The contribution of combinatorial chemistry to lead generation: an interim analysis.

In the process of finding new drug candidates medicinal chemists nowadays have a variety of options to choose from, one is to apply combinatorial chemistry techniques. Since the early 1990's synthetic and analytical methods as well as new technologies have been growing rapidly in the area of combinatorial chemistry. Applying these techniques have resulted in the production of large numbers of compounds. A trend is observed towards smaller libraries of compounds with more drug-like properties. An analysis is made to establish the contribution of combinatorial chemistry in providing new lead candidates for (pre)clinical development towards new pharmaceutical products. Ten representative examples are given to describe the impact of ombinatorial chemistry on different levels of the lead discovery and optimization process. Furthermore, reports on combinatorial chemistry products that are already in (pre)clinical development were traced back to their source. The interim analysis showed only limited success of combinatorial chemistry approaches in terms of delivering leads. Second generation libraries appear more drug-like and focussed and may result in more compounds entering clinical studies in the future.

Strategies and progress towards the ideal orally active thrombin inhibitor.

Thrombin plays a key role in the control of thrombus formation, for which reason its inhibition has become a target for new antithrombotics. Important issues in the profile of the ideal thrombin inhibitor are: potency, selectivity, oral bioavailability, half-life in the circulatory system and safety. Although many potent direct inhibitors of thrombin have been discovered, most of these inhibitors lack sufficient oral bioavailability. This is often associated with the presence of highly basic functionalities such as guanidine or amidine. These basic functionalities in the P1 moiety are preferred by thrombin and are present in the first generation of thrombin inhibitors. Recently, several orally active direct thrombin inhibitors have been disclosed. Most of these inhibitors originate from leads of the first generation. Two major optimization strategies could be identified to further improve these leads: A: maintain the highly basic P1 moiety and compensate its negative effects, and B: reduce the basicity of the P1 moiety and compensate for the decrease in inhibitory activity. The progress made using these strategies is evaluated. In addition, screening large sets of compounds yielded new structures that provide useful starting points for optimization. The optimization strategy used to convert leads from screening into potent orally active thrombin inhibitors is also be evaluated.

Solution-phase and solid-phase synthesis of novel transition state inhibitors of coagulation enzymes incorporating a piperidinyl moiety.

2-Amino-3-piperidin-4-yl-propionic acid containing peptidomimetics are potent protease inhibitors when combined with an appropriate keto-thiazole or keto-carboxylic acid moiety. A novel P1 residue in factor Xa and thrombin inhibitors has been found resulting in IC50 values as low as 0.048 microM, a factor of ten more potent than Argatroban. Starting with non-chiral synthetic routes, a new stereospecific route was developed as well as a new solid-phase method.

Novel acylguanidine containing thrombin inhibitors with reduced basicity at the P1 moiety.

Replacement of the noragmatine group in thrombin inhibitors with a beta-alanyl-guanidine group resulted in a nearly equipotent and more selective compound 8 despite the fact that the pKa of this P1 moiety is five orders of magnitude lower. Further modification resulted in a nonpeptide inhibitor with this beta-alanyl-guanidine group, compound 28. This is an active and selective thrombin inhibitor and in view of its nonpeptide/low basicity structure selected for further pharmacological studies.

Kinetic studies on rat liver microsomal glutathione transferase: consequences of activation.

Rat liver microsomal glutathione transferase is activated by sulfhydryl reagents and proteolysis. This property varies, however, depending on the combination, concentration and reactivity of the substrates. Thus, a multi-dimensional diagram can be envisioned in which the parameters affecting enzyme activity and activation are visualized. In principle activation could stem from an alteration in enzyme mechanism, transition-state complementarity, product release rate or pH-rate behaviour. These studies appear to rule out these possibilities and an alternate hypothesis is suggested based on the following experiments: (i) alternate substrate diagnosis of the kinetic mechanism of microsomal glutathione transferase indicates a random sequential mechanism. Non-activated and activated enzyme follow the same mechanism by these criteria. (ii) The microsomal glutathione transferase stabilizes a Meisenheimer complex between 1,3,5-trinitrobenzene and glutathione. The formation constants were similar for the unactivated and activated enzyme ((15 +/- 1).10(3) and (14 +/- 1).10(3) M-1, respectively, at pH 8). Inasmuch as the Meisenheimer complex resembles the transition state there is no evidence for an increased stabilization upon activation. (iii) The catalytic rate constant kcat does not vary with the viscosity in the assay medium. Thus, product release is not rate limiting for the unactivated and activated microsomal glutathione transferase (with saturating 1-chloro-2,4-dinitrobenzene and varying GSH). (iv) The pH dependence of the Kf-values for Meisenheimer complex formation exhibited pKa values close to 6 for both the activated and unactivated microsomal glutathione transferase. The pH profile of kcat (with saturating 1-chloro-2,4-dinitrobenzene and variable GSH concentrations) showed apparent pKa values of 5.7 +/- 0.5 and 6.3 +/- 0.4 for the unactivated and activated enzyme, respectively, indicative of a very similar requirement for deprotonation of the enzyme-GSH-1-chloro-2,4-dinitrobenzene complex. (v) Examination of the kinetic parameters (obtained with GSH as the variable substrate against increasingly reactive electrophilic substrates) in Hammett plots shows that the activation mechanism entails a more efficient utilization of GSH. It is suggested that a higher rate of formation of the glutathione thiolate anion occurs in the activated enzyme.

Evidence that rat liver microsomal glutathione transferase is responsible for glutathione-dependent protection against lipid peroxidation.

Evidence that rat liver microsomal glutathione transferase is responsible for the glutathione-dependent inhibition of lipid peroxidation in liver microsomes has been obtained. Activation of the microsomal glutathione transferase in microsomes by cystamine renders this organelle even more resistant to lipid peroxidation in the presence of glutathione compared with untreated microsomes. Upon examining the effect of seven glutathione analogues on lipid peroxidation, it was found that only those that serve as good substrates for the microsomal glutathione transferase (Glutaryl-L-Cys-Gly and alpha-L-Glu-L-Cys-Gly) can inhibit lipid peroxidation. The lack of inhibition by the other five analogues (alpha-D-Glu-L-Cys-Gly, gamma-D-Glu-L-Cys-Gly, beta-L-Asp-L-Cys-Gly, alpha-L-Asp-L-Cys-Gly and alpha-D-Asp-L-Cys-Gly) shows the specificity of the protection and rules out any non-enzymic component. Inhibitors of selenium-dependent glutathione peroxidase (mercaptosuccinate at 50 microM) and phospholipid hydroperoxide glutathione peroxidase (iodoacetate, 1 mM + glutathione, 0.5 mM) do not inhibit the glutathione-dependent protection of rat liver microsomes against lipid peroxidation. Purified microsomal glutathione transferase, NADPH-cytochrome P450 reductase and cytochrome P450 were reconstituted in microsomal phospholipid vesicles by cholate dialysis. The resulting membranes contained functional enzymes and did display enzymic lipid peroxidation induced by 75 microM NADPH and 10 microM Fe-EDTA (2:1). This model system was used to investigate whether microsomal glutathione transferase could inhibit lipid peroxidation in a glutathione-dependent manner. The results show that 5 mM glutathione did inhibit lipid peroxidation when functional microsomal glutathione transferase was included. This was not the case when the enzyme had been pre-inactivated with diethylpyrocarbonate. Furthermore, the protective effect of glutathione could be partly reversed by an inhibitor (100 microM bromosulphophtalein) of the enzyme. Apparently, rat liver microsomal glutathione transferase has the capacity to inhibit lipid peroxidation in a reconstituted system.

Inhibition of glutathione S-transferase 3-3 by glutathione derivatives that bind covalently to the active site.

In all, 13 GSH derivatives have been synthesized and tested for their potency to inhibit glutathione S-transferase (GST) 3-3. All of these derivatives contained a reactive group that could potentially react with the enzyme active site. Best results were obtained with the phenylthiosulphonate derivative of GSH, GSSO2Ph. Preincubation of GST 3-3 with a 100 microM concentration of this inhibitor resulted in a time-dependent loss of activity: after 30 min at pH 6.5 and 25 degrees C, 51% of the activity was lost. At more alkaline pH, the activity is more rapidly inhibited: at pH 8.0 the 90%-inhibition level is already reached after 10 min preincubation. Separation of enzyme and excess unbound GSSO2Ph after preincubation by gel-filtration chromatography did not result in a reappearance of enzyme activity. If 100 microM-GSH was added to the preincubation mixture at pH 7.4, inhibition was almost completely prevented. Addition of S-(hexyl)glutathione (20 microM) could delay the inhibition but, ultimately, not prevent it. The inhibited enzyme could be re-activated by addition of 10 mM-2-mercaptoethanol: 60 min after this thiol was added, the inhibited GST-3- activity was bacxk to the control level. GSH at the same concentration could not re-activate the enzyme. On the basis of these results, on the known reactivity of thiosulphonate compounds, and on current knowledge about the amino acid residues involved in GST catalysis, a covalent modification of an active-site cysteine residue by mixed-disulphide formation between enzyme and the cosubstrate GSH is postulated. Information on the synthesis and characterization of the GSH derivatives is given in Supplementary Publication SUP 50166 (5 pages) which has been deposited at the British Library Document Supply Centre, Boston Spa, Wetherby, West Yorkshire LS23 7BQ, U.K., from whom copies can be obtained on the terms indicated in Biochem. J. (1991) 273, 5.

Studies on the activity and activation of rat liver microsomal glutathione transferase with a series of glutathione analogues.

The substrate specificity of rat liver microsomal glutathione transferase toward glutathione has been examined in a systematic manner. Out of a glycyl-modified and eight gamma-glutamyl-modified glutathione analogues, it was found that four (glutaryl-L-Cys-Gly, alpha-L-Glu-L-Cys-Gly, alpha-D-Glu-L-Cys-Gly, and gamma-L-Glu-L-Cys-beta-Ala) function as substrates. The kinetic parameters for three of these substrates (the alpha-D-Glu-L-Cys-Gly analogue gave very low activity) were compared with those of GSH with both unactivated and the N-ethylmaleimide-activated microsomal glutathione transferase. The alpha-L-Glu-L-Cys-Gly analogue is similar to GSH in that it has a higher kcat (6.9 versus 0.6 s-1) value with the activated enzyme compared with the unactivated enzyme but displays a high Km (6 versus 11 mM) with both forms. Glutaryl-L-Cys-Gly, in contrast, exhibited a similar kcat (8.9 versus 6.7 s-1) with the N-ethylmaleimide-treated enzyme but retains a higher Km value (50 versus 15 mM). Thus, the alpha-amino group of the glutamyl residue in GSH is important for the activity of the activated microsomal glutathione transferase. These observations were quantitated by analyzing the changes in the Gibbs free energy of binding calculated from the changes in kcat/Km values, comparing the analogues to GSH and each other. It is estimated that the binding energy of the alpha-amino group of the glutamyl residue in GSH contributes 9.7 kJ/mol to catalysis by the activated enzyme, whereas the corresponding value for the unactivated enzyme is 3.2 kJ/mol. The importance of the acidic functions in glutathione is also evident as shown by the lack of activity with 4-aminobutyric acid-L-Cys-Gly and the low kcat/Km values with gamma-L-Glu-L-Cys-beta-Ala (0.03 and 0.01 mM-1s-1 for unactivated and activated enzyme, respectively). Utilization of binding energy from a correctly positioned carboxyl group in the glycine residue (10 and 17 kJ/mol for unactivated and activated enzyme, respectively) therefore also appears to be required for optimal activity and activation. A conformational change in the microsomal glutathione transferase upon treatment with N-ethylmaleimide or trypsin, which allows utilization of binding energy from the alpha-amino group of GSH as well as the glycine carboxyl in catalysis, is suggested to account for at least part of the activation of the enzyme.

Inhibition of rat liver glutathione S-transferase isoenzymes by peptides stabilized against degradation by gamma-glutamyl transpeptidase.

Inhibitors for glutathione S-transferase (GST) iso-enzymes from rat liver with high affinity for the glutathione-binding site (G-site) have been developed. In previous studies, a model was described for the G-site of GST (Adang, A. E. P., Brussee, J., van der Gen, A., and Mulder, G. J. (1990) Biochem. J. 269, 47-54) in terms of essential and nonessential interactions between groups in glutathione (GSH) and the G-site. Based on this model, compounds were designed that have high affinity for the G-site but cannot be conjugated. In the dipeptide gamma-L-glutamyl-D-aminoadipic acid (gamma-L-Glu-D-Aad), the L-cysteinylglycine moiety is replaced by D-aminoadipic acid. This dipeptide is an efficient competitive inhibitor (toward GSH) of mu class GST isoenzymes with Ki values of 34 microM for GST isoenzyme 3-3 and 8 microM for GST isoenzyme 4-4. Other GSH-dependent enzymes, such as gamma-glutamyl transpeptidase (gamma-GT), glutathione reductase, and glutathione peroxidase, were not inhibited by 1 mM of gamma-L-Glu-D-Aad. Inhibition is also highly stereospecific since gamma-L-Glu-L-Aad is only a poor inhibitor (Ki = 430 microM for GST 3-3). Gamma-L-Glutamyl-D-norleucine also had a much higher Ki value for GST 3-3. Thus, the presence of a delta-carboxylate group in D-Aad appears to be essential for a high affinity inhibitor. An additional hydrophobic group did not result in increased inhibitory potency. In a different approach, the gamma-L-glutamyl moiety in GSH was replaced by delta-L-aminoadipic acid; delta-L-Aad-L-Cys-Gly is an efficient cosubstrate analogue for GSTs with Km values comparable to GSH and Vmax values ranging from 0.24 to 57 mumol/min/mg for the different GSTs. The structures of the efficient inhibitor and the cosubstrate analogue were combined in delta-L-Aad-D-Aad, which had a Ki value of 68 microM with GST 3-3. In order to investigate their possible use in vivo studies, the degradation of gamma-L-Glu-D-Aad and delta-L-Aad-L-Cys-Gly by gamma-GT was investigated. The peptides showed no measurable hydrolysis rates under conditions where GSH was rapidly hydrolyzed. Thus, an efficient, mu class-specific GST inhibitor and a gamma-glutamyl-modified cosubstrate analogue of GSH were developed. Their gamma-GT stability offers the possibility to use these peptides in in vivo experiments.

The glutathione-binding site in glutathione S-transferases. Investigation of the cysteinyl, glycyl and gamma-glutamyl domains.

The GSH-binding site of glutathione S-transferase (GST) isoenzymes was studied by investigating their substrate-specificity for three series of GSH analogues; further, a model of the interactions of GSH with the G-site is proposed. Twelve glycyl-modified GSH analogues, four ester derivatives of GSH and three cysteinyl-modified GSH analogues were synthesized and tested with purified forms of rat liver GST (1-1, 2-2, 3-3 and 4-4). The glycyl analogues exhibited spontaneous chemical reaction rates with 1-chloro-2,4-dinitrobenzene comparable with the GSH rate. In contrast, the enzymic rates (Vmax.) differed greatly, from less than 1 up to 140 mumol/min per mg; apparently, a reaction mechanism is followed that is very sensitive to substitutions at the glycyl domain. No correlation exists between the chemical rates and Vmax. values for the analogues. Analogues of GSH in which L-cysteine was replaced by D-cysteine, L-homocysteine or L-penicillamine showed little or no capacity to replace GSH as co-substrate for the GSTs. GSH monomethyl and monoethyl esters showed Vmax. values greater than the Vmax. measured with GSH: the Vmax. for the monoethyl ester of GSH and GST 3-3 was 5-fold that for GSH. The data obtained in this and previous studies [Adang, Brussee, Meyer, Coles, Ketterer, van der Gen & Mulder (1988) Biochem. J. 255, 721-724; Adang, Meyer, Brussee, van der Gen, Ketterer & Mulder (1989) Biochem. J. 264, 759-764] allow a model of the interactions of GSH in the G-site in GSTs to be postulated. The gamma-glutamyl site is the main binding determinant: the alpha-carboxylate group is obligatory, whereas shifting of the amino group and shortening of the peptide backbone only decreased kcat./Km. Furthermore, the GSTs appear to be very critical with respect to a correct orientation of the thiol group of the GSH analogue. The glycyl site is the least restrictive domain in the G-site of GSTs: amino acid analogues all showed Km values between 0.2 and 0.6 mM (that for GSH is 0.2-0.3 mM), but large differences in Vmax. exist. The glycyl carboxylate group is not essential for substrate recognition, since decarboxy analogues and ester derivatives showed high activities. The possible mechanisms for an increased Vmax. in some analogues are briefly discussed.

Interaction of rat glutathione S-transferases 7-7 and 8-8 with gamma-glutamyl- or glycyl-modified glutathione analogues.

Analogues of GSH in which either the gamma-glutamyl or the glycyl moiety is modified were synthesized and tested as both substrates for and inhibitors of glutathione S-transferases (GSTs) 7-7 and 8-8. Acceptor substrates for GST 7-7 were 1-chloro-2,4-dinitrobenzene (CDNB) and ethacrynic acid (ETA) and for GST 8-8 CDNB, ETA and 4-hydroxynon-trans-2-enal (HNE). The relative ability of each combination of enzyme and GSH analogue to catalyse the conjugation of all acceptor substrates was similar with the exception of the combination of GST 7-7 and gamma-L-Glu-L-Cys-L-Asp, which used CDNB but not ETA as acceptor substrate. In general, GST 7-7 was better than GST 8-8 in utilizing these analogues as substrates, and glycyl analogues were better than gamma-glutamyl analogues as both substrates and inhibitors. These results are compared with those obtained earlier with GSH analogues and GST isoenzymes 1-1, 2-2, 3-3 and 4-4 [Adang, Brussee, Meyer, Coles, Ketterer, van der Gen & Mulder (1988) Biochem. J. 255, 721-724] and the implications with respect to the nature of their active sites are discussed.

Synthesis and nucleophilic reactivity of a series of glutathione analogues, modified at the gamma-glutamyl moiety.

A series of GSH analogues with modifications at the gamma-glutamyl moiety was synthesized and purified by following peptide chemistry methodology. Benzyl, benzyloxycarbonyl and t-butyloxycarbonyl protective groups were used to protect individual amino acid functional groups. The formation of peptide bonds was accomplished through coupling of free amino groups with active esters, generated by reaction of the carboxylate functions with dicyclohexylcarbodi-imide and 1-hydroxybenzotriazole. The protecting groups in the tripeptides were removed in a single step by using Na in liquid NH3. Precautions were taken in order to prevent oxidation of the thiol function in the cysteine residue. Thus GSH analogues containing both L- and D-glutamic acid and L- and D-aspartic acid, coupled to cysteinylglycine through both the alpha- and the omega-carboxylate group, were synthesized. Also, decarboxy-GSH and deamino-GSH, lacking one functional group in the glutamate moiety, were prepared. The spontaneous non-enzyme-catalysed nucleophilic reaction of these GSH analogues with the electrophilic model substrate 1-chloro-2,4-dinitrobenzene showed appreciable rate differences, indicating the importance of intramolecular interactions in determining the nucleophilic reactivity of the thiol function in the cysteine residue. In particular, the free amino group in the gamma-L-glutamic acid residue appears to play a crucial role in activating the thiol group in GSH. In an adjacent paper [Adang, Brussee, Meyer, Coles, Ketterer, van der Gen & Mulder (1988) Biochem. J. 255, 721-724] these results are compared with those obtained in a study on the ability of these GSH analogues to act as a co-substrate in the glutathione S-transferase-catalysed conjugation reaction with 1-chloro-2,4-dinitrobenzene.

Substrate specificity of rat liver glutathione S-transferase isoenzymes for a series of glutathione analogues, modified at the gamma-glutamyl moiety.

The substrate specificity of purified rat liver glutathione S-transferases (GSTs) for a series of gamma-glutamyl-modified GSH analogues was investigated. GST isoenzyme 3-3 catalysed the conjugation of 1-chloro-2,4-dinitrobenzene with six out of the nine analogues. alpha-L-Glu-L-Cys-Gly and alpha-D-Glu-L-Cys-Gly showed catalytic efficiencies of 40% and 130% that of GSH respectively. The GSH analogue with an alpha-D-glutamyl moiety appeared to be a highly isoenzyme-3-3-specific co-substrate: kcat./Km with GST isoenzyme 4-4 was only about 5% that with GST isoenzyme 3-3, and no enzymic activity was detectable with GST isoenzymes 1-1 and 2-2. GST isoenzyme 4-4 showed some resemblance to GST 3-3: five out of nine co-substrate analogues were accepted by this second isoenzyme of the Mu multigene family. Isoenzymes 1-1 and 2-2, of the Alpha multigene family, accepted only two alternative co-substrates, which indicates that their GSH-binding site is much more specific.

The microsomal metabolism of hexachlorobenzene. Origin of the covalent binding to protein.

The microsomal metabolism of hexachlorobenzene is studied, with special attention to the covalent binding to protein. The metabolites formed are pentachlorophenol and tetrachlorohydroquinone. In addition, a considerable amount of covalent binding to protein is detected (250 pmoles pentachlorophenol, 17 pmoles tetrachlorohydroquinone and 11 pmoles covalent binding in an incubation containing 50 mumoles of hexachlorobenzene). In order to establish the potential role of reductive dechlorination in the covalent binding, the anaerobic metabolism of hexachlorobenzene was investigated. At low oxygen concentrations no pentachlorobenzene was detected, and only very small amounts of pentachlorophenol as well as covalent binding, indicating a relationship between covalent binding and the microsomal oxidation of hexachlorobenzene. Incubations with 14C-pentachlorophenol at low concentrations showed that a conversion-dependent covalent binding occurs to the extent of 75 pmole binding per nmole pentachlorophenol. This is almost enough to account for the amount of label bound to protein observed in hexachlorobenzene incubations. This indicates that less than 10% of the covalent binding occurs during conversion of hexachlorobenzene to pentachlorophenol, and the remainder is produced during conversion of hexachlorobenzene to pentachlorophenol, and the remainder is produced during conversion of pentachlorophenol. The major product of microsomal oxidation of pentachlorophenol is tetrachlorohydroquinone, which is in redox-equilibrium with the corresponding semiquinone and quinone (chloranil). The covalent binding is inhibited by addition of ascorbic acid or glutathione to the hexachlorobenzene incubations. Ascorbic acid decreases the covalent binding with a simultaneous increase in formation of tetrachlorohydroquinone, probably due to a shift in the redox-equilibrium to the reduced side. Glutathione does not act as a reducing agent, since the inhibition of covalent binding is not accompanied by an increase in tetrachlorohydroquinone formation. Instead, glutathione reacts with chloranil, producing at least three stable products, probably in a Michael-type reaction. These results strongly indicate the involvement of chloranil or the semiquinone radical in the covalent binding during microsomal hexachlorobenzene metabolism.