Ab Initio Fragment Molecular Orbital Study of Molecular Interactions

Graduate School of Science and Technology, Kobe UniVersity, 1-1, ... Rikkyo UniVersity, 3-34-1 Nishi-Ikebukuro, Toshima-ku, Tokyo 171-8501, Japan, and...
0 downloads 0 Views 275KB Size
Download PDF
J. Phys. Chem. B 2007, 111, 3525-3533

3525

Ab Initio Fragment Molecular Orbital Study of Molecular Interactions between Liganded Retinoid X Receptor and Its Coactivator: Roles of Helix 12 in the Coactivator Binding Mechanism Mika Ito,*,†,‡ Kaori Fukuzawa,§ Yuji Mochizuki,‡,| Tatsuya Nakano,‡,⊥ and Shigenori Tanaka†,‡ Graduate School of Science and Technology, Kobe UniVersity, 1-1, Rokkodai, Nada, Kobe 657-8501, Japan, CREST, Japan Science and Technology Agency, Mizuho Information and Research Institute, Inc., 2-3 Kanda Nishiki-cho, Chiyoda-ku, Tokyo 101-8443, Japan, Department of Chemistry, Faculty of Science, Rikkyo UniVersity, 3-34-1 Nishi-Ikebukuro, Toshima-ku, Tokyo 171-8501, Japan, and DiVision of Safety Information on Drugs, Food, and Chemicals, National Institute of Health Sciences, 1-18-1 Kamiyoga, Setagaya-ku, Tokyo 158-8501, Japan ReceiVed: January 3, 2007; In Final Form: January 26, 2007

On the basis of the fragment molecular orbital method we addressed molecular interactions of liganded retinoid X receptor (RXR) with steroid receptor co-activating factor-1 (SRC1) coactivator to examine the contribution of helix 12 (H12), which contains the core of the transcriptional activation function 2 activating domain, to the coactivator binding of RXR. The interaction between H12 and SRC1 was proved to be the main cause for the stabilization of the coactivator binding. In particular, highly conserved charged (Glu453) and hydrophobic (Phe450) residues in H12 were found to have stronger electrostatic and dispersion interactions with SRC1 than the other charged and hydrophobic residues in H12, respectively. In addition, the charge transfer (CT) from RXR to SRC1 was found to occur mainly by the changes in charges of H12 residues. Large positive and negative charge changes were observed especially for Glu453 and for Lys631 and Ile632 in SRC1, respectively, indicating that Glu453 is an electron donor for Lys631 and Ile632 in this CT. Taken together, our findings quantitatively demonstrated that H12 and its highly conserved residues significantly contribute to the coactivator binding not only by the Coulomb and dispersion interactions but also by the CT described with the quantum-mechanical framework.

1. Introduction The retinoid X receptor (RXR) belongs to the nuclear receptor (NR) superfamily and acts as a ligand-inducible transcriptional factor that regulates expression of many genes involved in various activities of its ligands at the transcriptional level. Its natural ligand 9-cis-retinoic acid (9cRA) is the metabolite of vitamin A that controls morphogenesis, differentiation, and homeostasis. It is noteworthy that 9cRA is an effective inhibitor of tumor cell growth, and thus its antitumor activity is useful in therapy and prevention of cancers such as HIV-associated Kaposi’s sarcoma.1,2 Because RXR has important biological roles associated with human life and diseases, it has been one of the primary targets of drug discovery. A major goal of the RXR study has been to elucidate the transcriptional activation mechanism of RXR so as to efficiently exploit the functions of RXR. To date, many experimental studies have been devoted to achieve this goal. It is now widely accepted that the transcriptional activity of RXR, as well as many NRs, is induced by the binding of ligands to the ligandbinding domain (LBD), which contains transcriptional activation function 2 (AF-2), and controlled by the exchange of the * Address correspondence to this author at Kobe University. Phone: +8178-803-7991. Fax: +81-78-803-7761. E-mail: [email protected]. † Kobe University. ‡ CREST. § Mizuho Information and Research Institute, Inc. | Rikkyo University. ⊥ Division of Safety Information on Drugs, Food, and Chemicals, National Institute of Health Sciences.

bindings of the co-regulator proteins including coactivator and corepressor.3 In an early experimental study,4 the structure of the LBD of the unliganded RXR was determined by X-ray crystal structure analysis. Through comparison of this structure and the liganded retinoic acid receptor (RAR), a “mouse trap” mechanism that involves ligand-induced remarkable conformational changes especially in the orientation of helix 12 (H12) at the carboxyl terminus of the receptor LBD was proposed for the transcriptional activation mechanism, and H12 was identified that constitutes the core of the AF-2 activating domain (AF-2 AD), which plays a central role in the mechanism.5 This model mechanism was subsequently validated by comparison of the unliganded and liganded states of the same RXR LBD.6 On the basis of the model mechanism, H12 was proposed as a liganddependent “switch” that generates, in its ligand-induced configuration, a surface for binding of transcriptional coactivators.7 After the “mouse trap” mechanism was proposed, structural studies of NR LBDs complexes with its ligands and coactivators such as steroid receptor co-activating factor-1 (SRC1) revealed that a “charge clamp” formed by lysine and glutamic acid residues in H3 and H12, respectively, contacts with the helical LXXLL motif of a coactivator. These structural data suggested that a particular position of H12 is essential for coactivator binding.8,9 Afterward, the same structural character was also identified in the X-ray crystal structure of human RXRR (hRXRR) LBD complex with 9cRA and SRC1 (Figure 1).10 In a recent structural and fluorescence anisotropy study,11 com-

10.1021/jp070054w CCC: $37.00 © 2007 American Chemical Society Published on Web 03/09/2007

3526 J. Phys. Chem. B, Vol. 111, No. 13, 2007

Figure 1. Ribbon display of the hRXRR LBD (green) complexed with 9cRA (purple) and SRC1 peptide (blue). The position of H12 (red) is also displayed.

Figure 2. Sequence alignment of H12 in hRXRR with the corresponding region in NRs. The names of NRs, residue numbers of the first amino acid residue in H12, and amino acid sequences of H12 are displayed from left to right. Highly conserved hydrophobic and charged residues are in green and red boxes, respectively.

parison of structures of a 9cRA-bound RXR LBD complex with and without a coactivator peptide indicated that the particular position of H12 is supported by an “aromatic clamp” formed by three phenylalanine residues in H3, H11, and H12, and mutation of the phenylalanine residues in H3 and H11 revealed that the transcriptional activity of RXR is reduced by the destabilization of H12. In addition, it is known that H12 has highly conserved charged and hydrophobic residues among many NRs (Figure 2).12-15 Figure 2 shows the sequence alignment of H12 in hRXRR with the corresponding region in human (h) or mouse (m) NRs such as RAR, thyroid hormone receptor (TR), estrogen receptor (ER), peroxisome proliferator-activated receptor (PPAR), and constitutive androstane receptor (CAR). The highly conserved charged and hydrophobic residues of hRXRR are Glu453 and Phe450 that have been proposed to form the “charge clamp” and “aromatic clamp”, respectively.10,11 Biological studies12-15 have shown that the mutations of these conserved residues in H12 markedly reduced transcriptional activity of NRs due to impaired coactivator recruitment. On the basis of these structural and biological data, it is evident that H12 and its highly conserved residues are essential for coactivator recruitment associated with the transcriptional outcome. To better understand the transcriptional activation mechanism of RXR, it would be necessary to clarify the precise nature of molecular interactions of H12 and its highly conserved residues with a coactivator, though the details of these interactions have not been elucidated until now. Molecular interactions including charge transfers in the formation of complexes are

Ito et al. difficult to evaluate by experimental methods and classical mechanical methods based on empirical force fields. On the other hand, ab initio quantum mechanical calculations have succeeded in the analysis of such molecular interactions as well as of molecular structures and properties. Moreover, by taking advantage of the ab initio quantum mechanical calculations using the fragment molecular orbital (FMO) method,16-21 we can perform detailed and accurate analysis of biomacromolecules. In particular, using the interfragment interaction energies (IFIEs)22-25 evaluated by the FMO calculations, we can quantitatively estimate interactions at a residue level when the fragmentations are performed according to the amino acid unit for the protein. In addition, by means of electron-correlation methods beyond the Hartree-Fock (HF) method such as the second-order Møller-Plesset perturbation (MP2) method26,27 with the FMO procedure, we can also describe the dispersion energies appropriately. In previous studies with the MP2 method,24,25 it has been demonstrated that the dispersion energies play an important role in interactions of biomacromolecules. To date, the FMO calculations for biomacromolecules have been performed mainly under gas-phase conditions, because those calculations including numerous water molecules in solution are difficult to perform at present. However, previous FMO calculations24,25 under gas-phase conditions with X-ray crystal structures have provided much valuable information on molecular interactions and charge transfers (CTs) in biomacromolecules. In an effort to provide insights into the detailed molecular mechanism of the transcriptional activation of RXR, we have investigated the molecular interactions and CTs between 9cRA liganded hRXRR LBD and SRC1 coactivator at a residue level by the ab initio FMO calculations using the HF and MP2 methods, and examined how H12 and its highly conserved residues may contribute to the coactivator binding of RXR. It is now well-known in NRs that ligand binding globally stabilizes the receptor LBD, resulting in a more compact and rigid structure, and also stabilizes H12 in its active conformation, which in turn promotes coactivator binding.3,5-7 In other words, structural changes are small after the ligand binding and the subsequent conformational change of H12. Considering this fact, we think that FMO calculations for the fixed structure of the hRXRR LBD complex after the ligand and coactivator bindings could provide reasonable knowledge of the molecular interactions and CTs between liganded hRXRR LBD and its coactivator. For eventual understanding of the transcriptional activation mechanism, examination of roles of H12 through the ligandand coactivator-induced conformational changes in solution should be performed. However, in this work, we focus on the roles of H12 in the hRXRR LBD complex after these conformational changes, and performed FMO calculations for the fixed structure of the hRXRR LBD complex under gas-phase condition as a first step of the investigation to elucidate the electronic contributions. 2. Theoretical Calculations The initial atomic coordinates of the RXR-9cRA-SRC1 complex were obtained from the Research Collaboratory for Structural Bioinformatics (RCSB) Protein Data Bank (PDB),28 PDB code 1FM9 (Figure 1).10 The entire hRXRR LBD consisting of 232 amino acid residues (residues 227-458), 9cRA, and SRC1 peptide consisting of 10 amino acid residues (residues 630-639) were employed for simulations. Missing hydrogen atoms and side chains in the PDB file were complemented manually by using the molecular graphic software

Ab Initio Fragment MO Study of Molecular Interactions

J. Phys. Chem. B, Vol. 111, No. 13, 2007 3527

Molecular Operating Environment (MOE) Version 2005.06.29 Hydrogen atoms were added to both the N- and C-terminal residues of the peptide chains and all the dissociative and associative residues in their charged states. The total number of atoms in the complex is 3926, including hydrogen atoms. The charges of RXR, 9cRA, and SRC1 are 0e, -1e, and +3e, respectively, and the total charge of the complex is +2e. All the positions of hydrogen atoms and side chains added in the previous procedure were geometrically optimized by using AMBER99 force field30 with the other heavy atoms fixed at the positions given in the PDB data. The geometry of the RXR-9cRA-SRC1 complex used for the ab initio FMO16-21 calculations was prepared by the procedure mentioned above, and the geometries of the RXR9cRA complex and free RXR, 9cRA, and SRC1 were fixed at the geometry of the RXR-9cRA-SRC1 complex. The ab initio FMO calculations were carried out under gas-phase conditions at the HF level with the 6-31G basis set. The energies and electron densities were further improved at the MP2 level26,27 with the 6-31G basis set. To save computational time without losing significant accuracy, the approximations of electrostatic potentials considered as the Mulliken orbital charge (esp-aoc) and the fractional point charge (esp-ptc) were applied to fragments whose separations of the closest contact atoms were more then 0.0 and 2.0 in units of van der Waals radii (vdW), respectively.19 The Coulomb interaction approximation (dimeres) was also applied to fragments whose separation was more than 2.0 in vdW units. The fragmentation was performed according to the amino acid unit for the protein, and each amino acid residue of RXR and SRC1 and the 9cRA molecule were treated as a single fragment. The number of fragments in the RXR-9cRA-SRC1 complex is 243. The IFIE20-25 in FMO calculations is defined as follows,

∆EIJ ) (E′IJ - E′I - E′J) + Tr(∆PIJVIJ)

(1)

where ∆PIJ is a difference density matrix, VIJ is an environmental electrostatic potential for fragment dimer IJ from other fragments, and E′I and E′IJ are energies of fragment monomer I and dimer IJ without environmental electrostatic potential. The many-body effects are considered through the environmental electrostatic potentials. From ∆EIJ, the total energy E is calculated by

E)

∆EIJ + ∑E′I ∑ I>J I

(2)

The IFIEs in the RXR-9cRA-SRC1 complex were analyzed primarily for the interactions between the residues of H12 and SRC1. The IFIEs of the N- and C-terminal residues of RXR and SRC1 with the ligand or other residues are not listed in the tables in the following sections, because these residues are not ends of the peptide chains in the actual system. All the FMO calculations were performed with the ABINITMP program31 on 32 Dual AMD Opteron 2.0 GHz clusters (64 CPUs), and the visualization was carried out with the BioStation Viewer.32 In the following sections, we discuss the results calculated at the MP2/6-31G level, unless otherwise noted. 3. Results and Discussion 3.1. Ligand and Coactivator Binding Energies. The RXR9cRA-SRC1 complex is formed via two processes, that is, 9cRA ligand and SRC1 coactivator binding processes. The coactivator binding occurs after the ligand binding. To examine the effects of each process on the formation of the complex,

TABLE 1: Energy Differencesa between the Receptor-Ligand-Coactivator Complex (RLC), the Receptor-Ligand Complex and Coactivator (RL + C), and Individual Molecules (R + L + C) method

∆E1

∆E2

∆E3

HF/6-31G MP2/6-31G

-127.62 -181.27

-538.10 -586.95

-665.71 -768.22

a ∆E1 ) [E(RL) + E(C)] - [E(R) + E(L) + E(C)] ) E(RL) [E(R) + E(L)], ∆E2 ) E(RLC) - [E(RL) + E(C)], and ∆E3 ) E(RLC) - [E(R) + E(L) + E(C)]. Energies are in kcal/mol.

the ligand and coactivator binding energies were calculated by the HF and MP2 methods with the 6-31G basis set as shown in Table 1. The binding energies of ligand (∆E1) and coactivator (∆E2) and the total energy of the ligand and coactivator bindings (∆E3) were given as follows,

∆E1 ) E(RL) - [E(R) + E(L)]

(3)

∆E2 ) E(RLC) - [E(RL) + E(C)]

(4)

∆E3 ) E(RLC) - [E(R) + E(L) + E(C)] ) ∆E1 + ∆E2 (5) where E(RLC), E(RL), E(R), E(L), and E(C) are the energies of the receptor-ligand-coactivator complex (RLC), the receptor-ligand complex (RL), the receptor (R), the ligand (L), and the coactivator (C), respectively. A similar tendency was observed in the HF and MP2 calculations for these binding energies. The binding energy of the coactivator is larger than that of the ligand by both the HF and MP2 methods. The coactivator binding energy is larger than the ligand binding energy by 405.68 kcal/mol at the MP2 level. Through the ligand and coctivator binding processes, the RXR9cRA-SRC1 complex is greatly stabilized by -768.22 kcal/ mol at the MP2 level. These values show that the RXR-9cRASRC1 complex is primarily stabilized by the coactivator binding. However, the calculated ligand and coactivator binding energies above are perhaps too large to be considered literally. In a previous theoretical study33 on ligand binding affinity for ER by molecular dynamics simulations, it has been shown that, though each binding energy of various ligands calculated without the effects of solvent and entropy is larger than that calculated with these effects by about 50 kcal/mol, the tendency of the former ligand binding affinities is consistent with that of the latter and experiments. On the basis of this previous theoretical estimation, it is suggested that, if the effects of solvent and entropy are included in the calculations, the binding energies for RXR would be reduced significantly, but the qualitative tendency of the binding affinities for RXR would not be changed. In addition, the basis set superposition error (BSSE)34 could be one of the reasons for the overestimation of these binding energies, but the tendency of the binding affinities for RXR would not be changed by the BSSE as well. To evaluate how H12 of RXR contributes to the two processes, the IFIE analysis was performed. In this analysis, the total IFIEs of 9cRA or SRC1 with the whole H12 were calculated and compared to those of 9cRA or SRC1 with the whole RXR as shown in Table 2. A similar tendency was also observed in the HF and MP2 calculations for these IFIEs, as well as the binding energies (∆E). The interaction of SRC1 with RXR is larger than that of 9cRA with RXR by both the HF and MP2 methods. The SRC1-RXR interaction is larger than the 9cRA-RXR interaction by 126.86 kcal/mol at the MP2 level. Although the 9cRA-H12 interaction is calculated as small

3528 J. Phys. Chem. B, Vol. 111, No. 13, 2007

Ito et al.

Figure 3. Positions of the residues and the atoms in H12 and SRC1. The RXR, H12, and SRC1 are displayed by green, red, and blue ribbons, respectively. (a) Positions of Phe450 and Glu453 in H12. (b) Positions of CR atoms (gray balls) of the main chains in H12 and SRC1 residues. All of the residues except for the N- and C-terminal residues of SRC1 are displayed. (c) Positions of O (red balls) or N atoms (blue balls) of the side chains in H12 and SRC1 residues. Only charged or polarized residues are displayed. (d) Positions of C atoms (gray balls) at the tips of the side chains in H12 and SRC1 residues. Only hydrophobic residues are displayed.

TABLE 2: IFIEs of 9cRA and SRC1 with the Whole RXR and H12 in the Receptor-Ligand Complex (RL) and the Receptor-Ligand-Coactivator Complex (RLC)a RL method HF/6-31G MP2/6-31G

RLC

9cRA-RXR 9cRA-H12 SRC1-RXR SRC1-H12 -170.12 -238.54

26.62 26.41

-309.90 -365.40

-277.23 -300.03

a Interactions with the N- and C-terminal residues are not included. Energies are in kcal/mol.

positive values, the SRC1-H12 interaction is calculated as large negative values by both the HF and MP2 methods. These differences would be due to the charges of each component, where the charges of 9cRA, SRC1, and H12 are -1e, +3e, and -2e, respectively. Also, 82.1% of the SRC1-RXR interaction is occupied by the SRC1-H12 interaction at the MP2 level. These calculations indicate that the coactivator binding is mainly stabilized by the SRC1-H12 interaction. 3.2. Coactivator Interactions with H12 Residues. As shown in Figure 3a, the highly conserved residues (Glu453 and Phe450) in H12 are directed to SRC1 in contrast to the other residues in H12. To evaluate how each residue in H12 contributes to the coactivator binding, the IFIEs between H12 and SRC1 residues were calculated by the MP2 method as shown in Table 3. H12 of RXR has one noncharged polarized residue (Thr449), five hydrophobic residues (Phe450, Leu451, Met452, Met454, and Leu455), and two negative charged residues (Glu453 and Glu456). SRC1 has one noncharged polarized residue (Gln638), four hydrophobic residues (Ile632, Leu633, Leu636, and Leu637), and three positive charged residues (Lys631, His634, and Arg635) except for the terminal residues. The five amino

acids, Leu633, His634, Arg635, Leu636, and Leu637, correspond to the LXXLL motif of SRC1. Total1 and Total2 in Table 3 are sums of IFIEs of each H12 residue with the LXXLL motif and all the listed SRC1 residues, respectively. These total IFIEs have a similar tendency. Among the eight amino acid residues of H12, only four amino acid residues (Phe450, Glu453, Met454, and Glu456) have attractive interactions with SRC1, and SRC1 is stabilized mainly by the interactions with Glu453 and Glu456. It is noteworthy that the highly conserved charged (Glu453) and hydrophobic (Phe450) residues in H12 interact with SRC1 more strongly than the other charged and hydrophobic residues in H12, respectively. The electrostatic interactions of Glu453 with the charged residues in SRC are stronger than those of Glu456, and consequently, Glu453 has stronger interactions with SRC1 than Glu456 by 71.66 and 112.41 kcal/ mol of Total1 and Total2, respectively. Likewise, the dispersion interactions of Phe450 with the hydrophobic residues of SRC1 are stronger than those of Met454, and consequently, Phe450 has stronger interactions with SRC1 than Met454 by 3.73 and 9.94 kcal/mol of Total1 and Total2 values, respectively. To examine the relationship of IFIEs to distances between residues in H12 and SRC1, some distances were measured as shown in Table 3. Distance1 in Table 3 shows distances between CR atoms of the main chains of the residues in H12 and SRC1 (Figure 3b). Distance2 in Table 3 shows either of two types of average distances: average distances between O atoms of the side chains of the negatively charged or polarized residues in H12 and N atoms of the side chains of the positively charged or polarized residues in SRC1 (Figure 3c) or average distances between C atoms at the tips of the side chains of the hydrophobic residues in H12 and SRC1 (Figure 3d). Average1 and Average2 in Table 3 are averages of Distance1 and Distance2, respectively. First, the relationship of total IFIEs and average distances of positively charged residues Glu453 and Glu456 was analyzed. Average1 and Average2 of Glu453 are 3.98 and 4.18 Å shorter than those of Glu456, and these values are correlated with the results of IFIEs that Glu453 has stronger total interactions with SRC1 than Glu456. Second, the relationship of total IFIEs and average distances of hydrophobic residues Phe450 and Met454 was analyzed. Average1 and Average2 of Phe450 are 1.64 and 1.58 Å shorter than those of Met454, and these values are correlated with the results of IFIEs that Phe450 has stronger total interactions with SRC1 than Met454. These results thus indicate that total IFIEs are correlated with the average distances between the residues. On the other hand, some of individual distances between the residues in H12 and SRC1 are not correlated with the results of IFIEs. For one example of the relationship of IFIEs to distances between charged residues, Distance1 and Distance2 between Glu453 and Lys631 are 2.73 and 1.54 Å shorter than those between Glu453 and His634, respectively, but the attractive interaction of Glu453 with Lys631 is smaller by 2.54 kcal/mol than that with His634. The charges without charge redistributions of Glu453, Lys631, and His634 are -1e, +1e, and +1e, respectively, and the calculated fragment charges of these residues are -0.80e, +1.07e, and +0.89e at the MP2/631G level, respectively. In addition, the Glu453-Lys631 distances are shorter than the Glu453-His634 distances. On the basis of the information about these charges and distances, the Glu453-Lys631 interaction energy is anticipated to be larger than the Glu453-His634 interaction energy. However, the calculated interaction energies disagree with this anticipation, indicating that the interaction energies between charged residues are not determined only by the charge and distance of residues.

Ab Initio Fragment MO Study of Molecular Interactions

J. Phys. Chem. B, Vol. 111, No. 13, 2007 3529

TABLE 3: IFIEs and Distances between H12 and SRC1 Residuesa H12 residue IFIE

Distance1e

Distance2g

SRC1 residueb

Thr449

Phe450

Leu451

Met452

Glu453

Met454

Leu455

Glu456

Lys631 Ile632 Leu633 His634 Arg635 Leu636 Leu637 Gln638 Total1c Total2d Lys631 Ile632 Leu633 His634 Arg635 Leu636 Leu637 Gln638 Average1f Lys631 Ile632 Leu633 His634 Arg635 Leu636 Leu637 Gln638 Average2h

0.04 1.91 -0.18 -0.44 0.85 -0.06 -0.01 -0.01 0.17 2.11 10.21 8.46 10.67 13.62 13.48 13.33 16.13 18.48 13.05 9.85 17.33 13.86 18.23 14.82

0.42 -5.74 -1.42 -2.29 -1.73 -1.22 -0.47 -0.25 -7.13 -12.71 8.62 6.35 7.43 10.78 11.30 10.79 13.07 15.82 10.52 4.49 6.53 5.68 9.45 6.54

1.22 1.88 0.01 -0.98 1.01 -0.04 0.05 -0.06 0.05 3.09 12.04 9.94 10.11 13.66 14.68 13.74 15.57 18.72 13.56 11.15 13.61 14.02 16.97 13.94

1.48 0.32 0.11 0.28 1.35 0.04 0.08 0.05 1.86 3.71 11.50 10.52 11.36 14.36 15.50 15.46 17.35 20.09 14.52 13.86 17.13 17.31 21.07 17.34

-49.16 -27.16 -27.58 -51.70 -39.38 -4.70 -4.18 -2.16 -127.55 -206.02 8.14 7.72 8.29 10.87 12.40 12.86 14.41 16.93 11.45 8.12 9.66 10.57 11.46 9.95

0.19 0.52 -0.65 -2.54 0.07 -0.14 -0.13 -0.08 -3.40 -2.77 10.37 9.41 8.38 11.29 13.49 13.24 14.04 17.09 12.16 9.02 5.46 9.35 8.65 8.12

1.83 0.68 0.50 1.22 1.51 0.16 0.17 0.11 3.56 6.17 13.01 12.56 12.04 14.72 16.92 16.91 17.80 20.75 15.59 12.38 13.64 15.88 17.43 14.83

-33.44 -3.62 -2.40 -28.87 -21.62 -1.54 -1.46 -0.68 -55.89 -93.61 11.90 12.61 12.28 14.11 16.57 17.39 18.06 20.51 15.43 12.41 12.24 16.72 15.13 14.13

a Energies and distances are in kcal/mol and Å, respectively, calculated at the MP2/6-31G level. b All SRC1 residues except for the N- and C-terminal residues are listed. c Sum of IFIEs between each H12 residue and LXXLL motif in SRC1. d Sum of IFIEs between each H12 residue and all the SRC1 residues. e CR-CR distances of main chains in H12 residues and SRC1 residues. f Average of Distance1. g O-N or C-C distances of side chains in H12 residues and SRC1 residues. h Average of Distance2.

Further, for one example of the relationship of IFIEs to distances between hydrophobic residues, Distance1 and Distance2 between Met454 and Ile632 are 3.83 and 0.33 Å shorter than those between Met454 and Leu636, respectively, but Met454 has repulsive interaction with Ile632, whereas it has attractive interaction with Leu636. All of the charges without charge redistributions of Met454, Ile632, and Leu636 are 0e, and the calculated fragment charges of these residues are -0.08e, +0.06e, and +0.02e at the MP2/6-31G level, respectively. In addition, the Met454-Ile632 distances are shorter than the Met454-Leu636 distances. Although, on the basis of the information about these charges and distances, the Met454 is anticipated to have the larger attractive interaction with Ile632 than with Leu636, the calculated interaction energies disagree with this anticipation. This result indicates that the interaction energies between hydrophobic residues are not determined only by the charge and distance of residues. It should be noted that interactions between hydrophobic residues are sensitive to the detailed geometry. From these results, it is pointed out that IFIEs at a residue level are not completely correlated with the charge and distance of the residues, though total IFIEs are well correlated with the average distances between the residues. It is suggested that the interaction energies between residues are determined not only by charge and distance but also by the effect of CTs. In addition, the choice of the distance to measure, particularly the distance between side chains, is accompanied by some arbitrariness. Therefore, it is indicated that the interaction energies between residues are difficult to estimate or predict without theoretical calculations. In this analysis, the interaction energies between the residues in H12 and SRC1 were quantitatively evaluated by the FMO calculation.

TABLE 4: IFIEs between Glu453 and Each Amino Acid Residue in the Receptor-Ligand-Coactivator Complex (RLC)a residueb

positionc

IFIEMP2

IFIEHF

∆IFIEd

His634 Lys631 Arg302 Arg635 Lys440 Leu633 Ile632 Lys284 Lys381 Arg285

SRC1 SRC1 H4 SRC1 H11 SRC1 SRC1 H3 L8-9 L3-4

-51.70 -49.16 -40.11 -39.38 -27.78 -27.58 -27.16 -25.51 -20.70 -16.62

-51.46 -46.87 -39.98 -39.38 -27.72 -24.32 -22.30 -25.51 -20.70 -16.62

-0.23 -2.29 -0.13 0.00 -0.06 -3.26 -4.85 0.00 0.00 0.00

a

IFIEMP2 and IFIEHF are IFIEs (in kcal/mol) calculated at the MP2/ 6-31G and HF/6-31G levels, respectively. Only strong attractive interaction energies are shown in order of the IFIEMP2 values. b The N- and C-terminal residues are not listed. c The positions of residues in RXR are shown by helix (H) or loop (L) numbers. d ∆IFIE ) IFIEMP2 - IFIEHF.

The IFIEs of Glu453 with each amino acid residue in the RXR-9cRA-SRC1 complex were calculated by the HF and MP2 methods and only strong attractive interactions are listed in order of the MP2 values as shown in Table 4. The MP2 and HF calculations show that Glu453 mainly interacts with His634 and Lys631 in SRC1. The IFIEs of Glu453 with His634 and Lys631 are 11.59 and 9.05 kcal/mol stronger than that with Arg302, respectively, at the MP2 level. The differences between the MP2 and HF values for the Glu453-His634 and Glu453Lys631 interactions are small, and these values show that large Coulomb forces dominantly contribute to these interactions. As shown in Figure 4a, Glu453 is located between His634 and Lys631, and the distances r2(OGlu453-HHis634) and r1(OGlu453-

3530 J. Phys. Chem. B, Vol. 111, No. 13, 2007

Ito et al. TABLE 6: Charge Differencesa between the Receptor-Ligand-Coactivator Complex (RLC), the Receptor-Ligand Complex and the Coactivator (RL + C), and Individual Molecules (R + L + C) q RLC RL + C R + L + C

method HF/6-31G

RXR 9cRA SRC1 MP2/6-31G RXR 9cRA SRC1

Figure 4. Positions of (a) Glu453 in H12 (red), Lys631 in SRC1 (blue), and His634 in SRC1, and (b) Phe450 in H12 (red), Phe277 in H3 (green), and Ile632 in SRC1 (blue). The distances (dotted lines) are shown in Å.

TABLE 5: IFIEs between Phe450 and Each Amino Acid Residue in the Receptor-Ligand-Coactivator Complex (RLC)a residueb

positionc

IFIEMP2

IFIEHF

∆IFIEd

Phe277 Ile632 Asp448 Met454 Leu276 His634 Met452 Lys274 Arg635 Leu633

H3 SRC1 L11-12 H12 H3 SRC1 H12 H3 SRC1 SRC1

-6.82 -5.74 -3.24 -3.22 -2.48 -2.29 -2.26 -2.18 -1.73 -1.42

-0.91 -0.39 -2.04 -0.67 1.99 -2.10 -0.27 -1.65 -1.73 1.01

-5.91 -5.35 -1.20 -2.54 -4.46 -0.19 -1.99 -0.54 0.00 -2.44

a IFIE MP2 and IFIEHF are IFIEs (in kcal/mol) calculated at the MP2/ 6-31G and HF/6-31G levels, respectively. Only strong attractive interaction energies are shown in order of the IFIEMP2 values. b The N- and C-terminal residues are not listed. c The positions of residues in RXR are shown by helix (H) or loop (L) numbers. d ∆IFIE ) IFIEMP2 - IFIEHF.

HLys631) are short. These calculations indicate that Glu453 plays a role as a clamp of SRC1 in coactivator recruitment. The IFIEs of Phe450 with each amino acid residue in the RXR-9cRA-SRC1 complex were also calculated by the HF and MP2 methods as shown in Table 5. The MP2 calculation shows that Phe450 mainly interacts with Phe277 in H3 and Ile632 in SRC1. Note that, as well as Phe450, Phe277 corresponds to the residue that has been proposed to form the “aromatic clamp”.11 The IFIEs of Phe450 with Phe277 and Ile632 are 3.58 and 2.50 kcal/mol stronger than that with Asp448, respectively, at the MP2 level. The large differences of MP2 and HF values for the Phe450-Phe277 and Phe450Ile632 interactions show that large dispersion forces, which are elusive in the HF calculation, contribute to the interactions between these hydrophobic residues. As shown in Figure 4b, Phe450 is sandwiched between Phe277 and Ile632, and the distances r1(CPhe450-CPhe277) and r2(CPhe450-CIle632) are short. Moreover, it is shown that Phe450 and Phe277 are stabilized by the π-π stacking effect, which is describable by the MP2 method. These calculations indicate that Phe450 plays a role as a clamp of H12 and assists the role of Glu453 for coactivator recruitment. It should be noted that the interactions of Phe450 with hydrophobic residues in this model are not regarded as the hydrophobic interactions but as the dispersion interactions, because solvent effects are not included in the present calculation. If solvent effects are taken into account in the calculation, the role of Phe450 would be associated with the hydrophobic interactions.

-0.02 -0.78 2.80 -0.05 -0.69 2.74

-0.23 -0.77 3.00 -0.32 -0.68 3.00

0.00 -1.00 3.00 0.00 -1.00 3.00

∆q1

∆q2

∆q3

-0.23 0.23 0.00 -0.32 0.32 0.00

0.21 -0.01 -0.20 0.27 -0.01 -0.26

-0.02 0.22 -0.20 -0.05 0.31 -0.26

a ∆q ) [q(RL) + q(C)] - [q(R) + q(L) + q(C)] ) q(RL) - [q(R) 1 + q(L)], ∆q2 ) q(RLC) - [q(RL) + q(C)], and ∆q3 ) q(RLC) [q(R) + q(L) + q(C)]. Charges are in a.u.

3.3. Charge Transfers. The CTs between RXR, 9cRA, and SRC1 were estimated by the HF and MP2 methods as shown in Table 6. The charge changes on the ligand (∆q1) and coactivator (∆q2) binding processes and the total charge change through these binding processes (∆q3) were given as follows,

∆q1 ) q(RL) - [q(R) + q(L)]

(6)

∆q2 ) q(RLC) - [q(RL) + q(C)]

(7)

∆q3 ) q(RLC) - [q(R) + q(L) + q(C)] ) ∆q1 + ∆q2 (8) where q(RLC), q(RL), q(R), q(L), and q(C) are the Mulliken atomic charges of the receptor-ligand-coactivator complex (RLC), the receptor-ligand complex (RL), the receptor (R), the ligand (L), and the coactivator (C), respectively. Because the HF and MP2 calculations for the CTs give similar results, the following discussions are given based on the results calculated at the MP2 level. As we have mentioned above, the charges of RXR, 9cRA, and SRC1 are 0e, -1e, and +3e, respectively, and the total charge of the RXR-9cRA-SRC1 complex is +2e. The calculated charge difference ∆q1 shows that negative charge of 9cRA decreases and an equal amount of negative charge is induced on RXR by -0.32e, indicating that CT (electron transfer) occurs from 9cRA to RXR on the ligand binding. The change in the charge of SRC1 on the ligand binding is zero, since SRC1 does not bind to RXR without the ligand. On the other hand, the calculated ∆q2 shows that the induced negative charge of RXR decreases and almost the same amount of positive charge of SRC1 decreases by -0.26e, indicating that CT occurs from RXR to SRC1 on the coactivator binding. The change in the charge of 9cRA on the coactivator binding is very small. Through the ligand and coactivator binding processes, negative charge of 9cRA decreases by +0.31e of ∆q3 and a similar amount of positive charge of SRC1 decreases by -0.26e of ∆q3. However, it is difficult to consider that CT directly occurs from 9cRA to SRC1 through these processes, because 9cRA does not directly interact with SRC1. The changes in the charges of 9cRA and SRC1 through these processes would be due to the changes in the charge distributions of RXR on each ligand and coactivator binding process. To clarify which residues of RXR are involved in the CTs from 9cRA to RXR (CT19cRAfRXR) and from RXR to SRC1 (CT2RXRfSRC1), the charge differences ∆q1 and ∆q2 at a residue level were then calculated and discussed. Figure 5a shows the visualized charge differences ∆q1 at a residue level calculated by the MP2 method. The positive and negative charge differences are given in blue and red, respectively, and the magnitudes of the charge differences are

Ab Initio Fragment MO Study of Molecular Interactions

J. Phys. Chem. B, Vol. 111, No. 13, 2007 3531

Figure 5. Visualizations of the charge differences (a) ∆q1 and (b) ∆q2 calculated at the MP2/6-31G level: ∆q1 ) q(RL) - [q(R) + q(L)] and ∆q2 ) q(RLC) - [q(RL) + q(C)]. The positive and negative charge differences are showen in blue and red, respectively.

TABLE 7: Charge Changesa of H12 and SRC1 Residues qMP2 position H12

SRC1

qHF

residueb

RLC

RL + C

∆q2_MP2

RLC

RL + C

∆q2_HF

∆∆q2c

Thr449 Phe450 Leu451 Met452 Glu453 Met454 Leu455 Glu456 Total1d Lys631 Ile632 Leu633 His634 Arg635 Leu636 Leu637 Gln638 Total2e

0.07 -0.09 0.08 0.01 -0.82 -0.13 -0.05 -0.93 -1.86 0.97 0.06 0.07 0.84 0.88 0.01 -0.04 -0.10 2.69

0.08 -0.02 0.01 -0.04 -0.90 -0.16 -0.12 -0.86 -2.01 1.09 0.15 0.07 0.82 0.88 0.04 -0.06 -0.18 2.81

-0.01 -0.07 0.07 0.05 0.08 0.03 0.07 -0.07 0.15 -0.12 -0.09 0.00 0.02 0.00 -0.03 0.02 0.08 -0.12

0.06 -0.10 0.07 0.01 -0.89 -0.09 -0.03 -0.93 -1.81 0.95 0.05 0.06 0.87 0.91 0.01 -0.02 -0.08 2.75

0.07 -0.03 0.01 -0.03 -0.92 -0.12 -0.10 -0.86 -1.98 1.06 0.11 0.05 0.86 0.92 0.05 -0.04 -0.14 2.87

-0.01 -0.07 0.06 0.04 0.03 0.03 0.07 -0.07 0.17 -0.11 -0.06 0.01 0.01 -0.01 -0.04 0.02 0.06 -0.12

0.00 0.00 0.01 0.01 0.05 0.00 0.00 0.00 -0.02 -0.01 -0.03 -0.01 0.01 0.01 0.01 0.00 0.02 0.00

a ∆q ) q(RLC) - [q(RL) + q(C)]. q 2 MP2 and qHF are charges (in a.u.) calculated at the MP2/6-31G and HF/6-31G levels, respectively. ∆q2_MP2 and ∆q2_HF are charge differences (in a.u.) calculated at the MP2/6-31G and HF/6-31G levels, respectively. b The N- and C-terminal residues are not listed. c ∆∆q2 ) ∆q2_MP2 - ∆q2_HF. d Sum of charges or charge differences of H12 residues. e Sum of charges or charge differences of SRC1 residues.

represented by the deepness of hue. 9cRA is given in darkblue according to its calculated large charge difference of +0.32e. Only two residues (Arg316 and Leu326) of the ligand binding pocket are given in dark-red according to their calculated large charge differences of -0.08e and -0.07e, respectively, and the other residues of the ligand binding pocket are given in light-red or light-blue according to their calculated small charge differences. Thus, it is shown that about half of the CT19cRAfRXR occurs primarily from 9cRA to these two residues. Figure 5b shows the visualized charge differences ∆q2 at a residue level calculated by the MP2 method. The amino acid residues in SRC1 and the coactivator binding site of RXR are given in red or blue. In particular, two residues (Lys631 and Ile632) of SRC1 are given in dark-red, and three residues (Leu451, Glu453, and Leu455) of RXR are given in dark-blue on the dark-red side of SRC1. All three of these residues of RXR belong to H12, and Glu453 corresponds to the highly conserved charged residue. The calculated positive charge differences of Leu451, Glu453, and Leu455 are +0.07e, +0.08e, and +0.07e, respectively, and the calculated negative charge differences of Lys631 and Ile632 are -0.12e and -0.09e, respectively. The sum of the positive charge differences calculated for Leu451, Glu453, and Leu455

amounts to 81.5% of the total ∆q2 of RXR (+0.27e), and the sum of the negative charge differences calculated for Lys631 and Ile632 amounts to 80.8% of the total ∆q2 of SRC1 (-0.26e). Thus, it is shown that the CT2RXRfSRC1 occurs mainly from these three residues of H12 to these two residues of SRC1. Our results indicate that CT19cRAfRXR and CT2RXRfSRC1 are caused by the changes in charges of amino acid residues in the ligand binding pocket and the coactivator binding site of RXR, respectively, and that these CTs are phenomena independent of each other. Therefore, it is suggested that the ligand binding does not have a strong influence on the change in charge of the coactivator binding site formed by H12, though the ligand binding is known to have a strong influence on the change in the structure of H12.3,5-7 To examine more details of the CT on the coactivator binding, the charge changes of H12 and SRC1 residues in CT2RXRfSRC1 calculated at the MP2/6-31G level were analyzed and compared to those calculated at the HF/6-31G level as shown in Table 7. qMP2 and qHF are the Mulliken atomic charges of the receptorligand-coactivator complex (RLC) or the receptor-ligand complex and the coactivator (RL + C) calculated at the MP2/ 6-31G and HF/6-31G levels, respectively. ∆q2_MP2 and ∆q2_HF

3532 J. Phys. Chem. B, Vol. 111, No. 13, 2007 are the charge changes in CT2RXRfSRC1 calculated at the MP2/ 6-31G and HF/6-31G levels, respectively, and ∆∆q2 is the difference between ∆q2_MP2 and ∆q2_HF. Total1 and Total2 are sums of q, ∆q2, or ∆∆q2 of H12 residues and SRC1 residues, respectively. As shown in Table 7, Total1 and Total2 of ∆q2_MP2 are +0.15e and -0.12e and these values of ∆q2_HF are +0.17e and -0.12e, respectively. These values show that about the same amount of charge transfers from H12 to SRC1 by both MP2 and HF calculations. By the MP2 method, the large positive charge changes are calculated for Leu451, Glu453, and Leu455 of H12, and the large negative charge changes are calculated for Lys631 and Ile632 of SRC1. The absolute value of the sum of ∆q2_MP2 calculated for Leu451 (+0.07e), Glu453 (+0.08e) of H12, and Leu455 (+0.07e) is almost equal to that of the sum of ∆q2_MP2 calculated for Lys631 (-0.12e) and Ile632 (-0.09e) of SRC1. It is thus shown that the CT2RXRfSRC1 occurs mainly from these three residues of H12 to these two residues of SRC1 as mentioned above. In addition, it is also shown that Glu453 plays an important role as a probable candidate for the electron donor for Lys631 and Ile632 in CT2RXRfSRC1 with the largest positive charge change in the residues of H12. As well as the result of the MP2 calculation, the absolute value of the sum of ∆q2_HF calculated for Leu451 (+0.06e), Glu453 (+0.03e) of H12, and Leu455 (+0.07e) is almost equal to that of the sum of ∆q2_HF calculated for Lys631 (-0.11e) and Ile632 (-0.06e) of SRC1. It is thus shown that the CT2RXRfSRC1 would occur from these three residues of H12 to these two residues of SRC1. On the other hand, in contrast to the result of the MP2 calculation, the positive charge change calculated for Glu453 is smaller than that calculated for the other residues in H12 by the HF calculation. This difference between the MP2 and HF calculations appears in the large ∆∆q2 value for Glu453 (+0.05e). Our results indicate that the changes in charges of the amino acid residues in H12 of RXR, especially of the highly conserved charged reside Glu453, would be essential for CT2RXRfSRC1 on the coactivator binding, which can be appropriately described in terms of the calculations accounting for the electron correlation effects. 4. Conclusions The ab initio FMO calculations were performed for molecular interactions between 9cRA liganded RXR and SRC1 coactivator to examine how H12, which contains the AF-2 AD core, and its highly conserved residues may contribute to the coactivator binding of RXR. In the RXR-9cRA-SRC1 complex, SRC1 was proved to be stabilized mainly by the interaction with H12. In particular, the highly conserved charged (Glu453) and hydrophobic (Phe450) residues in H12 were found to interact with SRC1 more strongly than the other charged and hydrophobic residues in H12, respectively. Strong electrostatic interactions of Glu453 with His634 and Lys631 in SRC1 were revealed, indicating that Glu453 plays a role as a clamp of SRC1 in the coactivator recruitment. Strong dispersion interactions of Phe450 with Phe277 in H3 and Ile632 in SRC1 were also revealed, indicating that Phe450 plays a role as a clamp of H12 and assists the role of Glu453. These computational findings are in good agreement with the experimental proposals10,11 of a “charge clamp” and an “aromatic clamp” for Glu453 and Phe450, respectively. In addition, on the coactivator binding, the CT from RXR to SRC1 was found to occur mainly through the changes in charges of H12 residues. In this CT process, the largest positive charge difference was obtained for Glu453 in H12 residues and larger negative charge differences were obtained for Lys631 and Ile632 than the other SRC1 residues.

Ito et al. The charge change of Glu453 was found to have an influence on the charge changes of Lys631 and Ile632. These calculations indicate that Glu453 plays another role as an electron donor for Lys631 and Ile632. Taken together, our findings quantitatively demonstrated that H12 and its highly conserved residues have strong interactions with SRC1 by the Coulomb and dispersion forces, and that they have important roles as electron donors for SRC1 residues on the CT from RXR to SRC1. Furthermore, this work provided physical insights into why the same residues of H12 corresponding to Glu453 and Phe450 in RXR are highly conserved in NRs. To completely elucidate the roles of H12 and its highly conserved residues, computational comparisons among several receptors of NRs are hoped to be carried out in future. To obtain more quantitative understanding of these interactions, solvent effects and geometry optimization of the entire complex and its components should be taken into account in further theoretical studies. However, the knowledge obtained from this work based on the ab initio calculations could be helpful for our better understanding of the transcriptional activation mechanism of RXR and related NRs. Acknowledgment. We thank Prof. Kuniyoshi Ebina for useful comments, and Dr. Takeshi Ishikawa and Dr. Hirofumi Watanabe for technical assistance. This work was supported by the “Core Research for Evolutional Science and Technology” (CREST) project of the Japan Science and Technology Agency. References and Notes (1) Ross, S. A.; McCaffery, P. J.; Drager, U. C.; Luca, L. M. D. Physiol. ReV. 2000, 80, 1021-1054. (2) Miles, S. A.; Dezube, B. J.; Lee, J. Y.; Krown, S. E.; Fletcher, M. A.; Saville, M. W.; Kaplan, L.; Groopman, J.; Scadden, D. T.; Cooley, T.; Von Roenn, J.; Friedman-Kien, A. AIDS 2002, 16, 421-429. (3) Nagy, L.; Schwabe, J. W. Trends Biochem. Sci. 2004, 29, 317324. (4) Bourguet, W.; Ruff, M.; Chambon, P.; Gronemeyer, H.; Moras, D. Nature 1995, 375, 377-382. (5) Renaud, J. P.; Rochel, N.; Ruff, M.; Vivat, V.; Chambon, P.; Gronemeyer, H.; Moras, D. Nature 1995, 378, 681-689. (6) Egea, P. F.; Mitschler, A.; Rochel, N.; Ruff, M.; Chambon, P.; Moras, D. EMBO J. 2000, 19, 2592-2601. (7) Moras, D.; Gronemeyer, H. Curr. Opin. Cell Biol. 1998, 10, 384391. (8) Nolte, R. T.; Wisely, G. B.; Westin, S.; Cobb, J. E.; Lambert, M. H.; Kurokawa, R.; Rosenfeld, M. G.; Willson, T. M.; Glass, C. K.; Milburn, M. V. Nature 1998, 395, 137-143. (9) Darimont, B. D.; Wagner, R. L.; Apriletti, J. W.; Stallcup, M. R.; Kushner, P. J.; Baxter, J. D.; Fletterick, R. J.; Yamamoto, K. R. Genes DeV. 1998, 12, 3343-3356. (10) Gampe, R. T.; Montana, V. G.; Lambert, M. H.; Miller, A. B.; Bledsoe, R. K.; Milburn, M. V.; Kliewer, S. A.; Willson, T. M.; Xu, H. E. Mol. Cell 2000, 5, 545-555. (11) Pogenberg, V.; Guichou, J. F.; Vivat-Hannah, V.; Kammerer, S.; Pe´rez, E.; Germain, P.; Lera, A. R.; Gronemeyer, H.; Royer, C. A.; Bourguet, W. J. Biol. Chem. 2005, 280, 1625-1633. (12) Tone, Y.; Collingwood, T. N.; Adams, M.; Chatterjee, V. K. J. Biol. Chem. 1994, 269, 31157-31161. (13) Collingwood, T. N.; Rajanayagam, O.; Adams, M.; Wagner, R.; Cavaille`s, V.; Kalkhoven, E.; Matthews, C.; Nystrom, E.; Stenlof, K.; Lindstedt, G.; Tisell, L.; Fletterick, R. J.; Parker, M. G.; Chatterjee, V. K. K. Proc. Natl. Acad. Sci. U.S.A. 1997, 94, 248-253. (14) Gurnell, M.; Wentworth, J. M.; Agostini, M.; Adams, M.; Collingwood, T. N.; Provenzano, C.; Browne, P. O.; Rajanayagam, O.; Burris, T. P.; Schwabe, J. W.; Lazar, M. A.; Chatterjee, V. K. K. J. Biol. Chem. 2000, 275, 5754-5759. (15) Andersin, T.; Va¨isa¨nen, S.; Carlberg, C. Mol. Endocrinol. 2003, 17, 234-246. (16) Kitaura, K.; Sawai, T.; Asada, T.; Nakano, T.; Uebayasi, M. Chem. Phys. Lett. 1999, 312, 319-324. (17) Kitaura, K.; Ikeo, E.; Asada, T.; Nakano, T.; Uebayasi, M. Chem. Phys. Lett. 1999, 313, 701-706.

Ab Initio Fragment MO Study of Molecular Interactions (18) Nakano, T.; Kaminuma, T.; Sato, T.; Akiyama, Y.; Uebayasi, M.; Kitaura, K. Chem. Phys. Lett. 2000, 318, 614-618. (19) Nakano, T.; Kaminuma, T.; Sato, T.; Fukuzawa, K.; Akiyama, Y.; Uebayasi, M.; Kitaura, K. Chem. Phys. Lett. 2002, 351, 475-480. (20) Fedorov, D. G.; Kitaura, K. Theoretical Development of the Fragment Molecular Orbital (FMO) Method. In Modern Methods for Theoretical Physical Chemistry of Biopolymers; Starikov, E. B., Lewis, J. B., Tanaka, S., Eds.; Elsevier: Amsterdam, The Netherlands, 2006; pp 3-38. (21) Nakano, T.; Mochizuki, Y.; Fukuzawa, K.; Amari, S.; Tanaka, S. Developments and Applications of ABINIT-MP Software Based on the Fragment Molecular Orbital Method. In Modern Methods for Theoretical Physical Chemistry of Biopolymers; Starikov, E. B., Lewis, J. B., Tanaka, S., Eds.; Elsevier: Amsterdam, The Netherlands, 2006; pp 3952. (22) Amari, S.; Aizawa, M.; Zhang, J.; Fukuzawa, K.; Mochizuki, Y.; Iwasawa, Y.; Nakata, K.; Chuman, H.; Nakano, T. J. Chem. Inf. Model. 2006, 46, 221-230. (23) Nemoto, T.; Fedorov, D. G.; Uebayasi, M.; Kanazawa, K.; Kitaura, K.; Komeiji, Y. Comput. Biol. Chem. 2005, 29, 434-439. (24) Fukuzawa, K.; Mochizuki, Y.; Tanaka, S.; Kitaura, K.; Nakano, T. J. Phys. Chem. B 2006, 110, 16102-16110.

J. Phys. Chem. B, Vol. 111, No. 13, 2007 3533 (25) Fukuzawa, K.; Komeiji, Y.; Mochizuki, Y.; Kato, A.; Nakano, T.; Tanaka, S. J. Comput. Chem. 2006, 27, 948-960. (26) Mochizuki, Y.; Nakano, T.; Koikegami, S.; Tanimori, S.; Abe, Y.; Nagashima, U.; Kitaura, K. Theor. Chem. Acc. 2004, 112, 442-452. (27) Mochizuki, Y.; Koikegami, S.; Nakano, T.; Amari, S.; Kitaura, K. Chem. Phys. Lett. 2004, 396, 473-479. (28) Berman, H. M.; Westbrook, J.; Feng, Z.; Gilliland, G.; Bhat, T. N.; Weissig, H.; Shindyalov, I. N.; Bourne, P. E. Nucleic Acids Res. 2000, 28, 235-242. The RCSB Protein Data Bank (http://www.rcsb.org/). (29) MOE Version 2005.06, Chemical Computing Group, Montreal, Canada, 2005. (30) Wang, J.; Cieplak, P.; Kollman, P. A. J. Comput. Chem. 2000, 21, 1049-1074. (31) ABINIT-MP ver 3.0 is available from the website of RSS21 project: http://www.rss21.iis.u-tokyo.ac.jp/en. (32) BioStation Viewer ver 6.00 is available from the website of RSS21 project: http://www.rss21.iis.u-tokyo.ac.jp/en. (33) van Lipzig, M. M. H.; ter Laak, A. M.; Jongejan, A.; Vermeulen, N. P. E.; Wamelink, M.; Geerke, D.; Meerman, J. H. N. J. Med. Chem. 2004, 47, 1018-1030. (34) Sˆ poner, J.; Jurecˇka, P.; Hobza, P. J. Am. Chem. Soc. 2004, 126, 10142-10151.