Computational Analysis of Molecular Recognition in DNA Base-Sequence and Groove by Methidium Chloride Using Molecular Mechanics Calculation

Takeo KONAKAHARA, Harunobu KOMATSU, Norio SAKAI and Barry GOLD


1 Introduction

It is well known that condensed-heterocyclic compounds, such as ethidium bromide and proflavine, intercalate between DNA base pairs [1, 2]. It has been reported that intercalation of ethidium into a 5'-pyrimidine-purine sequence is favored over a 5'-purine-pyrimidine sequence. Moreover, intercalation at C-G rich sequences is favored over A-T [3 - 6]. Crystallographic X-ray studies using a dinucleotide DNA substrate [7, 8] concluded that ethidium intercalated between base pairs from a minor groove. In other words, ethidium recognizes CG base pair and intercalates into DNA particularly from the minor groove. Ethidium derivatives also stabilize G-quadruplexes [9]. The equilibrium binding properties of ethidium and methidium have been used to more efficiently deliver alkylating [10, 11] and DNA cleaving agents to DNA [12, 13], and many anticancer drugs that intercalate to DNA have been reported [14 - 16]. Recently, Morii et al developed a new method to prepare self-assembled DNA-films [17] and applied it to the preparation of dye-DNA hybrid films [18]. In addition, we have reported that this method is highly effective to determine the binding modes between drug molecules and DNA, using polarized ultraviolet-visible spectroscopy and/or polarized fluorescence spectroscopy of DNA films, which are chain-oriented in a magnetic field [19].
In this paper, we try to estimate a structure and a binding mode of a drug-DNA complex by a computational modeling. There are several papers that describe computational modeling of DNA-intercalator complexes. In most cases modeling is done in dimers [20, 21], although a 6mer has also been used [21 - 23]. However, the driving force for intercalation and the interactions between the intercalator and each base pair has not been extensively studied. Importantly, the positioning of ethidium in the base pair stack is still debated [24]. This paper, which models both the sequence and groove specificity of methidium in a B-DNA 12mer, addresses some of these issues.

2 Calculation

The computational modelings in this study were performed with CACheTM software (CAChe Scientific, Inc., USA). A sequence of DNA used in our modeling is 5'd-CATCCCGGGATG-3', B-form double helical oligonucleotide, that was used in our studies on methylation of DNA by N-methyl-N-nitrosourea linked to a methidium chloride analogue [11]. The initial structure of the 12mer oligonucleotide was constructed by combining appropriate fragments of DNA extracted from X-ray data libraries in a Protein Data Bank. Hydrogen atoms and double bonds were appended to them, and one of the oxygen atoms in each phosphate residue was changed to a monovalent anion, then a sodium cation was arranged to each phosphate as a counter ion.
This structure was stabilized to converge the steric energy to 0.01 kcal.mol-1 by molecular mechanics (MM2) method. Then a more stable structure was searched by molecular dynamics (MD) method. MM2 and MD calculation settings are shown in the following; MM calculation: force field parameters, MM2 parameters augmented by CACheTM; optimization method, conjugate gradient; convention condition, 0.0001 kcal.mol-1; calculation option, united atom model; MM calculation: force field parameters, MM2 parameters augmented by CACheTM; micro canonical ensemble, constant N, V, E; sampling period, 3 ps; sampling-time step, 1 fs; saving trajectory file, every 20 time step; calculation option, united atom model. In MD calculation, the structure was equilibrated at 300 K, and the potential energy was sampled for 2.5 ps. The snapshot of a fragment that has minimum potential energy was picked out, and the structure was finally re-optimized with convergence of 0.0001 kcal.mol-1 by the molecular mechanics method. Augmented MM2 parameters belonging to the CACheTM system were used as force filed parameters in both MM2 and MD calculation. The structure of methidium cation was constructed graphically on CACheTM system and optimized by MOPAC (ver.5, PM3).
In the modeling of the DNA-methidium complex, a methidium was intercalated between each base pair on DNA oligomer (Figure 1) from minor or major groove side referring to Tsai's data of X-ray crystallography [7]. A methidium cation was placed at the center between two base pairs so that the phenanthridine ring would be parallel with base pairs and the degree of overlap between the phenanthridine ring and base pairs would be maximum. The structures roughly optimized by MM2 calculation were employed as initial structures of DNA-methidium complexes and they were precisely optimized in a similar manner as the 12mer oligomer.
The binding constants Kass for the association of ethidium bromide with 7 dimer-duplex DNAs were calculated by Scatchard plot analysis [25] using Krugh's data [3].

3 Results & Discussion

3. 1 Model of the DNA-Methidium Complex Intercalated from the Minor Groove

It is difficult to decide exactly the configuration of DNA, methidium and the complex by the molecular mechanics; therefore the molecular dynamics method was employed to find the configuration having global minimum energy, in this study. Calculation was done without any restricted conditions, and a united atom model was used in order to save in time.
As a result, a methidium, that intercalated from the minor groove in every site (Figure 1), located between two base pairs, and intercalated position of methidium was nearly at the center of these hydrophobic spaces. The phenanthridine ring of methidium located in parallel to the plane of the base pairs. The phenyl group at the 9-position of methidium was perpendicular to the phenanthridine ring, and it prevents methidium from turning round or going through to a major groove side.

Figure 1. The sites intercalated by methidium

The interval of intercalated two base pairs was expanded to twice, namely from 3.4 to 6.4A, as shown in Figure 2. This agrees with experimental results, i.e. excluded site size is twice in the width of base pairs. In other words, double helical DNA was wound off as a result of intercalation by a methidium nucleus.

3. 2 Intercalation from the Major Groove

An initial structure of the complex, in which methidium was intercalating to each site from the major groove, was applied to the MD calculation. The methidium protruded from the space between the stacking base pairs for seven among eleven intercalation sites.
These seven cases were classified under the following three patterns as shown in Figure 3; (a) the methidium is widely out of place from DNA helix (site 7), (b) the methidium is parallel to base pairs (sites 5 and 11), and (c) the methidium is held between the major groove in parallel with the DNA helix axis in the form of having interaction between O-atom in phosphate and amino group at the 3- and 7-positions of methidium as reported by Monaco et. al [23] (sites 2, 3, 9, and 10). It is considered that the stabilization energy of DNA-methidium complex was not large when methidium intercalated from the major groove as described later. Therefore, methidium is easily released from the base pair space by adding kinetic energy in MD calculation, and the DNA-methidium complex has various configurations.

3. 3 Model of the DNA-Methidium

Steric energy Es of a molecule is given by the following equation in molecular mechanics in this work;

where EStr, Eq, Et, Eit, Evdw, Ee and Eh are bond stretching energy, bond angle bending energy, torsion energy, improper torsion energy, van der Waals energy, electrostatic energy, and hydrogen bonding energy, respectively. Each of them was defined originally by Allinger [26, 27].
Therefore, the difference (DE) between steric energy (ES) of the DNA-methidium complex and total steric energy (ES0) of DNA and methidium without interacting each other shows stabilization energy by intercalation (Table 1).

The absolute values of DE are not discussed in this paper because of their accuracy. The relative values, however, may be worthy of note. They varied with both the site and the groove in which intercalation occurs, and it showed a maximum (-28.4 kcal.mol-1) when methidium intercalated at site 6 (5'-dCG) from the minor groove as shown in Table 1. On the other hand, DE was only -2.8 kcal.mol-1 when methidium intercalated at site 6 from the major groove (Figure 3c). This value was much smaller than that from the minor groove at the same site. This result indicates that the DNA-methidium complex intercalated from the minor groove is more stable than that from the major groove, and is consistent with well-known experimental results [7, 8].

Figure 2. Exclusion of base pairs by intercalated methidium nucleus: (a), before intercalation; (b), after intercalation

Figure 3. MD structures of 12mer DNA-methidium complex intercalated from major groove: (a) the methidium nucleus is displaced from the DNA helix (site 7); (b) the methidium nucleus is parallel to base pairs (sites 5 and 11); and (c) the methidium nucleus is held between the major groove in parallel to the DNA helix axis with interaction between a non-bonding phosphate oxygen atom and the amino groups at 3- and 7-positions of methidium nucleus as reported previously [13] (sites 2, 3, 9, and 10).

Table 1. Stabilization energy (DE) of the 12mer DNA-methidium complex calculated by molecular mechanics
Intercalation sitea)ESb) / kcal.mol-1DE / kcal.mol-1
1, minor-374.09-11.06
2, minor-378.41-15.37
3, minor-370.59-7.56
4, minor-370.94-7.91
5, minor-374.82-11.79
6, minor-391.42-28.38
7, minor-378.44-15.41
8, minor-378.91-15.88
9, minor-377.67-14.64
10, minor-367.81-4.77
11, minor-380.35-17.32
6, major-365.79-2.76
1', minor-388.00-24.97
DNA + methidiumc)-363.03
a) Intercalation sites 1-11 are indicated in Figure 1 and site 1' in Figure 5.
b) ES: steric energy calculated by augmented MM2.
c) Total steric energy of DNA and methidium without mutual interaction (ES0).

Figure 4. Stabilization energy (DE) vs. log KassBinding constants Kass were calculated from data reported in Reference [3]. The numbers 1-11 and 1' mean the intercalation sites shown in Figures 1, 5.

Figure 5. The sites 1' and 2' intercalated by a methidium cation

Krugh and his coworkers reported that ethidium bromide binds most strongly with 5'-dCG in a study using photometric titration of various dinucleotides [3]. It was considered that differences in the DNA binding affinity between ethidium and methidium are small because the two molecules have virtually the same structure. Indeed, the logarithm of binding constants Kass, determined by Scatchard analysis using Krugh's data [3], gave an excellent linear relationship when plotted against the stabilization energy DE as shown in Figure 4 and Table 2. The only exception was observed for sites 1 and 2. The scattering of data for the sites 1 and 2 may result from conformational disorder of the minor groove of the 5'-terminal nucleotides due to so-called "end effects" (see Figure 2). In order to model the intercalation complex at site 1, two base pairs composing site 11 were deleted and then added to the other side of the complex to obtain a new complex at site 1' as shown in Figure 5. The DE value of this new complex at site 1', which was larger than that of site 1, is in good agreement with the regression analysis shown in Figure 4. Although the -DE value of a complex at site 2' decreased slightly, it was still scattered from the regression line (not cited in Table 1 and Figure 4). This may result from using a different sequence, 5'-d(CATC)-3' and 5'-d(GATG)-3', and the same binding constant of ethidium with 5'-d(AT)-3'.

Table 2. Binding constant Kass and DG of complexes between ethidium bromide and duplex-deoxydinucleotides at 25°Ca)
Sequence and binding siteb)Binding constant 10-2Kass/(L.mol-1)DG/(kcal.mol-1) at 25°C
5'-d(CA)-3'; 19.33-4.04
5'-d(AT)-3'; 22.99-3.37
5'-d(TC)-3'; 33.30-3.43
5'-d(CG)-3'; 612.6 -4.22
5'-d(GA)-3'; 95.46-3.73
5'-d(AT)-3'; 102.99-3.37
5'-d(TG)-3'; 116.09-3.79
a) Calculated from data reported in Reference [3].
b) Sequence of one of the deoxydinucleotides in the duplex and binding sites in Figure 1.

In conclusion, it is able to estimate by simple computational molecular modelings using MM and MD calculation that methidium chloride binds to DNA by the mode of intercalation from a minor groove to yield more stable complexes than those from a major groove, although the calculation is done for an ideal bimolecular reaction. Methidium chloride has a tendency to intercalate into site 6, composed by the hydrophobic space between 5'-d(CG)-3' and 3'-(CG)d-5' partial structures of the complementary double helical DNA 12mer, rather than to intercalate into site 3 [5'-d(TC)-3'] or site 10 [5'-d(AT)-3']. Computational molecular modelings of DNA-intercalator complexes in water are in progress.

The present work was partially supported by a Grant-in-Aid for Scientific Research from MEXT (16550148), 2004-2005, a grant from the Japan Private School Promotion Foundation, a fund for "High-Tech Research Center" Project for Private Universities (Frontier Research Center for Computational Science); a competitive fund subsidy from MEXT, 1996-2000 & 2001-2003, and a fund for "High-Tech Research Center" Project for Private Universities (Research Center for Advanced Materials); a competitive fund subsidy from MEXT, 2000-2004 & 2005-2007.


[ 1] Feigon, J., Denny, W. A., Leupin, W., Kearns, D. R., J. Med. Chem., 27, 450-465 (1984).
[ 2] Dervan, P. B., Becker, M. M., J. Am. Chem. Soc., 100, 1968-1970 (1978).
[ 3] Krugh, T. R., Wittlin, F. N., Cramer, S. P., Biopolymers, 14, 197-210 (1975).
[ 4] Kastup, R. V., Young, M. A., Krugh, T. R., Biochemistry, 17, 4855-4865 (1978).
[ 5] Reinhardt, C. G., Krugh, T. R., Biochemistry, 17, 4845-4854 (1978).
[ 6] Bresloff, J. L. Crothers, D. M., Biochemistry, 20, 3547-3553 (1981).
[ 7] Tsai, C.-C., Jain, S. C., Sobell, H. M., J. Mol. Biol., 114, 301-315 (1977).
[ 8] Jain, S. C., Tsai, C.-C., Sobell, H. M., J. Mol. Biol., 114, 317-331 (1977).
[ 9] Koeppel, F., Riou, J.-F., Laoui, A., Mailliet, P., Arimondo, P. B., Labit, D., Petitgenet, O., Helene, C., Mergney, C., J.-L., Nucleic Acid Research, 29, 1087-1096 (2001).
[10] Konakahara, T., Wurdeman, R. L., Gold, B., Biochemistry, 27, 8606-8613 (1988).
[11] Konakahara, T., Tanaka, Y., Gold, B., Nucleic Acids Symposium Series , 31, 113-114 (1994).
[12] Hertzberg, R. P., Dervan, P. B., Biochemistry, 23, 3934-3945 (1984).
[13] Hertzberg, R. P., Dervan, P. B., J. Am. Chem. Soc., 104, 313-315 (1982).
[14] Brana, M. F., Cacho, M., Gradillas, A., Pascual-Teresa, B. de, Ramos, A., Current Pharmaceutical Design, 7, 1745-1780 (2001).
[15] Tamura, S., Konakahara, T., Komatsu, T., Ozaki, H., Ohta, Y., Takeuchi, H., Heterocycles, 48, 2477-2480 (1998).
[16] Abe, N., Nabeshima, M., Fujii, H., Kakehi, A., Kageura, Y., Konakahara, T., Heterocycles, 54, 329-339 (2001).
[17] Mori, N., Kido, G., Suzuki, H., Morii, H., Biopolymers, 77, 163-172 (2005).
[18] Mori, N., Kido, G., Suzuki, H., Nimori, S., Morii, H., Biomacromolecules, 5, 2297-2307 (2004).
[19] Morii, N., Kido, G., Konakahara, T., Morii, H., J. Phys. Chem. B, 109, 15636-15644 (2005).
[20] Miller, K. J.,Brodzinsky, R., Hall, S., Biopolymers, 19, 2091-2122 (1980).
[21] Lybrand, T., Kollman, P., Biopolymers, 24, 1863-1879 (1985).
[22] Rudolph, B. R., Case, D. A., Biopolymers, 28, 851-871 (1989).
[23] Monaco, R. R., Hausheer, F. H., J. Biomol. Structure & Dynamics, 10, 675-680 (1993).
[24] Slobodyansky, E., Stellwagen, J., Steelwagen, N. C., Biopolymers , 27, 1107-1126 (1988).
[25] McGhee, J. D., von Hippel, P. H., J. Mol. Biol., 86, 469-489 (1974).
[26] Allinger, N. L., J. Am. Chem. Soc., 99, 3279 (1977).
[27] Burkert, U., Allinger, N. L., "Molecular Mechanics", American Chemical Society, Washington, D. C. (1982), ; translated by Osawa, E. and Takeuchi, Y. in Japanese, Keigaku Shuppan, Tokyo, (1986).