J. Phys. Chem. B 2008, 112, 4833-4840
4833
Unfolding Thermodynamics of DNA Pyrimidine Triplexes with Different Molecularities Hui-Ting Lee,† Santiago Arciniegas,† and Luis A. Marky*,†,‡,§ Department of Pharmaceutical Sciences, Department of Biochemistry and Molecular Biology, and Eppley Institute for Cancer Research, UniVersity of Nebraska Medical Center, 986025 Nebraska Medical Center, Omaha, Nebraska 68198-6025 ReceiVed: NoVember 15, 2007; In Final Form: January 17, 2008
Nucleic acid oligonucleotides (ODNs), as drugs, present an exquisite selectivity and affinity that can be used in antigene and antisense strategies for the control of gene expression. In this work we try to answer the following question: How does the molecularity of a DNA triplex affect its overall stability and melting behavior? To this end, we used a combination of temperature-dependent UV spectroscopy and calorimetric (differential scanning calorimetry) techniques to investigate the melting behavior of DNA triplexes with a similar helical stem, TC+TC+TC+T/AGAGAGA/TCTCTCT, but formed with different strand molecularity. We determined standard thermodynamic profiles and the differential binding of protons and counterions accompanying their unfolding. The formation of a triplex is accompanied by a favorable free energy term, resulting from the typical compensation of favorable enthalpy-unfavorable entropy contributions, i.e., the folding of a particular triplex is enthalpy driven. The magnitude of the favorable enthalpy contributions corresponds to the number and strength of the base-triplet stacks formed, which are helped by stacking contributions due to the incorporation of dangling ends or loops. Triplex stability is in the following order: monomolecular > bimolecular > trimolecular; this is explained in terms of additional stacking contributions due to the inclusion of loops. As expected, acidic pH stabilized all triplexes by allowing protonation of the cytosines in the third strand; however, the percentage of protonation increases as the molecularity decreases. The results help to choose adequate solution conditions for the study of triplexes containing different ratios of CGC+ and TAT base triplets and to aid in the design of oligonucleotide sequences as targeting reagents that could effectively react with mRNA sequences involved in human diseases, thereby increasing the feasibility of using the antisense strategy for therapeutic purposes.
Introduction The antigene and antisense strategies are two approaches that use oligonucleotides (ODNs) to modulate gene expression.1,2 In the antigene strategy, an ODN binds to the major groove of a DNA duplex forming a triple helix,3,4 whereas in the antisense strategy ODNs bind to messenger RNA forming a DNA/RNA hybrid duplex.5,6 In both cases what is needed is the formation of stable triplex or duplex molecules, which is dictated by the energy provided by the formation of base-triplet and base-pair stacks, respectively. For instance, the formation of a triplex can potentially inhibit transcription by competing with the binding of proteins that activate the transcriptional machinery.7 Triplex formation is sequence-specific, allowing the recognition of different DNA sequences by Hoogsteen base pairing.3,4 In general, the affinity in the stable formation of a triplex is provided by base stacking contributions, hydrogen bonding between the third strand bases and purine bases of the duplex, and electrostatic and hydration contributions.8-10 Triplexes can be classified based on the composition and/or orientation of the third strand. Parallel triplexes, or triplexes of the “Pyrimidine” motif are pyrimidine-rich strands that bind parallel to the purine strand of the duplex, whereas antiparallel triplexes, or triplexes of the “Purine” motif are purine-rich strands that bind antipar* To whom correspondence should be addressed. Tel.: (402) 559-4628. Fax: (402) 559-9543. E-mail:
[email protected]. † Department of Pharmaceutical Sciences. ‡ Department of Biochemistry and Molecular Biology. § Eppley Institute for Cancer Research.
allel to the duplex purine strand.3,11,12 Pyrimidine triplexes are characterized by the formation of T•A/T and C•G/C+ base triplets (“•” and “/” represent Watson-Crick and Hoogsteen base pairing, respectively). These triplets are isomorphous, i.e., superimposable, allowing a regular conformation of the sugarphosphate backbone.3 However, the stabilities of Pyrimidine triplexes containing exclusively TAT base triplets are pHindependent and salt-dependent, whereas those with exclusively CGC+, or combinations of these base triplets, are pH-dependent and salt-independent.12,13 Therefore, in order to develop ODNs as therapeutic agents and to know how ODNs can be used to target nucleic acid molecules, it is important to have a clear understanding on how the sequence, base composition, and solution conditions affect the stability of a nucleic acid triple helix. To this end, we have investigated the temperature unfolding of a set of four triplexes with a common triple-helical sequence and different molecularities.14-16 We determined standard thermodynamic profiles, ion and proton binding for each triplex. The results show that triplex stability follows the order monomolecular > bimolecular > trimolecular, and triplex formation is enthalpy driven. Materials and Methods Materials. All ODNs were synthesized by the Core Synthetic Facility of the Eppley Research Institute at the University of Nebraska Medical Center, HPLC-purified, and desalted by column chromatography. The solution concentration of each ODN was determined at 260 nm and 90 °C using the following
10.1021/jp710926h CCC: $40.75 © 2008 American Chemical Society Published on Web 03/22/2008
4834 J. Phys. Chem. B, Vol. 112, No. 15, 2008
Lee et al. thermoelectrically controlled Aviv 14-DS or a Perkin-Elmer Lambda-10 spectrophotometer, as a function of strand, pH, and salt concentration. The temperature was scanned at a heating rate of ∼0.6 °C/min. These melting curves allow us to measure transition temperatures, TM, which are the midpoint temperatures of the order-disorder transition, and van’t Hoff enthalpies, ∆HvH, from analysis of the shape of these melting curves (intramolecular transitions) and/or from the TM dependence on strand concentration for intermolecular transitions. We used the following relationships and procedures that have been described earlier:22
Figure 1. Cartoon of the structures and designations of triplexes and control duplexes.
∆HvH ) (2n + 2)RTM2(∂R/∂T)
(1)
molar extinction coefficients (in mM-1 cm-1 of strands): d(AGAGAGAC5TCTCTCTC5TCTCTCT),258;d(AGAGAGAC5TCTCTCT), 176; d(TCTCTCT), 54; d(TCTCTCTC5TCTCTCT), 145;d(AGAGAGA),92;d(CCTCTCTCT),66;andd(AGAGAGA C5TCTCTCTCC), 191. These values were calculated by extrapolation of the tabulated values of the dimer and monomer bases at 25 °C to high temperatures, using procedures reported earlier.17,18 All experiment were done using a 10 mM sodium phosphate buffer solution at pH 6.2 and adjusted to the appropriate sodium concentration with NaCl. All other chemicals, from Sigma Chemicals, were of reagent grade and used without further purification. Design of Triplexes with Different Molecularities. The sequence of all molecules and their corresponding designations are shown in Figure 1. The basis set of four triplexes contain a common triple-helical stem with sequence TC+TC+TC+T/ AGAGAGA/TCTCTCT. The 2:1 mixture of the pyrimidine and purine strands, respectively, yielded a trimolecular triplex, whereas the inclusion of one or two cytosine loops, joining two strands, yielded triplexes with lower molecularities. For instance, the 1:1 mixture of the homopurine strand with an ODN that is formed by joining two pyrimidine strands with a loop yielded the bimolecular BiAG7 triplex, whereas the inclusion of an additional cytosine loop to the 3′-end of the purine strand to join this other strand yielded a monomolecular triplex (MonoAG7). An additional triplex similar to BiHAG7 (Figure 1) but with two cytosines at the 3′- and/or 5′-ends (BiHAG7-CC) was designed to mimic the cytosine loop of the third strand, allowing us to determine the stacking contributions of the loop, if any.19-21 In addition, a set of two control molecules is included (Figure 1) to determine the overall thermodynamic contributions of their duplex component. Overall Experimental Protocol. We initially used temperature-dependent UV spectroscopy (UV melts) to investigate the temperature unfolding of each triplex as a function of strand concentration to determine their transition molecularity from the dependence of their transition temperature (TM) on strand concentration. Then, differential scanning calorimetry (DSC) was used to determine standard thermodynamic profiles from analysis of their heat capacity functions and the Gibbs equation.22 Additional UV melting was obtained at a salt concentration of 0.2 M NaCl as a function of pH, and at pH 6.2 as a function of salt concentration, to determine the TM dependences on pH and salt concentration, respectively. These TM dependences together with their corresponding unfolding DSC enthalpies are used to determine the differential binding of protons and counterions between their helical and coil states. Temperature-Dependent UV Spectroscopy. Absorbance versus temperature profiles (UV melting curves) for each triplex and their control molecules were measured at 260 nm with a
1/TM ) [(n - 1)R/∆HvH] ln(CT/2n) + ∆S/∆HvH
(2)
where R is the universal gas constant, R is the fraction of single strands in the helical state, and (∂R/∂T) is the slope of the R versus T curve measured around the TM; n is the molecularity of the transition, equal to 1, 2, or 3 for monomolecular, bimolecular, and trimolecular transitions, respectively. The TM’s of multiphasic melting curves were obtained from the peaks of the corresponding differential melting curves. To determine the molecularity of the transition(s) of each triplex, we carried out additional UV melting curves as a function of the total strand concentration. The TM for the unfolding of an intramolecular DNA structure is independent of strand concentration, whereas the TM of a DNA complex with higher molecularity is strand concentration-dependent. Differential Scanning Calorimetry. To thermodynamically investigate the helix-coil transition of each molecule, excess heat capacity functions (DSC melts) were measured with a Microcal VP-DSC differential scanning microcalorimeter (Northampton, MA). Two cells, the sample cell containing 0.7 mL of ODN solution and the reference cell filled with the same volume of buffer solution, were heated from 0 to 90 °C at a heating rate of 0.75 °C/min. Analysis of the resulting thermograms yielded standard thermodynamic profiles (∆Hcal, ∆Scal, and ∆G°cal). Each thermodynamic parameter was obtained from the following relationships, using procedures described previously:22 ∆Hcal ) ∫∆Cpa dT and ∆Scal ) ∫(∆Cpa/T) dT, where ∆Cpa represents the anomalous heat capacity during the unfolding process. The free energy change at any temperature, ∆G°T, is obtained from the Gibbs equation: ∆G°T ) ∆Hcal - T∆Scal. Shape analysis of the experimental DSC curves allows us to determine ∆HvH’s according to the relationship ∆HvH ) A/[(1/ T1) - (1/T2)], where A is a constant, equal to 7.0 (monomolecular transition), 10.14 (bimolecular), and 12.9 (trimolecular);22 T1 and T2 correspond to the lower and upper temperatures, respectively, at the half-height width of the DSC curve. The ∆HvH/∆Hcal ratio allows us to inspect if triplex or duplex unfolding takes place through two-state transitions or through the formation of intermediates. If the ∆HvH/∆Hcal ratio is in the range of 0.9-1.1 the transition takes place in an all-ornone fashion.22 Differential Thermodynamic Binding of Protons and Counterions. UV melting curves were carried out as a function of salt concentration and pH to measure the differential thermodynamic binding of counterions, ∆nNa+, and protons, ∆nH+, between the folded and unfolded states, respectively. These two linking numbers were determined with the following relationships:23
Unfolding Thermodynamics of DNA Triplexes
J. Phys. Chem. B, Vol. 112, No. 15, 2008 4835
∆nH+ ) (∆Hcal /RTM2)[dTM /d ln(H+)] ) -0.43(∆Hcal /RTM2)[dTM/d pH] (3) ∆nNa+ ) (∆Hcal/RTM2)[dTM/d ln(Na+)] ) 1.1(∆Hcal/RTM2)[dTM/d ln[Na+]] (4) The first term in parenthesis of eqs 3 and 4 was measured directly from DSC experiments, whereas the second term in brackets was measured from the slopes of the dependences of the TM on salt and pH, respectively; 1.1 is a factor for converting solution activities into concentrations, whereas 0.43 is the factor that converts natural logarithms into decimal logarithms. Results and Discussion Triplex Design. All ODNs were designed to form triplexes with a common triple-helical domain of three CGC+ base triplets interspaced by four TAT base triplets; therefore, their stability should depend on pH and more or less be independent of salt. The main experiments were conducted under optimized salt (0.2 M) and pH (6.2) conditions to stabilize both their short duplex component, such as AG7dup and TriAG7, and to protonate the three cytosines of the third strand.13,24-28 We also expect that if all six base-triplet stacks (TAT/C+GC and C+GC/TAT) formed 100%, their enthalpy contributions should be similar after appropriate subtraction of stacking contributions of the loops and/or dangling ends.13 We also expect that triplexes with lower molecularity to be more stable because of their lower entropy penalty.10,13 Temperature-Dependent UV Spectroscopy. UV melting experiments as a function of strand and salt concentrations and pH were used to characterize the helix-coil transition of the helical molecules shown in Figure 1. Figure 2a shows typical UV melts for each triplex. All transitions in these curves follow the characteristic sigmoidal behavior for the unfolding of a nucleic acid helical structure. The UV melting curves of BiAG7 are monophasic, BiHAG7 and TriAG7 exhibit clear biphasic transitions, whereas MonoAG7 shows a very broad first transition followed by a sharp second transition. However, the overall melting behavior of each molecule changes with solution conditions, as will be discussed in a later section. UV melts over a total strand concentration range of 6-70 µM were obtained to determine the transition molecularities of each molecule; the corresponding TM dependences are shown in Figure 2, parts b and c. The TM’s of the main transition of MonoAG7 and the second transition of BiHAG7 do not depend on strand concentration (Figure 2b), indicating that these transitions are intramolecular, whereas the TM’s of the single transition of BiAG7, BiHAG7 (first transition), and the two transitions of TriAG7 do depend on strand concentration (Figure 2c), indicating that these triplexes formed intermolecularly. The TM’s (at a total helical concentration of ∼80 µM) are shown in the first column of Table 1. The thermal stability of the first or single transitions of these triplexes are in the following order: BiHAG7 < TriAG7 < BiAG7 < MonoAG7, whereas the thermal stability of the second transition of TriAG7 is much lower than that of BiHAG7. We were able to obtain ∆HvH’s for the majority of the transitions; however, there were some biphasic transitions that were close together or transitions that took place at low temperatures without initial baselines that did not allow for this determination. For these reasons we have averaged the ∆HvH’s determined from the shape of the UV melts and/or from the TM dependence on strand concentration, using eqs 1 and 2,
Figure 2. (a) Typical UV melting curves of triplexes in 10 mM sodium phosphate buffer at pH 6.2, 0.2 M NaCl, and at the indicated helical concentration: MonoAG7 (b, 2.6 µM), BiAG7 (9, 3.4 µM), BiHAG7 (2, 23 µM), and TriAG7 (1, 28 µM). (b) TM dependences on strand concentration for intramolecular transitions: MonoAG7 (b), the second transition of BiHAG7 (∆), and AG7hp ([); (c) TM dependences on strand concentration for intermolecular transitions: BiAG7 (9), the first transition of BiHAG7 (2), and the first (1) and second (3) transition of TriAG7.
respectively. We obtained ∆HvH’s (data not shown) for the transition(s) of each triplex as follows: ∼50 and 36 kcal/mol (TriAG7), 48 and 47 kcal/mol (BiHAG7), 93 kcal/mol (BiAG7), and 42 kcal/mol (MonoAG7). On the other hand, the ∆HvH values for the control duplexes are ∼36 kcal/mol (AG7Dup) and 40 kcal/mol (AG7hp). The main observation is that the total ∆HvH’s for the overall unfolding of the three intermolecular triplexes range from 86 to 95 kcal/mol, which is higher than the ∆HvH range of 36-40 kcal/mol of their duplex components, confirming the formation of triple-helical structures at low temperatures. The enthalpy value for the intramolecular triplex (40 kcal/mol) is unexpectedly low and may be explained in terms of a lower melting cooperativity at this salt concentration.10,13 The comparison of the TM and ∆HvH values of each triplex with those of their duplex component, together with their TM dependences on strand concentration and observed number of transitions, allow us to explain their melting behavior, as follows.
4836 J. Phys. Chem. B, Vol. 112, No. 15, 2008
Lee et al.
TABLE 1: Thermodynamic Profiles for the Folding of Triplexesa TM °C
∆Hcal kcal/mol
∆HvH kcal/mol
first second total
31.2 52.5
-13.6 -110 -124.6
-47.1 -39.1
single
39.0
-105
transition
∆G°(5) kcal/mol
∆nH+ mol H+ per mol
MonoAG7 -12.4 -94.0 -106.4
-1.2 -16.0 -17.2
-3.3 ( 0.2
1.0 ( 0.2
BiAG7 -93.5
-11.4
-2.1 ( 0.2
-1.0 ( 0.2
BiHAG7 -37.2 -40.3 -77.5
-1.2 -7.8 -9.0
-1.3 ( 0.9 0 ( 0.2
0.6 ( 0.4 -1.5 ( 0.3
T∆Scal kcal/mol
-101
∆nNa+ mol Na+ per mol
first second total
13.9 59.0
-41.3 -48.1 -89.4
-46 -41.9
first second total
27.2 59.6
-52.1 -44.8 -97.0
-64 -41
BiHAG7-CC -48.2 -37.4 -86.2
-3.9 -7.4 -10.7
-2.1 ( 0.2 -0.2 ( 0.2
0.3 ( 0.1 -1.2 ( 0.4
first second total
18.4 32.0
-51.0 -38.0 -89.0
-71 -40
TriAG7 -48.7 -34.0 -82.7
-2.3 -4.0 -5.2
-1.5 ( 0.9 -0.1 ( 0.1
-1.2 ( 1.0 -1.4 ( 0.8
single
59.1
-44.9
-42
AG7hp -37.6
-7.3
-0.1 ( 0.1
-1.1 ( 0.3
single
26.4
-37.8
-39.9
AG7dup -35.1
-2.7
0.1 ( 0.1
a
All parameters measured in a 10 mM sodium phosphate buffer with 0.2 M NaCl, at pH 6.2, except AG7dup that was done at pH 7. Experimental errors are shown in parentheses: TM ((0.5 °C), ∆HvH ((10%), ∆Hcal ((5%), T∆Scal ((5%), and ∆G°(5) ((7%). The TM’s correspond to an extrapolated helical concentration of 80 µM. The larger errors shown in some of the values of ∆nH+ and ∆nNa+ are due to the poor resolution of the associated melting curves, making it difficult to obtain good values of the TM dependences on pH and salt, respectively.
TriAG7 and BiHAG7 unfold in biphasic transitions, corresponding to the sequential bimolecular removal of the third strand followed by the intermolecular and intramolecular unfolding of their duplex component, respectively. MonoAG7 unfolds in a monophasic transition that corresponds to the simultaneous removal of the third strand and unfolding of its duplex component. Interestingly, BiAG7 also unfolds through a monophasic transition that corresponds to the simultaneous removal of the second and third strand, consistent with its bimolecular formation. Overall, the unfolding of each triplex takes place according to their design. However, the single transition of BiAG7 is unexpected; one contributing factor is the double degeneracy of the pyrimidine strand, yielding an entropy effect. This degeneracy is due to the sequence symmetry of both pyrimidine and purine strands, forming triplexes with loops on the 5′-end or on the 3′-end of the purine strand; in addition, each of these triplexes can form Watson-Crick base pairs followed by the cooperative formation of Hoogsteen base pairs or vice versa. The net effect is a favorable free energy contribution of 1 kcal/mol. Furthermore, the transitions of MonoAG7 can be understood in terms of the enthalpy-entropy coupling reported previously.13 Differential Scanning Calorimetry. Figure 3 shows the calorimetric unfolding for each of the main triplexes and control duplexes. The unfolding of BiAG7 and MonoAG7 triplexes takes place mainly through monophasic transitions, TriAG7 and BiHAG7 unfold in biphasic transitions, whereas the control duplexes (AG7hp and AG7dup) show monophasic transitions; the curves are consistent with their UV melting curves. All DSC curves show similar pre- and post-transition baselines, indicating that triplex, or duplex, unfolding takes place with negligible heat capacity effects. The TM’s for the removal of the third strand follows the same trend as the TM’s of the UV melts and are in the following order (Table 1): BiHAG7 (13.9 °C) < TriAG7 (18.4 °C) < BiAG7 (39 °C) < MonoAG7 (52.5 °C). Furthermore, BiHAG7-CC unfolds similarly to BiHAG7 (DSC curve
not shown), but its first transition has a higher TM, by 13 °C. Integration of the DSC curves yielded endothermic enthalpies (Table 1 shows these enthalpies for the formation of each triplex) ranging from 89 kcal/mol (TriAG7) to 125 kcal/mol (MonoAG7), which are larger than the enthalpies of the control duplexes that ranged from 38 kcal/mol (AG7dup) to 45 kcal/mol (AG7hp). The comparison of the triplex versus duplex enthalpies yields heat contributions ranging from 51 to 79 kcal/mol, which correspond to the removal of the third strand. This enthalpy range can be explained in terms of additional stacking contributions due to both the incorporation of loops that might close helical frayed ends and base-base stacking contributions between the helical ends and the loops. The enthalpies for the unfolding of the duplexes are in excellent agreement with the enthalpy value of 40 kcal/mol predicted from nearest-neighbor parameters at 0.2 M NaCl.29,30 However, the enthalpies of the triplexes are lower than the predicted 138 kcal/mol, which has been determined using monomolecular triplexes in 10 mM NaCl;10,13 one exception is MonoAG7 that is similar within experimental error. The differences with the other triplexes can be explained in terms of the strength of each triplex, i.e., the absence of one of the loops yields a higher exposure to the solvent of the base triplets at the end without the loop, reducing base-triplet stacking. In addition, the use of higher salt concentration (0.2 M) stabilizes the duplex state and generates a weaker triplex,16 yielding a lower magnitude of the unfolding enthalpy, which is manifested indirectly with heat capacity effects.13 To determine the enthalpy contributions for the inclusion of cytosine loops, we compare the enthalpies for pairs of triplexes. For instance, the comparison of BiHAG7 with TriAG7 yielded the enthalpy contribution for the inclusion of the loop connecting the two duplex strands; we obtained a negligible enthalpy contribution, which is explained in terms of the difference in duplex strength, i.e., an enthalpy compensation takes place between the stronger AG7hp duplex that generates a weaker triplex.16 The comparisons between BiHAG7 and BiHAG7-CC
Unfolding Thermodynamics of DNA Triplexes
J. Phys. Chem. B, Vol. 112, No. 15, 2008 4837
Figure 3. DSC melting curves for the set of four triplexes, as indicated in each panel, in 10 mM sodium phosphate at pH 6.2 and 0.2 M NaCl and their corresponding duplexes AG7hp (0) in BiHAG7 and AG7dup (O or 0).
and BiHAG7-CC and MonoAG7 should yield the contributions for the inclusion of the loop connecting the duplex and the third strand; the two cytosine dangling ends of BiHAG7-CC mimic the contributions of the cytosine loop. The enthalpy difference between BiHAG7-CC and BiHAG7 is 7.6 kcal/mol (Table 1); this suggests base-base stacking contributions from the cytosines dangling ends and/or the closing of the frayed end at the 5′-end of the purine strand. The enthalpy difference between MonoAG7 and BiHAG7-CC is 27.6 kcal/ mol (Table 1), which is too large to be explained in terms of just base-base stacking contributions of the cytosine loops. One possible explanation for the large enthalpy of MonoAG7 is that the association of the third strand brings the loop closer, making this triplex more compact, i.e., optimization of base-triplet stacking contributions takes place.13 This observation is supported by the effect of the second loop on triplex formation for the BiAH7 and BiHAG7 pair that yielded a large enthalpy difference of 15.6 kcal/mol. Furthermore, the TM of BiAG7 is in between the two control duplexes and its enthalpy is ∼64 kcal larger than the average enthalpy of these duplexes, indicating that the incorporation of a single loop, joining the pyrimidine strands, is optimizing base-triplet stacking, by the simultaneous formation of base pairs at both duplex and triplex states. Its high stability provides a strategy for the use of homopyrimidine strands in the targeting of mRNA with a large association free energy. Table 1 also shows van’t Hoff enthalpies determined from the shape of the DSC curves, which are similar to the ones obtained from UV melting curves within experimental error. To determine the ∆HvH/∆Hcal ratios, we have averaged the
Figure 4. (a) UV melting curves of triplexes in 10 mM sodium phosphate and 0.2 M NaCl and as a function of pH (5.8-7.0). (b) TM dependence on pH: first (b) and second (O) transition of MonoAG7, BiAG7 (9), first (2) and second (∆) transition of BiHAG7, first (1) and second (3) transition of TriAG7, and AG7hp ([).
optical and calorimetric ∆HvH’s for the transitions of each triplex. We obtained ∆HvH/∆Hcal ratios in the range of 0.91.1 for TriAG7 (second transition), BiAG7, BiHAG7 (both transitions), AG7dup, and AG7hp, indicating that these conformations unfold in two-state transitions without the presence of intermediate states. The ∆HvH/∆Hcal ratio of 1.2 for the first transition of TriAG7 is high because of its larger experimental error in the ∆HvH determination. The small and main transition of MonoAG7 has ∆HvH/∆Hcal ratios of 3.5 and 0.4, respectively; the lower value indicated a non-two-state transition with the presence of intermediate states, whereas the unusual large ratio is due to a low ∆Hcal value that corresponds to the partial unfolding of this triplex.13,22 UV Melts as a Function of pH and Differential Binding of Protons. UV melting curves as a function of pH at the total Na+ concentration of 216 mM are shown in Figure 4a. All transitions involving the removal of the third strand are shifted to higher temperatures when the pH is decreased; this is consistent with protonation of the cytosines in the third strand. However, the overall melting behavior of each triplex as a function of pH depends on its molecularity, and the hyperchromicity value at the lowest pH qualitatively indicates the strength of a particular triplex. The intermolecular triplexes, TriAG7 and BiHAG7, unfold in single transitions at pH 7, and gradually a second transition builds up at the lower temperature side as the pH is decreased, yielding overall triplex hyperchromicities of
4838 J. Phys. Chem. B, Vol. 112, No. 15, 2008 17% (Figure 4a). The unfolding behavior of BiAG7 shows a single transition over the entire pH range and a small hyperchromicity change from 20% to 22% (at pH 6.2). On the other hand, MonoAG7 unfolds in a biphasic transition at pH 7; lowering the pH shifts the first transition (third strand removal) to higher temperatures, whereas the second transition (duplex unfolding) remains the same; the triplex unfolds in two separate transitions at pH 5.8 with an overall hyperchromicity of 17.5%. This melting behavior of MonoAG7 is similar to its DSC melt. One observation based on triplex hyperchromicity suggests that the strength of these triplexes is in the following order: BiAG7 > MonoAG7 > TriAG7 ∼ BiHAG7. We used well-defined UV melting curves through the pH range of 5.8-7 to determine the slope of the TM dependences on pH. These TM dependences are shown in Figure 4b, which together with the enthalpy term of eq 3 allowed us to measure the ∆nH+ linking number. As expected, the control duplexes yielded ∆nH+ values (Table 1) of 0.1 ( 0.1 (AG7dp) and -0.1 ( 0.1 (AG7hp). This shows that the differential binding of protons in triplex folding is associated with the protonation of the third strand cytosines and, to a smaller extent, to cytosine protonation of the loops.31,32 Therefore, the ∆nH+ values have been determined using the TM and enthalpy for the association of the third strand; for instance, in the case of MonoAG7 and BiAG7, we subtracted the folding enthalpy of the duplex from the enthalpy of the triplex. At a total salt concentration of 216 mM Na+, the association of the third strand from each triplex is accompanied by a net uptake of protons, see Table 1: -3.3 ( 0.2 (MonoAG7) < -2.1 ( 0.2 (BiAG7 and BiHAG7-CC) < -1.5 ( 0.9 (TriAG7) < -1.3 ( 0.9 (BiHAG7); the last two values were very difficult to determine. The expected ∆nH+ value for 100% formation of this triple-helical sequence is -3, that lowers in magnitude to -2.7 at pH 6.2 when the pKa of cytosine is considered27 and increases in magnitude, by -0.2 per loop, if loops are involved. Overall, the closer the ∆nH+ value to -3, the better the formation of the triplex. Therefore, the ranking of the proton uptake of these triplexes showed that the lower their molecularity the higher the optimization of base triplets and base-triplet stacks. UV Melts as a Function of Salt Concentration and Differential Binding of Counterions. UV melting curves as a function of salt concentration at pH 6.2 are shown in Figure 5a. All triplex f duplex transitions are shifted to lower temperatures, whereas the duplex f random coil transitions are shifted to higher temperatures when the salt concentration is increased. This is consistent with the exclusion of counterions due to cytosine protonation and with the stabilization of the duplex state by salt. The overall melting behavior of each triplex as a function of salt also depends on triplex molecularity. In this range of salt concentration, the intermolecular triplexes, TriAG7 and BiHAG7, unfold in clear biphasic transitions, BiAG7 unfolds in monophasic transitions, whereas MonoAG7 unfolds biphasically, one broad transition at lower temperatures and a clear transition at higher temperatures. The increase in salt concentration decreases (by 1%) the overall triplex hyperchromicity to 13% (BiHAG7) and to 17% (MonoAG7), whereas increases (by 2-5%) the hyperchromicity to 13% (TriAG7) and to 21% (BiAG7). In spite of the complicated melting behavior as a function of salt concentration, the increase in salt concentration at pH 6.2 has a small effect on the stability of the triplex state. The TM dependences on salt concentration are shown in Figure 5b, which together with the enthalpy term of eq 4 allow us to measure the associated ∆nNa+ values, shown in Table 1.
Lee et al.
Figure 5. (a) UV melting curves of triplexes in 10 mM sodium phosphate at pH 6.2 and as a function of Na+ concentration (16-216 mM). (b) TM dependence on salt concentration: MonoAG7 (b), BiAG7 (9), first (2) and second (∆) transition of BiHAG7, first (1) and second (3) transition of TriAG7, and AG7hp ([).
However, the measurement of ∆nNa+ for each triplex is difficult because of the competition between counterion exclusion by the triplex state and counterion binding by the duplex state. This effect was reported earlier with a similar, but shorter, doublehairpin triplex with sequence TCTCTC5AGAGAC5TCTC.13 The ∆nNa+ value for the unfolding of the intramolecular AG7hp duplex at pH 7 yielded an uptake of counterions of -1.2 mol Na+ per mol or 0.10 mol Na+ per phosphate, which is characteristic of short ODNs.33 We were unable to determine this parameter for the AG7dp control duplex because of its short length and the range of salt concentrations used that yielded melting curves without prebaselines (data not shown). Furthermore, the 1:1 strand ratio of this duplex also yielded a small percentage of triplex formation at pH 6.2. The increase in the strand ratio to 2:1 to form a triplex shifted the melting curves to higher temperatures, yielding a ∆nNa+ value of -1.4 ( 0.8 mol Na+ per mol for its duplex state. On the other hand, the biphasic transitions of BiHAG7 and BiHAG7-CC allowed us to obtain, in mol Na+ per mol, counterion releases of 0.6 ( 0.4 and 0.3 ( 0.1 (triplex folding), and counterion uptakes of -1.5 ( 0.3 and -1.2 ( 0.4 (duplex folding), respectively; both sets of values are consistent with previous observations.13 The folding of the entire MonoAG7 triplex is accompanied by an overall counterion release of 1.0 ( 0.2 mol Na+ per mol, whereas the folding of BiAG7 yielded an overall uptake of counterions of -1.0 ( 0.2 mol Na+ per mol. This apparent discrepancy may be explained in terms of their melting
Unfolding Thermodynamics of DNA Triplexes cooperativity of their triplex and duplex domains, see the top two panels of Figure 5a. MonoAG7 unfolded in several sequential transitions, whereas BiAG7 unfolded in single transitions, perhaps due to its symmetric sequence. Complete Thermodynamic Profiles for the Folding of Triplexes. Standard thermodynamic profiles for the folding of each molecule at 5 °C are summarized in Table 1; at this temperature all 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 a favorable enthalpy and unfavorable entropy contributions.34 In general, favorable heat contributions correspond to the formation of base pairs and basetriplet stacks, whereas unfavorable entropy contributions arise from the unfavorable ordering of strands and the putative uptake of counterions, protons, and water molecules. The main observation is that all triplexes fold with larger exothermic enthalpies, relative to their corresponding control duplexes, which indicates the incorporation of a third strand. This is consistent with the overall observation that the folding of a helical structure is enthalpy driven.34 Specifically, the ∆G°(5) terms follow the order (Table 1) MonoAG7 < BiAG7 < BiHAG7 < TriAG7, which indicates that their favorable formation increases with decreasing its molecularity. The ∆Hcal terms are in the order TriAG7 (-89.0 kcal/mol) ∼ BiHAG7 (-89.4 kcal/ mol) > BiAG7 (-105 kcal/mol) > MonoAG7 (-124 kcal/mol). This order shows that the magnitude of the enthalpy depends on the number and strength of the base-triplet stacks that formed within each triplex and stacking contributions from the loops, i.e., the compactness of a particular triplex. For instance, the inclusion of a cytosine loop to AG7dup to yield AG7hp results in more favorable ∆G° and ∆Hcal contributions of 4.6 and 7.1 kcal/mol, respectively, whereas the inclusion of two cytosine dangling ends in BiHAG7 to yield BiHAG7-CC, allows for the thermodynamic contributions of the cytosine loops, results in more favorable ∆G° and ∆Hcal terms, by -1.7 and -7.6 kcal/ mol, respectively. On the other hand, the more favorable formation of BiAG7 (by -2.4 kcal/mol), and relative to BiHAG7, is enthalpy driven (by -15.6 kcal/mol). This indicates that the inclusion of loops has a stabilizing effect that improves basetriplet stacking, and prevents the exposure of base triplets to the solvent. The overall T∆Scal terms are in the order (Table 1) BiHAG7 (-77.5 kcal/mol) < TriAG7 (-82.7 kcal/mol) < BiAG7 (-93.5 kcal/mol) < MonoAG7 (-106.4 kcal/mol). From the point of view of triplex molecularity the lower the number of strands that formed a particular triplex, the lower the entropy. However, the above ranking indicates that the magnitude of the unfavorable entropy contributions depends strongly on the extent of proton and counterion binding by the triple-helical state. Conclusion We used a combination of temperature-dependent UV spectroscopy and DSC techniques to thermodynamically characterize the melting behavior of a set of triplexes with a common helical domain as a function of their molecularity. Complete thermodynamic profiles, including the differential binding of protons and counterions, were obtained for their unfolding to their random coil states at pH 6.2 and 0.2 M NaCl. Triplex formation is accompanied by a favorable free energy term, resulting from the typical compensation of favorable enthalpyunfavorable entropy contributions, i.e., the folding of a particular triplex is enthalpy driven. The magnitude of the favorable enthalpy contributions correspond to the number and strength
J. Phys. Chem. B, Vol. 112, No. 15, 2008 4839 of the base-triplet stacks formed, which are helped by stacking contributions due to the incorporation of dangling ends or loops. The overall triplex stability is in the following order: monomolecular > bimolecular > trimolecular. Specifically, UV and DSC melting experiments revealed that the trimolecular triplex (TriAG7) unfolds in two bimolecular transitions which corresponded to the sequential transitions of the triplex f duplex f random coils. The inclusion of a loop to form a more stable hairpin duplex yielded a weaker bimolecular triplex (BiHAG7) that unfolded in a bimolecular transition, followed by an intramolecular transition. This is consistent with a report16 that a strong duplex yields a weaker triplex, whereas a weak duplex generates a stronger triplex. Furthermore, the inclusion of a second loop to BiHAG7 yielded an intramolecular and compact triplex (MonoAG7) that unfolded mainly in a monophasic transition at higher temperatures. In addition, the combination of a purine strand and its double-complementary pyrimidine strand yielded a strong bimolecular triplex (BiAG7) that unfolded in a monophasic transition at relative high temperatures. As expected, acidic pH stabilized all triplexes by allowing protonation of the cytosines in the third strand; however, the percentage of protonation depended on triplex molecularity. The folding of these triplexes is consistent with a model reported earlier in which the triplex to duplex transition is accompanied by a release of protons and the uptake of sodium ions, whereas the duplex to random coil transition is accompanied by a release of sodium ions.13 The results help to choose adequate solution conditions for the study of triplexes containing different ratios of CGC+ and TAT base triplets. Moreover, our results are likely to aid in the design of ODN sequences that could effectively target mRNA involved in human diseases, thereby increasing the feasibility of using the antisense strategy for therapeutic purposes. For example, the higher stability of the MonoAG7 and BiAG7 triplexes suggest that we can use MonoAG7 and the long single-stranded pyrimidine strand of BiAG7 as mRNA targeting reagents because they will form very stable DNA-RNA hybrid duplex and triplex, respectively. Acknowledgment. This work was supported by Grants MCB-0315746 and MCB-0616005 from the National Science Foundation. References and Notes (1) Helene, C. Eur. J. Cancer 1991, 27, 1466. (2) Juliano, R. L.; Astriab-Fisher, A.; Falke, D. Mol. InterVentions 2001, 1, 40. (3) Soyfer, V. N.; Potaman, V. N. Triple-Helical Nucleic Acids; Springer-Verlag: New York, 1996. (4) Frank-Kamenetskii, M. D.; Mirkin, S. M. Annu. ReV. Biochem. 1995, 64, 65. (5) Crooke, S. T. Biochim. Biophys. Acta 1999, 1489, 31. (6) Helene, C. Eur. J. Cancer 1994, 30A, 1721. (7) Maher, L. J., III; Wold, B.; Dervan, P. B. Science 1989, 245, 725. (8) Cheng, Y. K.; Pettitt, B. M. Prog. Biophys. Mol. Biol. 1992, 58, 225. (9) Soto, A. M.; Marky, L. A. Biochemistry 2002, 41, 12475. (10) Shikiya, R.; Marky, L. A. J. Phys. Chem. B 2005, 109, 18177. (11) Beal, P. A.; Dervan, P. B. Science 1991, 251, 1360. (12) Keppler, M. D.; Fox, K. R. Nucleic Acids Res. 1997, 25, 4644. (13) Soto, A. M.; Loo, J.; Marky, L. A. J. Am. Chem. Soc. 2002, 124, 14355. (14) Chen, F. M. Biochemistry 1991, 30, 4472. (15) Mundt, A. A.; Crouch, G. J.; Eaton, B. E. Biochemistry 1997, 36, 13004. (16) Roberts, R. W.; Crothers, D. M. Science 1992, 258, 1463. (17) Cantor, C. R.; Warshaw, M. M.; Shapiro, H. Biopolymers 1970, 9, 1059. (18) Marky, L. A.; Blumenfeld, K. S.; Kozlowski, S.; Breslauer, K. J. Biopolymers 1983, 22, 1247.
4840 J. Phys. Chem. B, Vol. 112, No. 15, 2008 (19) Senior, M.; Jones, R. A.; Breslauer, K. J. Biochemistry 1988, 27, 3879. (20) Riccelli, P. V.; Mandell, K. E.; Benight, A. S. Nucleic Acids Res. 2002, 30, 4088. (21) Williams, J. C.; Case-Green, S. C.; Mir, K. U.; Southern, E. M. Nucleic Acids Res. 1994, 22, 1365. (22) Marky, L. A.; Breslauer, K. J. Biopolymers 1987, 26, 1601. (23) Kaushik, M.; Suehl, N.; Marky, L. A. Biophys. Chem. 2006, 126, 154. (24) Asensio, J. L.; Lane, A. N.; Dhesi, J.; Bergqvist, S.; Brown, T. J. Mol. Biol. 1998, 275, 811. (25) Plum, G. E.; Breslauer, K. J. J. Mol. Biol. 1995, 284, 679. (26) Leitner, D.; Schroder, W.; Weisz, K. J. Am. Chem. Soc. 1998, 120, 7123.
Lee et al. (27) Leitner, D.; Schroeder, W.; Weisz, K. Biochemistry 2000, 39, 5886. (28) Sugimoto, N.; Wu, P.; Hara, H.; Kawamoto, Y. Biochemistry 2001, 40, 9396. (29) SantaLucia, J., Jr.; Allawi, H. T.; Seneviratne, P. A. Biochemistry 1996, 35, 3555. (30) Breslauer, K. J.; Frank, R.; Blocker, H.; Marky, L. A. Proc. Natl. Acad. Sci. U.S.A. 1986, 83, 3746. (31) Zimmer, C.; Luck, G.; Venner, H.; Fric, J. Biopolymers 1968, 6, 563. (32) Zimmer, C.; Venner, H. Biopolymers 1966, 4, 1073. (33) Rentzeperis, D.; Kharakoz, D. P.; Marky, L. A. Biochemistry 1991, 30, 6276. (34) Marky, L. A.; Kupke, D. W. Methods Enzymol. 2000, 323, 419.
