Cundari Th.R. -- Computational Organometallic Chemistry-0824704789
.pdf178 |
Maseras |
the single interatomic repulsions that contribute to it. Because of the difficulty of dealing with 6693 such interactions present in this particular complex, a filter was applied to choose only those differences larger than 0.05 kcal/mol. This reduced the number of interactions to 58, which were further grouped by the ligand to which each of the two atoms belonged. The result provided the contribution of each pair of ligands to the VdW repulsion in this complex. The leading terms happened to be those associated with the P(7)–Si(9) and P(8)–Si(10) ligand pairs (Fig. 7), which are related by symmetry, with values of 2.28 and 2.30 kcal/ mol, respectively. The next largest repulsions in size correspond to the P(7)–P(7) and P(8)–P(8) pairs, associated with intraligand reorganization, with significantly smaller values of 0.61 kcal/mol each. A similar analysis of the ‘‘bending’’ term showed that the differences were also concentrated within the P(7) and P(8) ligands. This is therefore an example of how QM/MM methods can allow the quantification of intraligand steric effects and can allow their separation into specific ligand-to-ligand contributions in a given organometallic complex.
A second application of QM/MM methods to the quantification of steric
effects is provided by the IMOMM(Becke3LYP:MM3) study on the origin of enantioselectivity in the dihydroxylation of styrene by the (DHQD)2PYDZ OsO4
complex (21). The size and complexity of the catalyst, which can be seen in Figure 8, preclude the possibility of accurate pure QM calculations on the problem, because the selectivity is decided precisely by the bulky substituents in the NR3 cinchona group. The selectivity is defined by the initial approach of the substrate to the catalyst to form an osmate ester intermediate, and consequently a number of possible paths were analyzed. In particular, there are 12 such paths, defined by the three possible regions (A, B, and C) of approach of the substrate
FIGURE 8 Schematic presentation of the (DHQD)2PYDZ OsO4 catalyst, with indication of the most significant areas of the cinchona ligand.
Quantifying Steric Effects via QM/MM Methods |
179 |
to the catalyst and the four possible orientations (I, II, III, and IV) of the phenyl ring of the substrate within each region, as shown in Figure 9. Each of these 12 possible paths was theoretically characterized through the location of the corresponding transition state, with its associated energy. The lowest-energy saddle point, therefore the most likely transition state for the reaction, was B-I, followed at close distance by B-III and B-IV, 0.1 and 2.7 kcal/mol above. The next saddle point in energy, A-IV, came 4.7 kcal/mol above B-I. The preference for paths B-I and B-III has an immediate consequence on the enantioselectivity of the reaction, because both lead to the R product. The minor S product must be obtained from a path through the B-IV saddle point. This is in excellent agreement with experimental data (51), which give a high enantiomeric excess of R product.
This agreement would in principle also be obtained by costly pure QM calculations. But the IMOMM method allows a further analysis of which regions of the catalyst affect the outcome of the reaction the most. In order to do that, the interaction energy between catalyst and styrene in the different saddle points B-I, B-III, and B-IV must be decomposed and compared. The first part of this decomposition consists of separating the interaction energy between substrate and catalyst into binding and distortion contributions. To do this, the process of formation of each saddle point from the separate reactants is divided into two imaginary steps: 1) a first step where the substrate and the catalyst are kept at infinite distance but distorted from their respective equilibrium geometries to the one they have in the saddle point (distortion energy), 2) and a second step where they are put together to yield the saddle point structure (binding energy). This division shows that the similar total interaction energies are reached through the addition of terms of different magnitude. For instance, a total interaction energy
FIGURE 9 Definition of the criteria for labeling the 12 possible reaction paths in the reaction of H2CCCHPh with (DHQD)2PYDZ OsO4.
180 |
Maseras |
of 3.3 kcal/mol for B-I is reached by adding a distortion of 15.4 kcal/mol and a binding of 18.7 kcal/mol; the interaction energy of 3.2 kcal/mol in B-III is reached from the addition of 13.0 kcal/mol (distortion) and 16.2 kcal/mol (binding). The fact that the total interaction energies are negative, with the transition state having an energy below the reactants, is explained through the presence of an intermediate in the reaction profile (13).
The analysis can be further refined to see which are the specific parts of the catalyst contributing to this binding energy. This can be done because the IMOMM partition in this particular system leaves most of the binding energy between catalyst and substrate in the MM part. The MM binding energies oscillate widely between 6.6 and 11.43 kcal/mol, while the QM change is comparatively much smaller, with changes between 6.9 and 7.6 kcal/mol. The analysis of the MM contribution to binding energies is straightforward, because it is composed mostly of ‘‘van der Waals’’ contributions, much in the same way as in the previous example. This VdW contribution comes again from a summation of atom-pair interactions. The number of atom pairs is very large, but a grouping of them in the regions depicted in Fig. 8 is instructive. In the case of B-I, 53% of the interaction happens with quinoline A, 21% with quinoline B, 12% with PYDZ, and 15% with the rest of the system. The respective values for B-III are 42%, 12%, 22%, 24%; B-IV gives 57%, 15%, 11%, 17%. The large role of the two quinoline and the pyridazine substituents in the selectivity of the reaction is clear from these results, which leave at most 24% of the interaction to the rest of atoms in the catalyst. The main role furthermore is played by one single group, quinoline A. These results indicate the most sensitive point of the catalyst for alterations in its selectivity.
A final comment on this example of catalyzed olefin dihydroxylation concerns the fact that ‘‘steric’’ effects appear to be stabilizing. This is no surprise if one realizes that they correspond to the parallel (or perpendicular) placement of aromatic rings, which is expected to yield a stabilizing interaction. The validity of the labeling of these interactions as steric effects is arguable, but it goes back to the discussion on the nature of steric effects at the beginning of this section. At any rate, these interactions are properly reproduced by the MM calculations and, therefore, correspond to steric effects according to our criteria defined earlier.
Another aspect concerning both examples presented in this subsection is the importance that van der Waals interactions appear to have in this topic. The dominance of this term, which can be surprising, very likely is affected by the choice of the MM3 force field. Other force fields grant a lesser importance to van der Waals terms and give more weight to electrostatic contributions, for instance. If such other force fields had been applied, the decomposition would likely be substantially different, and other terms should be more important in defining the difference. In any case, the total difference would have to be similar, as far as the different force fields would properly describe the same chemical
Quantifying Steric Effects via QM/MM Methods |
181 |
reality. So this result is merely used in the sense that the more significant MM contributions correspond to what MM3 calls van der Waals interactions, without entering in the real chemical meaning of such terms.
The two examples described in this subsection show how the use of hybrid QM/MM methods gives access to analysis tools that are absent in pure QM calculations, even if those are more accurate. They also show how the particular chemical problem under study can require slightly different handling of the QM/MM results.
6. CONCLUSION
This chapter has presented a summary of specific aspects of the application of QM/MM methods to organometallic chemistry that should be of help to the researcher interested in their use. Emphasis has been put on practical examples of each feature. Although most of the examples correspond to calculations with one particular method, IMOMM, the general features should have application to the new developments that will surely appear in the near future.
Hybrid QM/MM methods should certainly be expected to experience a fast expansion, in large part because any improvement in QM or MM techniques will have a direct effect on them. The natural progress in both methodology and computer power, allowing calculations on larger and larger QM systems, will probably bring a time when systems that now require QM/MM methods will be studied with more accurate full QM methods. There will, however, always be a place for QM/MM methods, as a perpetual attempt to bridge the still enormous chasm between what can be computed and what actually exists.
ACKNOWLEDGMENTS
Financial support is acknowledged from the Spanish DGES through Project No. PB95-0639-CO2-01 and from the Catalan CIRIT through grant No. 1997SGR00411. Thanks are due to Prof. Morokuma, Emory, for introducing me to this subject, and for continued discussions on methodological improvements. Prof. Lledo´s, Bellaterra, and Prof. Eisenstein, Montpellier, are also thanked for the steady flow of interesting chemical problems through the years, the solution to which has been the bulk of our application of QM/MM methods. Thanks finally to users of the mmabin program, Toshiaki Matsubara, Dima Khoroshun, Thom Vreven, Stephan Irle, Eric Clot, Hele`ne Ge´rard, John McGrady, Gregori Ujaque, Guada Barea, Jaume Toma`s, Jean-Didier Mare´chal, Lourdes Cucurull-Sa´nchez, Jordi Carbo´, Nicole Do¨lker, Isabelle Demachy, Nu´ria Lo´pez, Simona Fantacci, Jordi Va´zquez, Rainer Remenyi, Olivier Maresca, Fahmi Himo, and Antonio Morreale, because discussion with them has shaped the ideas that constitute this chapter.
182 |
Maseras |
REFERENCES
1.F Maseras. Topics Organomet Chem 4:165, 1999.
2.F Maseras, K Morokuma. J Comput Chem 16:1170, 1995.
3.A Warshel, M Levitt. J Mol Biol 103:227, 1976.
4.UC Singh, PA Kollman. J Comput Chem 7:718, 1986.
5.MH Field, PA Bash, M Karplus. J Comput Chem 11:700, 1990.
6.J Gao. Acc Chem Res 29:298, 1996.
7.J Gao. Rev Comput Chem 7:119, 1996.
8.D Bakowies, W Thiel. J Phys Chem 100:10580, 1996.
9.G Ujaque, F Maseras, A Lledo´s. Theor Chim Acta 94:67, 1996.
10.G Barea, F Maseras, Y Jean, A Lledo´s. Inorg Chem 35:6401, 1996.
11.M Ogasawara, F Maseras, N Gallego-Planas, K Kawamura, K Ito, K Toyota, WE Streib, S Komiya, O Eisenstein, KG Caulton. Organometallics 16:1979, 1997.
12.G Ujaque, F Maseras, O Eisenstein. Theor Chem Acc 96:146, 1997.
13.G Ujaque, F Maseras, A Lledo´s. J Org Chem 62:7892, 1997.
14.F Maseras, O Eisenstein. New J Chem 22:5, 1998.
15.G Ujaque, AC Cooper, F Maseras, O Eisenstein, KG Caulton. J Am Chem Soc 120: 361, 1998.
16.F Maseras. New J Chem 22:327, 1998.
17.G Barea, A Lledo´s, F Maseras, Y Jean. Inorg Chem 37:3321, 1998.
18.J Jaffart, R Mathieu, M Etienne, JE McGrady, O Eisenstein, F Maseras. Chem Commun: 2011, 1998.
19.G Ujaque, F Maseras, O Eisenstein, L Liable-Sands, AL Rheingold, W Yao, RH Crabtree. New J Chem 22:1493, 1998.
20.AC Cooper, E Clot, JC Huffman, WE Streib, F Maseras, O Eisenstein, KG Caulton. J Am Chem Soc 121:97, 1999.
21.G Ujaque, F Maseras, A Lledo´s. J Am Chem Soc 121:1317, 1999.
22.JD Mare´chal, G Barea, F Maseras, A Lledo´s, L Mouawad, D Pe´rahia. J Comput Chem 21:282, 2000.
23.T Matsubara, F Maseras, N Koga, K Morokuma J Phys Chem 100:2573, 1996.
24.Y Wakatsuki, N Koga, H Werner, K Morokuma. J Am Chem Soc 119:360, 1997.
25.RDJ Froese, DG Musaev, K Morokuma. J Am Chem Soc 120:1581, 1998.
26.DG Musaev, RDJ Froese, K Morokuma. Organometallics 17:1850, 1998.
27.L Deng, TK Woo, L Cavallo, PM Margl, T Ziegler J Am Chem Soc 119:6177, 1997.
28.TK Woo, PM Margl, PE Blo¨chl, T Ziegler J Phys Chem B 101:7878, 1997.
29.N Lopez, G Pacchioni, F Maseras, F Illas. Chem Phys Letters 294:611, 1998.
30.RDJ Froese, JM Coxon, SC West, K Morokuma. J Org Chem 62:6991, 1997.
31.RDJ Froese, K Morokuma. Chem Phys Letters 305:419, 1999.
32.S Humbel, S Sieber, K Morokuma. J Chem Phys 105:1959, 1996.
33.M Svensson, S Humbel, RDJ Froese, T Matsubara, S Sieber, K Morokuma. J Phys Chem 100:19357, 1996.
34.S Dapprich, I Koma´romi, KS Byun, K Morokuma, MJ Frisch. J Mol Struct (Theochem) 461:21, 1999.
35.TK Woo, L Cavallo, T Ziegler. Theor Chem Acc 100:307, 1998.
36.JR Shoemaker, LW Burggraf, MS Gordon. J Phys Chem A 103:3245, 1999.
Quantifying Steric Effects via QM/MM Methods |
183 |
37.U Eichler, CM Ko¨lmel, J Sauer. J Comput Chem 18:463, 1996.
38.RDJ Froese, K Morokuma. In: PvR Schleyer, ed. Encyclopedia of Computational Chemistry. Vol 4. New York: Wiley, 1998, p 1257.
39.J Gao. In: PvR Schleyer, ed. Encyclopedia of Computational Chemistry. Vol 4. New York: Wiley, 1998, p 1257.
40.KM Merz Jr, RV Stanton. In: PvR Schleyer, ed. Encyclopedia of Computational Chemistry. Vol 4. New York: Wiley, 1998, p 2330.
41.J Tomasi, CS Pomelli. In: PvR Schleyer, ed. Encyclopedia of Computational Chemistry. Vol 4. New York: Wiley, 1998, p 2343.
42.I Tun˜on, MTC Martins-Costa, C Millot, MF Ruiz-Lo´pez, J-L Rivail. J Comput Chem 17:19, 1996.
43.M Strnad, MTC Martins-Costa, C Millot, I Tun˜on, MF Ruiz-Lo´pez, J-L Rivail.
JChem Phys 106:3643, 1997.
44.M Brookhart, MLH Green, L Wong. Prog Inorg Chem 36:1, 1988.
45.L Stryer. Biochemistry. New York: Freeman, 1995.
46.MD Beachy, D Chasman, RB Murphy, TA Halgren, RA Friesner. J Am Chem Soc 119:5908, 1997.
47.NL Allinger, YH Yuh, J-H Lii. J Am Chem Soc 111:8551, 1989.
48.A Albinati, VI Bakhmutov, KG Caulton, E Clot, J Eckert, O Eisenstein, DG Gusev,
VVGrushin, BE Hauger, WT Klooster, TF Koetzle, RK McMullan, TJ O’Loughlin,
MPe´lissier, JS Ricci, MP Sigalas, AB Vymenits. J Am Chem Soc 115:7300, 1993.
49.M Ogasawara, F Maseras, N Gallego-Planas, WE Streib, O Eisenstein, KG Caulton. Inorg Chem 35:7468, 1996.
50.JAK Howard, PA Keller, T Vogt, AL Taylor, ND Dix, JL Spencer. Acta Crystallogr
B438:4338, 1992.
51.EJ Corey, MC Noe. J Am Chem Soc 118:319, 1996.
7
HIV Integrase Inhibitor Interactions with Active-Site Metal Ions: Fact or Fiction?
Abby L. Parrill, Gigi B. Ray,
Mohsen Abu-Khudeir, Amy Hirsh,
and Angela Jolly
The University of Memphis, Memphis, Tennessee
1. INTRODUCTION
The human immunodeficiency virus (HIV) is a retrovirus that is responsible for acquired immunodeficiency syndrome (AIDS) (1). Human immunodeficiency virus requires the activity of three enzymes during its life cycle (2). Pharmaceutical agents targeting two of these enzymes, reverse transcriptase and protease, are in clinical use. These pharmaceutical agents are not ideal, however, due to the rise of resistant viral strains after treatment is initiated (3,4). The third enzyme, integrase (IN), is the least explored enzyme target for HIV treatment (5), with one agent undergoing clinical trials, although evidence indicates that IN is not its primary target in vivo (6). Integrase is responsible for two essential (7,8) catalytic activities that result in incorporation of viral DNA into host DNA. These activities are shown schematically in Figure 1. The first is 3′ processing, in which the 3′ ends of viral DNA are recessed by removal of two bases. The second is strand transfer, in which the viral DNA is joined to the 5′ phosphate of a staggered cut in the host DNA. This activity places the IN enzyme into the polynucleotidyl-
185
186 |
Parrill et al. |
FIGURE 1 Schematic of reactions catalyzed by the integrase enzyme. (a) Processing of the 3′ ends of viral DNA to generate staggered ends. (b) Insertion of staggered viral DNA, shown as thick lines, into double-stranded host DNA, shown as thin lines.
transferase family, which also includes RNase H, the Mu transpose, and Ruv C (9,10). The polynucleotidyltransferases characteristically contain an organometallic active site, with a divalent metal cation playing a critical role in the catalysis of the DNA or RNA strand transfer reactions.
Pharmaceutical development of IN-targeting agents lags behind the development of agents targeting protease and reverse transcriptase, due to incomplete structural data on IN. Complexes of the reverse transcriptase and the protease with their inhibitors as characterized by X-ray crystallography (2) have been available for many years, thus providing the basis for rational design of additional therapeutic agents. The IN structure, however, has been characterized less thoroughly than other HIV enzymes, with crystallographic characterization of the catalytic core domain in the absence of inhibitors (9,11,12) and NMR characterization of the N-terminal and C-terminal domains (13–15). Only very recently has the first crystal structure of the HIV IN catalytic core in the presence of an
HIV IN Inhibitor Interactions with Active-Site Metal Ions |
187 |
inhibitor become available (16). Structural information for the intact enzyme and for regions connecting the three domains is not available.
Strand transfer activity of the IN enzyme is consistent with the polynucleotidyltransferase family requirement for a divalent metal ion cofactor. Manganese or magnesium can serve as this cofactor (1,10). Studies on the impact of metal ions on the binding of domain-specific antibodies have shown that the metal ion induces a conformational change (17). This work indicates that the conformational change involves a reorganization of the catalytic core and C-terminal domains based on two separate experiments. Interactions with antibodies specific for the apoenzyme are lost upon metal ion binding, and an increased resistance to proteolysis is observed. These changes were independent of the presence of the N-terminal domain.
The active-site metal ion plays an accepted role in the catalytic function of IN, although its role in inhibition is poorly understood. The role of the metal ion in inhibition has been a topic of speculation since the first inhibitors began to appear in the literature in 1993. Early work identified some DNA-binding and DNA-intercalating molecules as IN inhibitors (18). These early inhibitors include doxorubicin (Fig. 2, 1), caffeic acid phenethyl ester (CAPE, Fig. 2, 2), and quercetin (Fig. 2, 3). This work established, however, that there was no clear relationship between DNA-binding or DNA-intercalating activities and inhibition. It was noted that active inhibitors contained polyhydroxylated aromatic regions that
FIGURE 2 Structures of HIV IN inhibitor classes.