jm050494g_si_001.pdf (153.06 kB)
Download file

Biospectra Analysis:  Model Proteome Characterizations for Linking Molecular Structure and Biological Response

Download (153.06 kB)
journal contribution
posted on 03.11.2005, 00:00 by Anton F. Fliri, William T. Loging, Peter F. Thadeio, Robert A. Volkmann
Establishing quantitative relationships between molecular structure and broad biological effects has been a long-standing goal in drug discovery. Evaluation of the capacity of molecules to modulate protein functions is a prerequisite for understanding the relationship between molecular structure and in vivo biological response. A particular challenge in these investigations is to derive quantitative measurements of a molecule's functional activity pattern across different proteins. Herein we describe an operationally simple probabilistic structure−activity relationship (SAR) approach, termed biospectra analysis, for identifying agonist and antagonist effect profiles of medicinal agents by using pattern similarity between biological activity spectra (biospectra) of molecules as the determinant. Accordingly, in vitro binding data (percent inhibition values of molecules determined at single high drug concentration in a battery of assays representing a cross section of the proteome) are useful for identifying functional effect profile similarity between medicinal agents. To illustrate this finding, the relationship between biospectra similarity of 24 molecules, identified by hierarchical clustering of a 1567 molecule dataset as being most closely aligned with the neurotransmitter dopamine, and their agonist or antagonist properties was probed. Distinguishing the results described in this study from those obtained with affinity-based methods, the observed association between biospectra and biological response profile similarity remains intact even upon removal of putative drug targets from the dataset (four dopaminergic [D1/D2/D3/D4] and two adrenergic [α1 and α2] receptors). These findings indicate that biospectra analysis provides an unbiased new tool for forecasting structure−response relationships and for translating broad biological effect information into chemical structure design.