larger line-widths as compared with GG and AA duplexes. This variation in line-width correlates with the relative stability of the duplexes; such broad line-widths for duplexes containing TT and CC mismatches have also been seen previously (24).

A representative imino-imino region in the NOESY spectra of the duplex containing a GG mismatch is shown in Figure 2. Since the imino protons are stacked one above another, the interactions among them lead to a NOE walk of the imino-imino cross peaks along the length of the oligonucleotide. Furthermore, the two imino protons from the mismatched guanine (GG) base pair are positioned in close proximity to each other and exhibit a strong NOE build up between them (G5/G5′). The peak splitting for the G5 imino proton reflects an asymmetry in conformation for the two Gs at the mismatch site. One G of the GG mismatch has been seen to adopt the syn conformation with the alternate G in the anti form, with slow exchange on the NMR time scale (17–19). This asymmetry at the mismatch site extends out to other sites within the duplex; peak splitting is observed not only for G5, but also for T6 and G7 imino resonances. These peaks can easily be assigned from the NOESY spectra.

Figure 2.

Expanded imino-imino region of the two-dimensional NOESY contours in 90:10 H₂O/D₂O for the duplex d(GACAGTGTC)₂ containing an intervening GG mismatch. Solution is 1 mM in DNA duplex, 5 mM Na₂HPO₄ and 15 mM NaCl buffer at pH 7.0. Spectrum is acquired with a mixing time of 120 ms and at 277 K and collected at 600 MHz. The details of the NOESY experiment are outlined in Materials and Methods. Chemical shifts are reported relative to TMSP (0.00 p.p.m.). The lines illustrate the NOE walk along the imino protons, which are stacked above or below one another (at a distance of 3.4 Å) along the oligonucleotide.

The TT mismatch is known to be in intermediate exchange between two asymmetric wobble structures involving two imino to carbonyl hydrogen bonds (20–22), and we also observe two peaks for T5 with a faster rate of exchange. Note that one of the T5 peaks is much bigger than the other (T5′) indicating that one of the conformations exchanges with water at a faster rate relative to the other. For the TT mismatch-containing duplex, the asymmetry can be seen to a small extent on T6. Interestingly, for the CC mismatch-containing duplex, a similar asymmetry is evident in the shifts for T6; since the CC mismatch lacks imino protons, no asymmetry can be detected directly at the mismatch site. This has been further corroborated by the presence of a cross peak for the T6N₃H proton in the same phase as the diagonal peak in the two-dimensional ROESY spectra of the TT and CC duplexes (data not shown), which indicates chemical exchange involving that imino proton. A combination of NOESY and ROESY spectra can thus be used to distinguish between cross-relaxation and chemical exchange (49). Note that our assignments of the imino protons of the GG and TT pair are in close agreement with those observed for related 9mers containing GG and TT mispairs (24).

The one-dimensional spectrum for the AA mismatch-containing duplex is relatively simple and is also easily assigned from the imino-imino NOESY spectra. Only four resonances from the four matched base pairs are evident. This observation must reflect a primarily symmetric positioning of the As within the mismatch site.

Also shown are the one-dimensional spectra for the fully matched duplexes. For the GC duplex, the G5 imino is now revealed at the expected shift for a matched base pair. In the case of the GC duplex, there are also very small peaks evident at 13.46, 10.65 and 10.22 p.p.m.; we attribute these to the formation of a very small amount of GG mismatch-containing duplex during the hybridization process. Interestingly, sharper resonances are evident for the AT matched oligomer compared with the GC matched duplex. In addition, in the case of the AT matched duplex, peak doubling is evident for most of the resonances, consistent with the symmetry breaking by the TA base pair at the central site. Only one resonance at each position is apparent for the matched GC oligomer; perhaps this results from a more central positioning of the G5N₁H within the duplex.

Imino proton exchange

The exchange times of the imino protons were obtained through the measurement of the inversion recovery times of the NMR. The addition of an exchange catalyst, NH₃, yields, in the limit of infinite catalyst concentration, the kinetic parameters for the base-pair opening (equation 1). Upon titration with base, the general observation was broadening and the gradual disappearance of the imino proton resonances as the exchanges times became too fast to be observed on the NMR timescale. Figure 3 shows the variation in the one-dimensional spectrum for the GG mismatch-containing duplex as a function of increased base concentration. Different imino protons are seen to broaden out at differing levels of base concentration. For example, G7N₁H and T6N₃H broaden out at a much slower rate compared with the imino protons G9 and G5. This is consistent with the fact that the outermost protons (G9N₁H and T8N₃H), as well as the protons at the mismatch site (G5N₁H), are undergoing exchange with solvent at a faster rate than the other internally positioned imino protons and, hence, the fast relaxation is observed. The line-width of the G9N₁H imino proton is relatively broad compared with the other imino protons even before the addition of the catalyst due to fraying of the end base positions; that the imino resonance for the end position is seen at all is somewhat surprising.

Figure 3.

The titration of 1 mM d(GACAGTGTC)₂ with 4 M ammonia buffer in 90:10 H₂O/D₂O followed by ¹H NMR. Shown is the imino proton region of the spectra taken at 600 MHz and 277 K. Conditions are 1 mM DNA duplex, 5 mM Na₂HPO₄ and 15 mM NaCl buffer and 4 M in the titrant ammonia buffer. Chemical shifts are reported relative to TMSP (0.00 p.p.m.). [B] denotes the concentration of the ammonia buffer in the sample solution at different points of the titration.

In Figure 4, the exchange times at 277 K of the imino protons of the base pairs of the oligonucleotide containing the GG mismatch are displayed as a function of the inverse ammonia concentration. The exchange times display the linear dependence on the inverse base concentration, as expected from equation 1. Similar plots were drawn for the other mismatches (Supplementary Material) and these plots were found to be linear in all cases. Therefore, imino proton exchange in these duplexes containing single base mismatches does follow Linderstrøm–Lang kinetics (51). The base-pair lifetimes, τ_op, obtained from the linear fits are given in Table 3.

Figure 4.

Exchange of the imino protons of the duplex d(GACAGTGTC)₂ containing the intervening GG mismatch versus the inverse of the concentration of ammonia catalyst. The exchange times (τ_ex) of (A) G7N₁H, (B) G9N₁H, (C) G5/G5′N₁H, (D) T6N₃H and (E) T8N₃H imino protons are individually displayed as a function of inverse ammonia concentration (1/[B]). The data points used in the drawing of these lines are normalized based on at least three trials. The straight lines are obtained by fitting to equation 2, with the exchange times weighted according to their errors. The corresponding base-pair lifetimes (τ_op) thereby obtained from extrapolation are displayed in Table 3. Two independent line fittings are accomplished for the imino proton at the mismatch site (G5/G5′N₁H).

Table 3. The base-pair lifetimes^a (τ_op) of the oligonucleotides containing intervening mismatch (XY).

τop (ms)b Base pairc	T5NH	G5NH	T6NH	G7NH	T8NH	G9NH
G·Gd	X	2 ± 1	24 ± 5	15 ± 4	8 ± 2	7 ± 3
		2 ± 0.9
T·Te	0.5 ± 0.3	X	6 ± 4	2 ± 1	10 ± 3	14 ± 4
			0.5 ± 0.2
C·Cf	X	X	10 ± 3g	4 ± 2	8 ± 3	12 ± 4
A·Af	X	X	15 ± 3	12 ± 3	10 ± 2	10 ± 4
G·C*h	X	18 ± 4	12 ± 3	18 ± 5	11 ± 2	15 ± 5
A·T*h	1 ± 0.4	X	8 ± 2	3 ± 1	12 ± 3	12 ± 4

123456789

5′-GACAXTGTC-3′     where X, Y = A, T, G, C

3′-CTGTYACAG-5′

^aSamples for the determining the base-pair lifetime were prepared and the experiments based on imino proton exchange were performed as described in Materials and Methods. The samples were at a concentration of 0.5–1.2 mM duplex and taken in a buffer solution of 5 mM Na₂HPO₄, 15 mM NaCl, pH 7.0 in 90:10 H₂O/D₂O.

^bShown are the mean and standard deviation of the base-pair lifetime values (τ_op) of the mismatches/base pairs (T5/G5NH) and their flanking base pairs based on at least three trials.

^cDesignation of XY as shown in assembly.

^dTwo independent values of the base-pair lifetimes at the mismatch site are obtained for the duplexes containing GG mismatch for each trial.

^eIn the case of TT mismatch, two different lifetimes are obtained for the T6NH proton (Results and Discussion).

^fSince only guanine and thymine have imino protons, no base-pair lifetimes are obtained for CC and AA duplexes directly at the corresponding mismatch sites.

^gThe base-pair lifetime corresponding to the T6′NH of the CC mismatch could not be determined because of very fast relaxation on the addition of base.

^hXY* corresponds to the Watson–Crick paired sequence.

Base-pair lifetimes

Fast exchange rates and therefore short base-pair lifetimes are observed for the imino protons directly at the mismatch sites as compared with the Watson–Crick base pair GC (τ_op of G5N₁H = 18 ± 4 ms). This observation supports the fact that the mismatches produce local destabilization in the helical structure of the duplex leading to dynamics for the mismatches that are different from those of the standard Watson–Crick base pairs. Furthermore, the τ_op of the GG mismatch site is appreciably longer than that of the TT mismatch, 2 ms for G5N₁H in the GG pair, compared with 0.5 ms for T5N₃H in the case of TT. The value of τ_op obtained for the GG mismatch is similar to values previously reported for a GT mismatch (52).

The effect of the mismatches also propagates out to the base pairs adjacent to it, T6N₃H and G7N₁H, such that each duplex can be distinguished by its differential kinetics. As seen in the chemical shift changes, the dynamic effects of the mismatches are mostly localized in the vicinity of the mismatch site. This finding is consistent with the crystallographic studies, which show that the conformational differences between the normal B-form and the mismatched DNA are small and affect solely the local environment of the mismatched site (10–12). This observation is also consistent with previous ¹H NMR studies of DNA mismatches (16,28,52,53).

In fact, in the case of the duplexes containing CC and AA mismatches, the base-pair lifetime at the mismatch site clearly cannot be determined since there are no imino protons at the mismatch site to measure the relaxation times. In those cases, to compare the lifetimes with the other mismatches, the neighboring imino protons (T6N₃H) to the mismatch site were considered.

The exchange times of T5′N₃H of the TT mismatched duplex and T6′N₃H of the CC duplex could not be determined due to very fast exchange and the rapid disappearance of the peaks on the first addition of the exchange catalyst. This observation is consistent with the very short lifetime of the TT mismatched base pair (τ_op = 0.5 ± 0.3 ms) obtained independently from the other imino proton (T5N₃H) at the mismatch site. Unusually fast exchange is also observed at T6′N₃H (τ_op = 0.5 ± 0.2 ms) of TT duplex, providing evidence of fast exchange occurring around the vicinity of the mismatch.

In terms of overall kinetics, the duplex containing the GG mismatch displays the longest base-pair lifetime, followed by AA, CC and then TT with the shortest lifetime. We believe that the short base-pair lifetimes reflect an increased lateral motion of the base pair, leading to an increased disruption of the π stacked array. This disruption affects not only the imino protons at the mismatch site, but also the imino protons up to two neighboring base pairs away from the mismatch. In all of the sequences under study, the outermost GC and AT base pairs exhibit similar base-pair lifetimes.

Matched sequence

The base-pair lifetimes in the fully matched GC-containing oligonucleotide are comparable to the values reported previously for fully matched duplexes (42). However, quite contrary to our expectation, the base-pair lifetime for the AT base pair in the sequence context of 5′-AAT-3′ was found to be much shorter than AT base pairs in other sequence contexts (42,43) and even shorter than many mismatches (τ_op of T5N₃H = 1 ± 0.4 ms). Interestingly, the flanking base pairs of the central AT base pair also show unusually rapid lifetimes as compared with the fully matched GC sequence. Note the very fast lifetime of the G7 base pair (τ_op of G7N₁H = 3 ± 1 ms) in this sequence. Also note that, in this given AT duplex, different AT base pairs show different lifetimes. The one flanked by GC base pairs on both sides exhibits the longest lifetime (τ_op of T8N₃H = 12 ± 3 ms) which can be rationalized based on the fact that the GC base pairs can provide better stacking for the AT base pair. It is apparent that the value of the base-pair lifetimes depends upon the sequence context, as the sequence context imposes different local conformational flexibilities on the duplex DNA. These structural variations are reflected in the base-pair opening dynamics.

This observation is particularly interesting because it shows that the dynamics of a fully matched base pair in a given sequence context can be comparable to a mismatched site even though this may not be reflected in the melting temperatures. Thus, an individual Watson–Crick base pair within a DNA helix may not necessarily be the most stable and well stacked. The AT base pair in general has a longer base-pair lifetime (>1 ms but <5–6 ms) under other sequence contexts (44). In contrast, long tracts of AT base pairs show very long base-pair lifetimes (>100 ms) (43). The alternating sequence 5′-TATA-3′ is also known to be particularly flexible (54). Thus, the base-pair lifetime for a given base pair depends sensitively upon the sequence context and composition. A systematic study of this effect is in progress. It is interesting to note that we have also seen evidence of the local flexibility of this sequence in studies of long-range DNA-mediated charge transport (38).

Internal dynamics of mismatched and matched base pairs

One expects that the imino proton of a Watson–Crick base pair (e.g. a GC pair) within a double helix cannot exchange on a rapid timescale, since neither solvent nor catalyst has access to the proton. Exchange, therefore, proceeds via three consecutive steps: (i) base-pair opening, whose rate is 1/τ_op, followed by (ii) proton exchange from the open state and (iii) closing the base pair (55):

GH*C ↔ GH* + C ↔ GH^# + C ↔ GH^#C

(i)(ii)(iii)

The base-pair lifetimes of the mismatches are found in general to be shorter than the Watson–Crick base pairs. Indeed in many cases, no ‘opening’ of the mismatched base pair is required. This means that the lifetime of the mismatches in the closed state is greatly reduced relative to that of the normal Watson–Crick base pairs. It has been shown (52) that ∼0.1% of the GT mismatches are in the open solvent-accessible state and GT is one of the more thermodynamically stable mismatches. The mismatches, in general, are more accessible to solvent and other extraneous molecules than the Watson–Crick bases. This may have relevance to how the mismatches are recognized by the mismatch repairing proteins and how the mismatches may act as ‘hot spots’ in reactivity studies (56).

The imino exchange experiments measure primarily the lateral motions of the base pairs. These opening motions may occur by different pathways and may be coupled to bending (57–60), which in turn affects the stacking of the base pairs. Stacking is mostly a ‘vertical’ effect along the axis of the duplex while the opening motions are ‘horizontal’ effects. However, these two are clearly related, possibly one motion reinforcing the other. Molecular dynamics (MD) simulations have been performed on the lateral motion of duplexes containing mismatches (61). Consistent with the short lifetime of mismatches, opening fluctuations in MD runs are larger at mismatched base pairs 5–7°, than fully matched CG, 4°. It should, however, be emphasized that, by measuring the base-pair lifetime, we are essentially quantifying only one type of dynamic motion occurring in the DNA duplex. Mismatches can bring about changes in a variety of other internal motions in DNA like helical twisting, propeller twisting, base tilting, base rolling, etc. and these internal motions are not reflected in the study of base-pair lifetimes, although they may be related.

CONCLUSIONS

A systematic study of the base-pair dynamics of mismatch-containing oligonucleotides has been carried out using imino exchange measurements. We have observed very fast base-pair opening rates in the mismatch site, indicative of the local distortion generated around the mismatch. Some mismatches generate more distortion than others, and these differences are captured in the faster base-pair lifetimes. A relative ordering of the mismatches with respect to the base-pair lifetime has been achieved. In general, the mismatches are found to be kinetically destabilized relative to the Watson–Crick base pairs. However, certain Watson–Crick base pairs, under a given sequence context, may behave very much like a mismatch in terms of base-pair opening dynamics. The faster dynamics associated with the mismatches and with some Watson–Crick base pairs requires consideration in how these sites may be recognized and repaired by proteins.

SUPPLEMENTARY MATERIAL

Supplementary Material is available at NAR Online. Figures depicting the exchange of the imino protons of the duplexes containing the intervening AA, CC, TT, mismatch and the Watson–Crick bases GC and AT base pairs versus the inverse of the concentration of ammonia catalyst are given in Figures S1–S5.