Reduction in Dynamics of Base pair Opening upon Ligand Binding by the Cocaine-Binding Aptamer

Abstract

We have used magnetization transfer NMR experiments to measure the exchange rate constant (k_ex) of the imino protons in the unbound, cocaine-bound, and quinine-bound forms of the cocaine-binding DNA aptamer. Both long-stem 1 (MN4) and short-stem 1 (MN19) variants were analyzed, corresponding to structures with a prefolded secondary structure and ligand-induced-folding versions of this aptamer, respectively. The k_ex values were measured as a function of temperature from 5 to 45°C to determine the thermodynamics of the base pair opening for MN4. We find that the base pairs close to the ligand-binding site become stronger upon ligand binding, whereas those located away from the binding site do not strengthen. With the buffer conditions used in this study, we observe imino ¹H signals in MN19 not previously seen, which leads us to conclude that in the free form, both stem 2 and parts of stem 3 are formed and that the base pairs in stem 1 become structured or more rigid upon binding. This is consistent with the k_ex values for MN19 decreasing in both stem 1 and at the ligand-binding site. Based on the temperature dependence of the k_ex values, we find that MN19 is more dynamic than MN4 in the free and both ligand-bound forms. For MN4, ligand-binding results in the reduction of dynamics that are localized to the binding site. These results demonstrate that an aptamer in which the base pairs are preformed also experiences a reduction in dynamics with ligand binding.

Significance

Understanding how the dynamics of biomolecules change as they function is important in understanding how these molecules works. We used NMR methods to measure exchange rate constants (k_ex) of imino protons in the cocaine-binding aptamer to understand how the dynamics in the aptamers changes between the free and bound states. We also determined the thermodynamics of base pair dissociation as a function of ligand binding. We show that ligand binding results in a reduction in k_ex values near the binding site and strengthening of base pairs near the binding site. It is possible that a reduction in dynamics and strengthening of base pairs with ligand binding is a common trait in aptamers, but this will require further studies to confirm.

Introduction

An aptamer is a nucleic acid strand that can bind a particular target molecule. Aptamers occur in nature as part of a riboswitch or can be selected through a process known as SELEX (1, 2, 3, 4). The ligands of an aptamer can range from small molecules to cells, and aptamers often bind with high specificity and affinity. These traits make aptamers particularly useful in the development of biosensors (5, 6, 7, 8, 9). Understanding how aptamers function is important in their designed development in biotechnology applications, as well as obtaining a better understanding of how aptamers function in nature. However, few aptamers to date have been thoroughly characterized.

One aptamer that has been particularly well studied is the cocaine-binding DNA aptamer. This aptamer was first selected by Stojanovic and co-workers (10) and has become a model system for use in biosensor development (11, 12, 13, 14, 15, 16). The cocaine-binding aptamer has a three-way-junction structure formed around a tandem AG base pair (Fig. 1), with its high-affinity binding site located at the junction (17). This aptamer is distinctive in that its mode of binding changes from being largely preformed to having a ligand-induced structure-switching mechanism as the length of stem 1 is shortened from 6 bp (MN4; Fig. 1) to 3 bp (MN19; Fig. 1; (17,18)). The exact nature of the ligand-induced structure-switching mechanism in MN19 is not known and could result from a reduction of dynamics of preformed base pairs or wholescale folding of a region or helix of MN19. As shown by the essentially identical chemical shift changes with ligand binding, both MN4 and MN19 interact with their ligands in a similar manner, though with a lower affinity for MN19 (19,20). For MN4, the base pairs are essentially all formed in the free structure, though there may still be structural changes occurring on the tertiary level with ligand binding. For example, the angles between the stems may be altered with ligand binding, but this is not currently known.

Figure 1.

Secondary structure of MN4 and MN19. Circled is the high-affinity ligand-binding site. Boxed in MN19 are the base pairs in which imino protons are not observed in the free form because of rapid exchange. For ease of comparison, both aptamers follow the same numbering scheme.

The cocaine-binding aptamer is also unusual in that it has high specificity for cocaine and not for common cocaine metabolites (10,21, 22, 23, 24), whereas the aptamer has high binding affinity, tighter than for cocaine, to quinine and quinine-based antimalarial compounds (23,25, 26, 27, 28, 29). Another unusual feature of this aptamer is that it binds two copies of its ligand in an independent manner in low-NaCl conditions. At high NaCl concentrations, greater than 75 mM, only one ligand is bound (30). In this study, we work at 120 and 245 mM KCl, in which both MN4 and MN19 bind one molecule of cocaine or quinine.

One aspect of biomolecules that has become increasingly important is how their function is related to their mobility or dynamics (31,32). Dynamics has been well studied in proteins, but relatively less is known about how dynamics and function are related in nucleic acids (33, 34, 35, 36, 37). One method to study the dynamics in nucleic acids is by measuring the k_ex of the imino proton in base pairs with the bulk solvent (38). Previous studies have reported k_ex values between 1 and 124 s⁻¹ being detected by this method (39), with an upper exchange limit of 400 s⁻¹ given for detection of an imino proton (40). A recent approach developed by the Schwalbe lab is to measure the k_ex values as a function of temperature to obtain the thermodynamic parameters of base pair opening (41, 42, 43). In a previous study, we measured the k_ex values of the imino protons in free, cocaine-bound, and quinine-bound cocaine-binding aptamers at a single temperature, 5°C (44). We observed changes in the k_ex values with ligand binding. In the study presented here, we measured the k_ex value of the imino protons in MN4 and MN19 as a function of temperature in their free, cocaine-bound, and quinine-bound states to allow us to get a more in-depth view of how the dynamics differ between the two aptamers in the different ligand-bound states. We show that upon ligand binding in MN19, there is a reduction in dynamics in stem 1 and at the ligand-binding site. For MN4 there is a reduction in dynamics that is localized to nucleotides at the ligand-binding site. In addition, the experiments presented here were done in two different buffer conditions for MN4 to determine the enthalpy, entropy, and free energy of base pair opening as a function of ligand binding. We show that only the base pairs near the binding site increase in free energy and strengthen with ligand binding.

Materials and Methods

Materials

All DNA samples were obtained from Integrated DNA Technologies (Coralville, IA) with their sequences confirmed by mass spectrometry, as performed by the manufacturer. The purity of the DNA was additionally confirmed by the subsequent NMR studies. DNA samples were exchanged against 1 M NaCl four times and then exchanged against deionized water five times. Aptamer samples were reduced to 1000 μL in volume, and then the concentration of the aptamer solution was determined using Beer’s Law by measuring the A₂₆₀ with a Varian (Palo Alto, CA) Cary 100 Bio ultraviolet-visible light spectrophotometer and using the known extinction coefficient provided by Integrated DNA Technologies. Quinine hemisulfate monohydrate was obtained from Sigma-Aldrich (catalog number 145912; St. Louis, MO). Cocaine hydrochloride was also obtained from Sigma-Aldrich (catalog number C-5776). Ligand solutions were prepared by dissolving an appropriate amount of ligand in distilled deionized water.

NMR experiments

NMR experiments were conducted on a 600-MHz Bruker (Billerica, MA) Avance spectrophotometer at temperatures from 5 to 45°C. Before performing NMR experiments, samples were heated in a 90°C water bath for 3 min, then cooled in an ice-water bath to favor intramolecular folding of the aptamer. Spectra were acquired in either 245 mM KCl, 5 mM KHPO₄ (pH 6.8) in 10% ²H₂O, 90% H₂O (high-salt buffer with low catalyst concentration) or 120 mM KCl, 100 mM KHPO₄ (pH 6.8) in 10% ²H₂O, 90% H₂O (high-salt buffer with high catalyst concentration). NMR data were processed using TopSpin 3.7 (Bruker).

For both the MN4 and MN19 aptamers, an original sample was divided in half. Using one half of the sample, the NMR data for the free aptamer were acquired, then cocaine was titrated in to form a 1:1 complex using our previously published titrations as a guide (44). With the second half of the aptamer sample, quinine was titrated in to form a 1:1 complex, again using our previous titrations as a guide. Sample volumes were 550–600 μL. The concentrations of MN4 samples in low catalyst concentration buffer were 1.0 mM and in high catalyst concentration buffer were 1.3 mM. For MN19, the aptamer samples were 1.6 mM.

The imino hydrogen exchange rates were determined as described by Lee and Pardi (45). First, we performed inversion-recovery experiments to measure the R_1a- and R_1w-values. The water R₁ was measured using a DANTE pulse train designed so that the overall 90° pulse width was 90 μs. The peak intensities of the imino protons were measured using a water magnetization transfer experiment using a shaped 180° pulse along the imino region. To accomplish this, a RE-BURP pulse shape of 965 μs was used. The pulse was centered in the range of 9–16 ppm and uniformly covered this region. The magnetization transfer experiments were conducted with 40 different delay times ranging from 0 to 500 ms in a predetermined, nonsequential order in a manner described earlier (44). The order of temperature acquisition was randomized between 5 and 45°C. For the 10 and 120 ms delay times, we acquired these points three times each to gauge the level of variability in the data.

Determination of k_ex

The apparent relaxation constant of water (R_1w) and the apparent relaxation constant of each imino proton (R_1a) were obtained by measuring the intensity of either the water or imino peak in the inversion-recovery spectra and fitting the intensity to a single exponential curve. The intensity of the resonances in the magnetization transfer experiments were then measured, and the data were fitted to Equation 1 using SigmaPlot 11.0.

$$\frac{I\left(t\right)}{I_0}-1=-2\frac{k_{ex}}{R_{1w}-R_{1a}}\left(e^{-R_{1a}t}-e^{-R_{1w}t}\right),$$ (1)

where I(t) is the peak intensity as a function of delay time, I₀ is the peak intensity with a delay time of 0 ms, k_ex is the exchange rate constant for the nucleotide, and t is the delay time in seconds. Once the constants (R_1w and R_1a) for the equation were obtained, k_ex was determined by rearranging Eq. 1 to isolate for k_ex.

Determination of enthalpy and entropy of base pair dissociation

To obtain the enthalpy and entropy of base pair dissociation, we followed the method developed by Schwalbe and co-workers, in which the exchange rates were determined as a function of temperature (41,42). At low temperatures, the measured exchange is dominated by NOE effects (the term d in Eq. 2), so to correct for this, the data were fitted to Eq. 2 using SigmaPlot 11.0 as follows:

$$k_{ex}=\frac{1}{\frac{h}{k_{b}T}{e}^{\left(\frac{\Delta H_{TR}-T\Delta S_{TR}}{RT}\right)}\left(1+e^{\frac{\Delta H_{diss}-T\Delta S_{diss}}{RT}}\right)}+d,$$ (2)

where k_ex was the calculated exchange rate constant from previous fitting, h is Planck’s constant, k_b is Boltzmann’s constant, T is the temperature in Kelvin, ΔH_TR and ΔS_TR are the change in enthalpy and entropy of the open base pair, ΔH_Diss and ΔS_Diss are the change in enthalpy and entropy of the imino proton dissociating, and R is the gas constant. Once the k_ex values were corrected, the effect of solvent catalysis on the exchange was found using k_ex values in the high catalyst concentration buffer and low catalyst concentration buffer and the nucleoside triphosphate exchange rates (42).

The calculated k_ex has contributions from both the base pair opening and the imino proton dissociation and will be referred to as k_{ex, net} from this point forward. To determine the k_{ex, net} contribution of base pair opening, the ratio between k_{ex, ext}/k_{ex, int} was determined, where k_{ex, ext} is the exchange rate constant of the imino proton dissociation and k_{ex, int} is the exchange rate constant of the base pair dissociation. These two values are related as follows:

$$q=\frac{k_{ex,ext}}{k_{ex,int}}=\frac{k_{TR,ext}}{k_{TR,int}},$$ (3)

where k_{TR, ext} was obtained from previously published single nucleoside exchange rate experiments at 20°C (42), whereas q was determined using the k_{ex, net} imino proton exchange rate data at 20°C from both the high catalyst concentration and low catalyst concentration data according to Steinert et al. (42). The value of q was then used to find the k_{TR, int} by Eq. 3.

The data were fitted directly to the Eyring equation as follows (Eq. 4):

$$K_{TR,int}=\frac{k_b}{h}T{e}^{-\frac{\Delta H_{diss}}{TR}}{e}^{\frac{\Delta S_{diss}}{R}}.$$ (4)

Data were fitted using Igor Pro 8.04. k_{TR, int} was used as the as the y-value and temperature (T) was used as the x-value. ΔH_diss and ΔS_diss were set to an initial value of 150,000 and 100, respectively, and allowed to fit based on the data. k_b is Boltzmann’s constant, h is Planck’s constant, and R is the gas constant. The Gibbs free energy of base pair dissociation (ΔG_diss) was then determined using the following relation:

$$\Delta G_{diss}=\Delta H_{diss}-T\Delta S_{diss}.$$ (5)

Error values in ΔH_diss and ΔS_diss were obtained from the error in fit from Igor Pro, whereas the error in ΔG_diss was calculated based on the errors in ΔH_diss and ΔS_diss by calculating the quadrature of the errors and adjusting for covariance. The error in ΔΔG_diss was calculated based on the error in ΔG_diss by calculating the quadrature of the errors.

Results

Imino proton k_ex measurements

The ¹H-NMR spectra of the MN4 and MN19 cocaine-binding aptamer constructs (Fig. 1) in the free, cocaine-bound, and quinine-bound states were highly similar to what we observed previously, and the imino ¹H assignments were obtained from a comparison to the previously published spectra (Fig. 2; (44)).

Figure 2.

Sample spectra of MN4 (a) free, (b) cocaine bound, and (c) quinine bound and MN19 (d) free, (e) cocaine bound, and (f) quinine bound. The spectra were acquired in 5 mM KHPO₄, 245 mM, KCl (pH 6.8), 10% ²H₂O/90% H₂O.

NMR spectra were acquired every 5° between 5 and 45°C to observe the effect of temperature on the dynamics of the aptamer. For each aptamer-ligand combination, the R_1w and R_1a-values were measured using inversion-recovery experiments. Next, the imino resonance intensity was measured as a function of delay time after perturbation of the water signal using the magnetization transfer experiments with 44 one-dimensional spectra obtained with delay times varying from 5 to 500 ms (Fig. 3 a). The peak intensity of these spectra was measured and fit to an intensity curve to obtain k_ex (Fig. 3 b). We will note that the three separate experiments acquired at both 10 and 120 ms are essentially overlapped in this plot, showing very close repeatability of this analysis. The measured k_ex values are found in Tables S1–S3. Plots of k_ex values as a function of temperature in low catalyst concentration buffer are found in Fig. 4, and plots of k_ex values as a function of temperature in high catalyst concentration buffer are in Fig. S3.

Figure 3.

One-dimensional ¹H spectra of (a) the magnetization transfer experiment showing the imino proton resonances for cocaine-bound MN4 at 10°C are shown. (b) The normalized peak intensity, I(t)/I₀, of cocaine-bound T15 in MN4 at 10°C (boxed peak in part (a)) is shown. Shaded circles represent the raw data acquired from the NMR experiments, whereas the line shows the fit of the data to Eq. 1 to determine the k_ex values. The R²-value for this fit is 0.98.

Figure 4.

The k_ex data as a function of temperature for MN4 (a) free, (b) cocaine bound, and (c) quinine bound and MN19 (d) free, (e) cocaine bound, and (f) quinine bound. All data were acquired in 5 mM KHPO_4, 245 mM KCl (pH 6.8). The bases are shown in the following scheme: G2, pink squares; T4, red diamonds; T15, dark red triangles; T18, magenta circles; T19, orange squares; G24, yellow diamonds; T28, lime triangles; G29, green circles; G30, aqua squares; G31, blue diamonds; T32, purple triangles; and G34, black circles. Multiple points at the same temperature for the same nucleotide represent the results from independent analysis of the same data set. To see this figure in color, go online.

Because of overlap of some imino proton resonances in the NMR spectra, not all imino proton intensities could be accurately measured in all aptamer/ligand/temperature combinations. For the 16 imino protons in MN4, G1 has never been observed, likely because of stem breathing effects, and G9, G10, and G27 are almost always overlapped. We were able to measure k_ex values for a maximum of 12 separate resonances in MN4, with eight being observable in all three samples. For MN19, there are 13 potentially observable resonances, with the k_ex of 10 resonances measured in at least one of the states and three being measurable across all the different states. This low number observed in all three states is due to the free state being loosely folded or partially folded and the MN19 aptamer folding or rigidifying upon ligand binding (19,44).

Entropy and enthalpy of base pair dissociation

Imino ¹H k_ex values for free, cocaine-bound, and quinine-bound MN4 were measured at different temperatures and in both high catalyst concentration and low catalyst concentration buffers to calculate the enthalpy (ΔH_diss), entropy (ΔS_diss), and the free energy (ΔG_diss) of base pair opening by fitting these data directly to the Eyring equation (42). Data are presented in Table 1 and plotted in Figs. 5 and S4.

Table 1.

Calculated Enthalpy, Entropy, and Free Energy at 15°C of Dissociation for Bases in MN4

		Residues
		G2	T18	T19	T28	G29	G31	T32	G34
Free aptamer	ΔHdiss (×104) (J mol-1)	12 ± 2	12 ± 1	–	11 ± 2	10 ± 2	8.4 ± 0.8	19 ± 2	21 ± 5
	ΔSdiss (×101) (J mol-1 K-1)	15 ± 7	18 ± 4	–	14 ± 6	13 ± 5	5.4 ± 3	40 ± 7	45 ± 17
	ΔGdiss (×103) (J mol-1)	75 ± 2	65.0 ± 0.2	–	68.3 ± 0.8	61.4 ± 0.2	68.8 ± 0.4	77.8 ± 0.2	84 ± 3
Cocaine-bound	ΔHdiss (×104) (J mol -1)	12 ± 2	13.9 ± 0.3	12 ± 2	25 ± 1	9.6 ± 0.8	23 ± 1	17 ± 9	13.9 ± 0.6
	ΔSdiss (×101) (J mol-1 K-1)	17 ± 5	21 ± 1	16 ± 5	57 ± 5	11 ± 3	51 ± 4	32 ± 31	19 ± 2
	ΔGdiss (×103) (J mol-1)	74 ± 1	77.6 ± 0.3	74 ± 1	83.6 ± 0.7	64.0 ± 0.4	85 ± 1	78 ± 4	83.9 ± 0.4
Quinine-bound	ΔHdiss (×104) (J mol-1)	7.0 ± 0.8	16.7 ± 0.5	–	–	9.7 ± 0.9	14 ± 6	–	7.6 ± 0.4
	ΔSdiss (×101) (J mol-1 K-1)	0 ± 3	31 ± 2	–	–	11 ± 3	13 ± 2	–	4.1 ± 0.2
	ΔGdiss (×103) (J mol-1)	69.6 ± 0.7	78.2 ± 0.5	–	–	64.0 ± 0.4	77 ± 4	–	75.9 ± 0.3

The uncertainty represents in error in the fit of the data in ΔH_diss and ΔS_diss and is propagated to ΔG_diss by calculating the quadrature of the errors and adjusting for covariance.

Figure 5.

Plot of the exchange rate for G31 in MN4 to the Eyring equation. Shown are data in the free form (shaded squares with dashed line), the cocaine-bound form (diamonds with dotted line), and the quinine-bound form of the MN4 aptamer (open triangles with solid line).

The values of ΔH_diss and ΔS_diss vary among the different bases in the free and ligand-bound forms of MN4, with ΔH_diss-values ranging from 70 to 250 kJ mol⁻¹ and ΔS_diss-values ranging from 0 to 570 J mol⁻¹ K⁻¹. The ΔH_diss- and ΔS_diss-values appear to be linearly correlated (Fig. S5), as also observed in previous studies using this method (41, 42, 43).

The ΔG_diss-values are presented in Table 1 and range from 61.4 to 85 kJ mol⁻¹. We note that for free MN4, the largest ΔG_diss-values, corresponding to the strongest base pairs, belong to the Watson-Crick GC base pair G34, followed by the Watson-Crick AT base pair T32. For the cocaine-bound MN4, the largest ΔG_diss-value belongs to G31, followed by G34. For quinine-bound MN4, the largest ΔG_diss-value belongs to the Watson-Crick AT base pair T18. In all three forms of MN4, the lowest measured ΔG_diss-value, indicating the weakest base pair, belongs to G29, a nucleotide that is in a non-Watson-Crick AG base pair.

Upon ligand binding, there is a uniform trend among the ΔG_diss-values, for which the base pairs closest to the ligand-binding site (T18, T28, G29, G31, and T32; Fig. 1) become larger, indicating they become stronger. For the base pairs furthest from the ligand-binding site (G2, G34), their ΔG_diss-values decrease, indicating these base pairs are weaker when the aptamer is ligand bound. The ΔΔG_diss-values between bound and free MN4 are tabulated in Table S4. We see no uniformly clear trend in changes in ΔH_diss and ΔS_diss-values with ligand binding among the different bases in MN4. This is exemplified by G31, where ΔH_diss and ΔS_diss increase for cocaine and quinine binding, whereas for T18, ΔH_diss and ΔS_diss change much less (Table 1).

Discussion

The hydrogen exchange rates of the imino protons in the MN4 aptamer correspond to their placement within the helices in the aptamer, with residues at the ends having higher k_ex values than residues in the middle of stems. For example, G2 and G34, which are located in the middle of stem 1 (Fig. 1), have low k_ex values of (3 ± 2) and (3 ± 1) s⁻¹, respectively, at 10°C in the free MN4 aptamer, whereas T18 and T15, located at the edges of stem 2, have higher k_ex values of (13 ± 2) and (40 ± 4) s⁻¹, respectively, at the same temperature in the free MN4 aptamer (Table S1). G29 and G30, the A-G mismatches at the core of the aptamer, have high k_ex values of (50 ± 6) and (65 ± 8) s⁻¹ at 10°C in the free MN4 aptamer. These trends match our previously reported data, as well as data from reports from other groups (40,43,44).

The NMR data presented here on free MN19 give new insights into the nature of the structure of the unbound MN19 aptamer. Free MN19 is thought to be loosely structured and is predicted to undergo ligand-induced folding (21). In the ¹H-NMR of free MN19 acquired here, in the presence of low catalyst concentration buffer (245 mM KCl, 5 mM KHPO₄ [pH 6.8]), we observed more resonances for free MN19 than seen in our previous studies (Fig. 2). We were able to observe distinct signals from T18, T28, T15, and G24, as well as a group of overlapped signals from G9, G10, and G27. We observe no signals from residues in stem 1 or the two AG base pairs. This demonstrates that in free MN19, stems 2 and 3 are structured, and it is stem 1 and the AG base pairs that become structured with ligand binding. This is in agreement with the model proposed by Sigurdsson and co-workers (46).

We were able to measure some k_ex values for the free MN19 aptamer (Fig. 4, d–f; Table S2). For all the nucleotides that we can measure that are in both free MN4 and free MN19 (T15, T18, G24, T28), these nucleotides have higher k_ex values at the same temperature in MN19. Additionally, the imino proton NMR signal disappears at lower temperatures in free MN19 compared with free MN4 (Fig. 4). This implies that the folded regions of the unbound MN19 aptamer are more dynamic than the same regions in free MN4. Again, this is consistent with MN19 undergoing ligand-induced folding starting from a partially unfolded or flexible state.

Increasing the temperature increases the k_ex values of imino protons in all forms of both MN4 and MN19, but the rate of increase depends on the position of the nucleotide within the aptamer. Residues buried within the aptamer or stabilized by interactions with the ligand increase their k_ex values more slowly than those found on the periphery of the aptamer (Fig. 6). For example, in cocaine-bound MN4, G2, located near the end of stem 1, increased from (4 ± 3) s⁻¹ at 5°C to (27 ± 2) s⁻¹ at 40°C, whereas G4, which is located further into stem 1, remains fairly constant, with a k_ex value of (4 ± 1) s⁻¹ at 5°C and a k_ex value of (3 ± 2) s⁻¹ at 40°C. T15 and T18 are located at opposite ends of stem 2, with T18 being located close to the ligand-binding site. Here, T15 increases from (21 ± 2) s⁻¹ at 5°C to (6 ± 1) × 10¹ s⁻¹ at 15°C in the cocaine-bound MN4 aptamer. This residue is not seen in higher-temperature data sets because it exchanges with the solvent too quickly to be observed. T18, however, has a k_ex value of (1 ± 3) s⁻¹ at 5°C and remains fairly constant at 15°C, with a value of (3 ± 3) s⁻¹, and does not match the exchange rate of T15 at 15°C until T18 reaches 40°C, at which it has a k_ex value of (54 ± 6) s⁻¹. This nonuniform increase in k_ex values as temperature increases is consistent with what was seen previously in the Hsp17 RNA thermometer (43).

Figure 6.

Secondary structure diagram of MN4 and MN19 highlighting the dynamics of the free, cocaine-bound, and quinine-bound aptamers. Base pairs are shaded on a scale of red to purple to blue, depending on the temperature the imino proton in that base pair reached a k_ex value of 40 s⁻¹ based on the fit line of the k_ex values (Fig. 4). Bases that started with a k_ex value of 40 s⁻¹ or higher at 5°C were shaded 100% red (255 red, on an RGB scale). Bases that do not reach a k_ex value of 40 s⁻¹ at 45°C were shaded 100% blue (255 blue, on an RGB scale). Other bases were shaded purple (shaded so that the sum of the red and blue sliders on a RGB scale was 255), depending on when the k_ex value for its imino proton reached a value of 40 s⁻¹. Imino protons that reached that value at a lower temperature are shaded darker red, and imino protons that reached that value at higher temperatures are shaded darker blue. Bases that are shaded in a pale red box are unobserved in the free MN19 because of rapid exchange. To see this figure in color, go online.

The addition of ligand reduced the k_ex of imino protons close to the binding site, but it has little effect on resonances distant from the binding site. For example, G29 is located at the binding site of the aptamer, and its k_ex value reduces from (50 ± 6) s⁻¹ in the free state of MN4 at 10°C to 15 ± 1 s⁻¹ in the cocaine-bound MN4 aptamer also at 10°C and to 12 ± 2 s⁻¹ in the quinine-bound MN4 aptamer at the same temperature. Other resonances such as T15 and G34, which are far removed from the binding site, differ little in their k_ex values, with (40 ± 4) and (4 ± 1) s⁻¹, respectively, in the free state of MN4 at 10°C; (33 ± 3) and (2 ± 2) s⁻¹, respectively, in the cocaine-bound MN4 at 10°C; and (25 ± 3) and (3 ± 3) s⁻¹, respectively, in the quinine-bound MN4 at 10°C. This demonstrates that the changes in dynamics we observe in MN4 at the timescale monitored by k_ex values are localized to the binding site and are not felt globally throughout the molecule.

When comparing quinine- and cocaine-bound MN4, we and others have shown that quinine binds ∼50-fold tighter than cocaine (15,23,25,27,47, 48, 49). When looking at the k_ex values of quinine- and cocaine-bound MN4, we generally observe little difference between these two bound aptamers at low temperatures. As the temperature increases, the k_ex values of some nucleotides (T15, T18, G29) in the tighter-binding quinine sample do not increase as fast as in the cocaine-bound sample. Also, we see that in the quinine-bound MN4 ¹H-NMR spectra, there are more peaks visible at higher temperatures than seen in the cocaine-bound spectrum (Figs. S1 and S2). This demonstrates that the tighter-binding ligand does not result in significantly decreased k_ex values until the sample is at temperatures approaching those for which its ¹H-NMR signal disappears.

At 5°C, the k_ex values for residues in bound MN19 have similar values as in bound MN4 (Tables S1 and S2). However, these two ligand-bound aptamers differ in that as the temperature increases, the k_ex values increase more rapidly in MN19, and their peaks in the NMR data disappear sooner in MN19 than in MN4. This observation is consistent with both the cocaine- and quinine-bound forms of MN19 being a more dynamic version of the cocaine-binding aptamer and undergoing a structure-switching binding mechanism. It is also consistent with ligand-bound MN19 being less thermally stable than ligand-bound MN4 (17,18,20,24). For example, at 5°C in the cocaine-bound aptamers, T15 has a k_ex of (14.4 ± 0.5) s⁻¹ in MN19 and a k_ex value of (21 ± 2) s⁻¹ in MN4, T28 has a k_ex of (1 ± 3) s⁻¹ in MN19 and a k_ex of (2 ± 2) s⁻¹ in MN4, and G29 has a k_ex of (7.0 ± 0.6) s⁻¹ in MN19 and a k_ex of (14 ± 2) s⁻¹ in MN4. What does differ is that residues in MN19 disappear faster than their counterparts in MN4. For the cocaine-bound aptamers, T18 disappears after 20°C in MN19 compared with 40°C in MN4. T19 disappears after 15°C in MN19 compared with 35°C in MN4, and G31 disappears after 20°C in MN19 vs. 45°C in MN4 (Fig. 4; Tables S1 and S2).

Ligand binding by aptamers is generally thought to result in aptamers becoming more structured, which should result in a reduction of motion or dynamics in the aptamer (37,50). Previous imino ¹H exchange rate studies on the VEGF₁₆₅-targeting aptamer (Macugen) and the AMP-RNA aptamer both show a large reduction in dynamics upon binding at the ligand-binding site, which in both these aptamers is a loop or bulge region that is unstructured in the free state (40,51). In this study, for both MN4 and MN19, we also see a reduction in k_ex values at the ligand-binding site upon cocaine and quinine binding. This might be expected for MN19, in which signals from stem 1 are not observed in the free aptamer and only appear in the ligand-MN19 complex. For MN4, the base pairs in this aptamer are formed in the free state, and this aptamer also shows a reduction in dynamics upon ligand binding. This demonstrates that even aptamers with no large scale change in base pairing have reduced dynamics with ligand binding and suggest this may be a general feature of aptamer function and does not depend on the formation of new base pairs with ligand binding.

Our observation that in MN4, the base pairs close to the ligand-binding site strengthen with both cocaine and quinine binding complements the reduction in dynamics, as indicated by reduced k_ex values. Together, these data lead to a binding model for MN4 in which both cocaine and quinine addition lead to a reduction of dynamics and base pair strengthening near the binding site despite the base pairs in this aptamer being preformed. It is a possibility that a strengthening of base pairs upon ligand binding is a common trait in aptamers, but this will require further studies to confirm.

Conclusions

We have measured k_ex values for the imino protons in MN4 and MN19 in their free, cocaine-bound, and quinine-bound states as a function of temperature. In addition, from data acquired in two different buffers, the ΔH_diss-, ΔS_diss-, and ΔG_diss-values for base pair dissociation in MN4 were also determined. For MN19, with ligand binding we observe a reduction in k_ex values of iminos both near the binding site and in stem 1. This is consistent with its ligand-induced-folding mechanism, in which stem 1 folds or becomes more rigid with ligand binding. In MN4, upon ligand binding we saw a localized reduction of k_ex values for imino protons near the binding site, but not elsewhere in the aptamer. Additionally, from the ΔG_diss-values, base pairs near the ligand-binding site in MN4 strengthened with both cocaine and quinine binding. This means that both a prefolded aptamer and an aptamer loosely folded in the free state undergo a reduction in dynamics upon ligand binding.

Author Contributions

Z.R.C., H.N.H., and P.E.J. contributed to the conception and design of the experiments. Z.R.C., D.G., and P.E.J. performed research and analyzed the data. Z.R.C. and P.E.J. wrote the manuscript.

Editor: Scott Showalter.