Frequency and Effect of the Binding of Mg - American Chemical

Nov 4, 2009 - Internet at http://pubs.acs.org. References and Notes. (1) Pyle, A. M. Science 1993, 261, 709. (2) Pyle, A. M. J. Biol. Inorg. Chem. 200...
0 downloads 0 Views 2MB Size
Download PDF
15670

J. Phys. Chem. B 2009, 113, 15670–15678

Frequency and Effect of the Binding of Mg2+, Mn2+, and Co2+ Ions on the Guanine Base in Watson-Crick and Reverse Watson-Crick Base Pairs Romina Oliva† and Luigi Cavallo*,‡ Dipartimento di Scienze Applicate, UniVersita` di Napoli “Parthenope”, I-80143 Naples, Italy, and Dipartimento di Chimica, UniVersita` di Salerno, I-84084 Fisciano (SA), Italy ReceiVed: July 20, 2009; ReVised Manuscript ReceiVed: September 21, 2009

We performed MP2 calculations to elucidate the structure and energetics of the Mg2+, Mn2+, and Co2+ hexahydrated aquaions, and the effect of the metal binding to the N7 atom of (i) a single guanine, (ii) a guanine involved in a Watson-Crick pair, and (iii) a guanine involved in a reverse Watson-Crick base pair. Our comparative analysis of the three aquaions indicates a clear inverse correlation between the radius of the cation and the binding energy, that indeed increases in the order Mn2+ < Co2+ < Mg2+. The trend in the binding energies of the pentahydrated cations to the N7 atom of the guanine is instead Mg2+ < Mn2+ < Co2+, suggesting a rather different bonding scheme that, for the two transition metals, involves back-donation from the aromatic ring of the guanine to their empty d orbitals. In the gas phase, the three hydrated metals significantly stabilize both G-C base pair geometries, Watson-Crick and reverse Watson-Crick, we investigated. Inclusion of a continuous solvent model, however, remarkably reduces this additional stabilization, which becomes almost negligible in the case of the Mg2+ cation coordinated to the guanine in the standard Watson-Crick geometry. Conversely, all three metal ions sensibly stabilize the reverse Watson-Crick geometry, also in water. Our results are supported by a screening of the structures available in the Protein Data Bank, which clearly indicates that the two transition metals we investigated have a tendency greater than Mg2+ to coordinate to the N7 atom of guanines, and that there is no clear correlation between the number of guanines in experimental structures with a metal bound to N7 atom and their involvement in Watson-Crick base pairs. Introduction Interaction of nucleic acids with metals is central to their structure and function. Metal ions, and especially divalent metal ions such as Mg2+, have long been known to play a crucial role in RNA three-dimensional folding, structure stabilization, and catalytic activity.1-8 Metal ions can bind RNA and DNA molecules aspecifically, by neutralizing the negative charges on the ribose-phosphate backbone, but can also bind them specifically on nucleotide base atoms, such as the basic N7 of purines. Indeed, transition metal ions have the tendency to coordinate almost exclusively to the N7 position of purine bases, especially guanine.9,10 It is thus not surprising that the effect of the binding of a series of dications (Mg2+, Ca2+, Sr2+, Ba2+, Zn2+, Cd2+, and Hg2+) on nucleic acid base pairs has been investigated theoretically over the years, evidencing that metal binding usually reinforces the stability of the base pairing. However, improvement of the first models, from systems with a coordinated naked cation11-13 to systems that include explicit water molecules to saturate the first coordination shell of the metal, indicated that solvent effects can remarkably reduce this metal stabilizing effect on the base pairing.14-19 In the same line of research, we have shown previously that a Mg2+ ion, which is bound to the N7 atom of the G15 base in the best resolution (1.93 Å) crystallographic structure of yeast tRNAPhe,20 is crucial to the stability of its tertiary structure.21 Nucleotide G15 is in fact involved with nucleotide C48 in an * To whom correspondence should be addressed. Tel.: +39 081 969549. Fax: +39 089 969603. E-mail: [email protected]. † Universita` di Napoli “Parthenope”. ‡ Universita` di Salerno.

important tertiary base-pairing interaction, the Levitt base pair, at the elbow of the “L-shaped” structure.22 Guanine 15 and cytosine 48 give a noncanonical trans Watson-Crick base pair, also known as reverse Watson-Crick (RWC), which is poorly stable in comparison to an alternative “un-native” bifurcated geometry, and we have shown that Mg2+ binding substantially stabilizes the “native” RWC geometry.23 Interestingly, experiments also indicated that Mn2+ and Co2+ ions preferentially replace the Mg2+ ion at this binding site.20 Although less common than Mg2+ in nucleic acids structures, Mn2+ also has a well-defined role in the biochemistry of nucleic acids. For example, Mn2+ is required for the activity of ribozyme in the 3′-untranslated region of Xenopus Vg1 mRNA,24 and, together with Mg2+, for the Tetrahymena ribozyme catalysis to occur.25,26 It has also been shown that a RNA-cleaving bipartite DNAzyme is a distinctive metalloenzyme, able to use either Mn2+ or Mg2+ as cofactor.27 On the other hand, Co2+ may affect, at least in vitro, the nucleic acid structural organization. It has been shown to enhance stability and complexity of RNA aptamer structure,10 to boost significantly the formation of triplehelix DNA (whereas alkaline earth metal ions have no positive effects on triplex formation),28 to mediate the binding of antibiotics to DNA,29 and to act as a stabilizing link between symmetry-related guanine bases (guanine-Co2+-guanine bridges) in crystals of DNA oligomers.30,31 Finally, Mn2+ and Co2+ induce more effectively than Mg2+ aggregation of melted DNA32 and a transition of a DNA oligonucleotide from an antiparallel to a parallel G-quartet structure.33 These findings indicate a peculiarity in the interaction of Mn2+ and Co2+ ions with nucleic acid bases, but no high level

10.1021/jp906847p CCC: $40.75  2009 American Chemical Society Published on Web 11/04/2009

Ion Binding on Guanine in WC and RWC Base Pairs

J. Phys. Chem. B, Vol. 113, No. 47, 2009 15671 nucleic acids to assume peculiar teritiary structures has increased remarkably in recent years. Methods

Figure 1. Examples of occurrences of G-C RWC base pairs in (a) yeast tRNAPhe (PDB Code: 1EHZ20) and (b) 23S rRNA from T. thermophilus (PDB Code: 3F1F39). The Mn2+ ion (violet) bound to the tRNA N7(G15) atom and the oxygen atoms of the five coordinated water molecules (red) are also shown.

theoretical study has been performed to shed light on the binding of Mn2+ and Co2+, relative to that of Mg2+, although several excellent theoretical studies have clarified the interaction of several alkaline earth and transition metals with the N7 atom of purines, and their influence on the H-bond pairing with other bases.14-19 One of the reasons for the lack of studies focused on Mn2+ and Co2+ probably is the technical and computational difficulties to treat properly these open-shell metals in the context of the high level computational methods required to model accurately base-pairing interaction. Indeed, in this paper we try to fill this gap by a comparative analysis of the effect of Mg2+, Mn2+, and Co2+ on (i) a single guanine base, (ii) a guanine base involved in a G-C WC base pair, and (iii) a guanine base involved in a G-C RWC base pair. In all cases the first coordination sphere of the metal is saturated by water molecules, since previous studies have clearly indicated that severe geometric deformations occur when a naked cation is coordinated to the N7 atom of purines. For this reason, we initiate our analysis with a comparative study of the hexahydrated aqauions of generic formula [M(H2O)6]2+ (M ) Mg, Mn, Co). It is difficult to overemphasize the importance of the G-C WC geometry in nucleic acids, both for its frequency and for its stability; therefore a detailed comprehension of the influence of Mn2+ and Co2+ binding on its geometry and stability is of very general interest. As regards the G-C RWC base pair, it was first observed in tRNAs34,35 and until recently it seemed to be unique to such molecules. However, the recent resolution at atomic level of the structure of several ribosomes allowed evidence of multiple occurrences of G-C RWC base pairs. They were first observed in the 23S rRNA from H. marismortui,21,36,37 and more recently in that from D. radiodurans38 and from T. thermophilus,39 with participation in building an astonishing variety of tertiary structure motifs; see Figure 1. Therefore the importance of such geometry as a structural motif that allows

Computational Details. The Turbomole 5.10 package40 was used for all the calculations reported here. All the geometry optimizations were performed in the gas phase using the hybrid density functional B3LYP functional41-43 and the SVP basis set44 on H, C, N, and O, and the TZVP basis set45 on the metals. For better energetics we performed single point energy calculations on the B3LYP geometries at the second-order Møller-Plesset, MP2, level of theory46 in the framework of the Resolution of Identity approximation.47 For these calculations we used the more extended aug-cc-pVTZ basis sets44 on H, C, N, and O,44 while on the metal we used the def2-TZVPP basis set.48 In the RI-MP2 calculations the aug-cc-pVTZ auxiliary basis set was used for main group atoms,49 while the TZVPP auxiliary basis set was used for the metals.50 All the MP2 interaction energies were corrected for basis set superposition error (BSSE) using the counterpoise procedure.51 The MP2 calculations have been performed both in the gas phase and in solution (water). Solvent effects were included with the conductor screening model (COSMO) implemented in Turbomole.52 A dielectric constant ε ) 78 was used together with a solvent radius of 1.30 Å to model the solvent. The optimized radii of 1.30, 2.00, 1.83, and 1.72 Å were used for H, C, N, and O, while for the metals we used the Bondi radii53 of 1.638, 2.223, and 2.223 Å for Mg, Mn, and Co, respectively. In the MP2 calculations the molecular orbitals corresponding to the 1s core of C, N, and O, to the 1s2s core of Mg, and to the 1s2s2p core of Mn and Co were frozen. Finally, we also tested if inclusion of dispersion corrections improves the agreement with the MP2 results. To this end we performed single point calculations on the B3LYP geometries using the Turbomole 5.10 implementation of the B97-D functional,54 and we added dispersion corrections to the B3LYP values by using the $disp keyword of Turbomole. This last set of numbers is denoted as B3LYP+D. The total base pair interaction energy ∆E was calculated as in eq 1:

∆E ) EG-C - (EG0 + EC0) + BSSE

(1)

where EG-C is the energy of the optimized G-C base pair (with or without a hydrated metal coordinated to the guanine), EG0 and EC0 are the energies of the isolated and optimized guanine (with or without a coordinated hydrated metal) and cytosine bases, and BSSE is the basis set superposition error. This is the standard approach when the main interest is focused on the energetics of the base pairing.23,55-57 This is different from the energy decomposition scheme that partitions the total interaction energy as a sum of three pairwise interaction energies (metal-guanine, metal-cytosine, and guanine-cytosine) and a three-body term, and that proved very useful to understanding the physics of metal binding to nucleic acid base pairs.14-19 The reason for adopting the more classical expression of eq 1 is that we are indeed interested in the effect of the three selected metals on the energetics of the base pairing, more than in the detailed decomposition of the total interaction energy of the complex formed by a metal and a base pair. In the following we will compare the B3LYP results with the MP2 results, although the MP2 values are clearly much more reliable for these systems. Indeed, previous studies have clearly indicated that DFT functionals on the average underestimate

15672

J. Phys. Chem. B, Vol. 113, No. 47, 2009

interaction energies in H-bonded base pairs, and that more sophisticated post Hartree-Fock methods, at least at the MP2 level, are required.57 However, it is clear that such expensive post Hartree-Fock methods can be routinely used only with small model systems of a maximum of two to three bases interacting with a single metal atom. This would make quantum chemistry a substantially limited method, since larger systems could not be investigated. Thus, calibration of much faster DFT methods versus high level calculations remains a fundamental step in almost any quantum mechanical study of fundamental interactions in nucleic acids. We remark here that for the Mn2+ systems in the hextet state the calculated S2 value is never greater than 8.754, while for the Co2+ systems in the quartet state it is never greater than 3.756. These values are very close to ideal values of 8.75 and 3.75 for the pure hextet and quartet states, respectively, which indicates that spin contamination is not a problem in these calculations. Moreover, in all the Mn2+ and Co2+ systems the spin density on the metal atom is never smaller than 4.85e and 2.77e, respectively, which implies that the spin density distributed over all the other atoms is always smaller than 0.25e. For this reason we limit the discussion of the spin density to the aquaions only. The total spin densities on the metal atoms in all the systems are reported in the Supporting Information. Finally, in the following we will neglect contributions to the interaction energy that should be derived from normal-mode analysis, since the accuracy of the harmonic approximation when dealing with weakly bound systems is unclear. For this reason, and considering that the systems we have investigated are extremely similar, we believe it is reasonable to assume that these corrections are rather similar in the various systems and thus cancel out. Structure Analysis. The subset of PDB structures used in this work was collected from the wwPDB58 updated to June 2, 2009. It is composed of 804 structures solved at a resolution of 3.0 Å or better, all containing Mg2+ and/or Mn2+ and/or Co2+ ions in addition to at least one RNA and/or DNA molecule. Metal ions were considered directly bound to the DNA/RNA molecules when they were found at a distance below 2.50 Å from any nucleotide/base atom. Occurrences of G-C base pairs in the RWC geometry were searched as described in ref 21. To find occurrences of G-C WC base pairs, we searched the structure data set for G and C nucleotides presenting N2(G)-O2(C) andO6(G)-N4(C)H-bonds(threshold3.4Å)andaC1′(G)-C1′(C) distance above 8.0 Å. Results The Aquaions. Electronic States. For the three metals considered here, we assumed the well-accepted coordination number of 6,59-62 and thus we investigated hexahydrated aquaions of general formula [M(H2O)6]2+. As for the electronic state, the closed-shell singlet clearly is the only electronic state of low energy for the Mg2+ aquaion, whereas for the Mn2+ and Co2+ aquaions different electronic states of low energy are possible. For this reason we performed geometry optimization of the d5 Mn2+ aquaion in the doublet, quartet, and hextet electronic states, and of the d7 Co2+ aquaion in the doublet and quartet electronic states. For both metals we found the high spin state to be the most stable, with the doublet and the quartet respectively 59.2 and 40.7 kcal/mol higher in energy than the hextet for the Mn2+ aquaion, while for the Co2+ aquaion the doublet is 23.8 kcal/mol higher in energy than the quartet. These preferences are in agreement with the known preference for the high spin state in the case of first-row transition metal complexes

Oliva and Cavallo with weak-field ligands such as water.63,64 Therefore, in the following we were restricted to the high spin hextet and quartet electronic states for species containing the Mn2+ and Co2+ cations, respectively. Optimal Geometries and Natural Population Analysis (NPA) Charges. In agreement with previous results,60 the optimized geometries of the hexacoordinated aquaions present an almost regular octahedral geometry around the metal. The Mg-OH2, Mn-OH2, and Co-OH2 distances, 2.09, 2.20, and 2.12 Å, respectively, are in very good agreement with the experimental average values of 2.09, 2.19, and 2.11 Å for the hydrated Mg2+, Mn2+, and Co2+ cations in dilute aqueous solutions.59 These M-O distances follow the trend in the Pauling-type crystal ionic radii of hexacoordinated ions: Mg, 0.72 Å; Mn, 0.83 Å; and Co, 0.75 Å.65,66 Geometry optimizations performed at the MP2 level result in slightly longer distances of 2.11, 2.23, and 2.13 Å for Mg2+, Mn2+, and Co2+, respectively, which confirms the general idea that the B3LYP geometries are usually quite accurate,57 and that they can be used for single point calculations with more time-consuming approaches. The B97-D optimized aquaions, with M-water distances of 2.13, 2.21, and 2.13 Å for Mg2+, Mn2+, and Co2+, respectively, also result in excellent agreement with the MP2 geometries. The NPA charges of the three metals, Mg ) 1.82e, Mn ) 1.67e, and Co ) 1.61e, indicate that only a small fraction of electron density is transferred from the six water molecules to the formally dicationic metals. Nevertheless, roughly 0.20e is transferred to the Mg2+ cation, whereas 0.33e and 0.39e are transferred to the Mn2+ and Co2+ cations, respectively. As previously indicated,60 the higher charge transfer in the case of the two transition metals is due to the presence of accepting 3d orbitals of low energy, and the higher charge transfer in the case of Co2+ relative to Mn2+ can be reasonably ascribed to the shorter Co-O distance and to the higher electron affinity of the Co. The NPA charges also indicate that in the Mn2+ and Co2+ aquaions the spin density is almost completely localized on the metal, with the R-β electron density reduced by roughly 0.20e from the ideal values of 5e and 3e for Mn2+ and Co2+, respectively, which means that the spin density on each water molecule is only 0.03e roughly. The Wiberg bond index of the M-O bonds, Mg ) 0.056, Mn ) 0.082, and Co ) 0.095, clearly indicates a high ionic character for the M-O bonds, though the Mn-O and Co-O bonds present a slightly higher covalent character. Binding Energies. On the energetics side, the B3LYP gasphase binding energy of a single water molecule to the three hexa-aquaions are 34.8, 31.3, and 32.8 kcal/mol, for Mg2+, Mn2+, and Co2+, respectively (see Table 1), and follows an inverse trend with the Pauling-type ionic radii (i.e., the shorter the M-OH2 bond, the higher the binding energy). A similar trend is observed in the MP2 gas-phase binding energies. Of course, when the continuous solvent model is included in the calculations, all the binding energies are strongly reduced, although the decrease of the binding energy of the water molecule to the two transition metal cations, roughly 15.5 kcal/ mol, is about 1 kcal/mol smaller than that to Mg2+ cation, 16.6 kcal/mol. This difference is another indication of the higher ionic character of the bond between the alkaline earth metal cation and water relative to that involving the two transition metal cations. Since the B3LYP values already overestimate the MP2 binding energies, the addition of the dispersion term (see the B3LYP+D column in Table 1) clearly results in an even larger overestimation. On the other hand, the B97-D results are in excellent agreement with the MP2 values.

Ion Binding on Guanine in WC and RWC Base Pairs

J. Phys. Chem. B, Vol. 113, No. 47, 2009 15673

TABLE 1: Binding Energy (in kcal/mol) of a Water Molecule and of the Guanine to the [Mg(H2O)5]2+ Moiety B3LYP (gas)

B3LYP+D (gas)

B97-D (gas)

MP2 (gas)

MP2 (water)

Mg Mn Co

34.8 31.3 32.8

[M(H2O)5 · · · H2O]2+ 37.9 31.3 33.9 29.5 36.1 29.2

31.1 28.7 30.6

14.5 13.3 15.0

Mg Mn Co

82.1 80.5 85.4

[(H2O)5M · · · G]2+ 91.4 76.4 89.4 76.6 95.3 81.4

82.0 82.3 87.7

24.8 26.9 32.5

Mg Mn Co

93.1 91.3 97.6

[(H2O)5M · · · G-C]2+ (WC) 103.2 87.6 94.5 107.0 87.9 94.7 108.0 94.8 101.3

25.0 27.0 32.9

108.9 107.4 113.6

[(H2O)5M · · · G-C]2+ (RWC) 119.4 102.7 108.9 117.5 103.3 109.0 124.9 110.2 115.5

27.1 29.2 35.3

Mg Mn Co

The Guanine Base. Optimal Geometries. We first tested that the high spin hextet and quartet electronic states remain the most stable for Mn2+ and Co2+ when a water molecule of the hexahydrated aquaions is replaced by a guanine. The B3LYP optimizations result in the quartet [(H2O)5MnG]2+ and doublet [(H2O)5CoG]2+ species being 35.3 and 16.3 kcal/mol higher in energy than the corresponding hextet and quartet states, respectively. This indicates that replacing a water molecule with a guanine reduces the preference for the high spin state that, nevertheless, remains clearly favored. For this reason, in the following we will focus only on the high spin hextet and quartet electronic states for all the Mn and Co species with a coordinated guanine. The optimized geometries of the [(H2O)5MG]2+ species (M ) Mg, Mn, and Co) are reported in Figure 2. In agreement with previous studies on dication coordination to the N7 atom of the guanine, two of the water molecules are engaged in H-bonds with the O6 atom of the guanine (in all cases we use the simple term “cation/metal” to refer to the pentahydrated cation/metal).14-19 Coordination of the metal results in an elongation of the M-O bond trans to the N7 atom, while the other M-O distances are scarcely affected. Interestingly, the trend in the M-N7 distances, Co < Mg < Mn, does not replicate the trend in the M-water distances, Mg < Co , Mn. Moreover, while the Mg-N7 distance is 0.09 Å longer than the Mg-O distance in [(H2O)5MgG]2+, the Mn-N7 and Co-N7 distances are only 0.03 and 0.01 Å longer, respectively, than the M-O distance in the corresponding hexahydrated aquaions. The longer Mg-N bond relative to the Mg-O bond reflects the larger radius of N relative to O,67 whereas the almost identical M-N and M-O bonds when M is Mn or Co suggests a different bonding scheme involving empty d orbitals of the transition metal cation, which are able to accept electron density from aromatic molecular orbitals of the guanine.13,14 As concerns the guanine, the greatest effect is in the bond distances that involve the C6 atom. In fact, the C6-O6 distance is roughly 0.05 Å longer in the presence of the metals, due to the H-bond interaction of the O6 atom with two water molecules coordinated to the metal, which reduces the double-bond character of the C6-O6 bond. As a consequence, the minor involvement of the p(π) orbital of the C6 atom in the carbonyl bond results in a higher participation of this orbital to the aromatic system of the guanine, as indicated by the roughly

Figure 2. Structure of the G base in the absence (a) and in the presence of a metal atom coordinated to the N7 atom (b). Distances in Å.

0.05 Å shorter C6-N1 and C6-C5 bond distances in the presence of the metals. The increased electron density in the aromatic system of the guanine in the presence of the metals also results in a shortening of the C2-N2 bond of 0.03 Å. These structural changes result in a decreased Wiberg bond order for the C6-O6 bond from 1.712 in the isolated guanine to roughly 1.46 upon metal coordination, and in an increase of the same parameter from 1.100 to roughly 1.18 in the case of the C6-N1 bond, and from 0.993 to roughly 1.12 in the case of the C6-N5 bond. NPA Charge Analysis. The rather strong effect of the [(H2O)5M]2+ moiety on the C6-O6 and C2-N2 bonds is translated in sensible variation of the NPA charges of the O6 and H2 atoms involved in the H-bonding with the cytosine, in the standard WC as well as in the RWC geometries (see below). The NPA charge of the O6 donor atom is increased from -0.59e in the isolated guanine to roughly -0.70e in the presence of the [(H2O)5M]2+ moiety, which makes the O6 atom a better H-bond donor. At the same time, the NPA charge of the two H2 atoms is increased from 0.38-0.40e in the isolated guanine to 0.42-0.44e in the presence of [(H2O)5M]2+ moiety, which makes the N2 group a better H-bond acceptor. Finally, the NPA charge of the H1 atom, which is also involved in the H-bonding with the cytosine, is slightly increased from 0.41e in the isolated guanine to 0.43e in the presence of [(H2O)5M]2+ moiety. Focusing on the metals, replacing one of the water molecules of the hexahydrated cations with the N7 atom of guanine results in a small reduction of the NPA charge, by roughly 0.03-0.05e, independently of the specific metal considered. The Wiberg bond orders of the M-N7 bond, 0.082, 0.133, and 0.149 for Mg, Mn, and Co, respectively, are quite higher than the M-O Wiberg bond orders in the hexa-aquaions. Nevertheless, the increase of the Wiberg bond order of the M-O bonds involving the two transition metals, roughly 0.50, is remarkably greater than the increase involving the alkaline earth metal. Again, this suggests for a different bonding scheme that involves empty d orbitals in the case of the transition metal cations. Binding Energies. The binding energy of the guanine to the [(H2O)5M]2+ moiety is roughly 80 kcal/mol in the gas phase, at both the B3LYP and MP2 levels (see Table 1), and it decreases to roughly 25 kcal/mol when implicit solvent effects

15674

J. Phys. Chem. B, Vol. 113, No. 47, 2009

are included. The binding of Mn2+ is slightly favored over that of Mg2+, while binding of Co2+ is far more favored than that of Mg2+ and Mn2+. This finding is in qualitative agreement with experiments that indicated that Mn2+ and Co2+ can displace the Mg2+ ion at this coordination site.20 Our gas-phase binding energy for Mg2+ is roughly 7 kcal/mol smaller than the MP2 values calculated with the 6-31G(d) basis set some years ago,17 evidencing that accurate MP2 energies require that rather extended basis sets be used.57 The strong gas-phase binding energies are clearly influenced by the stabilizing H-bond interactions that are established between the water molecules coordinated to the metal and the O6 atom of the guanine. However, guanine coordination to naked M2+ cations were shown to induce severe geometric distortions,13,14 and thus it is almost impossible to calculate the strength of the specific M-N interaction in these systems. Moving to the dispersion corrected results, the addition of the dispersion term to the B3LYP values deteriorates the otherwise excellent agreement with the MP2 values. The B97-D results, instead, clearly underestimate the MP2 values. Comparison of the energetics of water displacement between the different metals shows that displacement of a water molecule by a guanine is more favorable for the Mn2+ and Co2+ aquaions than for the Mg2+ aquaion, by 1.9 and 5.2 kcal/mol respectively, with the B3LYP approach. The MP2 values, both in the gas phase, of 2.6 and 6.2 kcal/mol, respectively, and with the continuous solvation model, of 3.3 and 7.2 kcal/mol, respectively, substantially confirm the B3LYP values. As for the dispersion corrected results, the B3LYP+D values somewhat overestimate the MP2 values, which are almost in the middle between the B3LYP and the B3LYP+D values. The B97-D results, instead, clearly underestimate the MP2 values, a clear consequence of the relatively weak performance of the B97-D functional in describing the binding of the guanine to the metals. Overall, the increased preference for guanine coordination in the case of the two transition metals relative to the alkaline earth metal is consistent with the quite higher Wiberg bond order calculated for the Mn-N and Co-N bonds relative to the Mg-N bond, and further supports a more covalent character in the M-N bond when M is a transition metal. It is also of interest that the Co-N bond is roughly 5 kcal/mol stronger than the Mn-N bond (depending on the level of theory), while the Co-O bond is only 2 kcal/mol stronger than the Mn-O bond. The greater stability of the Co-N bond relative to the Mn-N bond can be related to the shorter Co-N bond, which results in a better overlapping between filled aromatic molecular orbitals on the guanine with empty d orbitals on the Co2+ atom. As a final remark, we note that the agreement between the absolute BSSE uncorrected B3LYP values and the BSSE corrected MP2 values is remarkable, considering that the B3LYP calculations, also due to the much smaller basis set used (SVP for the B3LYP calculations vs the aug-cc-pVTZ for the MP2 calculations) only take a fraction of the time needed for the MP2 calculations. The Watson-Crick G-C Base Pair. Optimal Geometries. The optimized geometries of the isolated G-C WC base pair and of the [(H2O)5MG-C]2+ species (M ) Mg, Mn, and Co), with the G-C base pair in the WC geometry, are reported in Figure 3. In agreement with previous studies, the isolated G-C base pair in the WC geometry is substantially planar, and the optimized H-bond distances are in excellent agreement with the MP2optimizeddistancesofN2(G)-O2(C))2.89Å,N1(G)-N3(C) ) 2.90 Å, and O6(G)-N4(C) ) 2.75 Å.57 The planarity of the WC base pair is maintained in the presence of the [(H2O)5M]2+

Oliva and Cavallo

Figure 3. Structure of the Watson-Crick G-C base pair in the absence (a) and in the presence of a metal atom coordinated to the N7 atom of the G base (b). Distances in Å.

moiety coordinated to the N7 atom of the guanine.14,17,21 Despite this similarity, in agreement with these previous studies we found a remarkable increase of the O6(G):N4(C) H-bond by roughly 0.25 Å, and a corresponding shrinking of the N2(G): O2(C) H-bond of roughly 0.22 Å, while the N1(G):N3(C) H-bond length is substantially unchanged. The elongation of the O6(G):N4(C) H-bond has been ascribed to repulsive electrostatic interaction between the cation and the N4(C) amino group,17 while the shrinking of the N2(G):O2(C) H-bond can be rationalized by considering that the H-bond acceptor capability of the N2(G) amino group is enhanced by metal coordination (see previous section). Interestingly, the C1′(G)-C1′(C) distance is reduced marginally by metal coordination, from 10.7 to 10.5 Å, which is fundamental to preserving the isosteric geometry requirements for insertion of this base pair into the double-helix motif typical of nucleic acids.68 The base pairing also affects the M-N7(G) distances, which are roughly 0.03 Å shorter than in the absence of the cytosine, while the H-bonds between the O6(G) atom and two of the water molecules coordinated to the metal are substantially unaffected. Finally, we tested the performance of the B97-D functional in reproducing the WC geometry in the absence of any coordinated metal. The three H-bond distances, N2(G)-O2(C) ) 2.93 Å, N1(G)-N3(C) ) 2.94 Å, and O6(G)-N4(C) ) 2.77 Å, slightly overestimate the MP2 optimized values of 2.89, 2.90, and 2.75 Å.57 Base Pair Interaction Energies. In agreement with previous studies,14,17,21 the standard G-C WC base pair in the gas phase is substantially stabilized by a metal coordinated to the N7(G) atom, and this additional stability amounts to roughly 10 kcal/ mol in the gas phase. The computationally cheap B3LYP values overestimate the MP2 pair interaction energies by roughly 2-4 kcal/mol, and consequently, addition of the dispersion term results in the B3LYP+D values overestimating the MP2 values even more. The B97-D values, instead, are in excellent agreement with the MP2 values. Overall, the remarkable stabilization in the base pairing after metal binding has been essentially ascribed to a three-body term due to a polarization effect of the M2+ cation on the proximal G base that strengthens the H-bonding interaction between the two bases.14,17,69 However, when solvent effects are considered, we found that the

Ion Binding on Guanine in WC and RWC Base Pairs additional stability of the G-C WC base pair is reduced to 0.1, 0.5, and 0.8 kcal/mol only for Mg2+, Mn2+, and Co2+, respectively. This finding is in line with previous studies showing that inclusion of explicit water molecules to saturate the first coordination shell of the cation bound to the N7(G) atom already reduces the three-body polarization term by roughly 50%,14 and seems to suggest that the accurate evaluation of the effect of metal binding on the base pair interaction energy requires the inclusion of solvent effects beyond explicit water molecules in the first coordination sphere of the metal. As for the influence of the G-C interaction on the M-N7(G) bond, formation of a G-C base pair in the WC geometry stabilizes the M-N7(G) bond by roughly 12 kcal/mol in the gas phase, a value that reduces to 0.2-0.5 kcal/mol in water. In conclusion, coordination of Mg2+ to the N7(G) atom seems to have a negligible effect on the stability of the G-C WC base pair when solvent effects (beyond the first coordination shell around the metal) are included, while a more relevant effect is predicted for the two transition metals. This quantitative reduction of the three-body interaction term when an implicit solvent model is used can be reasonably explained considering that the solvent strongly screens the component corresponding to the electrostatic polarization of the electron density,17 while the component corresponding to polarization through the M-N7(G) bond, particularly through back-donation from the π aromatic ring of the guanine to empty d orbitals on the metal, remains active. This explains the almost negligible stabilizing effect of the Mg2+ cation, for which no appreciable back-donation can be expected, and the increased stabilizing effect of the Mn2+ and Co2+ transition metals, where a sizable back-donation occurs. Finally, although reduced in magnitude compared to gas-phase calculations, the N7(G)-M bond is slightly stabilized (by roughly 0.2-0.5 kcal/mol) by formation of the G-C WC base pair, and thus the G-C WC base pair is slightly more effective that an isolated guanine to displace a water molecule from hexaydrated aquaions. The Reverse Watson-Crick G-C Base Pair. Optimal Geometries. The optimized geometries of the isolated G-C base pair and of the [(H2O)5MG-C]2+ species (M ) Mg, Mn, and Co), with the G-C base pair in the RWC geometry, are reported in Figure 4. In agreement with previous studies the isolated G-C RWC base pair strongly deviates from planarity, and the N2(G): N3(C) H-bond is remarkably long, almost broken. The high instability of the simple G-C RWC base pair in the gas phase is well-known23,70 and it has been ascribed to carbonyl-carbonyl and amino-amino repulsion.70 Depending on the level of theory, basis set, and also optimization algorithm used, the geometry optimization of the G-C RWC base pair can even collapse, as indicated in the Introduction, into a rather more stable geometry that presents a bifurcated H-bond pattern.21 Differently, the presence of the [(H2O)5M]2+ moiety coordinated to the N7 atom of the guanine results in a remarkably stable G-C RWC base pair. The optimized geometries are reported in Figure 4b, and in all cases the overall base pair is definitely planar. Independently from the metal, both the N2(G): N3(C) and the N3(G):O2(C) H-bonds are perfectly established, although both the H-bonds are marginally shorter in the presence of the Co2+ cation. The base pairing also affects the M-N7(G) distances, which are roughly 0.04 Å shorter than in the absence of the cytosine. The base pairing also results in a shortening of 0.04 Å roughly of the H-bonds between the O6(G) atom and two of the water molecules coordinated to the metal. Base Pair Interaction Energies. The energetics of the G-C RWC base pair with a coordinated metal reflects the remarkable

J. Phys. Chem. B, Vol. 113, No. 47, 2009 15675

Figure 4. Structure of the reverse Watson-Crick G-C base pair in the absence (a) and in the presence of a metal atom coordinated to the N7 atom of the G base (b). Distances in Å.

structural changes that occur upon metal coordination. In fact, all the metals stabilize the G-C RWC base pair geometry by roughly 25 kcal/mol in the gas phase at both the B3LYP and MP2 levels of theory; see Table 3. As for the WC base pairs, the B3LYP values slightly overestimate the MP2 pair interaction energies, by roughly 1-2 kcal/mol and, consequently, addition of the dispersion term results in the B3LYP+D values overestimating the MP2 values even more. Differently, also in this case the B97-D values are in excellent agreement with the MP2 values. Focusing on the metal, Co2+ is the cation that stabilizes most the RWC geometry, with Mg2+ and Mn2+ are roughly 1 kcal/mol less effective. As in the case of the G-C WC base pair, the interaction energy is strongly reduced when solvent effects are considered. Nevertheless, the stabilizing effect of the metal remains sizable, since the metals stabilize the G-C RWC base pair by roughly 2 kcal/mol when solvent effects are considered. The Co2+ cation is slightly more effective in stabilizing the G-C RWC geometry. As explained in previous work,17,21 metal coordination has a synergic effect. It depletes electron density of the O6(G) atom away from the O2(C) atom, which results in reduced electrostatic repulsion, and it depletes electron density from the N1(G) and N2(G) amino groups, making them better H-bond acceptors. As for the influence of the G-C interaction on the M-N7(G) bond, formation of a G-C RWC base pair stabilizes the M-N7(G) bond by roughly 25 kcal/mol in the gas phase, which reduces to roughly 2 kcal/ mol only in water. The stronger M-N7(G) bond in the G-C RWC geometry clearly makes metal coordination from the hexahydrated aquaion more favored than in the case of an isolated guanine base or of a G-C WC base pair; see Table 2. Indeed, comparison between the water displacement energies of Table 2 indicates that the metals have a slightly stronger preference for the G-C WC base pair with respect to an isolated guanine, although this effect is almost negligible, only 0.2-0.4 kcal/mol. This preference increases to 2.3-2.9 kcal/mol in the G-C RWC base pair. Coordination of the Co2+ is roughly 4 kcal/mol more favored relative to Mn2+ coordination, and it is favored by roughly 7 kcal/mol relative to Mg2+.

15676

J. Phys. Chem. B, Vol. 113, No. 47, 2009

Oliva and Cavallo

TABLE 2: Energy (in kcal/mol) of the [M(H2O)6]2+ + G S [(H2O)5MG]2+ + H2O and [M(H2O)6]2+ + G-C S [(H2O)5MG-C]2+ + H2O Substitution Reactions (M ) Mg, Mn, Co) B3LYP (gas)

B3LYP+D (gas)

B97-D (gas)

[M(H2O)6] + G S [(H2O)5MG] -47.1 -53.3 -45.0 -49.0 -55.4 -47.1 -52.3 -59.0 -52.2 2+

Mg Mn Co Mg Mn Co Mg Mn Co

2+

MP2 (gas)

MP2 (water)

+ H 2O -51.0 -53.6 -57.2

-10.3 -13.6 -17.5

[M(H2O)6]2+ + G-C S [(H2O)5MG-C]2+ + H2O (WC) -58.4 65.2 -56.2 -63.5 -10.5 -60.0 -66.9 -58.4 -66.0 -13.7 -64.8 -71.8 -65.6 -70.7 -17.9 [M(H2O)6]2+ + G-C S [(H2O)5MG-C]2+ -74.1 -81.5 -71.4 -76.0 -83.6 -73.8 -80.8 -88.8 -80.9

+ H2O (RWC) -77.8 -12.6 -80.3 -15.9 -85.0 -20.4

TABLE 3: Interaction Energy (in kcal/mol) of the G-C Base Pair in the WC and in the RWC Geometry in the Absence and in the Presence of a [M(H2O)5]2+ Moiety Coordinated to the N7 Atom of the Guanine (M ) Mg, Mn, Co) B3LYP (gas) no M Mg Mn Co

31.2 42.2 42.0 43.5

no M Mg Mn Co

15.7 42.5 42.5 44.0

B3LYP+D (gas)

B97-D (gas)

MP2 (gas)

MP2 (water)

27.1 39.2 39.9 41.0

11.8 11.9 12.3 12.6

Reverse Watson-Crick Geometry 19.2 13.1 14.4 47.3 39.4 40.6 47.3 39.8 40.5 48.9 41.8 41.5

7.4 9.1 9.1 9.6

Watson-Crick Geometry 36.7 27.3 48.6 38.1 48.3 38.6 49.4 40.6

As for a comparison between the two G-C geometries, for both B3LYP and the MP2 the gas-phase values suggest that the RWC geometry becomes very competitive and even favored with respect to the standard WC geometry, but this is an artifact of the gas-phase calculations. In fact, as solvent effects are included, the WC geometry with a coordinated metal is roughly 3 kcal/mol more stable than the RWC geometry. Nevertheless, metal coordination reduces the energy preference for the WC geometry by roughly 1.5 kcal/mol with respect to the same geometries in the absence of a metal. Frequency of Mg2+, Mn2+, and Co2+ Ions Bound to Guanine Bases in DNA and RNA Structures. In this section we report a statistical analysis of the binding of Mg2+, Mn2+, and Co2+ to DNA and RNA structures available in the wwPDB.58 Clearly, comparison with the calculations presented in the previous sections can be made on a qualitative basis only, since a full comparison would require the accurate calculation of the binding mode of the three metals to phosphates, to sugars, and also to the other bases besides guanine. Nevertheless, our calculations have indicated that guanine binding to Mn2+ and Co2+ is reinforced by sensible back-donation from the aromatic system of the guanine to empty d orbitals of the transition metals, a bonding contribution which is not possible for water in the aquaions and for phosphates and sugars as well. With this caveat in mind, we discuss the results of a comparative analysis of the preferred coordination sites of Mg2+, Mn2+, and Co2+ on nucleic acid molecules; see Table 4. Not surprisingly, 96% of the total three metal ions found in the PDB structures (13 071 in total) correspond to Mg2+, with Mn2+ and Co2+

TABLE 4: Frequency of Binding of Metals to Different Sites on Nucleic Acid Molecules Mg hits

Mn %

hits

Co %

M-any nucleotide 7598 100.0 208 100.0 M-any base 1140 15.0 112 53.8 M-G base 690 105 M-N7(G) 136 84 M-N7(G) in G-C WC 34 66 M-N7(G) in G-C RWC 1 1 -

hits

%

31 22 21 21 10 1

100.0 71.0 -

representing only 2.7% and 0.4% of the occurrences, respectively. However, despite their abundance, Mg2+ ions mostly bind to the ribose-phosphate backbone of DNA/RNA molecules, since only 15% of the Mg2+ is directly bound to a base; see Table 4. Differently, the two transition metals have a remarkable tendency (54% and 71% for Mn2+ and Co2+, respectively) to specifically coordinate to a base atom, and this tendency is clearly stronger for Co2+. When looking at the specific bases, 690 of the 1140 Mg2+ ions are bound to a guanine, whereas almost all the Mn2+ and Co2+ ions were found to be bound to a guanine; see Figure 5. Percentually, 61%, 94%, and 95% of the total Mg2+, Mn2+, and Co2+ directly coordinated to a base are coordinated to a guanine. Considering that guanine represents about 30% of all the bases in the DNA/RNA molecules in the structures we analyzed, it is clear that all three metals show a clear preference for binding to guanines, and this preference is remarkably stronger in the case of the two transition metals, which seem to bind almost exclusively to guanines. Out of all the metal ions which are bound to a guanine, 11% of Mg2+, 80% of Mn2+, and 100% of Co2+ are bound to the N7

Figure 5. Distribution of the binding sites of Mg2+, Mn2+, and Co2+ ions on nucleotide bases in DNA and RNA PDB structures. The fraction of ion bound to the N7 atom of a guanine is shown in dark green, the fraction of ion bound to any other atom of a guanine is shown in light green, and the fraction of ion bound to any base but a guanine is colored gray. The number of coordinated metal ions and relative percentages are also reported.

Ion Binding on Guanine in WC and RWC Base Pairs atom; see Figure 5. This result highlights a clear preference of the two transition metals for the N7 atom, whereas Mg2+ prefers the more electronegative O6 atom as a binding site. This can be easily explained considering that the driving force for the Mg2+ coordination is mainly electrostatic in nature, whereas the two transition metals can take advantage of back-donation from the guanine aromatic ring. Furthermore, 25%, 79%, and 47% of the guanines with a Mg2+, Mn2+, or Co2+ ion bound to N7 are engaged in a G-C WC base pair, respectively. In other words, a high percentage of the guanines with a Mn2+ bound to their N7 is found to form a WC base pair, whereas this percentage drops to 45% and 25% for Co2+ and Mg2+, respectively, although the small number of occurrences for Co2+ ions bound to a guanine, 21 structures in total, makes the statistics for Co2+ probably not converged. These findings are in qualitative agreement with our calculations (see above) that indicated that a transition metal coordinated to the N7 of a guanine reinforces the stability of the G-C WC base pair, while Mg2+ coordination has only a minor effect. As for the G-C RWC base pair, only one occurrence per metal ion is found.20 This is not surprising and reflects the fact that differently from the WC base pairs, which are of course widespread in both DNA and RNA structures, the G-C RWC base pair has been until now only found in a dozen tRNAs and in the 23S subunit of ribosomes from H. marismortui,37 D. radiodurans,38 and T. thermophilus.39 Although the three occurrences of metal ions found bound to a N7(G) in tRNA structures are not very significant from a statistical point of view, it is worth noting that the presence of a strong binding site for divalent metal ions on the tRNA G base involved in the RWC base pair has been widely documented.20,71-73 Conclusions Our comparative analysis of the Mg2+, Mn2+, and Co2+ hexahydrated aquaions has indicated a clear inverse relationship between the radius of the cation and the binding energy that increases in the order Mn2+ < Co2+ < Mg2+, with the B3LYP values reproducing reasonably well the MP2 values. Inclusion of dispersion correction does not result in consistently better or worse agreement with the MP2 results. For example, the B97-D functional poorly reproduces the MP2 binding energy of the guanine to the metals, while it excellently reproduces the MP2 base pair interaction energies. Further, the B97-D results in an M-water distance in the aquaions close to the MP2 value and slightly overestimate the MP2 H-bond distances in the WC base pair. Natural population analysis suggested a negligible charge transfer from the water molecules to the metals, and in the case of the Mn2+ and Co2+ aquaions the spin density is almost completely localized on the metal. As concerns the binding of pentahydrated Mg2+, Mn2+, and Co2+ to the N7 atom of guanines, our calculations indicated a different trend in the binding energies, which increase in the order Mg2+ < Mn2+ < Co2+, suggesting a rather different bonding scheme that involves back-donation from the aromatic ring of the guanine to empty d orbitals on the two transition metals. Binding of Mn2+ is slightly favored over that of Mg2+, while binding of Co2+ is far more favored than that of Mg2+ and Mn2+. This is in good agreement with the high frequency of Co2+ ions found bound to N7(G) in the available experimental structures for DNA and RNA molecules, and with experiments that indicated that Mn2+ and Co2+ ions can displace the Mg2+ ion at this coordination site.20 Binding of the three metals was also shown to affect the geometry of the guanine base, resulting in a sensible variation of the NPA charges of the O6 and H2

J. Phys. Chem. B, Vol. 113, No. 47, 2009 15677 atoms and consequently of the stability of the H-bonds they form with a C base in both the WC and RWC base pairs. Focusing on the G-C WC geometry, we found that in the gas phase this base pair is quite stabilized by a cation coordinated to the N7(G) atom. However, inclusion of a continuous solvation model strongly reduces this stabilizing effect, and we found that the G-C WC geometry in water is only marginally stabilized by Mg2+ (0.1 kcal/mol only), whereas both Mn2+ and Co2+ have a slightly greater effect (0.5 and 0.8 kcal/mol, respectively). This is also consistent with the absence of a clear correlation between the number of guanines in experimental structures with a metal bound to the N7 atom and their involvement in a WC base pair. This result clearly indicates that solvent effects beyond the first explicit solvation shell should be always considered when studying metal coordination to nucleic acids. As concerns the RWC geometry, all three metals stabilize it remarkably also in water, with Co2+ being the most effective (2.2 kcal/mol vs 1.7 kcal/mol of Mg and Mn). In light of these results one would expect to find more occurrences of metals bound to the N7 atom of guanine involved in RWC geometries. However, only a few occurrences of this geometry have been reported to date, although it might be predicted that with more and more complex RNA structures being solved at atomic resolution, more occurrences of this base pair will be observed. Moreover, it is worth noting here that whereas occurrences of metals in the PDB structures can be trustfully considered “true positives”, many occurrences of metals could have not been reported because of technical reasons. As final remark, we believe that our calculations outline the ability of metal ions, such as Co2+, to stabilize the otherwise marginally stable G-C RWC geometry. This could represent a guideline for both wet biologists working with complex (maybe structure-unknown) RNA molecules and crystallographers dealing with the problem of crystallizing such molecules. Acknowledgment. We thank Regione Campania, Progetto L.R. 5 2007, for financial support. We thank ENEA (www.enea.it) and the HPC team for support and for giving us access to the ENA-GRID and the HPC facilities CRESCO (www.cresco.enea.it) in Portici (Naples), Italy. Supporting Information Available: Cartesian coordinates and energies of all species considered in this work. Metal spin density on all the open-shell species. Occurrences of Mg, Mn, and Co ions coordinated to N7(G) in RNA and DNA PDB structures. This material is available free of charge via the Internet at http://pubs.acs.org. References and Notes (1) Pyle, A. M. Science 1993, 261, 709. (2) Pyle, A. M. J. Biol. Inorg. Chem. 2002, 7, 679. (3) Hanna, R.; Doudna, J. A. Curr. Opin. Chem. Biol. 2000, 4, 166. (4) Misra, V. K.; Draper, D. E. Biopolymers 1998, 48, 113. (5) Adams, A.; Lindahl, T.; Fresco, J. R. Proc. Natl. Acad. Sci. U.S.A. 1967, 57, 1684. (6) Ishida, T.; Snyder, D.; Sueoka, N. J. Biol. Chem. 1971, 246, 5965. (7) Lindahl, T.; Adams, A.; Fresco, J. R. Proc. Natl. Acad. Sci. U.S.A. 1966, 55, 941. (8) Madore, E.; Florentz, C.; Giege, R.; Lapointe, J. Nucleic Acids Res. 1999, 27, 3583. (9) Gao, Y. G.; Sriram, M.; Wang, A. H. Nucleic Acids Res. 1993, 21, 4093. (10) Wrzesinski, J.; Jozwiakowski, S. K FEBS J. 2008, 275, 1651. (11) Sponer, J.; Leszczynski, J.; Hobza, P. J. Phys. Chem. 1996, 100, 1965. (12) Sˇponer, J.; Burda, J. V.; Mejzlik, P.; Leszczynski, J.; Hobza, P. J. Biomol. Struct. Dyn. 1997, 14, 613.

15678

J. Phys. Chem. B, Vol. 113, No. 47, 2009

(13) Burda, J. V.; Sˇponer, J.; Leszczynski, J.; Hobza, P. J. Phys. Chem. B 1997, 101, 9670. (14) Sponer, J.; Burda, J. V.; Sabat, M.; Leszczynski, J.; Hobza, P. J. Phys. Chem. A 1998, 102, 5951. (15) Gresh, N.; Sˇponer, J. J. Phys. Chem. B 1999, 103, 11415. (16) Sˇponer, J.; Burda, J. V.; Leszczynksi, J.; Hobza, P. J. Biomol. Struct. Dyn. 1999, 17, 61. (17) Munoz, J.; Sponer, J.; Hobza, P.; Orozco, M.; Luque, F. J. J. Phys. Chem. B 2001, 105, 6051. (18) Munoz, J.; Gelpi, J. L.; Soler-Lopez, M.; Subirana, J. A.; Orozco, M.; Luque, F. J. J. Phys. Chem. B 2002, 106, 8849. (19) Moroni, F.; Famulari, A.; Raimondi, M.; Sabat, M. J. Phys. Chem. B 2003, 107, 4196. (20) Shi, H.; Moore, P. B. RNA 2000, 6, 1091. (21) Oliva, R.; Tramontano, A.; Cavallo, L. RNA 2007, 13, 1427. (22) Levitt, M. Nature 1969, 224, 759. (23) Oliva, R.; Cavallo, L.; Tramontano, A. Nucleic Acids Res. 2006, 34, 865. (24) Kolev, N. G.; Hartland, E. I.; Huber, P. W. Nucleic Acids Res. 2008, 36, 5530. (25) Liao, X.; Anjaneyulu, P. S.; Curley, J. F.; Hsu, M.; Boehringer, M.; Caruthers, M. H.; Piccirilli, J. A. Biochemistry 2001, 40, 10911. (26) Piccirilli, J. A.; Vyle, J. S.; Caruthers, M. H.; Cech, T. R. Nature 1993, 361, 85. (27) Feldman, A. R.; Leung, E. K.; Bennet, A. J.; Sen, D. ChemBioChem 2006, 7, 98. (28) Wan, C.; Cui, M.; Song, F.; Liu, Z.; Liu, S. J. Am. Soc. Mass Spectrom. 2009, 20, 1281. (29) Hou, M. H.; Robinson, H.; Gao, Y. G.; Wang, A. H. Nucleic Acids Res. 2004, 32, 2214. (30) Adams, A.; Guss, J. M.; Collyer, C. A.; Denny, W. A.; Wakelin, L. P. Nucleic Acids Res. 2000, 28, 4244. (31) Valls, N.; Wright, G.; Steiner, R. A.; Murshudov, G. N.; Subirana, J. A. Acta Crystallogr., Sect. D: Biol. Crystallogr. 2004, 60, 680. (32) Duguid, J. G.; Bloomfield, V. A. Biophys. J. 1995, 69, 2642. (33) Miyoshi, D.; Nakao, A.; Toda, T.; Sugimoto, N. FEBS Lett. 2001, 496, 128. (34) Brennan, T.; Sundaralingam, M. Nucleic Acids Res. 1976, 3, 3235. (35) Robertus, J. D.; Ladner, J. E.; Finch, J. T.; Rhodes, D.; Brown, R. S.; Clark, B. F.; Klug, A. Nature 1974, 250, 546. (36) Nagaswamy, U.; Larios-Sanz, M.; Hury, J.; Collins, S.; Zhang, Z.; Zhao, Q.; Fox, G. E. Nucleic Acids Res. 2002, 30, 395. (37) Ban, N.; Nissen, P.; Hansen, J.; Moore, P. B.; Steitz, T. A. Science 2000, 289, 905. (38) Harms, J. M.; Wilson, D. N.; Schluenzen, F.; Connell, S. R.; Stachelhaus, T.; Zaborowska, Z.; Spahn, C. M.; Fucini, P. Mol. Cell 2008, 30, 26. (39) Korostelev, A.; Asahara, H.; Lancaster, L.; Laurberg, M.; Hirschi, A.; Zhu, J.; Trakhanov, S.; Scott, W. G.; Noller, H. F. Proc. Natl. Acad. Sci. U.S.A. 2008, 105, 19684. (40) Ahlrichs, R.; Bar, M.; Haser, M.; Horn, H.; Kolmel, C. Chem. Phys. Lett. 1989, 162, 165. (41) Becke, A. D. J. Chem. Phys. 1993, 98, 5648.

Oliva and Cavallo (42) Becke, A. D. J. Chem. Phys. 1996, 104, 1040. (43) Lee, C.; Yang, W.; Parr, R. G. Phys. ReV. B 1988, 37, 785. (44) Dunning, T. H., Jr. J. Chem. Phys. 1989, 90, 1007. (45) Schaefer, A.; Huber, C.; Ahlrichs, R. J. Chem. Phys. 1994, 100, 5829. (46) Møller, C.; Plesset, M. S. Phys. ReV. 1934, 46, 618. (47) Weigend, F.; Ha¨ser, M. Theor. Chem. Acc. 1997, 97, 331. (48) Weigend, F.; Ahlrichs, R. Phys. Chem. Chem. Phys. 2005, 7, 3297. (49) Weigend, F.; Ko¨hn, A.; Ha¨ttig, C. J. Chem. Phys. 2002, 116, 3175. (50) Weigend, F.; Ha¨ser, M.; Patzelt, H.; Ahlrichs, R. Chem. Phys. Lett. 1998, 294, 143. (51) Boys, S. F.; Bernardi, F. Mol. Phys. 1970, 19, 553. (52) Klamt, A.; Schu¨u¨rmann, G. J. Chem. Soc., Perkin Trans. 2 1993, 5, 799. (53) Bondi, A. J. Phys. Chem. 1964, 68, 441–451. (54) Grimme, S. J. Comput. Chem. 2006, 27, 1787. (55) Hobza, P.; Sˇponer, J. Chem. ReV. 1999, 99, 3247. (56) Hobza, P.; Sˇponer, J. J. Am. Chem. Soc. 2002, 124, 11802. (57) Sˇponer, J.; Jurecka, P.; Hobza, P. J. Am. Chem. Soc. 2004, 126, 10142. (58) Berman, H. M.; Bhat, T. N.; Bourne, P. E.; Feng, Z.; Gilliland, G.; Weissig, H.; Westbrook, J. Nat. Struct. Biol. 2000, 7, 957. (59) Marcus, Y. Chem. ReV. 1988, 88, 1475. (60) Aakesson, R.; Pettersson, L. G. M.; Sandstroem, M.; Siegbahn, P. E. M.; Wahlgren, U. J. Phys. Chem. 1992, 96, 10773. (61) Rode, B. M.; Schwenk, C. F.; Hofer, T. S.; Randolf, B. R. Coord. Chem. ReV. 2005, 249, 2993. (62) Asthagiri, D.; Pratt, L. R.; Paulaitis, M. E.; Rempe, S. B. J. Am. Chem. Soc. 2004, 126, 1285. (63) Figgis, B. N.; Hitchman, M. A. Ligand Field Theory and Its Applications; Wiley-VCH: New York, 2000. (64) Orgel, L. E. An Introduction to Transition-Metal Chemistry: LigandField Theory; Methuen & Co.: London, 1960. (65) Shannon, D. R.; Prewitt, T. C. Acta Crystallogr., Sect. B 1969, 26, 1046. (66) Shannon, D. R.; Prewitt, T. C. Acta Crystallogr., Sect. B 1969, 25, 925. (67) Cordero, B.; Go´mez, V.; Platero-Prats, A. E.; Reve´s, M.; Echeverrı´a, J.; Cremades, E.; Barraga´n, F.; Alvarez, S. Dalton Trans. 2008, 2832. (68) Lescoute, A.; Leontis, N. B.; Massire, C.; Westhof, E. Nucleic Acids Res. 2005, 33, 2395. (69) Basch, H.; Krauss, M.; Stevens, W. J. J. Am. Chem. Soc. 1985, 107, 7267. (70) Zhanpeisov, N. U.; Sˇponer, J.; Leszczynski, J. J. Phys. Chem. A 1998, 102, 10374. (71) Hurd, R. E.; Azhderian, E.; Reid, B. R. Biochemistry 1979, 18, 4012. (72) Jack, A.; Ladner, J. E.; Rhodes, D.; Brown, R. S.; Klug, A. J. Mol. Biol. 1977, 111, 315. (73) Teeter, M. M.; Quigley, G. J.; Rich, A. AdV. Inorg. Biochem. 1981, 3, 233.

JP906847P