Overview of Rational Drug Design - ACS Symposium Series (ACS

Chapter 1

Overview of Rational Drug Design 1

M . R a m i Reddy and A b b y L. P a r r i l l 1

2

Metabasis Therapeutics, 9360 Towne Centre Drive, San Diego, CA 92121 Department of Chemistry, University of Memphis, Memphis, TN 38152

Downloaded by MAHIDOL UNIV on February 18, 2015 | http://pubs.acs.org Publication Date: July 7, 1999 | doi: 10.1021/bk-1999-0719.ch001

2

Traditional Drug Design. Drug discovery programs i n the pharmaceutical industry prior to the 1960's were based entirely on screening thousands o f natural and synthetic compounds for activity. Once a potential drug compound was selected by this process, medicinal chemists would then synthesize hundreds o f related compounds to develop the safest, most effective drug for patients use. However, the costs and risks associated with this process have become enormous; the cost o f completing the research and development process for a single new drug has more than doubled i n the last decade. Various sources estimate this cost to to be anywhere from $200-$500 million. Each year researchers test hundreds o f thousands o f chemical compounds, yet i n the United States only about 25 new drugs are introduced. Even worldwide the introduction o f new drugs only reaches 40-45 per year. M a n y o f these new drugs are only "me-too compounds", as the various companies attempt to apply their patented "molecular manipulations" to other companies' top selling drugs. A major limitation o f the drug screening strategy is that it does not reveal w h y a compound is active or inactive, or how it might be improved. It also provides no assurances that an active compound is specific for a given human target protein. The lack o f such specificity can be a major source o f undesirable side effects which can halt the clinical development o f a drug. Drug screening is essentially a blind process, indicating the reason for the need to test approximately 20,000 compounds i n order to find one that becomes a marketable drug. Drug screening is often followed by structural optimization o f lead compounds i n order to improve potency and other properties, but deciding when to move from screening to synthesis is a problem. Although screening has produced the vast majority o f existing drugs, it has not proven to be a wholly satisfactory strategy. There are many important therapeutic needs for which screening has failed. 1

Notable work began to appear i n 1962 which led to drastic changes i n the process used to optimize chemical structures

for medicinal

2,3

purposes.

This

© 1999 American Chemical Society

In Rational Drug Design; Parrill, A., et al.; ACS Symposium Series; American Chemical Society: Washington, DC, 1999.

work

1

2

established the foundation for the multi-parameter Q S A R methods i n common use today. Subsequent publications outlined how the Hansch approach could be applied to drug design.4-6


Challenges of Drug Design Researchers have worked for a long time to overcome the limitations o f screening by designing molecules to perform specific therapeutic tasks. This vision o f rational drug design was made more plausible by improvements i n our understanding o f the similarities between the actions o f different biologically active compounds. Almost all drug molecules achieve a biological response through interaction with a target or receptor biomolecule. Descriptions o f our earliest understanding o f this interaction compared the drug molecule entry into a crevice o f the target protein to a key i n a lock, thus inhibiting the protein's normal biological function. Current descriptions liken drug interactions with biomolecules to a handshake, where both the ligand? and the protein adjust somewhat to accommodate the other. The general drug - target scheme suggests that structure-based rational drug design can be accomplished by three basic tasks. First, the appropriate protein target for a given therapeutic need must be identified. Second, the distinguishing structure o f the target protein must be determined. Finally, the structure o f a drug must be designed to interact with the target protein. However, a number o f technical barriers have hindered work i n the area o f structure-based drug design. First, many important human diseases are not sufficiently well-understood at the molecular level to permit scientists to identify an appropriate drug target. Second, even when an appropriate target has been known, its molecular structure has generally not been known i n adequate detail for drug design. Finally, the design o f structures complementary to the target requires consideration o f both the three-dimensional as well as the functional aspects o f chemical structures. In cases where an appropriate biological target cannot be identified or characterized, rational drug design requires a different strategy. This alternate strategy makes use o f structural information about drugs that produce the same biological response at different doses. It is often reasonable to assume that such drugs interact with the same, albeit unknown, biological target. They must, therefore, have some common set o f structural features that are required i n order to evoke the aforementioned biological response. This common set o f structural features is the pharmacophore. This assumed similarity o f drugs with similar effect suggests an alternative set o f tasks that can accomplish rational drug design. First, the structural features important for biological activity must be determined. It is important that these features provide three-dimensional information either implicitly or explicitly. Second, optimal combinations for these features must be determined. Finally, the structure o f a drug must be designed which exhibits the optimal combination o f these features. Drug design efforts that seek to accomplish this alternative set o f tasks are classified as pharmacophore based approaches.^ Drug design efforts using pharmacophore based approaches have their own set o f challenges. First, determining important structural features when a variety o f chemical structures


3

demonstrate the same biological activity requires an understanding o f the structural correspondance. A n additional complication arises due to the fact that some drugs which elicit the same biological activity display multiple modes o f interaction with the target. Challenges for the structure design portion of the pharmacophore based approach are the same as the challenges during the corresponding activity when used in structure-based drug design approaches.


Rational Drug Design Since the early 1980's, advances i n molecular biology, protein crystallography, and computational chemistry have greatly aided Rational Drug Design ( R D D ) paradigms and the accuracy o f their binding affinity predictions.9-11 Figure 1 shows a flowchart that describes the different approaches that may be employed by drug discovery groups during R D D or ligand design. Further discussion o f R D D w i l l be organized into four main areas. T w o of these areas, pharmacophore based approaches and structure-based approaches depend on whether the three-dimensional structure o f the biological target is available. The other two areas, new lead generation and structure evaluation, w i l l be performed regardless o f whether the biological target structure is known.

Pharmacophore-Based Approaches. The path at the first decision point is determined by the availability o f the 3-dimensional structure o f the enzyme or complex. I f the structure o f the biological target is unknown, various methods that utilize active (and inactive) analogs can be used to develop a working model o f the requirements for biological activity, in other words, the pharmacophore. There are several evolving quantitative methods that utilize active compounds such as 2 D 1 2

1 5

1 6

17

QSAR, " 3 D - Q S A R and neural n e t w o r k s . Comparative Molecular Field Analysis ( C o M F A ) is a very widely used 3 D - Q S A R technique. 1°* C o M F A represents a significant achievement due to its ability to develop a three-dimensional quantitative model that relates steric and electrostatic fields to biological activity. A n initial problem with the method was the need to select both conformations and alignments o f the molecules to be modeled. Due to this problem, many initial uses o f the C o M F A method involved molecules with rigid ring systems. For example, A l l e n et. al. predicted the binding affinities o f six analogs o f beta-carbolines for the benzodiazepine receptor ( B z R ) prior to synthesis ^ using a previously published C o M F A model. 19 The standard error o f prediction for these six analogs is significantly lower than the standard error estimate o f the cross-validation runs on the training set, hence the predictions made using this model are much better than expected. Even now, nine years after the first description o f the C o M F A method, papers are appearing i n the literature that offer new solutions to the a l i g n m e n t ^ and conformer selection^! problems. In addition to such three-dimensional models, pharmacophore hypotheses may also be developed by more qualitative methods. 22 Using any o f these methods one could propose new analogs o f a lead compound based on the pharmacophore model.


In Rational Drug Design; Parrill, A., et al.; ACS Symposium Series; American Chemical Society: Washington, DC, 1999. Docking, FEP, Hydration Free Energy, I Regression Methods... I

\YES

Figure 1. A n overview of the many types of methods that provide an understanding of drug action and their integration in the process of rational drug design.

Synthesize/Test Best Candidates!

QSAR or 3D QSAR model,! Hydration Free Energy... |

NO

EVALUATE NEW STRUCTURES

Propose New Lead or Optimize Existing Lead

I

Characterize Active Site ^riaMgseo^^gr^^kj^grtiaL

Generate Working Model of Protein jj«

STRUCTURE-BASED APPROACHES

GENERATE NEW LEAD STRUCTURES

i

Generate Working Models of Ligandi

PHARMACOPHORE-BASED APPROACHES

YES



5

Structure-Based Approaches. The other branch at the first decision point is used when the 3-dimensional structure o f the enzyme or complex is known. The process typically begins by generating a working computational model from crystallographic data, but methods to develop models o f the binding site from active ligands are becoming more prevalent.23-26 Development o f the working model may include developing molecular mechanics force field parameters for non-standard residues consistent with the force field for standard residues, modeling any missing segments, assigning the protonation states o f histidines, and orienting carbonyl and amide groups o f asparagine and glutamine residues based upon neighboring donor and acceptor groups. Characterization o f the active site is then aided by a variety o f visualization tools. For example, hydrophobic and hydrophilic regions o f the active site are readily identified by calculating the electrostatic potential at different surface grid points, and hydrogen bond donor and acceptor groups can be highlighted i n the active site. The information gained by the characterization o f the active-site is very important for proposing new lead compounds or analogs o f a known leads. New Lead Generation. Generation o f new lead compounds can be accomplished using de novo design methods to design new structures27,28 y searching databases22,29-35 f known chemicals for particular structural features. De novo molecular design methods may design structures by sequentially adding or joining molecular fragments to a growing structure, 36-38 by adding functionality to an appropriately-sized molecular scaffold, or by evolving complete structures39-41 Some de novo design methods have concentrated on the design o f diverse molecular scaffolds 42 or on the development o f diverse substituents to place on a single scaffold.35 Database search methods have been developed that search based on separation o f molecular functionality by a particular number o f bonds or distance ranges. More chemically intuitive database search methods seek for chemicals with particular steric and electrostatic fields.33 o

r

D

0

A growing number o f drug leads are being generated by combinatorial methods in combination with high-throughput screening. Computational chemistry is currently being used to assist efforts in this area by ensuring that the library o f structures generated for use i n high-throughput screening assays incorporates a great deal o f molecular diversity.34,35,43-48 This ensures that a diverse set o f lead compounds can be found and optimized at much lower cost than i f the entire library o f possible structures were synthesized and tested. The diverse set o f leads that can be found by combinatorial chemistry can give important insight into the requirements for biological activity. This is particularly valuable for relatively new drug targets for which insufficient information is available for application o f the structure-based or pharmacophore based approaches. Structure Evaluation. With new drug leads proposed, rapid and accurate prediction of in vivo activities are needed in order to evaluate and thereby prioritize these structures prior to chemical synthesis. In reality, evaluation methodologies are limited to in vitro measurables such as binding affinity although an in vivo property



6

such as clinical effect would be ideal. The challenge within this limitation is to develop evaluation methods that rapidly and accurately predict absolute binding affinities o f the aforementioned large, diverse set o f potential ligands. Currently available evaluation methods can either provide qualitative rank ordering o f a large number o f molecules i n a relatively short period o f time^^ or generate quantitatively accurate predictions o f relative binding affinities for structurally related molecules using substantial computing power. 50,51 Consequently, biological activity evaluation techniques that increase speed without greatly compromising accuracy (or vice versa) are o f value to drug discovery programs. Methods o f ligand evaluation include graphical visualization o f the ligand i n the binding site,52 substitution o f parameters from the new ligand into S A R models, and calculation o f relative binding affinities.53,54 Usually about 50% o f proposed new leads or optimized analogs can be eliminated by evaluating their expected binding affinities based on docking, visualization, conformational analysis and desolvation costs. The remaining analogs w i l l be ranked for synthesis using one or all o f the following methods, depending on computational power, time and resources, namely; 1) Free Energy Perturbation (FEP) calculations, which give very accurate quantitative predictions, but are computationally very expensive,50,51 2) molecular mechanics calculations, which w i l l give only qualitative predictions, but these calculations are very fast, 10,49 3) regression methods55 that incorporate interaction variables (intra and intermolecular interaction energies, hydrophobic interactions) and ligand properties (desolvation, log P etc.), which w i l l give semi-quantitative predictions, and are much faster than F E P calculations, and 4) relative hydration free energies.56 Calculation o f relative hydration free energies is important i n the design and optimization o f molecules that act as enzyme inhibitors only after undergoing covalent hydration. For example, a class o f adenosine deaminase and cytidine deaminase inhibitors are known by X - r a y crystallography to bind i n the hydrated form. 57,58 Calculation o f both the relative hydration free energy and the relative binding free energy for the hydrated species provides an accuarte method for calculating relative inhibitor potencies since it accounts for differences in both hydration equilibrium and binding. Then, top scoring compounds are synthesized and tested for activity. Depending on the convergence criteria of the biological activity, the flow chart is repeated. Free Energy Perturbation Methods. Since F E P methods provide very accurate quantitative predictions, we discuss its use i n the comparison o f similar ligands binding to an enzyme. This task is o f particular value during the lead optimization phase o f drug design. W e considered two examples where F E P calculations were used successfully to predict binding affinities o f ligands to enzymes prior to synthesis. The first example considered was one o f the earliest successes o f F E P calculations and involves transition state ligands bound to thermolysin, carried out by M e r z and Kollman.51 In this work these authors predicted that the replacement o f an - N H group with a methylene group would not be detrimental to binding affinity despite a loss i n a hydrogen bond between the N H and an amide carbonyl. The principal reason was related to ligand desolvation. This prediction, which was made


7

ahead o f biochemical measurements, was later confirmed experimentally. In the second example, several research groups59-62 j the F E P method for calculating relative binding affinities for HIV-1 protease inhibitors and obtained good agreement with experimental results. More recently, Reddy et. al^ used a computer-assisted drug design method that combines molecular mechanics, dynamics, F E P calculations, inhibitor design, synthesis, and biochemical testing o f peptidomimetic inhibitors and crystallographic structure determination o f the protein-inhibitor complexes to successfully design novel inhibitors o f HIV-1 protease. This study involved a large set o f molecules whose relative binding affinities were predicted using F E P methods prior to synthesis, and were later confirmed by experimental measurements. Downloaded by MAHIDOL UNIV on February 18, 2015 | http://pubs.acs.org Publication Date: July 7, 1999 | doi: 10.1021/bk-1999-0719.ch001

u s e (

Fast Methods for Qualitative Binding Prediction of Binding Affinities of Ligands. Though the F E P method is theoretically more accurate and provides quantitative predictions between two similar ligands, it suffers from some practical limitations as applied to ligand design. Therefore, faster methods that can accommodate structural diversity are being developed. In some cases predictions o f ligand binding has been based on soley on a visual analysis of structures without any force field calculations.52 These methods relied on graphical analysis o f features such as steric and electronic complementary o f the docked inhibitor to the target protein, the extent o f buried hydrophobic surface and the number o f rotatable bonds in the ligand. Quantitative descriptors based on molecular shape ^ and grid-based energetics^ have also proved to be useful. More advanced methods have used an empirical scoring f u n c t i o n ^ derived from crystal structure data and experimental binding affinities. Though molecular mechanics methods appear to be more useful i n this regard, these methods met with only limited success i n i t i a l l y , ^ due to the large approximations involved i n the analysis (e.g., binding conformations, solvent model used, lack o f entropic terms etc.). Recently, Montgomery et. al. adopted some improvements to the molecular mechanics methods by using Monte Carlo techniques to derive the binding conformations o f inhibitors followed by energy minimizations. 10,67 This method allowed the prediction o f binding affinities for proposed purine nucleoside phosphorylase inhibitors prior to synthesis. The calculated results suggested that differences in solvation and entropy would contribute minimally to binding affinity. Although the binding conformations were accurately predicted i n this study, analysis o f interaction energies across the inhibitors was less informative, presumably because o f unaccounted factors such as desolvation and entropy. 6

Future Directions Rational drug design methods are continually improving, and a wider variety o f drug targets are being approached by these methods. A wide variety o f additional improvements can be anticipated in the future as well. Improved computer hardware w i l l allow the use o f more rigorous methods to be applied to large molecular systems. It w i l l not be surprising to see fully quantum mechanical docking studies appearing i n the future. A second trend i n computational methods that should continue i n the


8

future is the development o f both hybrid methods (currently known examples include genetic neural networks, 5 k-nearest neighbor genetic a l g o r i t h m s , among others) and integrated tools for drug design. Advances i n the modeling o f protein structures w i l l promote more widespread use o f structure-based drug design for drug targets that do not crystallize.


1

68

N e w experimental methods w i l l also lead rational drug design i n new directions. Combinatorial chemistry and high-throughput screening would not be as highly useful as they are today without solid-phase synthesis methods. Improvements i n areas such as catalyst design to allow rapid access to an ever-increasing range o f chemical structures, biological activity assays to allow the use o f a wider variety o f biological targets, and experimental structure determination methods to provide a wider selection o f structural information for structure-based approaches w i l l have significant impact on how rational drug design is performed i n the future. For lead optimization, the quantitative F E P methods provide an accurate prediction o f relative binding affinities between inhibitors only for structurally similar molecules, whereas the qualitative methods provide qualitative trends for relative binding affinities across a more structurally diverse set o f compounds. Ideally, methods that combine both o f those features w i l l greatly enhance the utility o f computational methods to drug design. Increased structural diversity, however, requires accurate calculation o f additional factors that significantly impact the compounds binding affinity. For example, the larger the difference in structure, the greater the chance that solvation, entropy, inter and intramolecular interaction energies o f ligand both in solvent and i n the complex, hydrophobic effects, conformational flexibility etc., w i l l influence relative binding affinities. Understanding the magnitude o f each contribution is key to an accurate prediction. Since an equation that incorporates each factor accurately has not been derived, we cannot expect accurate predictions using any o f above mentioned methods for the diverse set o f molecules. Therefore, regression equations which incorporates many o f the properties discussed above would greatly strengthen the rational drug desion methods for fast screening (prior to synthesis) o f diverse set o f inhibitors to an enzyme semi-quantitatively. In conclusion, rational drug design is an exciting and constantly growing field o f research. Its impact on quality o f life and health ensure the vitality o f the field. References

1. Young, S. S.; Sheffield, C. F.; Farmen, M . J. Chem. Inf. Comput.Sci.1997, 37, 892-899. 2. Hansen, C.; Maloney, P. P.; Fujita, T.; Muir, R. M . Nature 1962, 194, 178. 3. Hansch, C. Acc. Chem. Res. 1969, 2, 232-239. 4. Topliss, J. G. J. Med. Chem. 1972, 15, 1006-1011. 5. Topliss, J. G. J. Med. Chem. 1977, 20, 463-469. 6. Topliss, J. G. Perspect. Drug Disc. Design 1993, 1, 253-268. 7. Nicklaus, M . C.; Wang, S.; Driscoll, J. S.; Milne, G. W. A. Bioorg. Med. Chem. 1995, 3, 411-428.



9

8. Marshall, G. R.; Barry, C. D.; Bosshard, H. E.; Dammkoehler, R. A.; Dunn, D. A . The Conformational Parameter in Drug Design: The Active Analog Approach; Marshall, G. R.; Barry, C. D.; Bosshard, H. E.; Dammkoehler, R. A.; Dunn, D. A., Ed.; ACS: Washington, D.C., 1979; Vol. 112, pp 205-226. 9. Appelt, K.; Bacquet, R. J.; Bartlett, C. A.; Booth, C. L.; Freer, S. T.; Fuhry, M . A . M.; Gehring, M . R.; Hermann, S. M.; Howland, E. F.; Janson, C. A.; Jones, T. R.; Kan, C.; Kathardeker, V.; Lewis, K. K.; Marzoni, G. P.; Matthews, D. A.; Mohr, D. A.; Morse, C. A.; Oatley, S. J.; Ogden, R. O.; Reddy, M . R.; Reich, S. H.; Schoettlin, W. S.; Webber, S. E.; Welsch, K. M.; White, J. J. Med. Chem. 1991, 34, 1925-1934. 10. Montgomery, J. A.; Niwas, S.; Rose, J. D.; Secrist, J. A.; Babu, S. Y.; Bugg, C. E.; Erion, M . E.; Guida, W. C.; Ealick, S. E. J. Med. Chem. 1993, 36, 55-69. 11. Whittle, P. J.; Blundell, T. L. Annu. Rev. Biophys.Biomol.Struct. 1994, 23, 349375. 12. King, R. D.; Hirst, J.; Sternberg, M . J. E. Perspect. Drug Disc. Design 1993, 1, 279-290. 13. Luke, B. T. Journal of Chemical Information and Computer Sciences 1994, 34, 1279-1287. 14. Rogers, D.; Hopfinger, A. J. Journal of Chemical Information and Computer Sciences 1994, 34, 854-866. 15. So, S.-S.; Karplus, M . J. Med. Chem. 1996, 39, 1521-1530. 16. Cramer, R. D. I.; Patterson, D. E.; Bunce, J. D. J. Am. Chem. Soc. 1988, 110, 5959-5967. 17. Baskin, I. I. J. Chem. Inf. Comput. Sci. 1997, 37, 715-721. 18. Allen, M . S.; LaLoggia, A. J.; Dorn, L. J.; Martin, M . J.; Costantino, G.; Hagen, T. J.; Koehler, K. F.; Skolnick, P.; Cook, J. M . J. Med. Chem. 1993, 35, 40014010. 19. Allen, M . S.; Tan, Y.-T.; Trudell, M . L.; Narayanan, K.; Schindler, L. R.; Martin, M . J.; Schultz, C.; Hagen, T. J.; Koehler, K. F.; Codding, P. W.; Skolnick, P. J. Med. Chem. 1990, 33, 2343-2357. 20. Robinson, D. D.; Barlow, T. W.; Richards, G. W. J. Chem. Inf. Comput. Sci. 1997, 37, 943-950. 21. Jiang, H.; Chen, K.; Tang, Y.; Chen, J.; Li, Q.; Wang, Q.; Ji, R. J. Med. Chem. 1997, 40, 3085-3090. 22. Van Drie, J. H. J. Comput.-Aid. Mol. Design 1997, 11, 39-52. 23. Hahn, M . J. Med. Chem. 1995, 38, 2080-2090. 24. Crippen, G. M . J. Comput. Chem. 1995, 16, 486-500. 25. Walters, D. E.; Hinds, R. M . J. Med. Chem. 1994, 37, 2527-2536. 26. Crippen, G. M . J. Med. Chem. 1997, 40, 3161-3172. 27. Murcko, M . A. Recent Advances in Ligand Design Methods; Murcko, M . A., Ed.; VCH: New York, 1997; Vol. 11, pp 1-66. 28. Clark, D. E.; Murray, C. W.; Li, J. Current Issues in De Novo Molecular Design; Clark, D. E.; Murray, C. W.; Li, J., Ed.; VCH: New York, 1997; Vol. 11, pp 67125. 29. Ewing, T. J. A.; Kuntz, I. D. J. Comput. Chem. 1997, 18, 1175-1189. 30. Hahn, M . J. Chem. Inf. Comput. Sci. 1997, 37, 80-86.



10

31. Neamati, N.; Hong, H.; Mazumder, A.; Wang, S.; Sunder, S.; Nicklaus, M . C.; Milne, G. W. A.; Proksa, B.; Pommier, Y. J. Med. Chem. 1997, 40, 942-951. 32. Sadowski, J. J. Comput.-Aid. Mol. Design 1997, 11, 53-60. 33. Thorner, D. A.; Willett, P.; Wright, P. M . ; Taylor, R. J. Comput.-Aid. Mol. Design 1997, 11, 163-174. 34. Higgs, R. E.; Bemis, K. G.; Watson, I. A.; Wikel, J. H. J. Chem. Inf. Comput. Sci. 1997, 37, 861-870. 35. Murray, C. W.; Clark, D. E.; Auton, T. R.; Firth, M . A.; Li, J.; Sykes, R. A.; Waszkowycz, B.; Westhead, D. R.; Young, S. C. J. Comput.-Aid. Mol. Design 1997, 11, 193-207. 36. Böhm, H.-J. J. Comput.-Aided Mol. Design 1992, 6, 61-78. 37. Böhm, H.-J. Perspect. Drug Disc. Des. 1995, 3, 21-33. 38. Luo, Z.; Wang, R.; Lai, L. J. Chem. Inf. Comput. Sci. 1996, 36, 1187-1194. 39. Gehlhaar, D. K.; Moerder, K. E.; Zichi, D.; Sherman, C. J.; Ogden, R. C.; Freer, S. T. J. Med. Chem. 1995, 38, 466-472. 40. Westhead, D. R.; Clark, D. E.; Frenkel, D.; Li, J.; Murray, C. W.; Robson, B.; Waszkowycz, B. J. Comput.-Aided Mol. Design 1995, 9, 139-148. 41. Glen, R. C.; Payne, A. W. R. J. Comput.-Aided Mol. Design 1995, 9, 181-202. 42. Todorov, N. P.; Dean, P. M . J. Comput.-Aid. Mol. Design 1997, 11, 175-192. 43. Agrafiotis, D. J. Chem. Inf. Comput. Sci. 1997, 37, 841-851. 44. Agrafiotis, D. J. Chem. Inf. Comput. Sci. 1997, 37, 576-580. 45. Brown, R. D.; Martin, Y. C. J. Med. Chem. 1997, 40, 2304-2313. 46. Gillet, V. J.; Willett, P.; Bradshaw, J. J. Chem. Inf. Comput. Sci. 1997,37,731740. 47. Lewis, R. A.; Mason, J. S.; McLay, I. M . J. Chem. Inf. Comput. Sci. 1997, 37, 599-614. 48. Warr, W. A. J. Chem. Inf. Comput: Sci. 1997, 37, 134-140. 49. Holloway, K.; Wai, J. M.; Halgren, T. A.; Fitzgerald, P. M.; Vacca, J. P.; Dorsey, B. D.; Levin, R. B.; Thompson, W. J.; Chen, J. L.; deSolms, J. S.; Gaffin, N . ; Ghosh, A. K.; Guiliani, E. A.; Graham, S. L.; Guare, J. P.; Hungate, R. W.; Lyle, T. A.; Sanders, W. M . ; Tucker, T. J.; Wiggins, M . ; Wiscount, C. M . ; Woltersdorf, O. W.; Young, S. D.; Darke, P. L.; Zugay, J. A. J. Med. Chem. 1995, 38, 305-317. 50. Reddy, M . R.; Varney, M . D.; Kalish, V.; Viswanadhan, V. N . ; Appelt, K. J. Med. Chem. 1994, 114, 10117-10122. 51. Merz, K. M.; Kollman, P. A. J. Am. Chem.Soc. 1989, 111, 5649-5658. 52. Bohacek, R. S.; McMartin, C. J. Med. Chem. 1992, 35, 1671-1684. 53. McCammon, J. A. Current Opinion in Structural Biology 1991, 1, 196-200. 54. Beveridge, D. L.; DiCapua, F. M . Annu. Rev. Biophys. Chem. 1989, 18, 431-492. 55. Head, R. D.; Smythe, M . L.; Oprea, T. I.; Waller, C. L.; Green, S. M.; Marshall, G. R. J. Am. Chem. Soc. 1996, 118, 3959-3969. 56. Erion, M . D.; Reddy, M . R. J. Am. Chem. Soc. 1997 (accepted). 57. Wilson, D. K.; Rudolph, F. B.; Quiocho, F. A. Science 1991, 252, 1278-1284. 58. Sharff, A. J.; Wilson, D. K.; Chang, Z.; Quiocho, F. A. J. Mol. Biol. 1992, 226, 917-921.



11

59. Reddy, M . R.; Viswanadhan, V. N.; Weinstein, J. N . Proc. Natl. Acad. Sci. USA 1991, 88, 10297-10291. 60. Tropshaw, A. J.; Hermans, J. Prot. Eng. 1992, 5, 29-33. 61. Ferguson, D. M.; Radmer, R. J.; Kollman, P. A. J. Med. Chem. 1991, 34, 26542659. 62. Rao, B. G.; Tilton, R. F.; Singh, U. C. J. Am. Chem. Soc. 1992, 114, 4447-4452. 63. Kuntz, I. D.; Meng, E. C.; Shoichet, B. K. Acc. Chem. Res. 1994, 27, 117-123. 64. Goodford, P. A. J. Med. Chem. 1985, 28, 849-857. 65. Böhm, H.-J. J. Comput. Aid. Mol. Des. 1994, 8, 243-256. 66. Sansom, C. E.; Wu, J.; Weber, I. T. Protein Eng. 1992, 5, 659-667. 67. Erion, M . D.; Montgomery, J. A.; Niwas, S.; Rose, J. D.; Ananthan, S.; Allen, M.; Secrist, J. A.; Babu, S. Y.; Bugg, C. E.; Guida, W. C.; Ealick, S. E. J. Med. Chem. 1993, 36, 3771-3783. 68. Raymer, M . L.; Sanschagrin, P. C.; Punch, W. F.; Venkataraman, S.; Goodman, E. D.; Kuhn, L. A. J. Mol. Biol. 1997, 265, 445-464.


Overview of Rational Drug Design - ACS Symposium Series (ACS

Recommend Documents