How Do Size-Expanded DNA Nucleobases Enhance Duplex Stability

Computational Analysis of the Hydrogen-Bonding and Stacking Ability of ... were the first expanded nucleobases reported in the literature that were ge...
1 downloads 0 Views 264KB Size
Download PDF
J. Phys. Chem. B 2007, 111, 2999-3009

2999

How Do Size-Expanded DNA Nucleobases Enhance Duplex Stability? Computational Analysis of the Hydrogen-Bonding and Stacking Ability of xDNA Bases Tom L. McConnell† and Stacey D. Wetmore*,‡ Department of Chemistry, Mount Allison UniVersity, 63C York Street, SackVille, New Brunswick, E4L 1G8, Canada, and Department of Chemistry and Biochemistry, UniVersity of Lethbridge, 4401 UniVersity DriVe, Lethbridge, Alberta, T1K 3M4, Canada ReceiVed: October 25, 2006; In Final Form: December 22, 2006

Computational chemistry (B3LYP, MP2) is used to study the properties of size-expanded DNA nucleobases generated by inserting a benzene spacer into the natural nucleobases. Although the addition of the spacer does not significantly affect the hydrogen-bonding properties of natural nucleobases, the orientation of the base about the glycosidic bond necessary for Watson-Crick binding is destabilized, which could have implications for the selectivity of expanded bases, as well as the stability of expanded duplexes. Consideration of the (stacked) binding energies in the preferred relative orientation of natural and expanded nucleobases aligned according to their centers of mass reveals that the stacking within natural dimers can be increased by up to 50% upon expansion of one nucleobase and up to 90% upon expansion of two nucleobases. The implications of these findings to the stability of expanded duplexes were revealed by considering simplified models of natural and mixed duplexes composed of four nucleobases. Although intra- and interstrand interactions within double helices are typically less than those predicted when nucleobases are stacked according to their centers of mass, some nucleobases utilize their full stacking potential within double helices, where both intra- and interstrand interactions can be significant. Most importantly, increasing the size of nucleobases within the duplex significantly increases both intra- and interstrand stacking interactions. Specifically, some interactions are double the magnitude of the corresponding intrastrand interactions in natural helices, and even greater increases in interstrand interactions are sometimes found. Thus, our work suggests that mixed duplexes composed of natural bases hydrogen bound to expanded bases may exploit the increase in the inherent stacking ability of the expanded bases in more than one way and thereby afford duplexes with greater stability than natural DNA.

Introduction Modified DNA components (nucleobase, sugar, phosphate) have been studied for many years because of their medicinal applications. For example, the use of modified DNA in antisense therapies has been explored,1 where the modified components may yield improved binding, delivery, and drug resistance. Alternatively, modified DNA components have been used to treat infectious diseases by inhibiting essential enzymes involved with viruses, bacteria, or cancer. Indeed, modified nucleosides have proven to be valuable tools to combat the effects of AIDS.2 Interest in modified DNA components has also arisen because of their potential applications in biotechnology3 and nanotechnology.4 For example, they are used in polymerase chain reaction (PCR) to bind primers to the target DNA for replication.5 They may be promising hybridization probes, which are used to identify specific sequences of nucleobases within a sample, or fluorescent probes, where their use could eliminate the need for dyes and provide stronger, more selective, binding to target sequences. Novel DNA components have also been used as bioprobes to study proteins that interact with DNA. For example, Leonard and co-workers used expanded versions of * To whom correspondence [email protected]. † Mount Allison University. ‡ University of Lethbridge.

should

be

addressed.

E-mail:

adenine (Figure 1)6-9 to explore the active sites of ATP dependent enzymes.10-12 The nucleobases designed by Leonard and co-workers were pioneering in another way. Specifically, these were the first expanded nucleobases reported in the literature that were generated by insertion of a (benzene) spacer into the natural structure. More recently, Kool and co-workers have expanded the utility of these unique adenine derivatives by synthesizing the corresponding size-expanded versions of all four natural nucleobases (Figure 2)13-17 and by incorporating these into duplexes denoted as xDNA (“expanded” DNA).13,18-22 Alternatively, modified forms of expanded cytosine and thymine have also been synthesized by the Kool group, which are denoted as yDNA (“wide” DNA) bases.23-25 Since the work of Kool and co-workers, nucleobases expanded by other spacers, such as thiophenes,26 benzo[b]thiophene,27 and napthalene,28 have been synthesized. The initial driving force for the development of a full set of xDNA nucleobases was to create a new genetic system,29 which may allow for enhanced information storage30 or data encryption and processing.31 It was also hoped that size-expanded nucleobases could serve as useful tools for studying the structure of DNA, as well as proteins that interact with DNA (i.e., DNA polymerases),32 because of enhanced (i.e., fluorescence)14-16 properties. More recently, there has been interest in the ability of expanded DNA to act as improved nanowires.33-35

10.1021/jp0670079 CCC: $37.00 © 2007 American Chemical Society Published on Web 02/28/2007

3000 J. Phys. Chem. B, Vol. 111, No. 11, 2007

McConnell and Wetmore stacking interaction energies include BSSE39 corrections. Additional details of our methodologies will be provided in the appropriate Results and Discussion sections for clarity and to avoid redundancy.

Figure 1. Expanded versions of adenine developed by Leonard and co-workers.6-9

Initial experimental work reveals that expanded DNA (xDNA) helices composed of base pairs between one natural and one expanded nucleobase are more stable than the corresponding natural DNA helices. A significant portion of this increased stability is expected to arise because of enhanced stacking interactions upon incorporation of the benzene spacer.11,18,36 Indeed, dangling-end experiments indicate that the expanded nucleobases stack considerably stronger than their natural counterparts.13,16 However, it is difficult from experiments alone to gain insight into changes in individual nucleobase interactions because of base size. On the other hand, changes in binding strengths can be relatively easily obtained from computational studies. Previous computational studies have investigated some of the properties of the xDNA expanded nucleobases.33-35,37 For example, Fuentes-Cabrera and co-workers investigated the structure (nonplanarity) and the electronic properties (HOMOLUMO gap) of the isolated bases,33-35 while Huertas et al. focused on the tautomeric preferences of the expanded bases in the gas phase and in solution.37 Although the latter study also considered the Watson-Crick hydrogen-bonding interactions, only the total stacking interactions between select Watson-Crick (A:T, xA:xT, G:C or xG:xC) hydrogen-bonded pairs were considered using a low level of theory (AMBER).37 The present work complements previous experimental and computational studies on expanded DNA nucleobases by investigating the properties and interactions of the nucleobases that directly affect duplex stability. The energy barriers for rotation about the glycosidic bonds, the (Watson-Crick) hydrogen-bond strengths, and the stacking interactions of the natural and expanded nucleobases will be analyzed in detail. Particular attention will be given to the enhancement in stacking afforded by the benzene ring when geometries that reflect those within the DNA duplex are considered. Computational Details All calculations were performed using Gaussian 03 (Revision C.02).38 Optimization and frequency calculations of isolated and hydrogen-bonded dimers (C1 symmetry) were performed using B3LYP/6-31G(d, p) in the gas phase. Relative energies for these systems were obtained using B3LYP/6-311+G(2df,p), and all relative energies include zero-point vibrational energy (ZPVE) and basis set superposition error (BSSE)39 corrections (where applicable). This basis set has been widely used to study hydrogen-bond strengths and has been shown to accurately reproduce the binding strengths for the natural nucleobases obtained from higher levels of theory.40 Geometries of isolated (Cs) monomers used to study stacking interactions were obtained with MP2/6-31G(d), and subsequent stacking interaction energies were calculated with MP2/6-31G*(0.25), where the 6-31G*(0.25) basis set replaces the optimized d-exponent for heavy atoms (0.8) with a value of 0.25. This formulism has been previously used to study the stacking interactions between the natural nucleobases,41-45 as well as their stacking interactions with other molecules,46-48 and has proven to provide a reasonable compromise between accuracy and computational cost. All

Results and Discussion (1) Rotation about the Glycosidic Bond. It has been wellestablished that the natural nucleobases preferentially adopt an anti orientation with respect to the deoxyribose sugar moiety, which is defined as the nucleobase position that affords the maximum distance between H1′ of the sugar and H8 (purines) or H6 (pyrimidines) (for example, see Figure 3 for the anti and syn conformers of adenine).49 However, the hydrogen-bonding patterns of some non-Watson-Crick binding modes between the natural nucleobases,49,50 as well as the hydrogen-bonding modes of some modified nucleobases,51,52 exploit a syn orientation. For example, ease of rotation about the glycosidic bond is necessary for a series of purine derivatives51 and the azole carboxamides52 to function as universal nucleobases (i.e., bases that bind to all natural nucleobases with an equal affinity). Preference for the syn conformation could be detrimental to the design of an expanded genetic system with the same (hydrogen-bonding) selectivity as the natural nucleobases since the syn conformer does not afford the Watson-Crick binding face to the neighboring strand. Indeed, solution structures of select xDNA sequences show that the bases adopt an anti conformation within these helices.18 Therefore, it is of interest to determine the influence of an additional benzene ring within the nucleobase on the favored conformation about the glycosidic bond. Figure 4 compares the relative energy as a function of the torsional angle about the glycosidic bond (χ) for the natural and expanded nucleobases, where χ ) O4′C1′N9C4 in the natural purines, O4′C1′N9C3b in the expanded purines, O4′C1′N1C2 in the natural pyrimidines, and O4′C1′N1C6 in the expanded pyrimidines. Our DNA sugar-backbone model replaces the C3′ and C5′ phosphate groups of DNA with hydroxyl groups. The sugar puckering was set to C2′-endo, which corresponds to the conformation in B-DNA, and the HO5′C5′C4′ dihedral angle was fixed at 180°, which eliminates hydrogen-bonding contacts between the C5′ hydroxyl group and the nucleobase since these cannot occur within DNA duplexes. The preferred torsional angle about the glycosidic bond was scanned by performing a series of optimizations with χ fixed in 10° increments between 0 and 360°. Figure 4a and 4b (closed symbols) reveals that the natural purines prefer the anti orientation, where the syn orientation is approximately 11-13 kJ mol-1 higher in energy and is separated from the favored conformer by a 14 (A)-18 (G) kJ mol-1 barrier. The expanded purines, on the other hand, show little preference in the orientation about the glycosidic bond (Figure 4a and 4b, open symbols). The differences in the properties of the natural and expanded bases occur since the syn orientation of the natural purines involves unfavorable interactions between O5′ and N3, while the syn orientation of the expanded purines involves a favorable O5′‚‚‚H-C4b hydrogen bond (for example, Figure 5 compares A and xA). The preferred conformation about the glycosidic bond also differs for the natural and expanded pyrimidines (Figure 4c and 4d). In this case, the anti conformation is highly favored in the natural pyrimidines because of unfavorable interactions between O5′ and the O2 carbonyl in the syn conformer (see Figure 6a, χ ) 80). However, the syn conformation is more favorable for the expanded pyrimidines because of a strong O5′‚‚‚H-N1

DNA Nucleobases Enhancing Duplex Stability

J. Phys. Chem. B, Vol. 111, No. 11, 2007 3001

Figure 2. Structure and atomic numbering of the four (a) natural DNA nucleobases and (b) expanded bases developed by Kool and co-workers.13-17

Figure 3. The (a) anti and (b) syn conformations of the natural nucleobase adenine with respect to the deoxyribose sugar moiety.

hydrogen bond (see Figure 6b, χ ) 50). Also, a second rotational barrier is observed for the expanded pyrimidines at an O4′C1′N1C6 dihedral angle equal to approximately 260° because of repulsive contacts between H1′ and H-N1 or H2′ and H-C3b (Figure 6b, χ ) 260). Because of these repulsive interactions, the most stable anti orientation (Figure 6b, χ ) 310) has a distorted structure, which may destabilize expanded helices. In summary, our calculations suggest that the expanded nucleobases favor different orientations about the glycosidic bond in comparison to the natural nucleobases. This could have implications for the formation, and resulting stability, of duplexes containing expanded bases. Additionally, the selectivity of the expanded bases may be jeopardized since the base orientation required for Watson-Crick hydrogen bonding is destabilized compared with orientations that may yield other (mismatched) binding modes. (2) Watson-Crick Hydrogen-Bond Strengths. Base-paring experiments show that the expanded nucleobases selectively pair with their Watson-Crick partners.15,16,19,20 Optimizations of the hydrogen-bonded pairs between natural or expanded nucleobases can yield important information about the strength of the hydrogen bonds, as well as the distance between anomeric carbons, which directly relates to the width of DNA duplexes. In the present study, we replace the deoxyribose sugar moiety (dR, Figure 2) with a hydrogen atom or a methyl group, and calculate the hydrogen-bond strengths of pairs between two natural nucleobases, one natural and one expanded nucleobase, and two expanded nucleobases (Table 1). All hydrogen-bonded (Watson-Crick) pairs optimize to nearly planar structures. The anomeric carbon distances (R(C‚ ‚‚C) in dR ) CH3 models) suggest that pairs containing one expanded DNA nucleobase are approximately 2 Å larger than the corresponding natural base pair when an expanded pyrimidine is considered, while inclusion of an expanded purine leads to an approximate 2.5 Å increase in the width of the natural pair. These separations are consistent with interglycosidic distances conjectured for mixed duplexes13,14 as well as those determined from the solution NMR of select xDNA sequences.18 Expansion of both bases increases the width of the pair by 4.5 Å.

The hydrogen-bond strengths calculated for models with the deoxyribose replaced by a hydrogen or methyl group are very similar. There are relatively small changes (less than 2.5 kJ mol-1) in the binding strengths of the A:T pair when either or both nucleobases are expanded. This finding is consistent with suggestions that the experimentally observed increases in the stability of expanded DNA duplexes composed of A and T relative to the corresponding natural strand are primarily due to other (stacking) interactions.11,13,16,18,37 Larger deviations in hydrogen-bond strengths are noted upon expansion of the G:C pair. Interestingly, inclusion of xG leads to a 6.1 kJ mol-1 decrease in binding strength, while inclusion of xC leads to a 3.5 kJ mol-1 increase in binding strength. These effects are roughly additive, where the xG:xC pair binds 3.0 kJ mol-1 weaker than the G:C pair. Nevertheless, the difference in the binding within natural and expanded pairs are small, and thus our calculations support previous suggestions13,37 that expanding the natural bases will not significantly change the contribution of hydrogen bonding to duplex stability. (3) Stacking Interactions. (3.1) Stacked Nucleobases. Attempts to fully optimize stacked geometries (with MP2) generally lead to hydrogen-bonded arrangements, which occurs at least in part because of the absence of basis set superposition error (BSSE) corrections in the most commonly used optimization routines.41,43 Furthermore, distortions in the monomer geometries are often seen upon attempts to fully optimize stacked dimers.53 Computational studies of the stacking interactions involving DNA nucleobases have revealed that important information about the strength of the binding energies can be obtained by scanning the potential energy surface of stacked dimers.41-46,48 Therefore, similar potential energy surface scans were performed in the present work. Initial geometries for the scans were obtained by aligning nucleobase monomers (optimized in (planar) Cs symmetry) via their centers of mass and separating the bases vertically by a distance of 3.3 Å, which is the average separation in B-DNA and the favored calculated separation between expanded nucleobases.54 Since the stacking energies of various dimers have been determined to depend most strongly on the angle of rotation about an axis passing through the aligned centers of mass,41,42,46,48 we focus on scanning the potential energy surface with respect to this variable. Specifically, a 0° angle of rotation was defined for all dimers to be the orientation that aligns the glycosidic bonds (N1-H for the pyrimidines and N9-H for the purines). From these structures, the MP2/6-31G*(0.25) stacking interaction energies were calculated for each dimer generated by rotating one monomer in 30° increments from 0 to 360°. Dimers between one natural and one expanded nucleobase or two expanded nucleobases were considered. Although the stacking energies of dimers between the natural nucleobases

3002 J. Phys. Chem. B, Vol. 111, No. 11, 2007

McConnell and Wetmore

Figure 4. Comparison of the relative energy as a function of rotation about the glycosidic bond for natural (closed symbols) and expanded (open symbols) (a) adenine, (b) guanine, (c) thymine, and (d) cytosine.

Figure 5. Comparison of the favored anti and syn orientations about the glycosidic bond in (a) natural and (b) expanded adenine.

have been previously reported in the literature, we reconsider these dimers in the present work to ensure that the most accurate comparison is made. Two relative orientations of the nucleobases (denoted as intra- and interstrand) were investigated, where direct atomic overlap arises for the intrastrand orientation with

a 0 Å vertical separation and a 0° angle of rotation, while the interstrand orientation involves a 180° flip of one nucleobase about an axis passing through the (Cs) molecular plane. As anticipated on the basis of previous work,41,42,46,48 there is a very large dependence of the stacking interaction on the angle of rotation for all dimers considered (see ∆, Tables 2-4). The stacking energy of dimers composed of one natural and one expanded nucleobase is less dependent on the angle of rotation (∆ ranges between 13 and 30 kJ mol-1, Table 3) than dimers composed of the same type of base (∆ ranges between approximately 20 and 50 kJ mol-1 for natural-natural (Table 2) or expanded-expanded (Table 4) dimers). The minimum calculated stacking energy for all dimers typically occurs for structures with the highest degree of atomic overlap. For example, the minimum intrastrand stacking interactions between dimers of the same nucleobase generally occur at 0° angles of rotation because of repulsive interactions between the nuclei. Indeed, the stacking energies for these orientations in natural-natural dimers are often positive. On the other hand, the maximum calculated stacking energy typically arises for structures where the dipole moments of the monomers are aligned in opposing directions, as found previously for natural nucleobase dimers41 as well as dimers between natural nucleobases and modified nucleobases46b or amino acids.48 Therefore, there is little correlation between the preferred angle of rotation for different (natural or expanded) pairs. Focusing on the maximum possible interaction energies, the smallest stacking interactions between the (natural or expanded) nucleobases occur for dimers composed of (natural or expanded) pyrimidines, while the largest stacking interactions occur for

DNA Nucleobases Enhancing Duplex Stability

J. Phys. Chem. B, Vol. 111, No. 11, 2007 3003 TABLE 1: Hydrogen-Bond Strengths (kJ mol-1) and Anomeric Carbon Distances (R(C‚‚‚C), Å) for Dimers Involving Natural and Expanded Nucleobasesa dR ) CH3c dimer

dR ) Hb

A:T xA:T A:xT xA:xT G:C xG:C G:xC xG:xC

-43.8 -46.2 -43.5 -46.0 -96.1 -89.7 -99.0 -92.9

R(C‚‚‚C) -44.1 -46.6 -43.7 -45.8 -95.7 -89.6 -99.2 -92.7

10.6 13.1 12.7 15.1 10.8 13.2 12.8 15.2

a Relative energies calculated at the B3LYP/6-311+G(2df,p) level of theory and include ZPVE and BSSE corrections. b Deoxyribose replaced with a hydrogen atom (dR ) H). c Deoxyribose replaced with a methyl group (dR ) CH3).

Figure 6. Comparison of the anti and syn orientations about the glycosidic bond in (a) natural and (b) expanded thymine.

dimers composed of at least one (natural or expanded) purine. This trend makes sense on the basis of the relative size and polarizability of the individual nucleobases and has been previously noted in the literature for the natural bases.41-45 Perhaps the most important information obtained from the stacking scans is the differences in the magnitude of the maximum stacking interaction upon expansion of the nucleobase by a benzene ring. More specifically, the stacking interactions between two natural nucleobases range between 30 (T‚‚‚T) and 49 kJ mol-1 (G‚‚‚G), where small differences (