Ann M. Richard
Beyond Toxicity Prediction: SAR as a Means For Inquiry Into Mechanism
Environmental Carcinogenesis Division (Maildrop 68), US Environmental Protection Agency, Research Triangle Park, NC 27711, USA. E-mail: email@example.com
The term "SAR" in relation to toxicology has come to connote primarily development and use of SAR models for toxicity prediction. But by virtue of its implicit relationship to the reaction kinetics of a biological interaction, an SAR model can also serve as a valuable departure point for further research inquiry into molecular mechanisms. To achieve the twin goals of prediction and mechanistic insight, however, requires both empirical correlation and expression of the SAR problem in terms that relate to the underlying chemistry of the molecular interaction in the biological system. A series of examples will be presented, in brief, to illustrate many facets of the "practice" of SAR as they pertain to current studies of chemical toxicity, with emphasis placed on the questions that guide and emanate from SAR study.
The first example will consider the problem of an isolated chemical toxicity prediction from the standpoint, not of SAR model development, but of SAR model application. This particular example will focus on a problem of carcinogenicity prediction from two commercial programs -- TOPKAT and CASE -- that rely primarily on statistical algorithms. The predictions of such programs represent untested hypotheses in the absence of mechanistic data and efforts must be directed towards rationalizing the prediction at the local SAR level and placing it in the context of current knowledge of chemical and biological mechanisms. To assist in this process, the programs themselves offer a number of tools to allow a user to probe the basis for an SAR prediction, to construct analogy arguments, and to assess the utility of the prediction.
The second example will consider the problem of SAR model construction for a real-world example of a small, structurally diverse nasal toxicity data set. In a case such as this, where no prior SAR information exists and little is known of the mechanism of toxicity, very little guidance is provided in SAR model development. This example will serve to illustrate the types of questions that can be posed, SAR approaches that can be used, and results that can be achieved under these circumstances. The promise of SAR in this case is to impose some rational framework upon existing data, to generate useful hypotheses, and to possibly direct future experimental study.
Molecular modeling in conjunction with SAR study provides a means for bridging the gap between biological observation and molecular level information, offering a language for phrasing problems and answering questions in chemical reactivity terms. Whereas an empirically derived SAR usually provides a rather crude representation of the association between chemical structure and activity, when combined with key experimental data, such a model can serve as a departure point for future investigation into the underlying basis for a parameter association.
Two examples of hypothesis-driven, computational chemistry studies derived from empirical SAR associations will be presented to illustrate such an approach. In the first, computational methods were used to evaluate the relative energetics of two plausible glutathione mediated reaction pathways for mutagenic bioactivation of haloalkenes. The goal was to provide a molecular basis for an empirical SAR linking bromination with mutagenicity and to offer insight into the identity of the ultimate reactive intermediate(s) and its potential interaction with DNA. The second example will outline a current computational study probing the chemical reactivity mechanism underlying an SAR for the mutagenicity of MX (halooxyfuranones) and its analogues. These examples serve to illustrate how an empirical SAR, in conjunction with experimental data, can be used to constrain and define the computational problem. In addition, they illustrate how SAR and molecular modeling provide complementary tools for probing chemical mechanisms underlying biological activity.
This abstract does not reflect EPA policy.Back to Program Page