STRUCTURE-ACTIVITY RELATIONSHIPS FOR A CLASS OF INHIBITORS OF PURINE NUCLEOSIDE PHOSPHORYLASE

Adriano D. Andricopulo (PQ)¹; Rosendo A. Yunes (PQ)²;

Jack W. Frazer (PQ)¹; Eugene H. Cordes (PQ)¹

¹College of Pharmacy and Department of Chemistry, University of Michigan, Ann Arbor, MI, USA. ²Departamento de Química, Universidade Federal de

Santa Catarina, Florianópolis-SC

keywords: QSAR models, purine nucleoside phosphorylase, competitive inhibitors

The ability to efficiently design or discover novel, patentable molecules which are potent, specific inhibitors of enzymes or potent, specific agonists or antagonists at biological receptors is of great importance. Many such molecules contribute to the prevention of or therapy for human diseases. The pharmaceutical industry worldwide continues to search for and employ novel technologies to improve their ability to rapidly identify and characterize molecules in these classes.

Quantitative structure-activity relationships (QSAR) [or quantitative structure-property relationships (QSPR)] have been employed, and continue to be developed and employed, both to correlate information in data sets and as a tool to facilitate, for example, the discovery of enzyme inhibitors. Work employing classical QSAR technology has proved useful in any number of settings.¹ At the same time, we believe that QSAR technology has a substantial unexploited potential to facilitate molecular design.^2,3 Three specific goals include: (i) to improve the ability of QSAR statistical models to predict property values for molecules outside of the training set; (ii) to extrapolate to property values outside those included by members of the training set; and (iii) to qualify these models in a way that provides reliable measures of the accuracy of such predictions. Our specific objective is to create high-quality standard databases which can be employed (i) to test and refine the developing QSAR technology, and (ii) to provide the basis for discovery of novel molecules having promise of utility in clinical medicine.

Work reported herein is the result of an effort to build upon the foundation for discovery of structurally novel, potent inhibitors of human purine nucleoside phosphorylase (PNP). Such inhibitors have multiple, plausible utilities in clinical medicine. Scientists at BioCryst Pharmaceuticals have employed structure-based drug discovery technologies to create a strong scientific basis for drug discovery and have used this basis to discover several classes of promising inhibitors of PNP, one of which is currently in advanced clinical trials (1).

Values of inhibition constants, K_i, for one of these classes of inhibitors of calf spleen PNP, have been determined employing both inosine as substrate and a manual assay and 2-amino-6-mercapto-7-methylpurine ribonucleoside (MESG) as substrate and a robot-based enzyme kinetics facility.² Several of the values determined robotically were confirmed employing the same substrate and a manual assay. The family of 29 inhibitors examined has modest structural diversity. Twenty three of the 29 inhibitors are 9-substituted-9-deazaguanines; one is 9-substituted-8-aminoguanine; one is 8-aminoguanine; one is a 9-substituted-8-methyl-9-deazaguanine; one is a 9-substituted-8-amino-9-deazaguanine; and the last two are 9-substituted-9-deazahypoxanthines. Thus, the bulk of the structural diversity lies in the nature of the substituent linked to the 9 position of the purine ring. For 28 of these, we have determined values of K_i^MESG, these vary from 1.7 to 67 500 nM, a factor of about 40 000-fold. Values of K_i^ino for the complete set of 29 inhibitors vary from 4.0 nM to about 24 000 nM, a factor of 6 000-fold. For many of the inhibitors examined, values of K_i determined with MESG as substrate are smaller than those obtained employing inosine as substrate by a factor that varies from less than two to ten. It has been established through structural studies as well as kinetic characterization that these are competitive inhibitors,⁴ a result confirmed in the present work. Substrate-dependent values of K_i for a set of competitive inhibitors is a highly surprising result, for which we know of no precedent. Since values of K_i are equilibrium constants for dissociation of E·I complexes, this results demands that E is in some sense different when MESG and inosine are employed as substrates. Definition of the precise basis for substrate-dependent values of K_i for competitive inhibitors will require detailed further study. Values of K_i^ino and K_i^MESG for subsets of inhibitors were employed as training sets to create quantitative structure-activity relationships (QSAR) which have substantial power to predict values of K_i for inhibitors outside the training set. The results suggest that the current QSAR model developed has substantial power to predict values of K_i for novel, potent 9-substituted-9-deazaguanines and useful power to predict such values for reasonably potent outside this structural class. These QSAR models should be useful in guiding future medicinal chemistry efforts designed to discover inhibitors of PNP having increased potency. Creation of a more robust QSAR model will require greater structural diversity and a larger and a more uniform distribuiton of K_i values in the data set. Work to achieve these goals is underway.

References:

1. Hansch, C.; Hoekman, D.; Gao, H. Chem. Rev., 1996, 96, 1045-1075.

2. Farutin, V.; Masterson, L.; Andricopulo, A.D.; Cheng, J.; Riley, B.; Hakimi, R.; Frazer, J.W.; Cordes, E.H. J. Med. Chem., 1999, 42, 2422-2431.

3. Andricopulo, A.D., Yunes, R.A., Cechinel Filho, V., Nunes, R.J., Frazer, J.W., Cordes, E.H. Pharmazie, 1999, 54, 698-704.

4. Guida, W.C.; Elliott, R.D.; Thomas, H.J.; Secrist, J.A., III; Babu, Y.S.; Buff, C.E.; Erion, M.D.; Ealick, S.E.; Montgomery, J.A. J. Med. Chem., 1994, 37, 1109-114.

CAPES/BioCryst Pharmaceuticals