Targeting of multiple signalling pathways 90 molecular chaperone inhibitors

The last decade has seen the molecular chaperone heat shock protein 90 (HSP90) emerge as an exciting target for cancer therapy. This is because HSP90 is involved in maintaining the conformation, stability, activity and cellular localisation of several key oncogenic client proteins. These include, amongst others, ERBB2, C-RAF, CDK4, AKT/PKB, steroid hormone receptors, mutant p53, HIF-1a, survivin and telomerase hTERT. Therefore, modulation of this single drug target offers the prospect of simultaneously inhibiting all the multiple signalling pathways and biological processes that have been implicated in the development of the malignant phenotype. The chaperone function of HSP90 requires the formation of a multichaperone complex, which is dependent on the hydrolysis of ATP and ADP/ATP exchange. Most current inhibitors of HSP90 act as nucleotide mimetics, which block the intrinsic ATPase activity of this molecular chaperone. The first-in-class inhibitor to enter and complete phase I clinical trials was the geldanamycin analogue, 17-allylamino-17-demethoxygeldanamycin. The results of these trials have demonstrated that HSP90 is a valid drug target. Evidence of clinical activity has been seen in patients with melanoma, breast and prostate cancer. This article provides a personal perspective of the present efforts to increase our understanding of the molecular and cellular consequences of HSP90 inhibition, with examples from work in our own laboratory. We also review the discovery and development of novel small-molecule inhibitors and discuss alternative approaches to inhibit HSP90 activity, both of which offer exciting prospects for the future. Endocrine-Related Cancer (2006) 13 S125–S135 HSP90 as a cancer drug target As our understanding of the genetic and molecular biology of cancer has increased, there has been a shift over the last decade in the approaches used in the discovery of novel cancer therapeutics (Workman 2005). In contrast to the earlier development of cytotoxic agents, focus has moved to the development of treatments that target the pathways responsible for malignancy. Validation of this approach has been provided by the clinical activity and approval of smallmolecule kinase inhibitors such as imatinib (Gleevec), This paper was presented at the 2nd Tenovus/AstraZeneca Workshop, Cardiff (2006). AstraZeneca supported the meeting and the Welsh School of Pharmacy, Cardiff University has supported the publication of these proceedings. Endocrine-Related Cancer (2006) 13 S125–S135 1351–0088/06/013–S125 q 2006 Society for Endocrinology Printed in Grea gefitinib (Iressa) and erlotinib (Tarceva) as well as the therapeutic antibodies trastuzumab (Herceptin), cetuximab (Erbutix) and bevacizumab (Avastin). However, despite the success that these agents have enjoyed, it is likely that modulation of a single molecular target will be insufficient for optimal therapy (Workman 2003). Even where malignancies are driven by single genes or pathways, the development of resistance is a major concern. For example, resistance to imatinib has been shown to arise by acquisition of mutations within the kinase domain of BCR–ABL (Gorre et al. 2001, Shah et al. 2002). Furthermore, the majority of cancers involve multiple molecular abnormalities that are likely to be involved in malignant progression. These observations have reinforced the suggestion that inhibition of multiple targets will be required to cure t Britain DOI:10.1677/erc.1.01324 Online version via http://www.endocrinology-journals.org M V Powers and P Workman: Inhibition of the heat shock protein 90 most human cancers (Workman 2003). It is this concern which provides the foundation for the increasing amount of interest in targeting the heat shock protein 90 (HSP90) molecular chaperone (Workman 2004). HSP90 exerts its chaperone function to ensure the correct conformation, activity, intracellular localisation and proteolytic turnover of a range of proteins that are involved in cell growth, differentiation and survival (Maloney & Workman 2002, Whitesell & Lindquist 2005).Of particular importance is thatHSP90 is essential for the stability and the function of many oncogenic client proteins, which contribute to the hallmark traits of cancer (Fig. 1). These include ERBB2, BCR–ABL, AKT/PKB, C-RAF, CDK4, PLK-1, MET, mutant p53, HIF-1a, steroid hormone receptors (oestrogen and androgen), survivin and telomerase hTERT (Maloney & Workman 2002). Inhibition of HSP90 function has been shown to cause degradation of client proteins via the ubiquitin-proteasome pathway (Connell et al. 2001, Demand et al. 2001), which results in the simultaneous depletion of multiple oncoproteins, the combinatorial down-regulation of signals propagated through numerous oncogenic signalling pathways and modulation of all aspects of the malignant phenotype (Maloney & Workman 2002, Workman 2004). The ability to deliver a combinatorial effect through a single drug target may have promise in treating cancers driven by multiple molecular abnormalities and could also reduce the opportunity for resistance developing (Workman 2004, 2005). Figure 1 Schematic illustrating how inhibition of HSP90 may interfere with all of the six hallmark traits of cancer. Examples of client proteins involved in the various phenotypic aspects of malignancy are shown. S126 In this paper, we will focus on our interest in developing inhibitors of the HSP90 molecular chaperone family and the progress we have made in understanding the effects of this modulation in both the preclinical and clinical settings. Examples will be taken mainly from the work of our own laboratory. Presently, five isoforms of HSP90 have been identified, which differ in their cellular localisation. The two major cytoplasmic isoforms are HSP90a and HSP90b which share approximately 85% sequence identity at the protein level (Hickey et al. 1989, Gupta 1995). Other major isoforms are Grp94 in the endoplasmic reticulum (Argon & Simen 1999), TRAP1 in the mitochondrial matrix (Felts et al. 2000) and HSP90N, which has been suggested to be associated with cellular transformation via its association with RAF (Grammatikakis et al. 2002). It is currently believed that all the mammalian HSP90 isoforms described above share a similar overall structure, which comprises a C-terminal dimerisation domain, a middle domain that is implicated in client protein binding and also an N-terminal ATPase domain (reviewed in Pearl & Prodromou 2001, Prodromou & Pearl 2003), which is absent in HSP90N (Schweinfest et al. 1998). The chaperone activity of HSP90 is dependent on its transient N-terminal dimerisation, which stimulates the intrinsic and essential ATPase activity (Prodromou et al. 2000). This process is controlled by an orchestrated set of interactions with a range of accessory proteins referred to as co-chaperones (Pratt et al. 2004, Riggs et al. 2004; Fig. 2). Initially, client proteins interact with an HSP70/HSP40/HIP complex. The HSP70 and HSP90 chaperone systems are then linked by the adaptor protein HOP/p60, which interacts with the C-terminals of both HSP90 and HSP70 via its tetracopeptide repeat domain (Scheufler et al. 2000). HOP/p60 can only bind to ADP-bound-HSP90, which has an open conformation and a high affinity for hydrophobic substrates (Pratt et al. 2004). The present model for chaperone activity suggests that when HSP90 exchanges ADP for ATP, it undergoes a conformational change, which includes the transient dimerisation of the N-terminal domains (Prodromou et al. 2000). This leads to the dissociation of HSP70/HSP40/HIP and HOP, allowing the ATPdependent association of other co-chaperones (e.g. CDC37, p23 or immunophilins) to form the mature complex (reviewed in Whitesell & Lindquist 2005, Sharp & Workman 2006). CDC37 is involved specifically in the loading of kinase clients onto HSP90 (Roe et al. 2004) and p23 has recently been shown to stabilise HSP90 in the ATP-bound form, which extends the time in which HSP90 is in the www.endocrinology-journals.org Figure 2 Model of the HSP90 chaperone cycle for steroid hormone receptor client proteins and the effect of an HSP90 inhibitor. Endocrine-Related Cancer (2006) 13 S125–S135 conformation required for client protein activation (Ali et al. 2006). It is while the HSP90 chaperone cycle is in the mature state that the associated client protein becomes activated to either bind ligand (steroid hormone receptor) or be phosphorylated during signal transduction (AKT/PKB). Inhibition of ATP binding to HSP90 prevents the formation of the mature complex and results in the proteasome-dependent degradation of associated client proteins. This can occur by the recruitment of the E3-ubiquitin ligase, CHIP, which is a TPR protein that is able to interact with both HSP70 and HSP90 (Connell et al. 2001, Demand et al. 2001). Both of these molecular chaperones may be present in the immature complex, which has been stabilised by the presence of an inhibitor (Fig. 2). At first sight, HSP90 would not appear to be an obvious drug candidate for the design of novel cancer therapeutics. This is because it is not, to our knowledge, subject to mutation or amplification in cancer. It is, however, well-documented as being over-expressed in a range of human malignancies (Maloney & Workman 2002, Sreedhar et al. 2004). This may be a consequence of the hostile conditions created in tumour cells by the effects of deregulated oncogenes and tumour suppressor genes (many of which are HSP90 client proteins), along with the stressful microenvironmental features of solid tumours, which include nutrient deprivation, hypoxia and acidosis (Whitesell et al. 2003, Mosser &Morimoto 2004). Collectively, these factors may lead tumour cells to become highly stressed and much more reliant on HSP90 than cells from normal, non-malignant tissue (Whitesell & Lindquist 2005). This will increase the opportunity for therapeutic selectivity when HSP90 inhibitors are used clinically. This ‘stress hypothesis’ has been explored by Kamal et al. (2003) who www.endocrinology-journals.org demonstrated that HSP90 extracted from tumour cells exists in a high-affinity, activated super-chaperone complex which is approximately 100-fold more sensitive to HSP90 inhibitors when compared with the uncomplexed HSP90 isolated from normal cells (Kamal et al. 2003). In addition to the stressed nature of cancer cells, HSP90 inhibitors could exert therapeutic selectivity by exploiting multiple oncogene addiction and via the preferential dependence of certain oncoproteins on chaperoning by HSP90 (see later). It is known that the natural product geldanamycin exerts its antitumour effect by binding to the N-terminal ATPase domain of HSP90 to inhibit its chaperone function (Roe et al. 1999). The progress of geldanamycin into the clinic was stopped due to instability and the unacceptable hepatotoxicity seen at therapeutic doses during preclinical in vivo studies (Supko et al. 1995). Further analogues were developed for clinical use, which included 17-AAG (Schnur et al. 1995a,b). Preclinical studies with 17-AAG We have investigated the detailed molecular consequences of exposing cancer cells to 17-AAG in vitro in an attempt to identify genes and proteins that influence the sensitivity to HSP90 inhibitors. In addition, these studies have enabled us to identify and validate biomarkers of HSP90 inhibition, which could be of clinical use (Banerji et al. 2003, Maloney et al. 2003). It has been well documented that inhibition of HSP90 function induces the expression of HSP72 and degradation of client proteins (reviewed in Workman 2005, Clarke et al. 2006). We have demonstrated the principle of combinatorial inhibition of multiple signal transduction pathways via targeting HSP90 using a panel of human colon cancer cell lines (Hostein et al.