Contributions of the loops on the stability and targeting of DNA pseudoknots

Background: Pseudoknots have been found to play important roles in RNA function, examples include ribosomal frameshifting and in the 5’ UTR of mRNA as riboswitches. In RNA frameshifting, there is a local formation of base-triplet stacks within the pseudoknot, increasing the stability of the terminal stem. The interaction of this triplex structure with the ribosome might help with the high-efficiency of frameshifting. The main objective of this work is to mimic RNA pseudoknots using DNA oligonucleotides for the control of gene expression. Specifically, we have designed a pair of DNA pseudoknots with different length in one of the loops to mimic the formation of a local triple helix, shown within RNA pseudoknots of the human telomerase. Methods: We have used a combination of temperature-dependent UV spectroscopy and calorimetric techniques to determine the thermodynamics for the unfolding of the pseudoknots and their targeting with complementary strands. The unfolding data is then used to create thermodynamic (Hess) cycles that correspond to each of the targeting reactions. The resulting data is then compared with the thermodynamic enthalpy data obtained directly from isothermal titration calorimetry. Results: UV melting curves of each pseudoknot show transitions with TMs independent of strand concentration, which confirms their intramolecular formation. Analysis of the differential scanning calorimetry (DSC) curves shows the pseudoknot with the longer thymine loop (PsK-9) to be more stable, by -5.7 kcal/mol, and to unfold with a higher enthalpy of 27.4 kcal/mol. The targeting of each pseudoknot yielded favorable reaction free energy contributions that were enthalpy driven. However, the disruption reaction of PsK-9 took place with a less favorable free energy term, by 0.7 kcal/mol, and less favorable enthalpy term, by 4.5 kcal/mol. Conclusion: The thermodynamic unfolding data showed that PsK-9 is more stable or more compact, due to the involvement of three loop thymines of PsK-9 in forming three T*AT base-triplets, or two T*AT/ T*AT base-triplet stacks, in the stem of this pseudoknot. The targeting thermodynamic data indicated that each complementary strand is able to disrupt the pseudoknots. However, the disruption of PsK-9 takes place with a less favorable free energy contribution, confirming the formation of a short and local triplex.

ODNs, as drugs, present an exquisite selectivity and are able to discriminate targets that differ by a single base and can be used to control the expression of genes. [5][6][7]24]. There are three main approaches for the use of ODNs as modulators of gene expression: the antisense, antigene, and small interfering RNA strategies [24]. In the antisense strategy, an ODN binds to messenger RNA, forming a DNA/ RNA hybrid duplex that inhibits translation by blocking the assembly of the translation machinery or by inducing an RNase H mediated cleavage of their mRNA target [5]. In the antigene strategy, an ODN binds to the major groove of a DNA duplex, forming a triple helix [30] that inhibits transcription, by competing with the binding of proteins that activate the transcriptional machinery [7,31]. There are advantages and disadvantages in these two strategies. In the direct targeting of a gene, the antigene strategy offers some advantages over the antisense strategy. First of all, there are only two copies of a particular gene whereas there is a large continuous supply of the mRNA gene transcript. Moreover, blocking the transcription of the gene itself prevents repopulation of the mRNA pool, allowing a more efficient and lasting inhibition of gene expression [32,33]. The main disadvantage is that the ODN needs to cross the nuclear membrane and access its DNA target within the densely packed chromatin structure [34]. Common disadvantages of the use of ODNs for targeting purposes are that the oligonucleotide needs to cross lipid membranes; for instance, hydrophilic ODN duplexes do not cross lipid membranes [35], and the fast degradative action of nucleases. These disadvantages can be circumvented by using single strands and by chemically modification of its phosphate or sugar groups. The presence of unpaired nucleobases renders the ODN more hydrophobic, allowing them to interact better with polycationic micelles and/or enabling them to cross the cellular membranes. These polycations can be used as delivery vectors, protecting the ODN from the action of nucleases.
From a thermodynamic point of view, successful control of gene expression depends on the effective binding of a DNA oligonucleotide sequence to its target with tight affinity and specificity. This is provided by using a long sequence of 15-20 bases in length when targeting genes [5]; strong specificity is conferred by hydrogen bonding in the formation of Watson-Crick and/or Hoogsteen base-pairs, while high affinity is provided by the large negative free energy upon formation of a duplex or triplex products; thereby, competing efficiently with the proteins involved in transcription or translation. In the successful targeting of nucleic acid secondary structures with complementary strands, the strand must be able to invade and disrupt the secondary structure forming a large number of base-pair stacks in the duplex products. Our laboratory is using DNA oligonucleotides to mimic the secondary structures of RNA molecules and their targeting with complementary strands to create a library of thermodynamic targeting data [27][28][29]. The novelty of this approach is several fold, DNA oligonucleotides are less expensive than RNA oligonucleotides and more stable against hydrolysis, and most important the resulting DNA-DNA thermodynamic data is similar to the DNA-RNA thermodynamic data [36][37][38], which is obtained in the targeting of RNA molecules with DNA complementary strands.
In this work, we have designed a pair of DNA pseudoknots with different length in one of the loops to mimic the formation of a local triple helix shown within RNA pseudoknots of the human telomerase, explaining the increase efficiency of ribosomal frameshift. To this end, we have used a combination of UV spectroscopy and calorimetric techniques to determine the thermodynamics for both the unfolding of both DNA pseudoknots and their targeting with complementary strands. The results show that the pseudoknot with the longer loop is more stable and forming a local triplex structure, consistent of two base-triplet stacks, which is confirmed with the lower free energy obtained in the targeting of this pseudoknot.

Temperature-dependent UV spectroscopy
Absorbance versus temperature profiles were measured at 260 nm with a thermoelectrically controlled Aviv Spectrophotometer Model 14DS UV-Vis (Lakewood, NJ). The temperature was scanned at a heating rate of 0.6°C/min, and shape analysis of the melting curves yielded transition temperatures, T M s [40]. The transition molecularity for the

Differential scanning calorimetry (DSC)
The total heat required for the unfolding of each oligonucleotide (pseudoknot, single strand or duplex product) was measured with a VP-DSC differential scanning calorimeter from Microcal (Northampton, MA). Standard thermodynamic profiles and T M s are obtained from the DSC experiments using the following relationships [40,41]: ΔH=∫ΔC p (T)dT; ΔS=∫ΔC p (T )/TdT, and the Gibbs equation, ΔG°( T) =ΔH-TΔS; where ΔC p is the anomalous heat capacity of the ODN solution during the unfolding process, ΔH and ΔS are the unfolding enthalpy and entropy, respectively, assumed to be temperature-independent. ΔG°( T) is the free energy at a temperature T.

Isothermal titration calorimetry (ITC)
The heat for the reaction of a pseudoknot with its complementary strand was measured directly by isothermal titration calorimetry using the ITC 200 from GE Microcal (Northampton, MA). A 40 μL syringe was used to inject the titrant; mixing was effected by stirring this syringe at 1000 rpm. Typically, we used 5-7 injections of 2 μL of pseudoknot solution with at least 2-fold lower concentration than the solution of the complementary strand in the cell, and over a time of 4-8 minutes between injections. The reaction heat of each injection is measured by integration of the area of the injection curve, corrected for the dilution heat of the titrant, and normalized by the moles of titrant added to yield the reaction enthalpy, ΔH ITC [27][28][29]42]. All titrations ITC experiments were designed to obtain the heat, ΔH ITC , for each targeting reaction by averaging the reaction heat of at least five injections under unsaturated conditions. These ΔH ITC terms correspond to the formation of duplex products. To determine the free energy, ΔG ITC , for each targeting reaction, we use the following relationship, ΔG ITC = ΔG HC (ΔH ITC /ΔH HC ) [27][28][29], while the Gibbs equation is used to determine the TΔS ITC parameter, where T is the temperature of the ITC experiments.

Overall experimental protocol
We used initially UV melting techniques to characterize the helix-coil transition of each molecule as a function of strand concentration. DSC experiments are carried out to determine T M s and unfolding thermodynamic profiles for each pseudoknot and for the other reactants and products of each targeting reaction [40,43]. The DSC data is used to set up Hess cycles to yield thermodynamic profiles for these reactions. Then, ITC titrations are used to measure directly the heat (ΔH ITC ) of each targeting reaction, which are compared with the Hess cycle data [27][28][29].
This enthalpy comparison shows that the loops of PsK-5 are constrained while the longer loop of PsK-9 is actually releasing this tension. However, the main observation is PsK-9 unfolds with a higher T M , by 6.5°C, and higher unfolding enthalpy, by 27.4 kcal/mol ( Table 1). This indicates that PsK-9 is more stable with improved base-pair stacks. Alternatively, the observed additional heat suggests that the 9 thymine loop of PsK-9 is located in the ceiling of the major groove of its 7 A•T stem, and three thymines (starred thymines of Figure 1) are involved in the formation of three T*A•T base-triplets (or two T*AT/T*AT base-triplet stacks) [44,45]. Overall, the unfolding of each pseudoknot takes place through the typical unfavorable enthalpy-favorable entropy compensation. Unfavorable enthalpy contributions correspond to energy needed to break base pairing and base-pair stacking interactions, while favorable entropy contributions correspond to the higher disorder state of the single strand at high temperature and the putative release of ions and water molecules [27][28][29]46]. In summary, the folding of PsK-9 is thermodynamically more stable than PsK-5 (reverse signs of Table 1), by -5.7 kcal/mol, resulting in a more compact molecule.

Targeting of pseudoknots with complementary strands
To confirm PsK-9 is forming a more compact pseudoknot, we investigated the reaction of each pseudoknot with its corresponding complementary strand to form duplex products with dangling ends: Unfolding thermodynamics for the species of each targeting reaction Figure 3 shows the DSC thermograms of each pseudoknot, complementary strand, and duplex product, and Table 1 has the resulting thermodynamic profiles from the analysis of these thermograms. The DSC of each pseudoknot was discussed earlier; their unfolding curves are included in this figure for clarity and for explaining the unfolding of each product duplex. Based on their sequence, magnitude of their enthalpies, and assuming an enthalpy of 7-8 kcal per mol of base-pair stack, these single strands are forming hairpin loops with dangling ends and with 5 (CS-5) and 6 (CS-9) base-pair stacks in their stems.
UV melting experiments of the duplex products, as a function of strand concentration, show that the T M s increased slightly (data not shown), indicative of their bimolecular formation. These T M -dependences are consistent with the unfolding of duplexes with 19 and 23 base pairs, which are approaching the unfolding behavior of polymers. The DSC curves of each product duplex (Figure 3) show PsKDup-5 unfolds in an apparent monophasic transition with a T M of 49.6°C and ΔH of 135 kcal/mol, while PsKDup-9 unfolds in a biphasic transition with T M s of 47.7°C and 58.3°C and total ΔH of 161 kcal/mol. Each DSC profile corresponds to the unfolding of the duplex followed by the folding and sequential unfolding of the corresponding pseudoknot and hairpin (single strands). For instance, the monophasic unfolding of PsKDup-5 is due to the similar T M s of the structures formed by the reactants and product of this reaction, which are within 5°C. On the other hand, PsKDup-9 shows a biphasic transition (Figure 3b), the first transition corresponds to the unfolding of the duplex into partial folded of both PsK-9 and CS-9, followed by the simultaneous unfolding of CS-9 and pseudoknot. The higher ΔH term of PsKDup-9, by 26 kcal/mol, corresponds to the formation of two additional base-triplet stacks.
We created Hess cycles with this unfolding data to generate indirectly thermodynamic profiles for each targeting reaction i.e., we added the thermodynamic profiles of the pseudoknot and single strand (hairpin), and subtracted the thermodynamic profiles of the duplex. The resulting data is shown in the last two entries of Table 1. This exercise yielded ΔG°H C and ΔH HC of -5.1 kcal/mol & -37 kcal/mol (PsK-5), and -3.5 kcal/mol & -25 kcal/mol (PsK-9), respectively. Both reactions are favorable and enthalpy driven. However, the targeting of PsK-9 is less favorable, which is consistent with the higher stability of this pseudoknot. Furthermore, we obtained unfavorable TΔS HC terms of -31.4 (PsK-5), and -21.7 kcal/mol (PsK-9), which correspond to the net uptake of ions and water molecules by the duplex products of each reaction, since the conformational entropy change is considered similar for each reaction.

Targeting reactions measured directly by ITC techniques
The heat for each targeting reaction was measured by ITC under unsaturated conditions, using ODN concentrations and temperatures that guaranteed 100% formation of the final duplex products. ITC titrations were carried out at 5°C, the heat of each injection was corrected for its dilution heat and The ΔG°I TC at 5 °C for each targeting reaction were obtained from the ΔG°H C values by using a temperature factor (=ΔH ITC / ΔH HC ), which assumes ΔH HC s to be independent of temperature i.e., ΔCp=0. The TΔS ITC parameters were calculated using the Gibbs equation. The overall results are shown in Table 1. We obtained favorable ΔG°I TC contributions for each targeting reaction, each complementary strand is able to invade and disrupt the pseudoknot structure. However, the reaction for the targeting of PsK-9 is less favorable in spite of forming a more stable duplex, by 0.7 kcal/mol. This result is consistent with the higher stability PsK-9 and confirms the formation of a local triplex in the stem of PsK-9 consistent of three T*A•T base-triplets (or two T*AT/T*AT base-triplet stacks).

Conclusions
We have investigated the thermodynamic stability of two pseudoknots to determine the formation of a local and short triplex in the pseudoknot with a longer thymine loop. Specifically, we used a combination of UV, DSC, and ITC techniques to determine the unfolding thermodynamics of a pair of pseudoknots and their reaction with complementary strands. The favorable folding of DNA pseudoknots results from the typical favorable enthalpy-unfavorable entropy compensation, confirming the flexibility of DNA strands being able to form pseudoknots that can be used to mimic known RNA secondary structures. The folding data shows that PsK-9 is more stable due to a more favorable enthalpy. This enthalpy value corresponds to the partial folding of the thymine loop (third strand) on the major groove of the duplex, yielding a net formation of two T*AT/T*AT base-triplet stacks at the middle of its stem. The targeting thermodynamic data indicated that each complementary strand is able to disrupt the pseudoknots. However, the disruption of PsK-9 takes place with a less favorable free energy contribution, confirming the formation of a short and local triplex.
The main observation from this study is that a triplex is able to form in the pseudoknot if the loop length and sequence are appropriate. The favorable targeting of these pseudoknots depends on the length and sequence of the complementary strand. However, the favorable free energy term of these targeting reactions may well be increased by improving the stability of the duplex products, by using longer single strands with complementary sequences and/or DNA intramolecular structures with loops containing a larger number of un-paired bases. In general, the higher the number of base-pairs and base-pair stacks that are formed in the duplex product, the higher the free energy term; specifically, if a larger number of unpaired bases are involved in this base-pairing. This investigation of the targeting of DNA pseudoknots has enabled us to improve our method, based on physicochemical principles, to determine the thermodynamics of the targeting of nucleic acid secondary structures that can be used to control the expression of genes.