Neal Rosen: Small Molecule Inhibitors of Hsp90

Computer model of PU24F-Cl bound to the N-terminal of Hsp90beta. Enlarge Image Computer model of PU24F-Cl bound to the N-terminal of Hsp90beta.

Addition of ansamycins to cancer cells induces the proteasomal degradation of a small subset of proteins involved in signal transduction (i.e., steroid receptors, Raf kinase, Akt, certain transmembrane tyrosine kinases). The clinical development of ansamycins has been impeded by their toxicity and in vivo instability. Efforts to improve the therapeutic index of these molecules have been plagued by the difficult task of chemically modifying the compounds. A geldanamycin derivative 17-AAG has less liver toxicity than the parent compound and is now in a Phase I clinical trial. The compound is difficult to formulate, and no effective synthetic methodology for its construction is available. The fundamental problem with these compounds lies in their difficultly to modify the skeleton, thus making it hard to make improvements directed towards selectivity.

In our initial studies we, in collaboration with Dr. Gabriela Chiosis, have designed structures to bind the Hsp90 alpha ATP/ADP-binding site. The first designed derivative, PU3, showed a good theoretical fit and fulfilled all of the important interactions with the protein pocket. The ‘rule of 5’ (of Lipinsky) predicts that this molecule should have acceptable pharmacokinetics. PU3 has one hydrogen bond donor, the sum of Ns and Os is 8; has a molecular weight of 371; and a Mlog P-value of 2.2. This suggests that PU3 has ‘drug-like’ properties, and is a reasonable starting point for the development of orally available drugs. The lab has studied PU3 in more detail. It was determined to bind Hsp90 with a relative affinity of 15 to 20mM. Addition of PU3 to breast cells caused the degradation of Her2, estrogen receptor, and Raf, and caused G1 arrest of cellular proliferation, RB hypophosphorylation, and loss in D-cyclin expression. It also caused the differentiation of breast cancer cells. It thus has many of the biologic properties of ansamycins.

PU3 was further modified to increase its affinity to Hsp90 by enhancing its interaction with the hydrophobic side pockets of the ATP/ADP pocket. Seventy such compounds were synthesized by rationally modifying the PU3-scaffold. Their biological activity was evaluated, and structure activity relationships determined. One such compound, PU24FCl, was found to bind Hsp90 with 30-fold higher affinity than PU3, an affinity approximating that of 17AAG (0.5mM). PU24FCl induced the growth arrest and differentiation of the tested breast cancer cell lines with an IC50 of 1-5uM. Addition of PU24FCl to breast cancer cells leads to selective degradation of Her2, Akt, and Raf at concentrations that correlate with the inhibition of these cells. This compound was administered to MCF-7 xenograft mice, and a dose-dependent decline in Her2 expression was observed at doses of PU24FCl that were nontoxic to the animals. The ability of these synthetic molecules to induce growth arrest is similar for the cell lines tested and correlates with their Hsp90 binding potency. This observation points to a general role of Hsp90 in maintaining the malignancy in most cancer cells and to the importance of this target in developing anticancer therapies.

The skeleton of this class of molecules is easily amenable for derivatization, and one would predict that further modifications would bring activity in the nanomolar range. Additionally, the molecules are water soluble at the tested concentrations and are amenable for oral administration. Such molecules could have considerable advantages as drugs including the fact that oral availability would eliminate the inconveniences of prolonged intravenous administration.