Melting Behavior of a DNA Four-Way Junction Using Spectroscopic

Sep 18, 2017 - Chemical industry criticizes agency for limiting DINP. Utility sues chemical makers. Maryland water supplier alleges several firms cons...
1 downloads 15 Views 2MB Size
Download PDF
Article Cite This: J. Am. Chem. Soc. 2017, 139, 14443-14455

pubs.acs.org/JACS

Melting Behavior of a DNA Four-Way Junction Using Spectroscopic and Calorimetric Techniques Carolyn E. Carr and Luis A. Marky* Department of Pharmaceutical Sciences, University of Nebraska Medical Center, 986025 Nebraska Medical Center, Omaha, Nebraska 68198-6025, United States S Supporting Information *

ABSTRACT: Intramolecular four-way junctions are structures present during homologous recombination, repair of double stranded DNA breaks, and integron recombination. Because of the wide range of biological processes four-way junctions are involved in, understanding how and under what conditions these structures form is critical. In this work, we used a combination of spectroscopic and calorimetric techniques to present a complete thermodynamic description of the unfolding of a DNA four-way junction (FWJ) and its appropriate control stem-loop motifs (Dumbbell, GAAATT-Hp, CTATC-Hp, GTGC-Hp, and GCGC-Hp). The overall results show that the four-way junction increases the cooperative unfolding of its stems, although the reason for this is unclear, as the arms do not unfold as coaxial stacks, and thus its melting behavior cannot be accurately described by its control molecules. This is in contrast to what has been seen for two- and three-way junctions. In addition, the lack of base stacking and the ΔHvH/ΔHcal ratio seen at low salt indicate the four-way junction exists as a mixture of conformations, one of which is most likely the open-X structure which has unpaired bases at the junction. This was confirmed by single value decomposition of CD and UV spectra. This indicates that at low salt there is a third spectroscopically distinct species, while at higher salt there are only two species, folded and unfolded. Based on the enthalpy, Δnion, and ΔnW, the dominant folded structure at high salt is most likely the antiparallel stacked-X structure.

■

INTRODUCTION Intramolecular structures within single stranded DNA and RNA can be formed from perfect or imperfect inverted repeats (IR).1−5 These IR are common within DNA and RNA, and the varied structures formed (hairpins, intramolecular junctions, and cruciforms) are critical for a variety of biological processes including replication and gene regulation.6−10 Understanding the process by which these structures fold and the conditions that facilitate folding will allow us to better understand how they control gene regulation. A complete physical description of the folding is needed to accurately predict folding conditions and structures; this will lead to better development of targeted therapies. The physical parameters that dictate folding are dependent on base-pairing, base-stacking, hydration, and ion binding of the folded oligonucleotide structures. Four-way junctions are known to exist in two different conformations, open-X and stacked-X (Figure 1).5,10−14 The first crystal structures of a DNA-only stacked-X junction was published by Ortiz-Lombardia et al. in 199915 while the open-X junction, in complex with the recombinase Cre, was published by Guo et al. in 1997. 16 The open-X conformation predominates at low salt and consists of a fully extended conformation with no coaxial stacking due to electrostatic repulsion from the lack of counterions. The other conformation is the stacked-X structure which involves coaxial stacking, with one arm stacked onto another arm to form a continuous duplex. The stacked-X conformation exists either in the parallel © 2017 American Chemical Society

Figure 1. Crystal structures of DNA in the stacked-X (left, PDB ID 1P54) and open-X (right, PDB ID 3CRX) conformations.

or antiparallel orientation which refers to the 5′−3′ direction of the two coaxial arms in relation to each other. While the openX structure does not predominate at higher salt, it is thought to exist as an intermediate when flipping between the parallel and antiparallel orientations of the stacked-X structure, which exist in equilibrium in solution. The most well-known four-way junction is the Holliday junction which is a key intermediate during homologous recombination and repair of DNA doublestranded breaks and adopts an antiparallel stacked-X Received: June 20, 2017 Published: September 18, 2017 14443

DOI: 10.1021/jacs.7b06429 J. Am. Chem. Soc. 2017, 139, 14443−14455

Article

Journal of the American Chemical Society structure.4,12,17−20 There is a large host of proteins involved in the recognition and resolution of Holliday junctions during these critical biological processes.13,19−21 However, cruciform structures are thought to be involved in a variety of other biological processes. Stacked-X cruciform structures are thought to restrict the movement of plasmids, reducing the possibility of distant sites coming into contact, i.e. preventing intermolecular contacts of distant genes that are critical for their function.11,22 During rolling circle-replication a cruciform is extruded from the plasmid, projecting the nick site that begins replication.23−25 Cruciform formation over promotor sites is thought to be a common regulatory mechanism for transcription, as studies have shown numerous proteins, such as p53, bind to four-way junctions and that cruciform structures do not hinder transcription.1,3,7,8 This method of regulation has been confirmed in the N4 virion, where cruciform formation over promoters is critical for transcription of early genes.1,26,27 Cruciforms are also formed during integron recombination, which were first discovered for their role in conveying antibiotic resistance.28−32 The diagram of the four-way junction in Scheme 1 is an example of an immobile four-way junction. This structure is

the folding of the four-way junction cannot be adequately described by its control molecules. The four-way junction folds in such a way as to promote the cooperative unfolding of the stems, significantly affecting their thermal stability, although the reason for this is unclear; it is clear that this thermal affect is not dictated by the coaxial stacking of the arms. In addition, at low salt, our results suggest that FWJ adopts an intermediate structure, presumed to be an observed rapid conformational change between its two dominant forms (open and antiparallel stacked); at higher salt, no intermediate is detected. High salt has been shown to cause the stacked-X structure to predominate, and our experiments are unable to detect the minor conformational changes that are occurring. In addition, our results suggest that the junction of FWJ does not cause an increase in salt and water uptake compared to the control hairpins, which would otherwise need to be factored in when targeting such structures biologically.

■

MATERIALS AND METHODS

Materials. The oligonucleotides and their designations (Scheme 1): d(GAAATTC5AATTTC), GAAATT-Hp; d(CCTATCT5GATAGC), CTATC-Hp; d(CGTGCT5GCACA), GTGC-Hp; d(TGCGCT5GCGCC), GCGC-Hp; d(GCGCT5GCGCGTGCT5GCAC), Dumbbell; d(GAAATTGCGCT5GCGCCTATCT5GATAGGTGCT5GCACAATTTC), FWJ. All DNA molecules were synthesized by IDT (Coralville, IA), purified by reverse-phase HPLC, and further desalted by size exclusion chromatography using a G-10 Sephadex column. The molecules were then lyophilized to dryness prior to use in experiments. The sequences of the four-way junctions and the control hairpins and their hypothesized structures are shown in Scheme 1. The solution concentration of each oligomer was determined optically at 260 nm and 90 °C using the molar absorptivities, in mM−1 cm−1 of strands, reported as follows: 161.4 (GAAATT-Hp), 158.1 (CTATC-Hp), 132.8 (GTGC-Hp), 126.9 (GCGC-Hp), 222.9 (Dumbbell), 490.8 (FWJ). These values were obtained by extrapolation from the molar absorptivities at 25 °C of the tabulated values of the dimer and monomer bases to high temperatures, using a previously reported procedure.46,47 All measurements were performed in buffer solutions consisting of 10 mM sodium phosphate buffer at pH 7.0, adjusted with different salt concentrations up to 0.2 M NaCl. All chemicals used in this study were reagent grade and used without further purification. Temperature-Dependent UV Spectroscopy (UV Melting Curves). Absorbance versus temperature profiles were measured at 268 nm with a thermoelectrically controlled Aviv 14 DS UV/vis spectrophotometer (Lakewood, NJ); this wavelength allows monitoring of the unstacking of both GC and AT base pairs. The spectra were scanned at a heating rate of ∼0.6 °C/min from 1 to 100 °C. Shape analysis of the melting curves yielded transition temperatures, TM, which corresponds to the inflection point of the helix−coil transitions and van’t Hoff enthalpies, ΔHvH, which were obtained using the following relationship: ΔHvH = (2n + 2)RTM2(∂α/∂T)T=TM, where n is the transition molecularity, α is the fraction of strands in the helical state, and (∂α/∂T)T=TM is determined from the slope of the melting curve measured around the TM.48 We investigated the dependence of TM as a function of strand concentration in order to determine the molecularity of the transition(s) of each DNA molecule, over at least a 10-fold range of total strand concentration (2−80 μM). If the TM is independent of strand concentration it indicates a monomolecular or intramolecular transition, while the TM’s of higher molecularities depend on strand concentration.48 Additional UV melting curves were obtained as a function of salt and ethylene glycol concentration to determine the differential binding of ions and water, which is discussed later. UV spectra for FWJ as a function of temperature in 10 mM NaPi at pH 7.0 and with the addition of 0.2 M NaCl were obtained in a 1 cm path length cuvette at 2 °C intervals from 15 to 95 °C. Data was

Scheme 1. Cartoon of the Sequences of the DNA Four-Way Junction and Control Molecules

obviously an integral part of important biological processes and understanding the physical properties that dictate its folding mechanism is critical for better understanding how and when these structures are formed during biological processes. Significant work has been done on understanding how Holliday junctions are resolved during homologous recombination and DNA breaks13,17,19−21 and on detecting cruciform structures in vivo,33−39 but little has been done to understand the type of stacking involved and interaction between the arms. Current understanding of intramolecular junctions is based on the structure and stability of DNA and RNA obtained from previous thermodynamic investigations of their helix−coil transitions.40−43 The primary focus of this research is to understand the folding/unfolding of a single-stranded DNA oligomer that is designed to adopt an intramolecular four-way junction. Use of a single-stranded DNA oligomer is favored due to a lower entropy penalty that gives rise to transition temperatures higher than those that occur for a bimolecular interaction.44,45 This allows investigation of the physical properties of its 100% helical conformation over a wider range of temperatures. Here we present a thermodynamic description of the unfolding of a DNA intramolecular four-way junction. A combination of spectroscopic and calorimetric techniques was used to investigate the folding/unfolding thermodynamics of a four-way junction and appropriate control stem-loop motifs. Our results indicate that, unlike two- and three-way junctions, 14444

DOI: 10.1021/jacs.7b06429 J. Am. Chem. Soc. 2017, 139, 14443−14455

Article

Journal of the American Chemical Society

reasonableness of the outputted data was used to determine the best fit of the data. Determination of the Differential Binding of Counterions and Water. The helical and coil states of an oligonucleotide are associated with a different number of bound ions and water molecules and therefore the helix−coil transition is associated with an uptake of counterions and water during folding. UV melting curves were obtained as a function of salt concentration to determine the differential binding of counterions, ΔnNa+. This ΔnNa+ linking number is measured experimentally using the following relationship:52−55

obtained from 200 to 350 nm with an Agilent 8453 UV−vis spectrometer equipped with a programmable thermostat. The concentration of the sample was taken before and after the experiment to factor in any evaporation that may have occurred. The data was buffer subtracted, normalized based on concentration, and zeroed at 350 nm, which was at the baseline for all samples. Temperature-Dependent Fluorescence Spectroscopy. Fluorescence measurements were obtained using an Aviv ATF107 spectrofluorometer (Lakewood, NJ) equipped with a Peltier temperature control system. The fluorescence emission spectra of the 2aminopurine modified oligonucleotides were obtained from 325 to 420 nm at several temperatures with an excitation wavelength of 307 nm. The emission and excitation slits were set at 4 nm. Fluorescence melting curves were obtained by scanning from 5 to 95 °C at a heating rate of 1 °C/min and a temperature equilibration time of 1 min. Thermodynamic parameters were obtained from the melting curves using previously described procedures.48 Circular Dichroism (CD) Spectroscopy. Simple inspection of the CD spectra allowed analysis of the conformation of each oligonucleotide at temperatures at which the structure of each oligonucleotide is 100% helical. The CD spectra were obtained using an Aviv 202SF CD spectrometer (Lakewood, NJ) equipped with a Peltier temperature control system from 320 to 220 nm in 1 nm increments. CD melting curves were obtained at ∼250 nm, a wavelength that corresponds to base pair stacking of each individual oligonucleotide. The experimental CD melting curves were converted to α-curves with “α” being the fraction of strands in the helical state. Analysis of the CD melting curves yielded TM’s and ΔHvH using procedures reported previously.48 CD spectra for FWJ as a function of temperature in 10 mM NaPi at pH 7.0 and with the addition of 0.2 M NaCl were obtained in a 0.1 cm path length cuvette at 2 °C intervals from 15 to 95 °C. Data was obtained from 320 to 220 nm with an Aviv 202SF CD spectrometer (Lakewood, NJ) equipped with a Peltier temperature control system. The concentration of the sample was taken before and after the experiment to factor in any evaporation that may have occurred. The data was buffer subtracted, normalized based on concentration, and zeroed at 320 nm, which was at the baseline for all samples. The spectra at each temperature is an average of two scans which were determined to be identical before averaging. Differential Scanning Calorimetry (DSC). Heat capacity functions of the unfolding of each oligonucleotide were measured with a VP-DSC from Malvern MicroCal (Northampton, MA). These thermograms were obtained with a temperature ramp of ∼0.6 °C min−1 with oligomers ranging in concentration from 40 to 200 μM in total strands. The DSC thermograms were buffer subtracted, normalized by strand concentration, and deconvoluted using a non two-state zero ΔCp model with the MicroCal software. Analysis of the thermograms yielded TM’s and model-independent thermodynamic profiles using the following relationships:48,49 ΔHcal = ∫ ΔCp(T) dT; ΔScal = ∫ ΔCp(T)/T dT, and the Gibbs equation: ΔG°(T) = ΔH − TΔS; where ΔCp is the heat capacity of the oligonucleotide solution during the unfolding process, ΔHcal and ΔScal are the unfolding enthalpy and entropy, respectively, both assumed to be temperatureindependent and ΔG°(T) is the free energy at a temperature T. Alternatively, ΔG°(T) can be calculated using the equation ΔG°(T) = ΔHcal(1 − T/TM). The ΔHvH terms were also obtained from the DSC thermograms and the ΔHvH/ΔHcal ratio provides information about the nature of the transition.48,50 Single Value Decomposition (SVD) Analysis. The UV or CD spectra obtained as a function of temperature were assembled into a matrix with each row consisting of the spectroscopic signal at each wavelength and each column consisting of the spectroscopic signal at each temperature over 2 °C increments. Each matrix was analyzed by single value decomposition (SVD) using the Olis GlobalWorks data analysis package.51 In order to determine the number of species present, the eigenvector plots were inspected and used as a guide for determining the appropriate fitting mechanism. The standard deviations of the fits, the distribution of the residuals, and the

Δn Na+ = 1.11(ΔHcal /RTM 2)[∂TM /∂ln[Na +]]

(1)

The first term in parentheses (ΔHcal/RTM2) is a constant determined directly from DSC experiments, whereas the term in brackets is determined experimentally from the slopes of the plots of TM as a function of the concentration of salt, ranging from 16 to 216 mM. The 1.11 value in eq 1 is a constant, used to convert the activity of Na+ into its concentration term, i.e., (Na+) = 0.9[Na+], where 0.9 is the activity coefficient over this range of salt concentration. UV melting curves were also obtained as a function of osmolyte concentration to determine the differential binding of water, ΔnW. The ΔnW linking number is measured experimentally using the following relationship:

Δn w = (ΔHcal /RTM 2)[∂TM /∂ln[a w]]

(2)

The first term in parentheses (ΔHcal/RTM2) is a constant determined directly from DSC experiments, whereas the term in brackets is determined experimentally from the slopes of the plots of TM as a function of ethylene glycol concentration, ranging from 0 to 2.5 M.56 Ethylene glycol is a cosolute that does not interact specifically with DNA. The osmolality of these solutions was obtained with a UIC vapor pressure osmometer model 830 which was calibrated with standardized solutions of NaCl. These osmolalities were converted into water activities (aW) using the following relationship:57 ln[aW] = −Osm/MW where Osm is the solution osmolality and MW is the molality of pure water (55.5 mol/kg of H2O). The measurement of the binding of water and counterions assumes that the helical and coil states of oligonucleotides have similar types of binding. This is a good assumption for counterions but may not be correct for water, since the helical state of DNA binds primarily electrostricted water and the random coil state binds structural water.

■

RESULTS AND DISCUSSION The folding/unfolding of an intramolecular four-way junction (FWJ) is investigated to understand how these structures are formed during biological processes. The folding of the FWJ was confirmed to be intramolecular by obtaining UV melting curves as a function of strand concentration. Varying the strand concentration did not alter the melting temperature (TM) of the junction, consistent with the formation of a monomolecular complex. Changes in base-pair stacking were monitored by UV and CD spectroscopy and the conformation of the oligonucleotides was confirmed by simple inspection of the CD spectra. We investigated the unfolding of all molecules using DSC spectroscopy and confirmed the equilibrium of the FWJ unfolding event by obtaining the folding thermogram using DSC. We then obtained UV and CD spectra at 2 °C increments and performed SVD analysis for the purpose of discerning intermediates in the unfolding of the FWJ. We performed fluorescence spectroscopy in order to determine the TM of the individual arms of FWJ and we obtained the differential binding of counterions and water during the folding of these molecules using the combined results from UV and DSC analysis in order to see if the junction caused a significant effect on the immobilization of sodium or water. 14445

DOI: 10.1021/jacs.7b06429 J. Am. Chem. Soc. 2017, 139, 14443−14455

Article

Journal of the American Chemical Society Table 1. TM’s and ΔHvH Obtained from UV and CD Melting Curvesa UV transition

TM (°C)

first second

40.6 (55.3) N/A (66.3) 58.6 (71.8) 30.9 (42.0) 41.0 (54.4) 59.6 (69.1) 71.4 (78.4)

CD ΔHvH (kcal/mol) FWJ −58 ± 9 (−65 ± 10) N/A (−39 ± 6) Dumbbell −30 ± 5 (−39 ± 6) GAAATT-Hp −40 ± 6 (−37 ± 6) CTATC-Hp −39 ± 6 (−36 ± 5) GTGC-Hp −42 ± 6 (−42 ± 6) GCGC-Hp −43 ± 6 (−44 ± 6)

TM (°C)

ΔHvH (kcal/mol)

53.3 N/A

−27 ± 4 N/A

63.3

−25 ± 4

30.5

−37 ± 6

39.8

−25 ± 4

57.8

−31 ± 5

70.2

−35 ± 5

All experiments were done in 10 mM NaPi buffer at pH 7.0. Experimental errors are as follows: TM (±0.5 °C), ΔHvH (±15%). Values in parentheses correspond to experiments done in 10 mM NaPi buffer, 0.2 M NaCl at pH 7.0. a

All Molecules Folded Intramolecularly. Typical UV melting curves in 10 mM sodium phosphate (NaPi) buffer at pH 7.0 are shown in Figure S1 (top and middle). All the melting curves are sigmoidal and show cooperative unfolding of base-pairs and base-pair stacks. The TM’s for the transitions of each oligonucleotide remains constant across a 10-fold increase in strand concentration (Figure S2) which is consistent with intramolecular folding. This intramolecular folding is important, as it reduces the entropy cost and supports the complete folding of the molecule. While some unstable sequences may have incomplete folding profiles across a measurable temperature range even for a monomolecular transition, intramolecular structures should allow observation of all possible transitions. The melting curves for the hairpins and Dumbbell have TM’s increasing from 30.9 °C (GAAATT-Hp) < 41.0 °C (CTATCHp) < 58.6 °C (Dumbbell) < 59.6 °C (GTGC-Hp) < 71.4 °C (GCGC-Hp). Based on nucleic acid sequence, Dumbbell should unfold at a TM in-between GCGC-Hp and GTGC-Hp. The low experimental TM of Dumbbell indicates that incorporation of the hairpins into a single structure decreases the thermal stability by ∼6.9 °C. Shape analysis of the UV melting curves yielded averaged ΔHvH values of 40 kcal/mol (GAAATT-Hp), 39 kcal/mol (CTATC-Hp), 42 kcal/mol (GTGC-Hp), 43 kcal/mol (GCGC-Hp), and 30 kcal/mol (Dumbbell). The ΔHvH, or unfolding of the cooperative unit, for Dumbbell is less than half of the predicted total enthalpy value, indicating that the two domains of Dumbbell melt simultaneously as a whole unit instead of as two sequential transitions. This has been seen previously in studies of doublehairpin structures.58 The van’t Hoff enthalpy is determined from the shape of an optical melting curve,48 so a broad transition would yield a small enthalpy while a sharp transition yields a larger enthalpy. This can be seen in the derivative spectra, where the hairpins have broader peaks than that of FWJ, which has a sharp transition that yields a higher ΔHvH (Table 1). Because the transitions of both sides of the Dumbbell melt as a single transition, the ΔHvH is much smaller than the expected total enthalpy obtained from the breaking of base-pair stacks, ΔHcal, which is obtained from DSC (Table 2). Because UV is not sensitive enough to distinguish two transitions less than 15 °C apart, it is possible the two transitions are separated and are simply not observable using

this technique. At higher salt, all three parameters follow the same order for the control molecules. FWJ Alters the Thermal Stability of Its Stems, Causing It to Unfold As a Condensed Unit. Based on its stem sequences, FWJ is expected to have four separate transitions. However, analysis of the UV melting curve shows only one resolved transition with a TM of 40.6 °C and a ΔHvH of 58 kcal/ mol. The TM is approximately that of the CTATC-Hp, with a ΔHvH that is higher than that of one hairpin but much lower than what would be expected for four hairpins, an indication that all four transitions are melting as one large unit. This would indicate stabilization of the GAAATT stem and destabilization of the GTGC and GCGC stems. Increasing the salt concentration to 0.2 M NaCl causes the TM to increase to 55.3 °C with a ΔHvH of 65 kcal/mol. At this salt concentration, an additional transition becomes visible with a TM of 66.3 °C and a ΔHvH of 39 kcal/mol. Based on sequence it is likely that this second transition corresponds to the GCGC stem. The first transition maintains a similar ΔHvH while the second transition has a ΔHvH corresponding to that of one hairpin. In total, the magnitude of the enthalpy accounts for roughly three of the four arms and is lower than the expected enthalpy based on nearest-neighbor parameters (Table 2).59 The enthalpy values indicate that the fourth arm is either melting simultaneously with one of these two transitions, or is completely unformed. While the second possibility is highly unlikely, due to the intramolecular formation leading to a reduced entropy cost and stabilization by the higher salt concentration, it cannot be ruled out based on UV analysis. Figure S3 highlights the discrepancy between the experimentally observed single transition of FWJ and the expected number of transitions by comparing FWJ with combined UV curves of all four hairpins (termed Fit 1 for the remainder of the paper) in addition to the combined curve of Dumbbell with GAAATT-Hp and CTATC-Hp (termed Fit 2 for the remainder of the paper) at low (Figure S3 top) and high (Figure S3 bottom) salt concentrations. At both salt concentrations, there is a large difference between the transitions in the combined curves of the control molecules as compared to the FWJ, indicating that factors other than the base-pair stacks of the stems are influencing the structure. In both graphs, the combined curves melt over a large temperature range in contrast to the sharp transition of FWJ, indicating that the 14446

DOI: 10.1021/jacs.7b06429 J. Am. Chem. Soc. 2017, 139, 14443−14455

Article

Journal of the American Chemical Society

difference between the hyperchromicity observed in melting curves and ellipticity of base-pair stacks in CD spectra. UV melting curves are a visible representation of the melting of individual transitions, while the CD spectra are a cumulation of static spectra and thus are more comparable. The CD melting curves of all molecules in 10 mM NaPi buffer at pH 7.0 are shown in Figure S4 in terms of α as a function of temperature. In the case of the control molecules, the TM from CD matches those obtained from UV. Analysis of the FWJ melting curve yielded one transition, like the UV analysis, with a TM of 53.3 °C, which was corroborated by the differentiated curve, and a ΔHvH of 27.3 kcal/mol. The melting temperature and enthalpy obtained from CD is significantly different from that obtained from UV analysis. Derivation of the UV melting curves at low salt reveal a shoulder at higher temperature; fitting this shoulder gives a TM of 53 °C (Figure S5), the TM obtained from the CD melting curve. Because CD appears to be observing both transitions in the raw curve, the TM is much higher than what is seen in UV. However, because the transition seen in CD is so broad, it yields a much smaller ΔHvH than what was seen in the UV melting curves. Calorimetric Unfolding of a Four-Way Junction as a Function of Salt Concentration. DSC curves are shown in Figure 3 for each oligonucleotide and the resulting thermody-

structure of FWJ is significantly altering the thermal stability of its arms. This indicates that the actual structure of the four-way junction may be different than what is shown in Scheme 1 and that increased salt concentration, which mimics biological conditions, does not alter this discrepancy although it does increase base-pair stacking based on the hyperchromicity values (21% at low salt and 23% at high salt). FWJ Has Less Base Pair Stacking Contributions than the Sum of Its Component Molecules. The top and bottom panels of Figure 2 show the CD spectra of all molecules at 5

Figure 2. (Top) CD spectra of GAAATT-Hp (black), CTATC-Hp (red), GTGC-Hp (green), GCGC-Hp (blue), and Dumbbell (orange) at 5 °C in 10 mM NaPi at pH 7.0. (Bottom) CD spectra at 5 °C of FWJ in 10 mM NaPi at pH 7.0 (solid line), FWJ with the addition of 0.2 M NaCl (short dash), GAAATT-Hp + CTATC-Hp + GTGC-Hp + GCGC-Hp (dashed line) and Dumbbell + GAAATT-Hp + CTATC-Hp (dotted line).

°C. At this temperature, all oligonucleotides are fully folded and both visible bands are approximately equivalent in magnitude, indicating that all molecules have a B-like conformation. All spectra show a positive band centered at ∼280 nm which is related to the sugar puckering of the DNA, and a negative band at ∼250 nm which corresponds to the extent of base-pair stacking. When all molecules present adopt a similar secondary structure (stem-loop motifs vs G-quadruplex), the magnitude of the band at 250 nm can be viewed in a similar manner as hyperchromicity, with a more positive band indicating less contributions from base-pair stacking. For instance, Dumbbell, as the largest of the control molecules, is expected to have the most base-pair stacking which is evident in Figure 2 (top, orange). As expected, the magnitude of FWJ is much greater than any of the control molecules. Figure 2 (bottom) shows the CD spectrum of FWJ compared against Fit 1 and Fit 2 at low salt. The combined spectra are similar to each other, but compared to FWJ at 250 nm are significantly more negative signifying FWJ has a lower number of base-pair stacks, presumably at the junction. This is an interesting contrast to the UV results, where Figure S3 shows that FWJ has a higher hyperchromicity value than the control molecules, indicating a greater number of base-pair stacks. It is unclear why there is significant difference, although it could be related to the

Figure 3. (Top) DSC thermograms of FWJ in 10 mM NaPi at pH 7.0 at different salt concentrations; (middle) DSC thermograms in 10 mM NaPi at pH 7.0; (bottom) DSC thermograms in 10 mM NaPi, 0.2 M NaCl, pH 7.0 of GAAATT-Hp (black), CTATC-Hp (red), GTGC-Hp (green), GCGC-Hp (blue), and Dumbbell (orange).

namic parameters are shown in Table 2. The unfolding of each molecule undergoes highly reproducible monophasic or biphasic transitions. The TM’s from the calorimetric melts obtained at higher strand concentrations are like those measured from UV melts, consistent with their intramolecular formation. We obtained a total unfolding enthalpy, ΔHcal, of 134.9 kcal/mol at low salt and 148.8 kcal/mol at 0.2 M NaCl concentration for FWJ. The value at high salt is in excellent agreement with the nearest-neighbor enthalpy of 156.8 kcal/ 14447

DOI: 10.1021/jacs.7b06429 J. Am. Chem. Soc. 2017, 139, 14443−14455

Article

Journal of the American Chemical Society Table 2. Thermodynamic Profiles for the Folding of Intramolecular Hairpins and Junctionsa buffer condition

transition

TM (°C)

ΔHcal (kcal/mol)

ΔHNN (kcal/mol)

ΔHvH (kcal/mol)

ΔG°5 (kcal/mol)

TΔScal (kcal/mol)

−104 −30

−7.5 −10.0 −17.5 −13.1 −11.8 −24.9

−55.5 −61.9 −117.4 −73.2 −57.7 −130.9

−37

−6.6 −6.3 −12.9 −14.1

−36.8 −27.7 −64.5 −57.8

−40 −40

−6.3 −6.9

−32.0 −29.9

−43 −43

−7.2 −8.4

−30.2 −39.0

−35 −35

−5.1 −5.9

−38.7 −34.3

−42 −40

−3.6 −4.7

−40.4 −36.3

FWJ 16 mM Na+

216 mM Na+

first second total first second total

42.7 49.9 54.6 69.5

−63.0 −71.9 −134.9 −86.3 −62.5 −148.8

−156.8 −88 −43 Dumbbell

16 mM Na+

216 mM Na+

first second total

54.8 68.0 72.9

−43.4 −34.0 −77.4 −71.9

16 mM Na+ 216 mM Na+

60.0 69.7

−38.3 −36.8

16 mM Na+ 216 mM Na+

71.4 80.4

−37.4 −39.4

16 mM Na+ 216 mM Na+

41.3 53.1

−43.8 −40.2

16 mM Na+ 216 mM Na+

29.7 40.6

−44.0 −41.0

−39 −45 −75.8 GTGC-Hp −30.5 GCGC-Hp −33.9 CTATC-Hp −37.8 GAAATT-Hp −39.1

All experiments were done in 10 mM NaPi buffer at pH 7.0. Experimental errors are as follows: TM (±0.5 °C), ΔHcal (±5%), ΔHvH (±15%), TΔScal (±5%), ΔG°(5) (±7%). a

ΔHcal value indicates ∼2−3 unformed base-pair stacks which accounts for the difference in experimentally determined versus calculated ΔHcal. The second transition is expected to consist of the GTGC and GCGC stems and the ΔHcal value of the second transition of FWJ matches both the experimentally and theoretically determined values for these stems. Upon increasing the salt concentration to 0.1 M NaCl (Figure 3 (top)), both transitions are shifted to higher temperatures but the increase in TM is different for each transition, leading to higher separation between transitions. Both transitions are further resolved upon increasing the salt concentration to 0.2 M NaCl. The ΔHcal of the first transition matches the predicted enthalpy for the CTATC and GAAATT stems, indicating that higher salt has increased base stacking around the junction leading to fully folded stems. However, the increase may also be due to part of the GTGC or GCGC stems melting in the first transition, signified by the decrease in ΔHcal for the second transition. Nevertheless, the total enthalpy is greater than in low salt buffer indicating stabilization of base-pair stacks by salt. However, if the unfolding is dictated by sequence as described above, then coaxial stacking is not affecting the TM’s and something else unique to the structure of FWJ is leading to significant alterations to the thermal stabilities. The ΔHvH/ΔHcal ratio can tell us about the nature of each transition observed; a ratio of 0.9−1.1 indicates a two-state, all or none transition. The ΔHvH/ΔHcal ratio of the first transition at low salt is >1, indicating aggregation60,61 or more than one equilibrium is involved,62 while the ratio of the second transition is