cardiovascular drugs that have been experimentally proved to work well on mouse were withdrawn because they caused adverse side effects in human.

Methods: In this study, we investigate whether binding patterns of withdrawn cardiovascular drugs are conserved

between mouse and human through computational dockings and molecular dynamic simulations. In addition, we also measured the level of conservation of gene expression patterns of the drug targets and their interacting partners by analyzing the microarray data.

Results: The results show that target proteins of withdrawn cardiovascular drugs are functionally conserved between human and mouse. However, all the binding patterns of withdrawn drugs we retrieved show striking difference due to sequence divergence in drug-binding pocket, mainly through loss or gain of hydrogen bond donors and distinct drug-binding pockets. The binding affinities of withdrawn drugs to their receptors tend to be reduced from mouse to human. In contrast, the FDA-approved and best-selling drugs are little affected.

Conclusions: Our analysis suggests that sequence divergence in drug-binding pocket may be a reasonable

explanation for the discrepancy of drug effects between animal models and human.