Printer Friendly

Quantitative structure-activity relationships in chemistry.

Developments in this area are traced through to current day uses.

Interest in quantitative structure-property and structure-activity relationships (QSARs) in chemistry dates back to the works by Dmitri Mendeleev, John Newlands, and Lothar Meyer, who developed theories ordering the chemical elements into periods and groups which allowed the prediction of the existence of the then undiscovered elements scandium and gallium. At the turn of the century, Overton [1] published systematic studies on the narcotic effects of congeneric series of alcohols, hydrocarbons, aldehydes, ketones, nitriles, esters, and nitroparaffins on a variety of aquatic species, notably tadpoles of the frog species Rana temporia. Among his discoveries was that the narcotic potency of such series of substances increases with the length of the carbon chain - up to a limit - or, by replacing halogen substitutents, in the order chlorine [less than] bromine [less than] iodine.

A major stimulus to the field came in the 1930s with Hammett's equation relating rate and equilibrium constants to the substituents in meta or para position of the active moiety on a benzene ring, such as hydrolysis rates of esters [2]. In the 1960s, Swain and Lupton solidified and advanced Hammett's work by calculating field and resonance components for the substituents [3]. Thus, the electronic effects of substituents were quantitatively recognized and they are still in wide use for the prediction of the effects of untested substances. In the 1950s, Taft and coworkers successfully correlated the volume of substituents with hydrolysis rate constants, thus recognizing the steric component of QSARs [4]. Verloop and coworkers later advanced Taft's contribution by developing multi-dimensional substituent parameters.

The third, and probably most influential contribution to the field of QSAR came from the quantitative description of hydrophobicity or lipophilicity as independent parameter in quantitative structure-activity relationships. Already at the turn of the century, Meyer and Overton recognized the relationship of fat solubility with narcotic potency of their chemicals. Since then, a variety of two-phase systems has been used to measure water/membrane equilibrium constants, but the n-octanol/water system has proven to be most useful and lipid-like. Hansch and Leo developed the field in great depth and published the first comprehensive collation of octanol/water partition coefficients (commonly abbreviated as logP or log[K.sub.ow]), and their underlying electronic, steric, and hydrophobic substituent constants [5].

Parallel with these developments, computational power and methodologies in chemistry advanced greatly. A major development was brought on by Weininger in the 1980s when he developed the SMILES (simplified molecular line entry system) notation for chemicals [6]. Since then, several companies have developed personal computer (PC) programs which allow the two-dimensional projection, search for substructures, and computation of physico-chemical properties from such SMILES strings or from the two-dimensional molecular structure graph. For example, there are now PC programs available to compute molar refractivity, molar volume, parachor, index of refraction, surface tension, density, dielectric constant, polarizability, dipole moment, acid constant, octanol/water partition coefficient, and other physicochemical properties directly from the SMILES string or molecular structure graph. Other PC programs predict environmental fate and effects, such as degradation rates in air and soil, aquatic bio-concentration factors and acute toxicity to aquatic organisms. In the pharmaceutical and health fields, models exist to calculate the potency of substances to cause skin sensitisation, irritancy, corrosivity, lachrymation, methaemoglinaemia, mutagenicity, carcinogenicity, teratogenicity, and many other effects. With the rapidly increasing computing power of PCs, applications which were exclusively the domain of mainframes and workstations, are now also becoming available for PCs. For example, molecular dynamics programs are computing torsion angles, conformational energy minima, ionization potentials, and derived properties.

Many of the models predicting the magnitude of a biological effect rely heavily on the computation of octanol/water partition coefficients which then are used to predict the derived activities as a linear function of logP, or related parameters, such as the connectivity indices developed by Kier and Hall [7]. This works generally well with sets of congeneric molecules, such as esters with increasing chain length, but fails when there is a change of toxicological mechanism and non-congenericity. More recently, non-linear models, frequently employing expert systems, genetic algorithms, or neural networks, are being used to develop more generalized models which can handle the large variety of chemical structures, functional groups, and mechanisms of action possible. Naturally, each system has strengths and weaknesses; a useful evaluation and review of the major computational expert systems has recently been published [8].

At the National Water Research Institute, we have been using traditional QSAR methods [9], inter-species correlations [10], and most recently, neural network algorithms to model the toxicity of chemicals to aquatic organisms, such as the fathead minnow (Pimephales promelas) and the luminescent bacterium Vibrio fischeri. Both organisms have sizable numbers of measured data for all kinds of substances and neural networks can use non-linear relationships between structural properties of chemicals and their effects, such as changes in mechanism of action. For example, the acute toxicity of over 800 chemicals with a large variety of structures, was modelled successfully with a set of approximately 50 simple structural descriptors using a probabilistic neural network [11]. Moreover, we found the octanol/water partition coefficient not to be a useful input parameter, thus eliminating the need for its computation. This type of modelling is of interest for the prediction of environmental and health effects of the 66,000 chemicals in commerce, most of which lack any such measured data.

Over the last two decades, the development of many new and powerful drugs has been made possible by applying classical QSAR methods to the effect-optimization process for new principal-structure compounds. The application of such QSAR methods is likely to continue in future. For the development of new structures and deeper insight into chemicals-enzyme interactions, advanced algorithms, expert systems, and 3-dimensional structure analysis, including computation of conformations, steric and electronic energy minima and maxima, of both the modelled substance and the receptor molecules are now driving the development to new achievements [12], but the importance of reliable and compatible experimental data as a basis must not be overlooked. The wise use of models will reduce the need for animal testing and contribute to the development of new desirable and useful substances. QSAR is an exciting field to work in, 100 years ago and now.


1. Lipnick, R.L., 'Charles Ernest Overton: Narcosis studies and a contribution to general pharmacology', Trends Pharmaceut. Sci., 7:161, 1986.

2. Hammett, L.P., Chem. Rev., 17:125, 1935.

3. Swain, C.G. and E.C. Lupton, JACS, 90:4328, 1968.

4, Taft, R.W., JACS, 81:5343, 1959.

5. Hansch C. and A.J. Leo, Substituent Constants for Correlations in Chemistry and Biology, John Wiley & Sons, NY, 1979.

6. Weininger, D.J., Chem. Inf. Comput. Sci., 28:31, 1988.

7. Kier, L.B and L.H. Hall, Molecular Connectivity in Structure-Activity Analysis, Research Studies Press, Letchworth, UK, 1986.

8. Dearden, J.C, et al., ATLA, 25:223, 1997.

9. Kaiser, K.L.E. (ed.)., QSAR in Environmental Toxicology -- II, D. Reidel Publ. Co., Dordrecht, NL, 1986.

10. Kaiser, K.L.E., Environ. Health Persp., suppl. 2, 106:583, 1998.

11. Kaiser, K.L.E. and S.P. Niculescu, Chemosphere, in press; see also

12. Kubinyi, H., Folkers, G., and Y.C. Martin, eds., 3D QSAR in Drug Design, Vol. 3, Kluwer Academic Publ., NL, 1998.

Website References

The QSAR and Modelling Society:

Some commercial sites of related interest:

Klaus L.E. Kaiser, FCIC is a research scientist with Environment Canada's Centre for Inland Waters in Burlington, ON since 1972. He holds a PhD from the Technical University of Munich, Munich, Germany. He is currently Director, Environment and Rubber Chemistry Divisions on the Canadian Society for Chemistry's Board of Directors. Kaiser also serves as Division Coordinator on that Board. He is also editor-in-chief of the Water Quality Research Journal of Canada, website at
COPYRIGHT 1999 Chemical Institute of Canada
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 1999 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Kaiser, Klaus L.E.
Publication:Canadian Chemical News
Geographic Code:1USA
Date:Jan 1, 1999
Previous Article:Water based coatings.
Next Article:Lobbying the federal government on your behalf.

Terms of use | Privacy policy | Copyright © 2020 Farlex, Inc. | Feedback | For webmasters