Perspective JJournal of Medicinal Chemistry.1994.Vol.37.No.8 1047 Figure 11.The bound conformation of S-2 within the active site of HCA II as determined by X-ray crystallographic studies might not be optimal be se of even small errors in th nterna and de penalties required de th ons of atoms in the proteir ture. As the des e ligands nce the geometry of desired interactions has bindin n ide of a novel Thymidylate synthas of a comb nat onoftheaboredesigncriterainadaitio bitor Series a of intere has beer TS is the rate oun en sh to be broad-spectr ound mu t in the cry ds the the onstant.It ismor etrahydrofolate.All design wor tha that it ave a tight et u sing the structure of the E.coli TS complexed 】 with inhibit po tha 100 the methyl probe grids are d ompo sh ours shown in as a guid ne nt tode the of the pos ute ingand the aph ation pr ces ere in the 8-D ons to complement P be tein ds is highly with Lin water mo ally only accur 0.4 ups gave filled ty space can incre the likelihood of opt alkvlated nit gen atom A nit atom used a tions that are placed too rigidly atom Perspective Journal of Medicinal Chemistry, 1994, Vol. 37, No. 8 1047 Figure 11. The bound conformation of S-2 within the active site of HCA I1 as determined by X-ray crystallographic studies. performed on each design idea in order to determine both the internal and desolvation penalties required for the new ligand to attain its binding conformation inside the protein. The internal energy that is required for the small molecule to reach its binding conformation is energy lost in binding. New design ideas are prioritized on the basis of a combination of the above design criteria in addition to ease of synthesis and classical medicinal chemistry principles. The ones with the highest scores are synthe￾sized. In a structure-based design program, the criteria for defining a lead compound is different from a more traditional approach. In order to be useful, an initial lead compound must have a solubility in the crystallization drop of at least 10 times its inhibition constant. It is more important that a compound be amenable to crystallization with the target protein than that it have a tight binding constant. However, from a practical point of view, compounds with inhibition constants lower than 100 pM are desired. In addition, early compounds in a series should be somewhat flexible and have positions for further elaboration. One does not want to design a lead compound that has all of the positions on a ring substituted and therefore have no room to place substituents during the optimization process. The design of somewhat flexible leads is highly desired because the positions of protein atoms, even in a highly refined structure, are generally only accurate to 0.2-0.4 A. Allowing the ligand and the protein to adjust during the binding process can increase the likelihood of optimal overlap. Designated interactions that are placed too rigidly might not be optimal because of even small errors in the positions of atoms in the protein structure. As the design and synthesis cycles progress, rigidity can be built into new ligands once the geometry of desired interactions has been well defined. De Novo Design of a Novel Thymidylate Synthase Inhibitor Series. The initial area of interest has been in the design of novel inhibitors against the enzyme thymidylate synthase (TS).85,86 TS is the rate-limiting step is the conversion of deoxyuridylate monophosphate (dUMP) to thymidylate monophosphate. Inhibitors of TS have been shown to be broad-spectrum antiprolifera￾tives and in particular antitumor agents. TS binds two ligands, the substrate dUMP and the cofactor 5,lO￾methylenetetrahydrofolate. All of the design work which will be described was conducted in the cofactor binding pocket using the structure of the E. coli TS complexed with 5-fluoro-2’-deoxyuridylate. Using a Connolly surface and the methyl probe GRID87 contours shown in Figure 15 as a guide, a naphthalene ring was positioned into the deep part of the active site such that a maximum overlap occurred between the aromatic ring and the GRID map. Next, the naphthalene ring was used as a scaffold, and hydrogen bonding groups were placed in the 1- and 8-positions to complement an aspartic acid with an NH and a bound water molecule with a carbonyl oxygen. Linkage of the two hydrogen￾bonding groups gave a benz [cd] indole ring. Empty space off the 6-position of benzindole ring was filled with a dialkylated nitrogen atom. A nitrogen atom was used as the linker atom because it would not create a chiral center