Increased Flexibility between Stems of Intramolecular Three-Way Junctions by the Insertion of Bulges

Abstract

Intramolecular junctions are a ubiquitous structure within DNA and RNA; three-way junctions in particular have high strain around the junction because of the lack of flexibility, preventing the junctions from adopting conformations that would allow for optimal folding. In this work, we used a combination of calorimetric and spectroscopic techniques to study the unfolding of four intramolecular three-way junctions. The control three-way junction, 3H, has the sequence d(GAAATTGCGCT₅GCGCGTGCT₅GCACAATTTC), which has three arms of different sequences. We studied three other three-way junctions in which one (2HS₁H), two (HS₁2HS₁), and three (HS₁HS₁HS₁) cytosine bulges were placed at the junction to allow the arms to adopt a wider range of conformations that may potentially relieve strain. Through calorimetric studies, it was concluded that bulges produce only minor effects on the enthalpic and thermal stability at physiological salt concentrations for 2HS₁H and HS₁HS₁HS₁. HS₁2HS₁ displays the strongest effect, with the GTGC stem lacking a defined transition. In addition to unfolding thermodynamics, the differential binding of counterions, water, and protons was determined. It was found that with each bulge, there was a large increase in the binding of counterions; this correlated with a decrease in the immobilization of structural water molecules. The increase in counterion uptake upon folding likely displaces binding of structural water, which is measured by the osmotic stress method, in favor of electrostricted waters. The cytosine bulges do not affect the binding of protons; this finding indicates that the bulges are not forming base-triplet stacks. These results indicate that bulges in junctions do not affect the unfolding profile or the enthalpy of oligonucleotides but do affect the number and amount of molecules immobilized by the junction.

Introduction

Inverted repeats within DNA and RNA form basestacks and extrude as a secondary structure known as a hairpin (1, 2, 3). Multiple hairpins or stems branching from a single point are known as junctions. Three-way junctions in particular are common within RNA riboswitches (4, 5, 6) and ribozymes (6, 7, 8, 9) and are important structures in DNA replication (10, 11, 12) and recombination (13, 14, 15). The tertiary structure of three-way junctions is important, as three-way junctions are known to be involved in loop-loop kissing interactions (6, 16) as well as docking of a loop sequence to another stem (17, 18), imparting stability or generating a structure recognized by proteins. These structures are very common, but the thermodynamics of their unfolding is still being established, as it is influenced by loop and stem sequence, loop and stem size, number of arms, and flexibility.

The effect of bulges on the flexibility of three-way junctions and the importance of this feature have been well documented. It has been determined from molecular modeling as well as experiments that three-way junctions without bulges are not capable of attaining ∼20–50% of interhelical orientations (19) and that true coaxial stacking, as seen in four-way junctions, is not attainable for three-way junctions (20, 21); instead, three-way junctions adopt nonplanar pyramid structures (22). Bulges have been shown to improve binding of proteins by acting as a unique structural feature that differentiates the smooth planes of the major and minor grooves (23, 24). Bulges have been shown to form interactions with distant sites that assist with tertiary structure formation in ribozymes (25, 26), and bulges at different locations either generate or abolish directional movement, i.e., bulges in specific locations either allow or prevent the movement of DNA or RNA in specific directions (27, 28). All of this reveals bulges to be simple yet critical structural features in oligonucleotides. In addition to their presence in biological systems, DNA three-way junctions are now being widely used as nanostructures (29, 30, 31, 32, 33), multichromophore arrays (34, 35), and aptamers (36, 37, 38). In addition, junctions are being used as metal sensors (39) and assembly scaffolds (40, 41). Although the structure of bulged three-way junctions has been studied and used in studies previously, very little has been determined of the effect of bulges on the thermodynamics or the binding of small molecules (ions, water, and protons). Our research should shed light on the thermodynamic behavior of these biological junctions, leading to a better understanding of their possible tertiary structures and the effects these bulges have on their overall stability. Our data should improve theoretical simulations on junction dynamics by expanding the known behavioral parameters and allow a greater understanding of the results obtained from both simulations and experiments designed to assess structural dynamics. In addition, for scientists utilizing three-way junctions as a synthetic structural framework, our work should improve their understanding of these scaffolds, leading to a wider variety of junctions that may be more suitable for a given experiment.

Current understanding of the unfolding of intramolecular junctions is based on previous investigations of intramolecular junctions and hairpins (42, 43). The primary focus of this research is to understand the effect that bulges have on the unfolding thermodynamics of intramolecular junctions. The intramolecular junctions studied here are experimentally favorable because of their lower entropy cost, which will affect the transition temperatures, leading to unfolding at temperatures higher than what would be observed for oligonucleotides of higher molecularity (44, 45). This approach ensures that the oligonucleotides are in a 100% helical conformation.

This research gives a complete thermodynamic description of the unfolding of DNA three-way junctions with increasing numbers of cytosine bulges (Fig. 1). Cytosines were chosen as the bulges because previous research has shown that pyrimidines have more conformational variability than purines, which always loop in (46, 47, 48). Pyrimidines tend to loop out, but this depends on temperature and flanking base sequence (46, 49). Looped-out bases do not cause significant kinks in DNA like looped-in bases do, which may twist the junctions into more strained conformations. In addition, bulges flanked by identical bases are more stable, leading to cytosine being chosen over thymine (50). A combination of ultraviolet (UV)-visible and differential scanning calorimetry (DSC) techniques were used to assess bulge effects on the unfolding of intramolecular junctions. In addition to unfolding thermodynamics, the differential binding of counterions, water, and protons is also discussed. The results suggest that in the native environment of the cell, bulges produce little effect on the unfolding profile and thermodynamics of a junction. However, increasing the number of bulges and thus the flexibility does significantly increase the number of counterions bound to the oligonucleotides, presumably at the junction point, which can now adopt various strain-relieving conformations. In conjunction with an increase in counterion uptake, there is a decrease in the immobilization of structural water molecules (water molecules hydrating polar and nonpolar/hydrophobic groups). This is likely due to the high amount of counterions, which preferentially bind electrostricted water molecules rather than structural water molecules. Finally, increasing the number of cytosine bulges linearly increases the heat capacity of the system, suggesting a more fully folded molecule that sequesters a higher amount of hydrophobic groups from solvent. This suggests that bulges do indeed impart flexibility that allows a three-way junction to relieve strain around the junction, resulting in a change in structure and conformation.

Figure 1.

Cartoon of the sequences of the three-way junctions and control hairpins drawn according to their hypothesized structures.

Materials and Methods

Materials

The oligonucleotides and their designations are as follows: d(GAAATTC₅AATTTC), GAAATT-Hp; d(GCGCT₅GCGCGTGCT₅GCAC), Dumbbell; d(GCGCT₅GCGCCGTGCT₅GCAC), 1C-Dumbbell; d(GAAATTGCGCT5GCGCGTGCT₅GCACAATTTC), 3H; d(GAAATTGCGCT₅GCGCCGTGCT₅GCACAATTTC), 2HS₁H; d(GAAATTCGCGCT₅GCGCGTGCT₅GCACCAATTTC), HS₁2HS₁; and d(GAAATTCGCGCT₅GCGCCGTGCT₅GCACCAATTTC), HS₁HS₁HS₁. All DNA molecules were synthesized by Integrated DNA Technologies (Coralville, IA), purified with reverse-phase high-performance liquid chromatography, desalted on a G-10 Sephadex column, and lyophilized to dryness before use in experiments. The sequences of the dumbbells and three-way junctions along with their hypothesized structures are shown in Fig. 1.

The concentration of each oligomer solution was determined from absorbance measurements at 260 nm at 90°C using the molar absorptivities, in mM⁻¹ cm⁻¹ of strands, reported as follows: 161.4 (GAAATT-Hp), 222.9 (Dumbbell), 231.2 (1C-Dumbbell), 348.6 (3H), 355.5 (2HS₁H), 371.0 (HS₁2HS₁), and 375.4 (HS₁HS₁HS₁). These values were calculated by extrapolation of the tabulated values of the dimer and monomer bases from 25°C to high temperatures, using the procedures reported earlier (51, 52). Buffer solutions consisted of 10 mM sodium phosphate buffer and pH 7.0, adjusted to different salt concentrations, ethylene glycol concentrations, and pH with NaCl, ethylene glycol, and HCl, respectively. All chemicals used in this study were reagent grade and used without further purification.

Temperature-dependent UV spectroscopy

Absorbance versus temperature profiles (UV melting curves) were measured at 268 nm with a thermoelectrically controlled Aviv 14 DS UV visible spectrophotometer (Lakewood, NJ). 268 nm was chosen because this wavelength follows the absorbance changes of both AT and GC basepairs. The temperature was scanned from 1 to 100°C at a heating rate of ∼0.6°C/min. Analysis of the shape of the melting curves yielded transition temperatures, T_M, which correspond to the inflection point of the helix-coil transition and van’t Hoff enthalpies, ΔH_vH, using standard procedures (44). To determine the molecularity of the transition(s) of each DNA molecule, we investigated the dependence of T_M over at least a 20-fold range of total strand concentration. If the T_M remains constant in this range of strand concentration, it indicates a monomolecular or intramolecular transition (44). Additional UV melting curves were obtained as a function of salt concentration to determine the differential binding of ions, ethylene glycol concentration to determine the differential binding of water molecules, and pH to determine the differential binding of protons.

Differential scanning calorimetry

The total heat required for the unfolding of each oligonucleotide was measured with a VP-DSC from Malvern MicroCal (Northampton, MA). These thermograms were obtained with a temperature ramp of ∼0.6°C/min with oligomers ranging in concentration from 30 to 80 μM in total strands. Analysis of the thermograms yielded T_Ms and standard thermodynamic profiles using the following relationships (44, 53): ΔH_cal = ∫ΔC_p(T)dT, ΔS_cal = ∫ΔC_p(T)/TdT, and the Gibbs equation, ΔG°_(T) =ΔH−TΔS, where ΔC_p(T) is the anomalous heat capacity of the oligonucleotide solution during the unfolding process, ΔH_cal and ΔS_cal are the unfolding enthalpy and entropy, respectively, both of which are assumed to be temperature-independent, and ΔG°_(T) is the free energy at a temperature T. Alternatively, ΔG°_(T) can be calculated using the equation ΔG°_(T) = ΔH_cal(1−T/T_M). Additional DSC experiments were obtained at several salt concentrations to indirectly obtain the associated heat capacity contributions, ΔC_p, which were determined from the slopes of the lines of the ΔH_cal versus T_M plots.

The DSC thermograms for all oligonucleotides at all buffer conditions (pH, salt, and ethylene glycol) were highly reproducible except when running in buffer at pH 5.2. At this pH, all oligonucleotides except GAAATT-Hp degraded after the first scan. However, the oligonucleotides were stable for long periods at pH 5.2 in low temperatures, and the thermodynamic profiles at pH 5.2 were similar to those obtained at pH 7.0 (i.e., total enthalpy is maintained, indicating complete formation of basepair stacks), and so they were assumed to be accurate unfolding thermograms. The first scans of at least three different runs were compared and analyzed to ensure reproducibility.

Determination of the differential binding of counterions

The following equations were used to measure the thermodynamic uptake of counterions, Δn_Na+, water molecules, Δn_W, and protons, Δn_H+, upon the folding of each hairpin, dumbbell, and three-way junction (54, 55):

$$\mathrm{\Delta }{n}_{\text{Na}}^{+}=1.11\left[\mathrm{\Delta }{H}_{\text{cal}}/\text{R}{T}_{\text{M}}^2\right]\left(\partial T_{\text{M}}/\partial \ln \left[{\text{Na}}^{+}\right]\right)$$ (1)

$$\mathrm{\Delta }{n}_{\text{W}}=\left[\mathrm{\Delta }{H}_{\text{cal}}/\text{R}{T}_{\text{M}}^2\right]\left(\mathrm{\partial }{T}_{\text{M}}/\partial \ln \left[{\text{a}}_{\text{W}}\right]\right)$$ (2)

$$\mathrm{\Delta }{n}_{\text{H}}^{+}=-0.434\left[\mathrm{\Delta }{H}_{\text{cal}}/\text{R}{T}_{\text{M}}^2\right]\left(\partial T_{\text{M}}/\partial \text{pH}\right),$$ (3)

where 0.434 and 1.11 are correction factors that correspond to the conversion of decimal logarithms into natural logarithms and of concentrations into ionic activities, respectively. The [ΔH_cal/RT_M²] term is a constant determined from DSC experiments in which the enthalpy is model independent, and R is the gas constant. The values in parentheses are determined from UV melting curves by measuring the T_M dependencies on the concentration of counterions (by varying the salt concentration), water molecules (by varying the osmolyte, ethylene glycol, concentration) and protons (by varying the pH).

In determining Δn_Na+, the UV experiments were carried out with [NaCl] ranging from 10 to 200 mM at pH 7.0. For Δn_H+, the UV melting curves were obtained from a pH range of 5.2 to 7.0 with a concentration of 10 mM NaCl. For Δn_W, the UV melting curves were carried out at pH 7.0 in 10 mM NaCl. The activity of water was varied by increasing the concentration of ethylene glycol, an osmolyte that does not interact with DNA, from 0.5 to 2.5 M (56). The osmolality of the ethylene glycol solutions was measured with a Model 830 UIC vapor pressure osmometer that was calibrated with standardized NaCl solutions. The osmolalities were converted into water activity using the equation lna_W = −Osm/M_W (57), where Osm is the measured solution osmolality and M_W is the molality of pure water. Measurement of the differential binding of water using the osmotic stress method yields only the immobilization of structural (water hydrating polar and nonpolar groups) and not electrostricted (water hydrating charged groups) water.

Results and Discussion

Experimental design

All the oligonucleotides in this study were designed to form an intramolecular hairpin, dumbbell, or three-way junction. The three-way junctions were designed to be a combination of Dumbbell and GAAATT-Hp (Fig. 1). To assess the effects of flexibility on the thermodynamic profile of a three-way junction, cytosine bulges were inserted at different junction sites, or where two stems are joined. Nomenclature for the junctions follows previously established naming methods, in which H refers to a double-stranded helix and S corresponds to a single strand (58). The numbers before the H correspond to the number of unbroken helices, whereas the subscripts refer to the number of nucleotides in the single strand, which in this study are single bulges. GAAATT-Hp and Dumbbell describe the hairpin and dumbbell structure that make up the three-way junction, 3H. 1C-Dumbbell refers to the dumbbell with a cytosine bulge between the stems on either side, directly below the nick site. 2HS₁H refers to the three-way junction with a cytosine bulge between the hairpins on the dumbbell structure, in the same location as 1C-Dumbbell. HS₁2HS₁ refers to the three-way junction with two cytosine bulges where GAAATT-Hp connects to Dumbbell in 3H. Finally, HS₁HS₁HS₁ refers to the junction with a cytosine bulge at all three junction positions. The oligonucleotides with cytosine bulges will be compared to the molecules with no bulges.

All oligonucleotides fold intramolecularly

Fig. 2 (left) shows typical UV melting curves for all oligonucleotides in 10 mM sodium phosphate (NaPi) buffer at pH 7.0. These melting curves were obtained over a total strand concentration range of 1–23 μM to determine the transition molecularity of each oligonucleotide. The corresponding T_M dependencies on concentration are shown in Fig. 2 (right). All T_Ms remain constant over a 20-fold range in strand concentration, indicating that all molecules are forming intramolecular complexes as designed.

Figure 2.

(Left) UV melting curves at 268 nm in 10 mM NaPi at pH 7.0 and (right) T_M dependence on total strand concentration at 268 nm in 10 mM NaPi at pH 7.0 of (top) GAAATT-Hp (solid line), Dumbbell (dashed line), and 1C-Dumbbell (dotted line) and (bottom) 3H (solid line), 2HS₁H (dashed line), HS₁2HS₁ (dotted line), and HS₁HS₁HS₁ (small dotted line). First transitions are circles and second transitions are squares.

The calorimetric unfolding of the oligonucleotides

DSC thermograms at three different salt concentrations for the hairpin and dumbbells are shown in Fig. 3 (left) and for the three-way junctions in Fig. 4 (left). The resulting thermodynamic profiles are shown in Table 1 for the thermograms in 10 mM NaPi, 0.1 M NaCl at pH 7.0. The unfolding of all molecules was highly reproducible. The identity of the stem that gives rise to each transition seen in the thermograms is assumed in our analysis based on a combination of enthalpy (GC, CG, and GG basepair stacks have more enthalpy than AT, TA, and AA basepair stacks) (59) and thermal stability (there is a significant >30°C thermal difference between a GCGC hairpin and an ATAT hairpin) (60).

Figure 3.

(Left) DSC thermograms in 10 mM NaPi at pH 7.0 (solid lines), with the addition of 0.1 M NaCl (dashed lines) and 0.2 M NaCl (dotted lines) and (right) associated heat capacity plots of the first (solid lines) and second (dashed lines) transition of GAAATT-Hp (top), Dumbbell (middle) and 1C-Dumbbell (bottom).

Figure 4.

(Left) DSC thermograms in 10 mM NaPi at pH 7.0 (solid lines), with the addition of 0.1 M NaCl (dashed lines) and 0.2 M NaCl (dotted lines) and (right) associated heat capacity plots of the first (solid lines), second (dashed lines), and third (dotted lines) transition of 3H (top), 2HS₁H (middle top), HS₁2HS₁ (middle bottom), and HS₁HS₁HS₁ (bottom).

Table 1.

Thermodynamic Profiles for the Unfolding of Intramolecular Hairpins and Junctions

Transition	TM (°C)	ΔHcal (kcal/mol)	TΔScal (kcal/mol)	ΔG°(5) (kcal/mol)	ΔCp (cal/mol-K)
GAAATT-Hp
	37.4	42.9	38.4	4.5	140
Dumbbell
1st	65.9	38.8	31.8	7.0	−600
2nd	75.3	33.9	27.1	6.8
Total		72.7	58.9	13.8
1C-Dumbbell
1st	60.5	37.3	31.1	6.2	−720
2nd	73.3	33.8	27.1	6.7
Total		71.1	58.2	12.9
3H
1st	33.6	33.7	30.6	3.1	290
2nd	45.8	38.2	33.3	4.9
3rd	70.3	45.7	37.0	8.7
Total		117.6	100.9	16.7
2HS1H
1st	25.9	6.1	5.7	0.4	660
2nd	31.7	32.6	29.7	2.9
3rd	52.9	48.2	41.1	7.1
4th	68.8	23.7	19.3	4.4
Total		110.6	95.8	14.8
HS12HS1
1st	30.3	49.0	44.9	4.1	890
2nd	54.1	18.4	15.6	2.8
3rd	75.6	46.6	37.2	9.4
Total		114.0	97.7	16.3
HS1HS1HS1
1st	29.4	29.2	26.8	2.4	1250
2nd	53.1	53.5	45.6	7.9
3rd	69.4	29.4	23.9	5.5
Total		112.1	96.3	15.8

All experiments were done in 10 mM NaPi buffer, 0.1 M NaCl at pH 7.0. Experimental errors are as follows: T_M (±0.5°C), ΔH_cal (±5%), TΔS (±5%), ΔG° ₍₅₎ (±7%), and ΔC_p (±20%). NA, not applicable.

A cytosine bulge in the dumbbell causes slight destabilization

The DSC thermogram of Dumbbell (Fig. 3, middle) displays only one peak, which cannot be fitted by a single curve. Fitting yields two transitions at 65.9 and 75.3°C and a total unfolding enthalpy of 72.7 kcal/mol. Thus, the two stems of Dumbbell unfold sequentially. In 1C-Dumbbell (Fig. 3, bottom), where a cytosine bulge was placed between the stems, there are two T_Ms at 60.5 and 73.3°C with a total ΔH_cal of 71.1 kcal/mol. Not only are the T_Ms lower than in Dumbbell but there is also greater separation between them. For both oligonucleotides, the first transition is assumed to be the GTGC stem and the second is assumed to be the GCGC stem, as the addition of an A/T basepair decreases the thermal stability. It can be concluded that addition of a cytosine bulge in a dumbbell structure causes destabilization, with the bulge having a greater impact on the GTGC stem than the GCGC stem, as evidenced by its lower T_M and enthalpy.

A single cytosine bulge in a three-way junction causes slight destabilization

The unfolding of 3H (Fig. 4, top) was determined in a previous study (42) and is described here for the unfolding in 0.1 M NaCl. The unfolding occurs triphasically at 0.1 M NaCl, with transitions at 33.6, 45.8, and 70.3°C. The enthalpy of these transitions in conjunction with the predicted thermal stability based on sequence (59) indicates that the first transition (ΔH_cal of 33.7 kcal/mol) corresponds to the unfolding of the GAAATT stem (having the most A/T basepairs would yield a significantly lower thermal stability than the other two stems), whereas the second transition (ΔH_cal of 38.2 kcal/mol) pertains to the unfolding of the GTGC stem (having only one A/T basepair would yield a thermal stability approximately in between the more thermally unstable GAAATT stem and the stable GCGC stem). The third transition (ΔH_cal of 45.7 kcal/mol) corresponds to the unfolding of the GCGC stem. At this salt concentration, all three transitions unfold separately, as evidenced by the three transitions with enthalpies corresponding roughly to the enthalpies obtained from nearest-neighbor calculations of the individual hairpins (61).

2HS₁H (Fig. 4, middle top) unfolds tetraphasically, with T_Ms of 25.9, 31.7, 52.9, and 68.8°C and a total enthalpy of 110.6 kcal/mol. Although the data must be fit with four transitions because of the shape, the first transition has a ΔH_cal of 6.1 kcal/mol, which is less than a single basepair stack (59). This may be due to stacking interactions caused by the cytosine bulge. The next three transitions are the GAAATT stem, the GTGC stem, and then the GCGC stem, as determined based on their sequence compositions (the more A/T basepair stacks, the more thermally unstable the stem will be). The main observation is that there is a slight decrease in thermal stability, and after removing contributions from the first minor transition, there is an overall decrease in the enthalpy of 13.1 kcal/mol. This is greater than one basepair stack and is likely the disruption of the center stack of the dumbbell structure within 2HS₁H as well as destabilization of the surrounding basepair stacks.

Two cytosine bulges disrupt the sequential unfolding of 3H

HS₁2HS₁ (Fig. 4, middle bottom) unfolds triphasically with T_Ms of 30.3, 54.1, and 75.6°C. The total ΔH_cal of 114.0 kcal/mol is within experimental error of 3H, indicating that the bulges do not disrupt the total number and/or strength of the basepair stacks. It can be seen in Fig. 4 (middle bottom) that HS₁2HS₁ has two almost entirely separate unfolding events, one at 30.3°C and one at 75.6°C. The most likely cause of this is that the bulges cause extreme destabilization of the GTGC stem. The reasoning for this is that the first unfolding event in HS₁2HS₁ occurs at 30.3°C, whereas GAAATT-Hp unfolds at 37.4°C. This same stem melts at 33.6 and 31.7°C in 3H and 2HS₁H, indicating that this stem is destabilized by the cytosine bulges but not significantly compared to the other junctions. The second major transition in HS₁2HS₁ occurs at 75.6°C, approximately the same temperature as the second transition of Dumbbell (75.3°C) and the transition corresponding to the GCGC stem in 3H (70.3°C) and 2HS₁H (68.8°C). This indicates that the bulges do not significantly alter these two stems; it is only the central transition corresponding to the GTGC stem that is now missing. The enthalpy of both transitions is slightly higher than that of a single stem, indicating that more than the individual stems is melting in each transition. This can be explained as the unfolding of the GTGC stem over a wide temperature range. In essence, the bulges allow the GAAATT stem to unfold with gradual unfolding of the GTGC stem until ∼70°C, at which point the GCGC stem unfolds.

HS₁HS₁HS₁ melts similarly to 2HS₁H

HS₁HS₁HS₁ (Fig. 4, bottom) unfolds triphasically with T_Ms of 29.4, 53.1, and 69.4°C. The total ΔH_cal, 112.1 kcal/mol, indicates that the total number and/or strength of the basepair stacks is the same. All three transitions have substantial enthalpy, indicating the sequential unfolding of the three stems. It can be seen from the T_Ms and the shape of the melting curve for HS₁HS₁HS₁ that despite having three cytosine bulges, HS₁HS₁HS₁ has a melting profile similar to what was seen for 2HS₁H. The last two transitions, corresponding to the GTGC and GCGC stems, have T_Ms and total enthalpy similar to those of 1C-Dumbbell, indicating that the cytosine bulge is causing a slight disruption of the dumbbell structure in terms of the thermal stability. Interestingly, a third cytosine bulge rescues the GTGC stem from the instability seen in HS₁2HS₁.

Cytosine bulges cause a significant increase in the binding of counterions

The differential binding of sodium ions was calculated using Eq. 1. The (∂T_M/∂ln[Na⁺]) term was obtained from the slope of the T_M dependence on salt concentration graph shown in Fig. S1. The ΔH_cal/RT_M² term was obtained by analyzing DSC thermograms at several salt concentrations. The Δn_Na+ values are summarized in Table 2 and Fig. 5. For most oligonucleotides, their unfolding is associated with a release of counterions due to the lower charge density parameter of unfolded nucleotides (62). This is the case for all molecules in this study. The Δn_Na+ term has been normalized per helical phosphate in column 4 by considering the total number of helical phosphates, including the two loop phosphates adjacent to each stem. GAAATT-Hp has a Δn_Na+ value of 0.088 mol Na⁺/mol hPi, which is consistent with previously obtained values of simple hairpins.

Table 2.

Differential Binding of Counterions, Water, and Protons for the Unfolding of the Hairpins and Junctions

	∂TM/∂ln[Na+]	ΔnNa+ (per mol)	ΔnNa+ (per hPi)	∂TM/∂ln[aW]	ΔnW (per mol)	ΔnW (per bp)	∂TM/∂pH	ΔnH+ (per mol)
GAAATT-Hp
	4.2	1.1	0.088	108.3	26	4	−5.29	0.52
Dumbbell
	5.7	2.0	0.120	112.6	39	5	1.42	−0.22
1C-Dumbbell
1st	4.0	1.03	NA	103.6	21	NA	NA	NA
2nd	5.2	0.62	NA	143.2	21	NA	NA	NA
Total	NA	1.65	0.103	NA	42	5	0.48	−0.07
3H
1st	2.4	0.97	NA	79.2	25	NA	−5.07	0.76
2nd	8.1	1.86	NA	135.4	29	NA	−0.58	0.06
Total	NA	2.83	0.098	NA	54	4	NA	0.82
2HS1H
1st	7.1	1.32	NA	101.6	28	NA	−5.94	0.86
2nd	8.6	3.10	NA	90.2	20	NA	0.44	−0.04
Total	NA	4.42	0.158	NA	48	3	NA	0.82
HS12HS1
1st	7.9	2.29	NA	73.1	16	NA	−4.12	0.39
2nd	9.7	0.92	NA	35.0	3	NA	−3.97	0.15
3rd	8.4	1.85	NA	102.9	22	NA	−0.05	0.01
Total	NA	5.06	0.187	NA	41	3	NA	0.55
HS1HS1HS1
1st	6.3	1.19	NA	0	0	NA	−4.61	0.40
2nd	10.9	2.54	NA	0	0	NA	−1.75	0.11
3rd	9.1	1.43	NA	0	0	NA	−0.93	0.07
Total	NA	5.16	0.198	NA	NA	0	NA	0.58

Experimental errors are as follows: Δn_H+ (±12%), Δn_Na+ (±12%), and Δn_W (±12%). NA, not applicable.

Figure 5.

Effect of cytosine bulges on (top) Δn_ion, (middle) Δn_W, and (bottom) ΔC_p.

Dumbbell has a value of 0.120 mol Na⁺/mol hPi, which is less than what has been obtained for DNA polymers, i.e., 0.17 mol Na⁺/mol phosphate (63, 64, 65), but more than that of a single hairpin, indicating that Dumbbell is a continuous helix (42, 45). 1C-Dumbbell has a Δn_Na+ value of 0.103, which is less than that of Dumbbell, likely because of partial disruption of the continuous helix by the cytosine bulge. In contrast, by adding the cytosine bulges, the Δn_Na+ value of the 3H derivates increases dramatically. 3H has a value of 0.098 mol Na⁺/mol hPi, which is slightly more than GAAATT-Hp but less than Dumbbell. However, even by adding a single cytosine bulge in 2HS₁H, there is a dramatic increase in the counterion release during unfolding with a value of 0.158 mol Na⁺/mol hPi. These values increase even more for HS₁2HS₁ and HS₁HS₁HS₁. The most likely explanation is that three-way junctions are sterically strained, which has been previously shown to be true (20, 21, 66). Unlike four-way junctions, which can adopt multiple conformations that relieve strain around the junction site, three-way junctions are more rigid. By placing cytosine bulges at the junction site, greater mobility is allowed, leading to a more tightly packed junction. The higher charge density around the junction then leads to higher counterion uptake.

Cytosine bulges cause a decrease in the amount of immobilized water molecules

The differential binding of water was calculated using Eq. 2. The (∂T_M/∂ln[a_W]) term was obtained from the slope of the curves displaying T_M dependencies on ethylene glycol concentration, taken from the analysis of UV-melting curves shown in Fig. S2. The average ΔH_cal/RT_M² term was obtained from analysis of DSC thermograms at several ethylene glycol concentrations. The unfolding of each oligonucleotide is accompanied by a release of water molecules due to a shift in the equilibrium toward a conformation that has a lower hydration state (55, 67). The calculated Δn_W values are shown in Table 2 and Fig. 5 and are normalized per basepair in column 7. It can be seen from the control molecules GAAATT-Hp and Dumbbell that, on average, there are four to five water molecules released per basepair. This value is unchanged for 1C-Dumbbell, indicating that the bulge affects counterion binding but not water immobilization. 3H also releases four water molecules per basepair. However, when adding a single nucleotide bulge for 2HS₁H, the Δn_W value decreases, releasing only three waters. The Δn_W value of HS₁2HS₁ is even further reduced, although this value averages out to three waters per basepair. However, HS₁HS₁HS₁ had no measurable T_M dependence on ethylene glycol concentration (Fig. S3), indicating that there are no, or very few, immobilized structural water molecules around this junction and that release is not measurable. It is possible that the high density of bound counterions is displacing the structural waters such that none or very few are immobilized during folding. However, it should be stated that the osmotic stress method measures only structural and not electrostricted water molecules. The decrease in structural water molecules may be counterbalanced by an increase in electrostricted waters, which would be bound by the increased amount of counterions or merely displaced by the much higher uptake of counterions compared to standard double-helical nucleic acids.

Cytosine bulges do not cause a change to proton uptake

The differential binding of protons was calculated using Eq. 3. The (∂T_M/∂pH) term was obtained from the slopes of the T_M dependence on pH curves, which were determined by analysis of UV melting curves at different pHs, as seen in Fig. S4. The average ΔH_cal/RT_M² term was taken from DSC thermograms at different pHs. The resulting Δn_H+ values can be seen in the last column of Table 2. Although protonation was not expected, as cytosines are not expected to be protonated in double-helical structures, there was a possibility that the bulges would form base triplets. In addition, the lack of water immobilization prompted other avenues of investigation. It can be seen that the GAAATT-Hp has a release of protons when unfolding, which is unexpected because it is a standard double-helical nucleic acid and should not be protonated as a hairpin. However, it does have a cytosine loop that could be protonated. This release of counterions is highlighted in Fig. S5, which displays the DSC thermograms of GAAATT-Hp, Dumbbell, and 1C-Dumbbell at different pHs. GAAATT-Hp is stabilized by low pH to a similar degree as higher salt. Both Dumbbell and 1C-Dumbbell have an uptake of protons upon unfolding, although the value for 1C-Dumbbell is so small as to be negligible. This is consistent with what is expected of double-helical nucleic acids. 3H and 2HS₁H have identical Δn_H⁺ values, i.e., 0.82 mol H⁺/mol oligomer. This is a release similar to GAAATT-Hp, although slightly higher. It can be seen from the values per transition that the proton release predominantly comes from the first transition, which corresponds to the unfolding of the GAAATT stem in the three-way junctions. This can also be seen in Fig. S6, which shows the DSC thermograms of the junctions at different pHs. From these thermograms, it can be seen that the first transition containing the GAAATT stem is stabilized by low pH, but the second transition is unaffected. Thus, it is not the protonation of the cytosine loop in GAAATT-Hp that is releasing protons, as the GAAATT stem of the three-way junction does not contain the cystosine loop. Regardless of the reason for protonation, it can be concluded that a single cytosine bulge does not change the release of protons during unfolding compared to 3H. In a similar vein, HS₁2HS₁ and HS₁HS₁HS₁ have nearly identical values of 0.55 and 0.58 mol H⁺/mol oligomer. These values mirror what was obtained for GAAATT-Hp, indicating that proton uptake comes from the GAAATT stem and is not affected by cytosine bulges.

Cytosine bulges increase the heat capacity of the oligonucleotide

The heat capacity of an oligonucleotide can provide information about the folded state of the molecule. Previous research has determined that, on average, the heat capacity of a single basepair is ∼30–80 cal/K-mol (68, 69, 70, 71). Because the heat capacity is dependent on the type of water molecules hydrating an oligonucleotide, the folded state of the molecule will dictate the hydration state, and thus, the heat capacity will change depending on the folded structure. Standard double-helical oligonucleotides typically have a positive heat capacity, although there are exceptions (45, 60). This is because hydration by nonpolar waters has a positive effect on the heat capacity, which is larger than the negative effect on the heat capacity caused by hydration of polar waters (72, 73, 74, 75). The surface area of an oligonucleotide has both nonpolar and polar areas of similar size, and when unfolded, there is exposure of both polar and nonpolar groups to solvent (76). It can be seen in Table 1 that the heat capacity of 3H is small for its size, smaller even than that of Dumbbell. This is likely because of the strain at the junction; 3H is poorly folded, exposing more polar groups to the solvent. As the number of cytosine bulges at the junction is increased, there is a linear increase in the heat capacity of the system (Fig. 5, bottom). This is due to the strain relief of the bulges. As the number of bulges increases, the conformations available to the junction increase, and the junction becomes more tightly folded, i.e., it makes complete basepair stacks with all possible neighboring nucleotides, sequestering more hydrophobic groups from solvent, leading to a more positive heat capacity.

A larger heat capacity, i.e., sequestration of hydrophobic groups from solvent, is not necessarily related to stability. The folded state of the molecule is linked to enthalpy, but enthalpically there is no statistically significant change when increasing the number of cytosine bulges. This could be because although stacks at the junction are now solvent inaccessible, the individual contacts are weaker, leading to a slight reduction in enthalpy offset by the higher number of basepair stacks. This is especially possible when considering that a bulge will naturally weaken the interaction of neighboring nucleotides. Even when the bulge base is flipped out and the nucleotides on either side form a complete stack, this stack will be weaker than if the bulge were not present. Thus, the overall enthalpy remains the same, but there is a greater sequestration of hydrophobic groups from solvent. In addition, although the T_Ms appear to decrease, this is not precisely related to the heat capacity. T_Ms are related to the entropy; an increase in cytosine bulges increases the flexibility of the stems and increases the associated counterions, thus altering the entropy of the system. Flexibility and entropy are not directly related to the folded state of the molecule with regard to the amount of exposed nucleotide, and therefore, a molecule can have lower thermal stability while simultaneously sequestering more hydrophobic groups.

Conclusions

We have reported here a complete thermodynamic description of the unfolding of intramolecular three-way junctions, including the differential binding of counterions, water, and protons. Our results suggest that the addition of bulges at the junction does not significantly impact the unfolding profiles of intramolecular junctions at physiological salt concentrations.

Standard thermodynamic profiles, including entropy and free energy, for the formation of each molecule are summarized in Table 1. The ΔG°₍₅₎ and TΔS terms are estimated at 5°C, where all the molecules are in the helical state. Inspection of Table 1 indicates that the folding of each molecule is accompanied by a favorable free-energy term, which results from the characteristic compensation of favorable enthalpy and unfavorable entropy contributions (77). The favorable enthalpy term is caused by the formation of basepairs and basepair stacks, whereas the unfavorable entropy term is due to the order of the strands, which is minimized for intramolecular complexes, and the immobilization of water, protons, and counterions. In the case of the three-way junctions with bulges, the total free energy and entropy of folding follow the same trend as that of the enthalpy: 3H > HS₁2HS₁ > HS₁HS₁HS₁ > 2HS₁H. HS₁2HS₁ and HS₁HS₁HS₁ have lower enthalpy, free energy, and entropy compared to 3H, but these values are not statistically significant.

The bulge effect on the differential binding of counterions, water, and protons was also investigated for all oligonucleotides. It was seen in this study that adding cytosines increases the amount of counterions bound, most likely at the junction site. This is possibly due to the bulges allowing the junction to adopt various strain-relieving conformations that lead to a more tightly folded oligonucleotide at the junction, resulting in a higher-charge density parameter. This is in line with the results of the heat capacity values, which increase linearly with the number of cytosines, indicating that the junctions are adopting a more tightly folded conformation as the number of cytosine bulges is increased, sequestering the hydrophobic groups of the nucleotide bases from solvent. In opposition to the higher counterion uptake, there is a decrease in the amount of immobilized structural water as the number of cytosines increases. Because the osmotic stress method measures only structural water, the high amount of counterions likely displaces structural water molecules. Thus, it is likely that the unmeasured electrostricted waters increase as the number of cytosine bulges increase. Finally, the differential binding of protons was measured to ensure that triplet basestacks were not forming from the cytosine bulges. It was determined that triplet basestacks were not forming, as the cytosine bulges do not increase the binding of protons. Instead, the GAAATT-Hp was found to bind protons both as a single hairpin and as part of the junctions, indicating that the stem sequence and not the loop binds protons.

Author Contributions

L.A.M. designed the experiments and assisted with writing the main article. C.E.C. performed the experiments and wrote the main article.

Editor: David Lilley.