A molecular guidance system based upon target genes, nuclear receptors and ligands applied to drug discovery and prediction of toxicity.

A molecular guidance system useful in drug design is described in which nuclear receptors position ligands in intercalation sites in responsive genes. Evidence is based upon positions of agonists in receptors and the transcriptional activity of a designed estrogen that is 3 times more potent than the steroid hormone estradiol.

Small molecule intercalation with double stranded DNA: implications for normal gene regulation and for predicting the biological efficacy and genotoxicity of drugs and other chemicals.

The binding of small molecules to double stranded DNA including intercalation between base pairs has been a topic of research for over 40 years. For the most part, however, intercalation has been of marginal interest given the prevailing notion that binding of small molecules to protein receptors is largely responsible for governing biological function. This picture is now changing with the discovery of nuclear enzymes, e.g. topoisomerases that modulate intercalation of various compounds including certain antitumor drugs and genotoxins. While intercalators are classically flat, aromatic structures that can easily insert between base pairs, our laboratories reported in 1977 that a number of biologically active compounds with greater molecular thickness, e.g. steroid hormones, could fit stereospecifically between base pairs. The hypothesis was advanced that intercalation was a salient feature of the action of gene regulatory molecules. Two parallel lines of research were pursued: (1) development of technology to employ intercalation in the design of safe and effective chemicals, e.g. pharmaceuticals, nutraceuticals, agricultural chemicals; (2) exploration of intercalation in the mode of action of nuclear receptor proteins. Computer modeling demonstrated that degree of fit of certain small molecules into DNA intercalation sites correlated with degree of biological activity but not with strength of receptor binding. These findings led to computational tools including pharmacophores and search engines to design new drug candidates by predicting desirable and undesirable activities. The specific sequences in DNA into which ligands best intercalated were later found in the consensus sequences of genes activated by nuclear receptors implying intercalation was central to their mode of action. Recently, the orientation of ligands bound to nuclear receptors was found to match closely the spatial locations of ligands derived from intercalation into unwound gene sequences suggesting that nuclear receptors may be guiding ligands to DNA with remarkable precision. Based upon multiple lines of experimental evidence, we suggest that intercalation in double stranded DNA is a ubiquitous, natural process and a salient feature of the regulation of genes. If double stranded DNA is proven to be the ultimate target of genomic drug action, intercalation will emerge as a cornerstone of the future discovery of safe and effective pharmaceuticals.

DNA intercalative potential of marketed drugs testing positive in in vitro cytogenetics assays.

We have previously noted that the Physicians' Desk Reference (PDR) contains over 80 instances in which a drug elicited a positive genotoxic response in one or more in vitro assays, despite having no obvious structural features predictive of covalent drug/DNA interactive potential or known mechanistic basis. Furthermore, in most cases, these drugs were "missed" by computational genotoxicity-predicting models such as DEREK, MCASE and TOPKAT. We have previously reported the application of a V79 cell-based model and a 3D DNA docking model for predicting non-covalent chemical/DNA interactions. Those studies suggested that molecules that are very widely structurally diverse may be capable of intercalating into DNA. To determine whether such non-covalent drug/DNA interactions might be involved in unexpected drug genotoxicity, we evaluated, using both models where possible, 56 marketed pharmaceuticals, 40 of which were reported as being clastogenic in in vitro cytogenetics assays (chromosome aberrations/mouse lymphoma assay). As seen before, the two approaches showed good concordance (62%) and 26 of the 40 (65%) drugs exhibiting in vitro clastogenicity were predicted as intercalators by one or both methods. This finding provides support for the hypothesis that non-covalent DNA interaction may be a common mechanism of clastogenicity for many drugs having no obvious structural alerts for covalent DNA interaction.

Transcriptional inhibition of the estrogen response element by antiestrogenic piperidinediones correlates with intercalation into DNA measured by energy calculations.

The energy of interaction of antiestrogenic ligands bound to DNA derived from molecular modeling was compared to the capacity of the ligands to directly inhibit the transcriptional activity of an estrogen responsive gene. 3-Phenylacetylamino-2,6-piperidinedione (A10) and related compounds were intercalated into a partially unwound DNA site in a canonical estrogen response element (ERE). The piperidinedione/ERE complexes were subjected to energy minimization and the strength of interaction of the ligands with the DNA was measured. The ability of the ligands to inhibit transactivation was assessed using a reporter gene constructed with the ERE of the vitellogenin gene promoter (ERE(v)-tk-Luc) transiently transfected into the human estrogen receptor-positive MCF-7 breast cancer cell line. The results demonstrate a direct correlation between the calculated energetic fit of the compounds in the ERE and inhibition of ERE(v) transactivation. The order of potency of the compounds to suppress estrogen-dependent reporter gene activity was identical to that previously shown for inhibiting the growth of MCF-7 cells. To our knowledge, these results provide the first direct experimental evidence that the predicted fit of a class of compounds into a defined DNA binding site correlates with the ability of the compounds to modulate specific gene functions regulated at that site.

Toward a greater appreciation of noncovalent chemical/DNA interactions: application of biological and computational approaches.

Noncovalent DNA interactions, e.g., DNA intercalation and DNA groove-binding, have not been well studied relative to covalent interactions largely due to the inability of predicting and detecting such events in intact cells. We have adapted an in vitro bleomycin amplification method for DNA intercalation for use in cultured V79 Chinese hamster cells and have validated this approach through the use of a three-dimensional DNA computational docking model that quantifies potential strength of DNA intercalative binding based on electrostatics and hydrogen bonding. For many structural classes of molecules, DNA intercalation is necessary but not sufficient for genotoxicity. The present article reviews our progress to date in predicting and confirming noncovalent binding of drugs and other chemicals and in understanding the mechanistic relationship between intercalation and genotoxicity.

Evaluation of DNA intercalation potential of pharmaceuticals and other chemicals by cell-based and three-dimensional computational approaches.

To what extent noncovalent chemical-DNA interactions, in particular weak nonbonded DNA intercalation, contribute to genotoxic responses in mammalian cells has not been fully elucidated. Moreover, with the exception of predominantly flat, multiple-fused-ring structures, our ability to predict intercalation ability of novel compounds is nearly completely lacking. Computational programs such as DEREK and MCASE recognize primarily those molecules that can form irreversible covalent adducts with DNA since their learning sets, for the most part, have not been populated by compounds for which a relationship between noncovalent interaction and genotoxicity exists. We describe here a novel three-dimensional (3D) computational DNA-docking model for prediction of DNA intercalative activity of molecules with both classical and nonclassical intercalating structures. The 3D docking results show a remarkable concordance with results obtained from testing these molecules directly in the Chinese hamster V79 cell-based bleomycin amplification system suggesting that either or both of these approaches may have utility in defining noncovalent chemical-DNA interactions. The ability to predict and/or demonstrate cellular DNA intercalation of novel molecules may well provide fresh insights into the nature and mechanistic basis of structurally unexpected genotoxicity observed during safety testing.

Androgenic Activity of Antiandrogens Predicted by Fit into DNA.

Computer modeling including graphics and energy calculations were employed for the first time to examine the stereochemical fit of antiandrogens into double-stranded DNA. In this study, we assessed the relative fit of antiandrogens in the cavity between base pairs known to accommodate androgens. When compared to testosterone which was given a normalized value of 100%, the antiandrogens manifested the following order of fit: RU23908 (88%) > hydroxyflutamide (71%) > cyproterone acetate (41%). A correlation was observed between the relative fit of the antiandrogens and reported agonistic properties as assessed by the ability to increase nuclear androgen receptor levels in the rat ventral prostate. These findings may be useful in the design and development of androgen antagonists without agonistic activity.

Conjugated linoleic acid blocks estrogen signaling in human breast cancer cells.

Conjugated linoleic acid (CLA), a mixture of positional and geometric isomers of linoleic acid found in dairy products and meat from ruminants, has been widely shown to possess anticarcinogenic activity against breast cancer both in vitro and in animal models. However, little information is available concerning the mechanisms of the antitumor effects of these compounds. In this study, we investigated whether CLA has direct antiestrogenic activity in estrogen receptor positive (ER+) breast cancer cells. Treatment of the ER+ cell line, MCF-7, with 5 purified CLA isomers as well as "mixed" CLA showed a dose-dependent growth inhibition with the 9cis,11cis and 9cis,11trans being the most and least potent isomers, respectively. In assessing effects on a number of variables that play obligatory roles in the estrogen signaling pathway, we determined that CLA treatment downregulated ERalpha expression at both mRNA and protein levels and decreased binding activity of nuclear protein to a canonical estrogen response element (ERE(v)). Using a reporter gene construct (ERE(v)-tk-Luc) that was transiently transfected into MCF-7 cells, we also demonstrated inhibition of promoter activity by CLA that was directly mediated by blockage of activity through the ERE. The results indicated that the order of potency of the CLA isomers for inhibiting activation of ERE(v) was similar to that demonstrated for their antiproliferative activity on MCF-7 cells. Taken together, these findings demonstrate that CLA compounds possess potent antiestrogenic properties that may at least partly account for their antitumor activity on breast cancer cells.