5700
J. Phys. Chem. B 2007, 111, 5700-5707
An ab Initio Investigation of the Interactions Involving the Aromatic Group of the Set of Fluorinated N-(4-Sulfamylbenzoyl)benzylamine Inhibitors and Human Carbonic Anhydrase II Kevin E. Riley, Guanglei Cui, and Kenneth M. Merz, Jr.* Quantum Theory Project and Department of Chemistry, The UniVersity of Florida, GainesVille, Florida 32611-8435 ReceiVed: NoVember 6, 2006; In Final Form: March 13, 2007
In this work we investigate the interactions that occur between the aromatic portion of the set of fluorinated N-(4-sulfamylbenzoyl)benzylamine (SBB) inhibitors and two residues of Human Carbonic Anhydrase II (HCAII), namely Phe-131 and Pro-202. Calculations were carried out at the MP2/aug-cc-pVDZ level of theory and the counterpoise scheme of Boys and Bernardi was employed to account for the basis set superposition error. The most striking result obtained here is that the SBB phenyl ring interacts at least as strongly with the proline pyrrolidine ring as with the phenylalanine phenyl ring, which is surprising because aromatic-aromatic interactions have long been thought to be particularly favorable in protein and protein-ligand structure. Comparison of the MP2 binding energies to those obtained with the Hartree-Fock method indicates that the attraction between the proline pyrrolidine ring and the SBB phenyl ring is largely attributable to dispersion forces. These favorable interactions between pyrrolidine and phenyl rings may have important implications in protein structure because there is potential for proline residues to interact with phenylalanine residues in a fashion analogous to that seen here. A preliminary protein data bank search indicates that the prolinephenylalanine contacts are about 40% as common as those between two phenylalanines. It is also found here that the number and pattern of fluorine substituents on the SBB phenyl ring is much less important in determining the SBB-HCAII binding energy than the relative geometric configuration of the interacting pairs.
1. Introduction There are many forces that contribute to the stability of a bound protein-ligand complex, among these are dispersion interactions, electrostatic interactions, and hydrophobic interactions. Recent studies have suggested that dispersion interactions, although they are generally weak and nonspecific, play an important role in the structure and stability of proteins and protein-ligand systems.1-7 Another group of interactions that are also thought to play a role in the stability of proteins and protein ligand systems are those between two aromatic rings; these have been the focus of many studies in the past several years.8-17 One example of a set of protein-ligand systems that demonstrate aromatic-aromatic interactions is the set of fluorine-substituted N-(4-sulfamylbenzoyl)benzylamine (SBB) inhibitors bound to Human Carbonic Anhydrase II (HCAII) (see Figure 1).8,10,18-20 HCAII is a zinc metalloenzyme that catalyzes the hydration of carbon dioxide releasing bicarbonate and a proton. Inhibition of HCAII is of clinical importance and can be useful in the treatment of various diseases such as glaucoma.21 Kim et al. have done high-resolution X-ray crystallographic work on five SBB-HCAII complexes with varying degrees of fluorination on the aromatic portion of the SBB ligand.10 In these studies it was found that various fluorine-substitution patterns modulated the enzyme-inhibitor affinity by a factor of 10. This study, as well as several others, showed that one of the key contacts leading to the overall binding of the set of (fluorinated and unfluorinated) SBB ligands to HCAII is the
Figure 1. SBB interactions with Phe-131 and Pro-202: 2-fluoro-SBB (green), 2,3-difluoro-SBB (blue), 2,6-difluoro-SBB (red), and 2,3,4,5,6pentafluoro-SBB (yellow).
interaction between the aromatic portion of SBB and the aromatic side chain of Phe-131.10,19,20 These interactions geometrically approximate the edge-to-face (T-shaped) benzenedimer interaction, which is known to be a local minimum for this complex. The observed range of separations between the centroids of the two aromatic rings was 3.4-6.5 Å, a range of distances associated with quadrupole-quadrupole interactions.10 Another contact that seems to be of importance in the SBBHCAII complex is that between the SBB aromatic ring and Pro202; this interaction is probably largely based on dispersion
10.1021/jp067313m CCC: $37.00 © 2007 American Chemical Society Published on Web 05/03/2007
Interactions between Phe-131 and Pro-202 forces. It has also been suggested that a dipole-induced dipole interaction between the SBB aromatic ring and the Pro-202 side chain may contribute to the overall SBB-Pro-202 binding.10 The substitution of fluorine atoms for hydrogen atoms is commonly exploited in medicinal chemistry to enhance ligand binding to proteins.10 Fluorine is an isosteric substitution for hydrogen and is isoelectronic with a hydroxyl group which, along with its relative stability, makes it a very suitable atom for substitution into small molecule ligands. Hence, a fluorinated small molecule ligand should be capable of binding in general the same location as a non-fluorinated ligand, but the chemical properties of the modified fluorine-substituted ligand may impact protein-ligand affinity and selectivity.22 The substitution of fluorine atoms in the benzyl group of the SBB inhibitor modifies the interaction of the benzyl group of SBB with both the Phe131 and Pro-202 residues, and thus influences the affinity of the SBB inhibitor for HCAII. The edge-to-face aromatic interaction is an example of a quadrupole-quadrupole interaction in which the partial positive charge on the ring hydrogen of the upper (edge) benzene (see Figure 1) interacts favorably with the partial negative charge above the aromatic ring of the lower (face) benzene.23 Substitution of fluorine onto the lower benzene molecule diminishes the partial negative character of the π cloud above the ring, thereby reducing the attractive forces between the two benzene molecules. Using this model, one might expect that the addition of further fluorine substituents would result in an increase in the intercentroid separation and a decrease in the binding energy of the benzene dimer system. In a recent study Sinnokrot and Sherrill carried out studies in which several substituents were introduced to a benzene dimer system by using the MP2 and CCSD(T) methods.24 Fluorine substitution onto the “face” ring generally resulted in a negligible change in intercentroid separation and a decrease in the binding energy by ca. 0.5 kcal/ mol. In another work Riley and Merz studied the effects of multiple fluorine substitutions onto the “face” ring of a benzene dimer; in this study it was shown that an increase in fluorine substituents generally yields a decrease in the binding energy of the complex and a small increase in the intercentroid separation between the two rings.9 Surprisingly, it was found in the studies of Kim et al. that the fluoroaromatic ring of each inhibitor tended to shift closer to Phe-131 with increasing fluorination.10 This observation contradicts the computationally predicted behavior of these types of fluorinated aromatic systems. Gaining insight into the energy of interaction of a proteinligand system at the molecular level is a significant challenge. The data obtained with experimental techniques are generally of a macroscopic nature. Theoretical methods lend themselves naturally to the computation of interactions at the molecular or atomic level. It is now possible to make calculations on large systems such as proteins by using semiempirical, density functional, and Hartree-Fock methods.25-28 These techniques, however, lack the precision needed to describe weak interactions such as those of dispersion bound systems. To understand the behavior of aromatic complexes in proteins it becomes necessary to carry out higher level calculations on smaller model systems chosen to mimic the aromatic moieties found within the system of interest. The behavior of these smaller model systems can then be related back to the original, larger, systems. The main objective of this study is to investigate the binding of the set of SBB inhibitors to the HCAII enzyme by calculating interaction energies of model systems that mimic the behavior of the SBB benzyl ring, the Phe-131 phenyl ring, and the Pro-
J. Phys. Chem. B, Vol. 111, No. 20, 2007 5701
Figure 2. Model systems for interactions of Phe-131 with 2-fluoroSBB (a), 2,3-difluoro-SBB (b), 2,6-difluoro-SBB (c), and 2,3,4,5,6pentafluoro-SBB (d).
202 ring. The analysis of the interactions between these model systems will give some insight into the roles of the SBB-Phe131 and SBB-Pro-202 contacts in the overall binding of the SBB-HCAII enzyme-inhibitor complex; it can also elucidate the effects of fluorine substitution and geometry on the individual interactions. Another goal of the current work is to determine the role that solvation effects play in the SBB-Phe131 and SBB-Pro-202 interactions. 2. Methods To make high-level calculations on interactions found within proteins or protein-ligand systems it is necessary to choose relatively small model systems that mimic the behavior of the chemical moieties in which we are interested. The system used to model the interactions of the SBB benzyl group with the phenyl ring of Phe131 is the dimer formed by the phenylalanine amino acid and toluene (representing the SBB aromatic group, see Figure 2). For the SBB-Pro-202 interaction toluene is used to model the benzyl ring of SBB while the proline residue is modeled as a pyrrolidine ring along with several of the backbone atoms found near proline (see Figures 3and 4). The coordinates of the heavy atoms (C, O, N, etc.) were taken directly from the X-ray structure data and held fixed at these positions, the hydrogen atoms associated with our model structures were added to the systems and their positions were optimized at the B3LYP/ 6-31+G* level of theory. There are two groups of crystal structures utilized in this study, those based on the native protein and those based on a mutant form of the protein in which Phe-131 has been replaced with valine (Phe-131 w Val). The first of these groups comprises the interactions of the HCAII enzyme with the 2-fluoro-SBB, 2,3-difluoro-SBB, 2,6-difluoro-SBB, and pentafluoro-SBB inhibitors. The second group of crystal structures contains the mutated HCAII enzyme bound to the (unfluorinated) SBB, 2-fluoro-SBB, 2,3-difluoro-SBB, 2,6-difluoro-SBB, and pentafluoro-SBB inhibitors. It should be noted that the crystal structure of the (unfluorinated) SBB inhibitor complexed with the native HCAII protein is not available. Crystal structures of
5702 J. Phys. Chem. B, Vol. 111, No. 20, 2007
Riley et al. TABLE 1: Experimental Binding Energies and PDB ID Numbers for Each of the SBB Inhibitors Considered in This Work
Figure 3. Model systems for interactions of Pro-202 with 2-fluoroSBB (a), 2,3-difluoro-SBB (b), 2,6-difluoro-SBB (c), and 2,3,4,5,6pentafluoro-SBB (d).
a
Units: kcal/mol.
schematic drawings, and experimental total binding energies for each of the protein-ligand systems studied in this work. To gain insight into the effects of geometry and fluorination pattern on the strengths of the individual interactions considered here, we have computed the interaction energies of model structures of SBB-Phe-131 and SBB-Pro-202 pairs with varying fluorination patterns of the SBB benzyl group for a given stationary geometry. These new model structures keep the heavy atom geometries of a given structure but the fluorines are substituted according to the patterns of the other SBB inhibitors within that crystal structure group. For example, the 2-fluoro, 2,6-difluoro, and pentafluoro fluorination patterns are substituted onto the geometry of the 2,3-difluoro-SBB-Phe131 complex. The geometries of the substituted fluorines were optimized at the B3LYP/6-31+G* level of theory. All ab initio calculations in this work were carried out with the Gaussian 03 program. Second-order perturbation theory (MP2) methods were employed to account for electron correlation effects.29 The basis set used for all MP2 calculations in this work is aug-cc-pVDZ.30 To correct for the basis set superposition error (BSSE) the counterpoise method of Boys and Bernardi was employed.31 To account for solvation effects the solvation contribution to the interresidue interaction energy is estimated to be the difference between the interaction energy of the solvated system and that of the system in vacuum
∆∆Gsolv ) ∆Gsolv - ∆Eno-CP vac The total interaction energy in solvent is given by Figure 4. Model systems for interactions of Pro-202 (Phe-131 w Val) with SBB (a), 2-fluoro-SBB (b), 2,3-difluoro-SBB (c), 2,6-difluoroSBB (d), and 2,3,4,5,6-pentafluoro-SBB (e).
all HCAII-SBB protein-ligand systems were obtained from the protein data bank (PDB). Table 1 gives the PDB ID numbers,
CP ∆ECP solv ) ∆Evac + ∆∆Gsolv
It should be noted here that ∆ECP vac is the interaction energy of the amino acid dimer as calculated in vacuum and corrected for the basis set superposition error (BSSE), whereas ∆Evac and ∆Gsolv are interaction energies with no BSSE corrections.
Interactions between Phe-131 and Pro-202
J. Phys. Chem. B, Vol. 111, No. 20, 2007 5703
Figure 5. Interaction between Phe-131 and 2-fluoro-SBB (green), 2,3difluoro-SBB (blue), 2,6-difluoro-SBB (red), and 2,3,4,5,6-pentafluoroSBB (yellow).
TABLE 2: Intercentroid Separation Distances for Each of the SBB-Phe-131 and SBB-Pro-202 Complexes Considered in this Worka
Figure 6. Phe-131-SBB interactions in vacuum, water, and ether.
2,3,4,5,62,62,32-fluoro- difluoro- difluoro- pentafluoroSBB SBB SBB SBB SBB Phe-131 (native) Pro-202 (native) Pro-202 (mutant)b 4.82 a
5.96 5.16 4.95
5.50 5.76 5.04
6.75 4.79 4.87
5.08 6.53 5.14
Distances in angstroms.
Solvation effects are modeled by using the polarizable continuum method (PCM),32 a method that has been well validated and is the subject of several reviews.33-35 To gain insight into the effects of solvation within protein and nucleic systems we have carried out our calculations with two different solvents: the first of these is ether ( ) 4.335) and the second is water ( ) 78.39). Ether is used because its dielectric constant is taken to represent the overall dielectric constant in the interior of a protein.36-38 3. Results and Discussion 3.1. Interactions of Wild Type HCAII with SBB Inhibitors. i. SBB-Phe-131 Interactions. Figure 5 shows the interactions of the phenyl ring of Phe-131 with the variously fluorinated SBB inhibitors, here it can be seen that the 2-flouro-SBB, 2,3difluoro-SBB, and 2,3,4,5,6-pentafluoro-SBB benzyl groups all interact with the Phe-131 phenyl ring with conformations that are approximately of the edge-to-face (T-shaped) type (see also Figure 2, parts a, b, and d). The 2,3,4,5,6-pentafluoro-SBBPhe-131 complex represents the aromatic interaction that most closely resembles the edge-to-face interaction while the other three complexes have aromatic interactions that deviate from an edge-to-face configuration more significantly. The geometry between the aromatic groups in the 2,6-difluoro-SBB-Phe-131 complex does not represent a T-shaped interaction. There is a great deal of variation in the separation distances between the aromatic groups in Phe-131 and the SBB inhibitors, Table 2 gives the intercentroid distances for each of the complexes considered in this work. The 2,3,4,5,6-pentafluoro-SBB-Phe131 complex exhibits the shortest intercentroid separation of 5.08 Å, and this distance is very close to the optimum intercentroid separation of an edge-to-face benzene dimer, which has a value of 5.0 Å.9 With a value of 6.75 Å, the intercentroid separation of the 2,6-diflouro-SBB-Phe-131 complex is the largest among all the systems considered in this work. Figure 6 gives the interaction energies in vacuum, water, and ether for the various fluorinated phenylalanine-toluene dimers (see Figure 2), which mimic the interaction between the phenyl rings of the Phe-131 enzyme and the SBB inhibitor. Here it can be seen that, in each medium, 2,3,4,5,6-pentafluoro-SBB is predicted to have the strongest interaction with Phe-131 (vacuum: -2.82 kcal/mol; water: -1.73 kcal/mol; ether:
Figure 7. Phe-131-SBB interaction energies for native SBB-HCAII. Histogram columns refer to geometries and colors refer to fluorine substitution pattern.
-1.63 kcal/mol). This result is somewhat surprising because generally the edge-to-face aromatic interaction becomes weaker with the addition of fluorine substituents onto the “face” group.9 A possible explanation for this result is that the near-ideal geometry of the phenyl groups in the 2,3,4,5,6-pentafluoroSBB-Phe-131 complex compensates for the detrimental effects of having five substituted fluorines. The 2,6-difluoro-SBBPhe-131 complex exhibits the weakest interactions in vacuum, water, and ether; this is an expected result as the relative geometry of the pair of phenyl rings in this complex does not assume a conformation that resembles a minimum energy conformation of the benzene dimer. As would be expected, the introduction of both water and ether results in a significant destabilization of the SBB-Phe131 interactions. The greatest increase (destabilization) in the binding energy is 1.19 kcal/mol (from -2.82 kcal/mol to -1.63 kcal/mol), which occurs for the 2,3,4,5,6-pentafluoro-SBB-Phe131 complex with the introduction of ether. The smallest increase in the interaction energy occurs for the 2,6-difluoroSBB-Phe-131 complex upon introduction of water, and this interaction is destabilized by 0.62 kcal/mol (from -1.07 kcal/ mol to -0.45 kcal/mol) Figure 7 shows SBB-Phe-131 interaction energies as a function of geometry and fluorination pattern. The most prominent feature of these data is that the relative orientation of these toluene dimers has a much greater impact on the strength of their interaction than the pattern (or number) of the fluorine substituents. As one might expect the geometric configuration of 2,3,4,5,6-pentaflouro-SBB-Phe-131 complex, which represents a nearly perfect aromatic edge-to-face interaction, is the most stable configuration among these dimers. At this geometry the 2,3-F (analogous to 2-fluoro-SBB) fluorine
5704 J. Phys. Chem. B, Vol. 111, No. 20, 2007
Riley et al.
Figure 8. Interactions of Pro-202 with 2-fluoro-SBB (green), 2,3difluoro-SBB (blue), 2,6-difluoro-SBB (red), and 2,3,4,5,6-pentafluoroSBB (yellow).
substitution pattern yields the most favorable interaction with a binding energy of -3.45 kcal/mol, the least stable fluorine pattern at this geometry is penta-F with a binding energy of -2.82 kcal/mol. The least stable geometry configuration is that of 2,6-difluoro-SBB-Phe-131, which exhibits no edge-to-face characteristics, the most stable dimer at this geometry is 2-F with a binding energy of -1.12 kcal/mol. For each configuration the penta-F substitution pattern produces the highest binding energy, and this result is generally consistent with expectations. Somewhat surprisingly, the 2-F substitution pattern exhibits the lowest binding energies for only two of the four geometric configurations, 2,3-F and 2,6-F. It is an especially unexpected result that the 2-F is not the most stable fluorination pattern at the 2,3,4,5,6-pentafluoro-SBB-Phe-131 geometry, it should be noted, however, that the 2,3-F substitution pattern is favored above 2-F by only 0.09 kcal/mol at this geometry. ii. SBB-Pro-202 Interactions. Figure 8 shows the interactions of the Pro-202 with the variously fluorinated phenyl rings of the SBB inhibitor, and here it can be seen that the interaction between the 2-flouro-SBB aromatic ring and the pyrrolidine ring of Pro-202 assumes a configuration that can be likened to an edge-to-face aromatic interaction; this is also shown (more clearly) in Figure 3a. The phenyl rings of the other three inhibitors, 2-fluoro-SBB, 2,6-difluoro-SBB, and 2,3,4,5,6-pentafluoro-SBB, diverge substantially from any type of edge-toface interaction. It is interesting to note that the interaction between the Pro-202 pyrrolidine ring and the 2,6-difluoro-SBB phenyl ring is very crowded (also see Figure 3c): the smallest distance between hydrogen atoms located on each of the monomers is 2.01 Å. Table 2 gives the centroid separation distances for the four (wild type) SBB-Pro-202 complexes considered in this work. Here it can be seen that the 2,3,4,5,6pentafluoro-SBB-Pro-202 complex exhibits the longest intercentroid separation of 6.53 Å. With a value of 4.79 Å, the intercentroid separation of the 2,6-diflouro-SBB-Pro-202 complex is the shortest among these systems. Figure 9 gives the interaction energies in vacuum, water, and ether of the SBB-Pro-202 complexes with various fluorination patterns. As might be expected, the 2-fluoro-SBB inhibitor is predicted to form the most stable complex with Pro-202 in each of the media considered here (vacuum: -2.75 kcal/mol; water: -1.74 kcal/mol; ether: -1.82 kcal/mol). This result can most likely be explained by the seemingly favorable geometry of the pyrrolidine-phenyl interaction, which is approximately of a T-shaped type; it should be noted, however, that, to our knowledge, there have been no studies characterizing the nature of the interactions between aromatic and pyrrolidine groups. The weakest interaction occurs between 2,6-difluoro-SBB and Pro-202, which can probably be attributed to steric effects, as the SBB phenyl ring and the proline pyrrolidine group are very close to one another. As in the case of the SBB-Phe-131 interactions the introduction of both water and ether results in a destabilization of all of
Figure 9. Pro-202-SBB interactions in vacuum, water, and ether.
Figure 10. Pro-202-SBB interactions for native SBB-HCAII. Histogram columns refer to geometries and colors refer to fluorine substitution pattern.
the SBB-Pro-202 interactions. For each SBB inhibitor, with the exception of 2-fluoro-SBB, the interaction with Pro-202 is stronger in water than in ether. The largest solvation binding destabilization occurs for the 2-fluoro-SBB-Pro-202 complex with the introduction of water: the binding energy increases by 1.01 kcal/mol (from 2.75 kcal/mol to 1.74 kcal/mol). The smallest increase in the binding energy upon solvation is 0.36 kcal/mol (from 0.81 kcal/mol to 0.45 kcal/mol), which occurs for the 2,6-difluoro-SBB-Pro-202 complex upon introduction of water. Figure 10 gives the SBB-Pro-202 binding energies as a function of configurations and fluorination patterns. As in the case of the SBB-Phe-131 interactions, geometric parameters seem to influence the strengths of interaction for these complexes much more strongly than the fluorination patterns. The most stable geometry is that of the 2-fluoro-SBB-Pro-202 complex, with the penta-F substitution pattern yielding the lowest overall binding energy of -2.82 kcal/mol and the 2,6-F substitution pattern giving the highest binding energy of -1.67 kcal/mol. Not surprisingly the 2,6-difluoro-SBB configuration produces the least stable complexes and the least stable fluorination pattern at this geometry is penta-F with a binding energy of -0.49 kcal/mol. For three of the four configurations, 2-fluoro-SBB, 2,3difluoro-SBB, and 2,3,4,5,6-pentafluoro-SBB, the penta-F substitution pattern gives the most favorable interactions. The lowest binding energy for the 2,6-difluoro-SBB geometry was -1.00 kcal/mol, which was obtained with the 2,3-F fluorination pattern. It should also be noted that at this geometry the penta-F fluorination pattern yields a particularly weak interaction (-0.49 kcal/mol), which is most likely due to an increased steric
Interactions between Phe-131 and Pro-202
Figure 11. Sum of Phe-131-SBB and Pro-202-SBB interaction energies in vacuum, water, and ether.
Figure 12. Pro-202 interactions (Phe-131 w Val) with SBB (blue), 2-fluoro-SBB (purple), 2,3-difluoro-SBB (orange), 2,6-difluoro-SBB (yellow), and 2,3,4,5,6-pentafluoro-SBB (green).
hindrance caused by the close contact between a pyrrolidine hydrogen and a phenyl fluorine (para to the SBB methyl group). iii. OVerall Binding of the SBB Aromatic Group with HCAII Phe-131 and Pro-202 Residues. Figure 11 gives the sums of the interaction energies of the aromatic ring of SBB with the Phe-131 and Pro-202 residues for each of the SBB inhibitors considered in this work. Here it can be seen that the 2-fluoroSBB aromatic group is predicted to interact most favorably with the HCAII enzyme. Not surprisingly, 2,6-difluoro-SBB binds very weakly compared to the other three inhibitors. In each medium considered here the predicted binding energies of 2,3difluoro-SBB and 2,3,4,5,6-pentafluoro-SBB are slightly higher (less stable) than those of 2-fluoro-SBB. It should be noted that these data are not in complete agreement with experimental evidence. For instance, experimental data show that the 2,6-difluoro-SBB (experimental binding energy 12.30 kcal/mol) inhibitor is bound more strongly to HCAII than 2,3,4,5,6-pentafluoro-SBB (12.00 kcal/mol). This is not necessarily a surprising result because the geometries derived from the crystal structures of these protein-ligand complexes represent only one of many conformations available to these structures. It should also be pointed out that the model systems employed here are designed to study the direct interaction of the SBB aromatic ring with the Phe-131 and Pro202 side chains, and the SBB inhibitor is quite large, and exhibits several contacts with the HCAII enzyme, any one of which may be strengthened or weakened through changes in the relative geometries of the protein and ligand. In particular, in an earlier study we highlighted the role of the interaction of the zincbinding group (sulfonamide) with its surroundings and how the fluorination pattern might affect these interactions in subtle ways.8 3.2. Interactions of Mutant (Phe-131 w Val) HCAII with SBB Inhibitors. Figure 12 shows the interactions of Pro-202, from the mutated (Phe-13 w Val) form of the HCAII enzyme, with the variously fluorinated phenyl rings of the SBB inhibitor
J. Phys. Chem. B, Vol. 111, No. 20, 2007 5705
Figure 13. Pro-202-SBB interactions in vacuum, water, and ether for mutated HCAII (Phe-131 w Val-131).
(one of these is unfluorinated). When compared to wild type SBB-Pro-202 complexes, there is much less variation in the geometrical configurations of the SBB phenyl rings with respect to the pyrrolidine ring of Pro-202. As in the case of the interaction of 2-fluoro-SBB with wild type Pro-202, all of these SBB-Pro-202 complexes assume a configuration that is similar to an edge-to-face aromatic interaction. In Table 2 it can be seen that there is a relatively small variation in the intercentroid separations of these five SBB-Pro-202 complexes, with the shortest distance being 4.82 Å (SBB) and the largest separation being 5.14 Å (2,3,4,5,6-pentafluoro-SBB). The interaction energies of each of the SBB-Pro-202 complexes in vacuum, water, and ether are given in Figure 13 (for Phe-13 w Val). One of the most salient aspects of these data is that the SBB-Pro-202 interactions are significantly stronger for mutated HCAII than for the wild type enzyme. This can most likely be attributed to the fact that, because of the absence of Phe-131, the SBB phenyl group has more freedom to orient itself in a more favorable position in relation to Pro202. Not surprisingly there is relatively little variation in the binding energies of these complexes; this seems reasonable because the geometries of all five SBB-Pro-202 pairs are so similar. In vacuum the (nonfluorinated) SBB inhibitor forms the most stable complex with Pro-202 (-3.23 kcal/mol), while the 2,3-difluoro-SBB-Pro-202 pair has the strongest interaction in both water and ether (water: -2.02 kcal/mol; ether: -2.08 kcal/mol). The weakest interactions, in each of the media considered here, occur for the 2,6-difluoro-SBB-Pro-202 complexes. For all of the SBB-Pro-202 complexes the introduction of solvent results in a destabilization of the interaction between the SBB phenyl ring and the pyrrolidine of Pro-202. The interaction between the SBB, 2-fluoro-SBB, and 2,3-difluoroSBB inhibitors and Pro-202 is destabilized to a greater extent in water than in ether while the opposite is true for the complexes involving the 2,6-difluoro-SBB and 2,3,4,5,6-pentafluoro-SBB inhibitors. Overall the largest destabilization of 1.33 kcal/mol (from -3.23 kcal/mol to 1.90 kcal/mol) occurs for the SBB-Pro-202 pair upon introduction of water. The smallest increase in the binding energy upon solvation occurs for the 2,3-difluoro-SBB-Pro-202 complex with a value of 0.49 kcal/mol (from -1.91 kcal/mol to -1.42 kcal/mol) Figure 14 gives the SBB-Pro-202 (Phe-131 w Val) interaction energies as a function of geometric parameters and fluorination patterns. Generally speaking, the 2,3-difluoro-SBBPro-202 geometry is the most stable among the configurations considered here, although the SBB-Pro-202 geometry yields
5706 J. Phys. Chem. B, Vol. 111, No. 20, 2007
Riley et al.
Figure 14. Pro-202-SBB interactions for mutated HCAII (Phe-131 w Val-131). Histogram columns refer to geometries and colors refer to fluorine substitution pattern.
TABLE 3: Hartree-Fock and MP2 Interaction Energies for the SBB-Pro-202 Complexes
SBB HF 1.59 MP2 -3.23
2,3,4,5,62-fluoro- 2,3-difluoro- 2,6-difluoro- pentafluoroSBB SBB SBB SBB 1.31 -1.91
0.47 -2.70
3.14 -3.05
0.12 -2.84
comparable binding energies for most fluorine-substitution patterns. Overall the most stable complex corresponds to a 2,3SBB-Pro-202 geometry with a 0-F fluorination pattern, the binding energy for this configuration is -3.36 kcal/mol. As in the case of the wild type SBB-Pro-202 interactions, the 2,6difluoro-SBB geometry yields the least stable complexes, and at this geometry the 2,6-F fluorination pattern produces the smallest binding energy of -1.91 kcal/mol. Rather surprisingly the 0-F and penta-F fluorination patterns produce the two lowest interaction energies for all of the geometry configurations considered in this work with the exception of the 2-fluoro-SBBPro-202 geometry. 3.3. Nature of the Interaction between the Aromatic Group of the SBB Inhibitor and the Pyrrolidine Ring of Pro202. To establish a qualitative idea of the nature of the interaction between the phenyl ring of the SBB inhibitors and the pyrrolidine ring of Pro-202 we have carried out HartreeFock calculations on the SBB-Pro-202 (Phe-131 w Val) complexes and compared the results of these calculations to those of the MP2 method. The Hartree-Fock method, because it does not contain terms that describe electron correlation, describes dispersion interactions very poorly. This method does, however, describe noncovalent interactions involving charged pairs fairly well. If the Hartree-Fock method predicts a SBBPro-202 interaction energy that is similar to that of the MP2 method then it would follow that this noncovalent interaction is largely due to electrostatic forces, whereas an HF interaction energy much higher than that of MP2 would indicate an interaction largely attributable to dispersion forces. Table 3 gives the Hartree-Fock and MP2 interaction energies for the SBBPro-202 (Phe-131 w Val) complexes. In this table it can be seen that all of the Hartree-Fock interaction energies are positive indicating that dispersion forces are likely to be dominant in the interactions between phenyl rings and pyrrolidine rings. 3.4. Implications of Pyrrolidine-Phenol Interactions in Protein Structure. The interaction that occurs between phenyl rings and pyrrolidine rings may also have important implications in protein structure as the interactions between phenylalanine and proline residues may play a role in the stabilization of
proteins. To get an idea of how common phenylalanine-proline interactions are we analyzed 92 protein structures in the protein data bank with resolutions of 1.0 Å or less. In this search it was found that there are 77 unique phenylalanine-proline interactions with intercentroid distances of 6.0 Å or less; among these there are 24 phenylalanine-proline pairs that exhibit approximate edge-to-face interactions similar to those of the HCAII-SBB complexes. For the purpose of comparison we have also conducted a search for phenylalanine-phenylalanine interactions for the same set of protein structures, and it is widely believed that these interactions play an important role in protein stabilization.12,14,15,17 Here it was found that there are 195 phenylalanine-phenylalanine interactions with intercentroid distances of 6.0 Å or less. Given that the interaction strength for a phenyl-pyrrolidine complex is similar to that of a phenylphenyl complex and that preliminary data indicate that the occurrence of proline-phenylalanine pairs in proteins is substantial (about 40% as common as phenylalanine-phenylalanine pairs), it seems that these interresidue interactions are very likely to have a non-negligible influence on protein stability. 4. Conclusions In this work we have shown that there are significant attractive interactions between the terminal aromatic SBB group and both the Phe-131 and Pro-202 residues of Human Carbonic Anhydrase II. The strengths of these interactions exhibit a pronounced dependence on the relative geometry between the SBB aromatic ring and the side chains of the residues and, to a much lesser extent, on the pattern of fluorine substituents on the SBB aromatic ring. The introduction of both water and ether solvents destabilizes both the SBB-Phe-131 and SBB-Pro202 interactions significantly. One of the key aspects of these interactions that is brought to light herein is the fact that the binding energies for the SBBPro-202 complexes are actually lower in some instances than those of the SBB-Phe-131 complexes. The lowest interaction energy for a SBB-Pro-202 complex is -3.36 kcal/mol (vacuum), which is for a geometry corresponding to the 2,3-difluoro-SBB inhibitor (Phe-131 w Val) along with a 0-F fluorine substitution pattern. The most stable SBB-Phe-131 complex has a binding energy of -3.10 kcal/mol (vacuum), for the 2,3,4,5,6-pentafluoro-SBB geometry with the 2-F fluorine-substitution pattern. As stated in the introduction, over the past several years there have been many studies focusing on the role of aromatic interactions in protein and protein-ligand structure;8-17 to our knowledge, however, there have been no studies carried out seeking to understand the role of interaction involving a pyrrolidine ring and an aromatic group. The results of this work indicate that these types of interactions are indeed important in the structure of biological molecules and that they may stabilize the structures of biomolecules, such as proteins, to at least the same extent as aromatic-aromatic interactions. Thus, in light of this it is clearly worthwhile considering other types of “underappreciated” noncovalent interactions that could play a role in stabilizing protein-protein or protein-ligand complexes. It may be found that these interactions play as significant a role as do the widely studied and relatively well-understood aromaticaromatic noncovalent interaction motif. Another key point about the data presented here is that, compared to geometrical parameters, the fluorine-substitution pattern has a very small effect on the interaction energies for both SBB-Phe-131 and SBB-Pro-202 complexes. This is not to say that fluorination patterns are not important in determining the binding energy of these types of protein-ligand interactions,
Interactions between Phe-131 and Pro-202 as the number of fluorine substituents and the specific positions of these substituents on an aromatic ring are likely to play an important role in determining the relative geometries of these aromatic groups. This is evidenced by the fact that each of the SBB aromatic groups assumes a different geometry, depending on its fluorination state, in the crystal structures used in this study. Acknowledgment. We thank the NIH (GM066859) for providing the funding for this research. References and Notes (1) Vondrasek, J.; Bendova, L.; Klusak, V.; Hobza, P. J. Am. Chem. Soc. 2005, 127, 8232. (2) Vondrasek, J.; Bendova, L.; Klusak, V.; Hobza, P. J. Am. Chem. Soc. 2005, 127, 2615. (3) Otyepka, M.; Sklenovsky, P.; Horinek, D.; Kubar, T.; Hobza, P. J. Phys. Chem. B. 2006, 110, 4423. (4) Jurecka, P.; Sponer, J.; Cerny, J.; Hobza, P. Phys. Chem. Chem. Phys. 2006, 8, 1985. (5) Fukuzawa, K.; Komeiji, Y.; Mochizuki, Y.; Kato, A.; Nakano, T.; Tanaka, S. J. Comput. Chem. 2006, 27, 948. (6) Dobes, P.; Otyepka, M.; Strnad, M.; Hobza, P. Chem.-Eur. J. 2006, 12, 4297. (7) Riley, K. E.; Merz, K. M. J. Phys. Chem. B 2006, 110, 15650. (8) Raha, K.; van der Vaart, A. J.; Riley, K. E.; Peters, M. B.; Westerhofft, L. M.; Kim, H.; Merz, K. M. J. Am. Chem. Soc. 2005, 127, 6583. (9) Riley, K. E.; Merz, K. M. J. Phys. Chem. B 2005, 109, 17752. (10) Kim, C.-Y.; Chang, J. S.; Doyon, J. B.; Baird, T. T., Jr.; Fierke, C. A.; Jain, A.; Christianson, D. W. J. Am. Chem. Soc. 2000, 122, 12125. (11) Meyer, E. A.; Castellano, R. K.; Diedrich, F. Angew. Chem. 2003, 42, 1210. (12) Aravinda, S.; Shamala, N.; Das, C.; Sriranjini, A.; Karle, I. L.; Balaram, P. J. Am. Chem. Soc. 2003, 125, 5308.
J. Phys. Chem. B, Vol. 111, No. 20, 2007 5707 (13) Chelli, R.; Gervasio, F. L.; Procacci, P.; Schettino, V. J. Am. Chem. Soc. 2002, 124, 6133. (14) Burley, S. K.; Petsko, G. A. Sience 1985, 229, 23. (15) Burley, S. K.; Petsko, G. A. FEBS Lett. 1986, 203, 139. (16) Befort, K.; Tabbara, L.; Kling, D.; Maigret, B.; Kieffer, B. L. J. Biol. Chem. 1996, 271, 10161. (17) Waters, M. L. Biopolymers 2004, 76, 435. (18) Chandra, P. P.; Jain, A.; Sapse, A. M. J. Mol. Model. 2004, 10, 1. (19) Doyon, J. B.; Hansen, E. A. M.; Kim, C. Y.; Chang, J. S.; Christianson, D. W.; Madder, R. D.; Voet, J. G.; Baird, T. A.; Fierke, C. A.; Jain, A. Org. Lett. 2000, 2, 2557. (20) Doyon, J. B.; Jain, A. Org. Lett. 1999, 1, 183. (21) Surgrue, M. F. Prog. Ritinal Eye Res. 2000, 19, 87. (22) Abeles, R. H.; Alston, T. A. J. Biol. Chem. 1990, 265, 16705. (23) Williams, J. H. Acc. Chem. Res. 1993, 26, 593. (24) Sinnokrot, M. O.; Sherrill, C. D. J. Phys. Chem. A 2004, 108, 10200. (25) Dixon, S. L.; Merz, K. M. J. Chem. Phys. 1996, 104, 6643. (26) Importa, R.; Barone, V.; Kudin, K. N.; Scuseria, G. E. J. Am. Chem. Soc. 2001, 123, 3311. (27) Li, X.; Millam, J. M.; Scuseria, G. E.; Frisch, M. J.; Schlegel, H. B. J. Chem. Phys. 2003, 119, 7651. (28) Van Alsenoy, C.; Yu, C.-H.; Peeters, A.; Martin, J. M. L.; Schafer, L. J. Phys. Chem. A 1998, 102, 2246. (29) Møller, C.; Plesset, M. S. Phys. ReV. 1934, 46, 618. (30) Dunning, T. H. J. Chem. Phys. 1989, 90, 1007. (31) Boys, S. F.; Bernardi, F. Mol. Phys. 1970, 19. (32) Mennucci, B.; Tomasi, J. J. Chem. Phys. 1997, 106, 5151. (33) Cramer, C. J.; Truhlar, D. G. Chem. ReV. 1999, 99, 2161. (34) Tomasi, J.; Cammi, R.; Mennucci, B.; Cappelli, C.; Corni, S. Phys. Chem. Chem. Phys. 2002, 4, 5697. (35) Tomasi, J.; Persico, M. Chem. ReV. 1994, 94, 2027. (36) Dwyer, J. J.; Gittis, A. G.; Karp, D. A.; Lattman, E. E.; Spencer, D. S.; Stites, W. E.; Garcia-Moreno, B. Biophys. J. 2000, 79, 1610. (37) Scheiner, S.; Kar, T. J. Phys. Chem. B 2005, 109, 3681. (38) Simonson, T.; Perahia, D. Proc. Natl. Acad. Sci. U.S.A. 1995, 92, 1082.
