Investigation of structural and electronic biases in mutagenic compounds

The use of high throughput screening as the main means to discover lead compounds continues to require the design of libraries for screening. In this work some insights that may permit building chemical libraries that are biased toward compounds that are less likely to be mutagenic are shown. Toward this goal, the study of the distribution of substructural and electronic properties found in compounds that have been reported to be mutagens and nonmutagens is discussed. Mutagenic data retrieved for each of four Salmonella strains, TA-98, TA-100, TA-1535, and TA-1537, were compiled and divided depending on whether the compounds underwent metabolic activation. For each of the eight compound databases, an exhaustive enumeration of all molecular substructures within it was carried out. Comparison of the fragment distributions in sets of molecules that were Ames positive to a set of compounds found to be negative and vice versa allowed the deduction of moieties that are statistically overrepresented in active versus inactive or inactive versus active compounds for each strain and metabolic state. Our results show that compounds containing nitro functionalities are overrepresented in the Ames test in all strains whether the compounds had been metabolically activated or not. However, presence of a nitro functionality does not ensure that the compounds being Ames positive. Therefore, we computed a series of steric and electronic properties for 270 nitroaromatic compounds using semiempirical techniques. The electron affinity of the compounds offers a statistically significant discrimination between positive and negative compounds. © 2002 Wiley Periodicals, Inc. Int J Quantum Chem 88: 107–117, 2002