Correlation between in vitro and in vivo activities of Hsp90 inhibitors

5636

Hsp90 is a protein chaperone that acts to stabilize and/or activate a number of proteins required for cellular signaling pathways, including several known determinants of carcinogenesis such as Her2 (ErbB2), B-Raf, c-Kit, Flt-3, and Raf-1. Hsp90 in cells is complexed with various cochaperones and client proteins and is known to undergo a variety of post-translational modifications that regulate its activity, including phosphorylation, acetylation, nitrosylation, and sumoylation. Hsp90 thus represents a complex target for drug discovery and development. We report here the results of experiments designed to investigate the correlation between in vitro and in vivo activities of a set of non-benzoquinone Hsp90 inhibitors spanning a range of potencies. While each molecule shows tight binding to the purified N-terminal domain of Hsp90α (K_d < 10 nM), their growth inhibitory potency varies widely and does not correlate with binding affinity. In vitro cell uptake experiments reveal that cellular accumulation is explainable using a simple equilibrium binding model with the cellular concentration of Hsp90 and the growth inhibition IC₅₀ as parameters. Further, the growth inhibitory potency of these compounds correlates with rates of ErbB2 degradation and Hsp70 induction in SKBr3 cells. These data strongly support the idea that cellular growth inhibition (IC₅₀) is a better measure of the thermodynamic stability of cellular Hsp90 complexes than is binding to purified Hsp90α N-terminal domain. Differences in the properties of the various cellular Hsp90•inhibitor complexes are noted even between apparently equipotent inhibitors. KOS-2539 (IC₅₀ = 140 nM in MCF-7) and KOS-2484 (IC₅₀ = 120 nM in MCF-7) are equipotent in standard 72-hour cytotoxicity assays, and these compounds show the expected behavior in in vitro cell uptake experiments, yet KOS-2484 is highly efficacious in mouse xenograft models whereas KOS-2539 is ineffective. PK/PD measurements using mouse HT-29 xenografts demonstrate that KOS-2484 accumulates in tumor tissue and is retained with a half-life of ~30 hours, whereas KOS-2539 is rapidly cleared. Consistent with these in vivo results, KOS-2484 was found to be well-retained in vitro in HT-29 cells up to 24 hours after a transient exposure to the drug, whereas KOS-2539 was >60% eliminated within 24 hours. Thus, while equipotent given long-term exposure, KOS-2484 and KOS-2539 appear to form complexes with cellular Hsp90 that differ greatly in kinetic stability. Application of this assay to MRC-5 mouse fibroblasts shows that both KOS-2484 and KOS-2539 are rapidly eliminated from this normal cell line, illustrating a difference in the kinetic stabilities of Hsp90•inhibitor complexes in normal and tumor cells. These data suggest structural differences between the Hsp90•inhibitor complexes in normal and tumor cells that manifest as differences in kinetic rather than thermodynamic stability.