HSP90 and GRP94 are homologous cellular chaperones found in cytosol and endoplasmic reticulum, respectively. Several years ago, we discovered that members of the benzoquinone ansamycin class of antibiotics, including herbimycin A and geldanamycin (GA) bound to HSP90 and GRP94 and disrupted certain multi-molecular complexes of which these proteins were a part. We have utilized pharmacologic disruption of HSP90 and GRP94 activity to study the function of these chaperones in cellular signal transduction. Multiple signal transduction proteins interact with these charperones, including the kinases src, erbB2, c-raf-1, Akt, Kit, Met, Bcr-Abl, the transcription factor HIF-1alpha, and mutated (but not wild type) p53. A general consequence of pharmacologic disruption of the chaperone/signal protein complex is the resultant marked instability and incorrect subcellular localization of the signalling protein. The instability is due to stimulation of targeted degradation of the signalling protein by the 26S proteasome proteolytic complex following chaperone dissociation. We made the novel observation that HSP90 associates with the cytosolic kinase RIP, a key component of the TNF signalling pathway which leads to NFkB activation. We have determined that disruption of RIP stability by geldanamycin prevents NFkB induction by TNF, but not TNF signalling to Jnk, thus sensitizing cells to the apoptotic properties of TNF. We have additionally observed that another kinase associated with cell survival, Akt, is sensitive to geldanamycin. Geldanamycin blocks NFkB induction by a wide variety of stimuli other than TNF, including chemotherapeutic drugs and IL-1. Its ability to do this may relate to its destabilization of Akt. Benzoquinone ansamycins (geldanamycin) had been the only agents capable of specifically interfering in HSP90/GRP94 function. Recently, we identified radicicol as representing a novel class of natural product capable of binding to HSP90. Both radicicol and the ansamycins bind to HSP90 at an amino terminal nucleotide pocket. Most recently, we have identified a third class of natural product, novobiocin, which also binds to HSP90, although at a lower affinity than either benzoquinone ansamycins or radicicol. Nonetheless, novobiocin appears to cause the same biologic effects on """"""""client proteins"""""""" as ansamycins and radicicol. Surprisingly, novobiocin appears to interact with a carboxyl terminal region on HSP90, which is in fact a previously unrecognized second nucleotide binding site. Preliminary animal testing has revealed no toxicity after twice daily administration of novobiocin for more than one month. This regimen demonstrates significant anti-tumor activity in a transgenic murine model of erbB2-driven breast cancer. We have observed that geldanamycin reverses beta-catenin tyrosine phosphorylation in melanoma cells, probably due to the rapid loss of erbB2 from these cells. In untreated cells, erbB2 and beta-catenin can be readily co-precipitated. Loss of beta-catenin tyrosine phosphorylation leads to an increased association with E-cadherin and decreased cell motility in vitro. This is the first indication that modulation of the tyrosine phosphorylation status of beta catenin in melanoma cells is associated with decreased motility. The fact that beta-catenin tyrosine phosphorylation seems to be mediated, in 3/3 melanoma cell lines examined, by erbB2 - a geldanamycin-sensitive tyrosine kinase - suggests that geldanamycin treatment may be anti-metastatic. This hypothesis is currently being tested in an in vivo metastasis model. The ErbB family of receptor tyrosine kinases contains four members. We have found that ErbB2, the only ligandless member of the family, is one of the most sensitive geldanamycin substrates. Since the ErbBs are transmembrane proteins, they are likely to come in contact with both Hsp90 and Grp94, and one or both chaperones may be responsible for ErbB2's geldanamycin sensitivity. Our current data demonstrate that the kinase domain of ErbB2 mediates its geldanamycin responsiveness and that Hsp90 binds to this domain in the mature protein. In contrast, mature ErbB1, much less sensitive to geldanamycin than ErbB2, does not associate with Hsp90. Geldanamycin-induced instability of nascent ErbB2 is also mediated by its kinase domain, and we can find little evidence to support a role for Grp94 in ErbB2 maturation. We recently identified the novel E3 ubiquitin ligase Chip as being recruited to ErbB2-associated chaperone complexes in the presence of geldanamycin. Chip mediates geldanamycin induced ErbB2 ubiquitination, which is necessary for, and precedes, its degradation by the proteasome. We have recently identified the site within the kinase domain of ErbB2 at which Hsp90 binds and we have proposed a model to explain remodeling of ErbB2-associated chaperone complexes in the presence of geldanamycin and other Hsp90 inhibitors. We have demonstrated that combination of low doses of geldanamycin and a proteasome inhibitor currently in clinical trial increases the toxicity toward tumor cells compared to that observed with each agent alone. We showed that this property was unique to tumor cells, in that the combination was not toxic to non-transformed cells at the concentrations tested. Further, we proposed a model to explain these results that invokes proteasome overload and deposition in the cell of insoluble proteins, leading to initiation of apoptosis. These data have led to initiation of a phase I combination clinical trial of an Hsp90 inhibitor and a proteasome inhibitor, in collaboration with investigators at the Mayo Clinic. We demonstrated that the trans-membrane endoplasmic reticulum signaling proteins, IRE1 and PERK, are Hsp90 client proteins and are thus sensitive to Hsp90 inhibitors. These data were the first to link Hsp90 to proper function of the unfolded protein response - the stress response as it occurs in the endoplasmic reticulum. The data further underline the crucial role of Hsp90 in allowing cells to survive stressful stimuli. We recently demonstrated that HIF-1alpha, a transcription factor whose expression is upregulated by hypoxia and loss of VHL, is an Hsp90 client protein and is sensitive to Hsp90 inhibition. These data were the first to identify an oxygen- and VHL-independent pathway regulating HIF stability - a pathway that is amenable to pharmacologic manipulation using Hsp90 inhibitors. These data have led to initiation, in the Urologic Oncology Branch, of a phase II clinical trial of benzoquinone ansamycins in clear cell kidney cancer (lacking VHL and expressing HIF-1alpha at constitutively high levels). We identified kinase-mutated KIT protein to be sensitive to Hsp90 inhibition. Kinase-mutated KIT is resistant to imatinib and other KIT inhibitors currently in the clinic. Kinase-mutated KIT is characteristic of mastocytosis and mast cell leukemia. Based on our findings, a phase II study of an Hsp90 inhibitor to treat mastocytosis has been initiated as an NCI/NIAID/Mayo Clinic collaborative trial. We demonstrated that MET and mutated MET protein, characteristic of papillary renal cell cancer, is sensitive to Hsp90 inhibition. Based on these findings, a phase II clinical trial of an Hsp90 inhibitor in papillary renal cell cancer is being initiated in the Urologic Oncology Branch.
Showing the most recent 10 out of 62 publications
