Signalling profile and antitumour activity of the novel Hsp90 inhibitor ...

Leukemia (2008) 22, 1604–1612 & 2008 Macmillan Publishers Limited All rights reserved 0887-6924/08 $30.00 www.nature.com/leu

ORIGINAL ARTICLE Signalling profile and antitumour activity of the novel Hsp90 inhibitor NVP-AUY922 in multiple myeloma T Stu¨hmer1, A Zo¨llinger1, D Siegmund2, M Chatterjee1, E Grella1, S Knop1, M Kortu¨m1, C Unzicker1, MR Jensen3, C Quadt4, P Che`ne3, J Schoepfer3, C Garcı´a-Echeverrı´a3, H Einsele1, H Wajant2 and RC Bargou1 1

Department of Internal Medicine II, Division of Hematology, University Hospital Wu¨rzburg, Wu¨rzburg, Germany; 2Department of Molecular Internal Medicine, Division of Hematology, University Hospital Wu¨rzburg, Wu¨rzburg, Germany; 3Oncology Research, Novartis Institutes for BioMedical Research, Basel, Switzerland and 4Exploratory Clinical Development Oncology, Novartis Pharma AG, Basel, Switzerland

We as well as others have recently shown that Hsp90 is overexpressed in multiple myeloma (MM) and critically contributes to tumour cell survival. Pharmacologic blockade of Hsp90 has consistently been found to induce MM cell death. However, most data have been obtained with MM cell lines whereas knowledge about the molecular effects of pharmacologic Hsp90 blockade in primary tumour cells is limited. Furthermore, these investigations have so far focused on geldanamycin derivatives. We analysed the biochemical effects of a novel diarylisoxazole-based Hsp90 inhibitor (NVP-AUY922) on signalling pathways and cell death in a large set of primary MM tumour samples and in MM cell lines. Treated cells displayed the molecular signature and pharmacodynamic properties for abrogation of Hsp90 function, such as downregulation of multiple survival pathways and strong upregulation of Hsp70. NVP-AUY922 treatment efficiently induced MM cell apoptosis and revealed both sensitive and resistant subgroups. Sensitivity was not correlated with TP53 mutation or Hsp70 induction levels and stromal cells from the bone marrow microenvironment were unable to abrogate NVP-AUY922induced apoptosis of MM cells. Thus, NVP-AUY922 may be a promising drug for treatment of MM and clinical studies are warranted. Leukemia (2008) 22, 1604–1612; doi:10.1038/leu.2008.111; published online 15 May 2008 Keywords: multiple myeloma; Hsp90; survival pathways; primary tumour cells

Introduction Multiple myeloma (MM) is a malignancy of the terminally differentiated B cell (plasma cell).1 Its main clinical manifestation, intramedullary myeloma entails extensive bone destruction and impairs haematopoiesis whereas tumour cells enjoy growthpromotive and anti-apoptotic support from the bone marrow (BM) microenvironment. Although novel drugs and new treatment regimens are currently being co-opted to increase the number and extent of remissions, the disease remains generally incurable and virtually all patients are eventually confronted with therapy-resistant tumour subclones.2 More and better treatment options are thus still required. Correspondence: Dr T Stu¨hmer, Department of Internal Medicine II, Division of Hematology, University Hospital Wu¨rzburg, Medical Clinic and Policlinic II, Haus C11, Josef-Schneider-Strasse 2, Wu¨rzburg, Bavaria 97080, Germany. E-mail: [email protected] Received 15 January 2008; revised 27 March 2008; accepted 1 April 2008; published online 15 May 2008

Recently, the heat shock proteins of 90 kDa (Hsp90a and Hsp90b) have emerged as attractive targets in cancer therapy. They are required for the stabilization of a large set of proteins, termed Hsp90 clients,3 of which a conspicuous number is involved in intracellular signalling cascades that promote proliferation and/or survival.4–6 They have also been implicated in the stabilization of anti-apoptotic proteins7 and in tumour angiogenesis.8 We and others have shown that abrogation of Hsp90 function, either through pharmacologic blockade or through siRNA-mediated knockdown, leads to apoptosis of MM cells.9–12 A prominent feature of Hsp90 inhibition in MM cell lines is downregulation of cytokine-triggered signalling pathways, such as those leading to activation of signal transducer and activator of transcription 3 (STAT3), mitogenactivated protein kinase (MAPK), protein kinase B (PKB/Akt) and nuclear factor-k B (NF-kB).9,10 Very little, however, is currently known about the molecular consequences of Hsp90 blockade in primary MM cells. Successful pharmacologic inhibition of Hsp90 has been achieved with the ansamycin antibiotic geldanamycin13 but the compound proved too toxic for medical use.14 Derivatives of geldanamycin, such as 17-allylamino-17-demethoxy-geldanamycin (17-AAG), 17-(dimethylaminoethylamino)-17-demethoxygeldanamycin (17-DMAG) and 17-allylamino-17-demethoxygeldanamycin hydroquinone hydrochloride (‘IPI-504’) have since been developed. The safety profiles of these have been thoroughly evaluated preclinically.15,16 Histopathological analysis of 17-AAG-treated animals demonstrated drug related lesions in liver, BM, thymus and caecum in the rat, as well as in liver, stomach, intestine, gallbladder, lymphoid tissues, BM and adrenal gland in the dog.16 However, preclinical data indicate that establishment of a therapeutic window for Hsp90 inhibition in myeloma may be achieved with these compounds,9–11 and a number of clinical Phase Ia/b trials have successfully been conducted with MM patients. In clinical trials with 17-AAG, hepatic toxicity has been described to be dose limiting, partially due to formulation issues.17 Different formulations, and combination therapies that employ 17-AAG and the proteasome inhibitor bortezomib are currently being evaluated in clinical Phase II/III and Phase III trials in myeloma (www.clinicaltrials. gov). Nevertheless, novel Hsp90 inhibitors with improved hepatotoxicity properties and convenient clinical formulations are desired. Here, we have analysed the pharmacologic and molecular effects of a new synthetic inhibitor of Hsp90, NVP-AUY922, on a large cohort of primary MM tumour samples and on established MM cell lines. The drug, which is based on a diarylisoxazole resorcinol structure,18 very effectively induced

Hsp90 inhibition in multiple myeloma T Stu¨hmer et al

1605 MM cell apoptosis in the large majority of samples. In primary and in cell line-derived MM cells alike the molecular signature of NVP-AUY922-mediated Hsp90 blockade included simultaneous downregulation of various signalling pathways relevant to myeloma biology. However, our analysis also revealed a good measure of heterogeneity between samples and suggests that resilience to Hsp90 inhibition is neither a result of failed survival pathway blockade, p53 mutation or upregulation of Hsp70.

Materials and methods

Hsp90 inhibitor NVP-AUY922 The chemical structure and properties of NVP-AUY922 have recently been described elsewhere.18 Stock solutions (10 mM) were prepared in H2O-free dimethyl sulphoxide (DMSO) and stored at 20 1C. Working dilutions were always freshly prepared.

Cell lines Human MM cell lines AMO-1, INA-6, KMS-12-BM, MM.1 s, MOLP-8, OPM-2, RPMI-8226 and U266 were maintained in RPMI 1640 medium (PAA Laboratories, Pasching, Austria), supplemented with 10% fetal bovine serum (FBS; Biochrom, Berlin, Germany), 100 U/ml penicillin, 100 mg/ml streptomycin (PAN Biotech, Aidenbach, Germany), 1 GlutaMAX-I (Invitrogen, Karlsruhe, Germany) and 1 mM Na-pyruvate (PAN Biotech). Cultures of INA-6 cells were grown in the presence of 2 ng/ml recombinant human interleukin-6 (IL-6). All cells were grown at 37 1C and 5% CO2. INA-6 was obtained from Prof. Gramatzki (University Clinic Schleswig-Holstein, Kiel, Germany), other MM cell lines were purchased at the German Collection of Microorganisms and Cell Cultures, Braunschweig, Germany.

Preparation and culture of primary cells BM aspirates from myeloma patients were subjected to Ficoll (Biochrom) density centrifugation. The mononuclear cell fraction was collected, washed with phosphate-buffered saline (PBS) and cold separation buffer (PBS containing 0.5% FBS and 2.5 mM EDTA), and incubated with 20 ml CD138 Microbeads (Miltenyi Biotech, Bergisch Gladbach, Germany) for at least 10 min at 4 1C. CD138-positive cell selection was carried out on MACS Large Cell Columns (Miltenyi Biotech). Myeloma cells were flushed from the column with full RPMI medium, counted and immediately seeded onto BM stromal cells (BMSCs) or taken into culture in medium supplemented with IL-6. Primary BMSCs were obtained from the column run-throughs cultured in Dulbecco’s modified Eagle’s medium (DMEM; supplemented with 20% FBS, Penstrep, GlutaMAX-I and Na-pyruvate as described above) for between 1 and 4 months. BMSCs were trypsinized and seeded at a density of 1500 per well (96-well plates) prior to their use in co-cultures. All primary materials were obtained from routine diagnostic samples after informed consent of patients and permission by the local ethics committee. Peripheral blood mononuclear cells (PBMCs) were obtained from commercial buffy coats after Ficoll density centrifugation.

Co-culture of myeloma cells with BMSCs and application of drugs Old medium was removed from BMSCs seeded in 96-well plates and exchanged for 1 104 myeloma cells suspended in 100 ml full

RPMI medium per well. Myeloma cells were allowed to attach to the BMSCs overnight and 100 ml of doubly concentrated drug solutions in full RPMI medium was added per well. PBMCs and myeloma cells that were not co-cultured with BMSCs were handled identically except that fresh wells were used. Primary myeloma cell cultures without BMSCs were supplemented with 10 ng/ml IL-6. Drugs were freshly diluted from frozen aliquots in DMSO and pure solvent controls corresponding to the level of DMSO for the highest drug concentration were included in every experiment. Experiments that required addition of z-VAD-fmk (75 mM; Biomol, Hamburg, Germany), followed by NVP-AUY922 (100 nM; provided by Novartis Pharma AG, Basel, Switzerland), were conducted in 24-well plates with 1 ml of medium, so that drug addition from 1000-fold concentrated solutions led to negligible differences in volume.

Treatment of BMSCs with NVP-AUY922 BMSCs from a strongly proliferating culture were thinly seeded in a 48-well plate (5000 cells per well), left to settle for 2 days (by which time the first hints of renewed proliferation after splitting are visible) and treated with NVP-AUY922 in concentrations ranging from 25 to 400 nM, or with DMSO for control purposes. After 3 days the drugged medium was carefully and completely removed, replaced with full DMEM and changed again for fresh medium after another 6 h. Cell culture was then continued until the cells were killed for photography by staining with 0.5% crystal violet solution in 20% methanol.

Apoptosis assay Cell death was assessed through annexin V-FITC/propidium iodide (PI) staining as described before.19

Western analysis Cells were lysed for 15 min on ice (buffer composition: 30 mM Tris-HCl, 120 mM NaCl, 10% glycerol, 1% Triton X-100 (pH 7)), supplemented with complete protease inhibitor cocktail (Roche, Mannheim, Germany) and phosphatase inhibitor cocktails I and II (Sigma, Deisenhofen, Germany). After centrifugation (14 000 g; 15 min), western analysis was performed as described before.20 Primary antibodies against the following targets were used: Akt (Cell Signaling Technologies (CST), Frankfurt am Main, Germany; 9272), caspase 3 (R&D Systems, Wiesbaden, Germany; MAB707), caspase 8 (a gift from Dr Schulze-Osthoff, Du¨sseldorf, Germany), caspase 9 (CST; 9502), extracellular signal-regulated kinase (ERK)1/2 (CST; 9102), phospho-ERK1/2 (CST; 9101), phospho-GSK-3b (CST; 9336), Hsp70 (Biomeda, Foster City, CA, USA; V10130), Hsp90b (Millipore, Schwalbach, Germany; AB3468), IkBa (CST; 9242), IkBb (CST; 9248), IKKa (Becton Dickinson (BD), Heidelberg, Germany; 556532), IKKb (BD; 551920), Raf-1 (Santa Cruz, Heidelberg, Germany; sc-133), receptor-interacting protein (RIP, BD; 610458), STAT3 (CST; 9132), phospho-STAT3 (CST; 9131), a-tubulin (NeoMarkers, CA, USA; MS-581-P0). Secondary antibodies used were anti-mouse (GE Healthcare, Little Chalfont, UK; NA9340V), anti-rabbit (GE Healthcare; NA9310V).

Intracellular epitope staining in myeloma cells

Co-cultured primary MM cells (5 104 per well) were harvested after 18 h of treatment with either 50 nM NVP-AUY922 or a suitable amount of DMSO and fixed with an equal volume of Fix Buffer I (BD). Fixation times varied between antigens: 10 min for phospho-STAT3 and STAT3; 20 min for Akt, phospho-Akt and Leukemia


1606 Hsp70 and 50 min for ERK1/2 and phospho-ERK1/2 at 37 1C. After centrifugation for 5 min at 350 g cells were permeabilized in 90% cold methanol with vigorous mixing at 4 1C for 30 min. Cells were washed twice with 3% bovine serum albumin in PBS (PBS/B), and primary antibodies were added: anti-phosphoSTAT3 coupled to Alexa Fluor 488 (BD; 557814, 1:10), antiSTAT3 (CST; 9132 (lot 3), 1:100), anti-phospho-ERK1/2 coupled to Alexa Fluor 647 (BD; 612593, 1:20), anti-ERK1/2 (CST; 9102 (lot 16), 1:100), anti-phospho-Akt (CST; 4058 (lot 6), 1:100), anti-Akt (CST; 9272 (lot 13), 1:100), anti-Hsp70 (Biomeda; V10130, 1:100). Cells were incubated (in the dark for Alexa Fluor conjugated antibodies) for 1 h at room temperature (RT), and washed twice in PBS/B. Unconjugated primary antibodies required additional incubation with Alexa Fluor 647-conjugated secondary antibodies (Invitrogen; A21237 (anti-mouse) or A21244 (anti-rabbit); each at 1:100 in PBS/B) for 30 min at RT in the dark, followed by two rinses in PBS/B. Finally, cells were measured by fluorescence-activated cell sorting (FACS), and the live population at the time of fixation (clearly demarcated in forward/sideward scatter plots) gated for further analysis. Shifts in median fluorescence value (MFV) were quantified as percent of the MFV of NVP-AUY922-treated samples versus the cognate DMSO-treated controls (thus, decreases in signal intensity are denoted by values lower than 100%, whereas increases are represented by values above 100%). For control purposes, a range of myeloma cell lines were tested with rabbit immunoglobulin G (IgG; Jackson ImmunoResearch, Newmarket, UK; 011-000-003) or mouse IgG2a (ImmunoTools, Friesoythe, Germany; 21275521) substituting for primary antibodies at similar concentrations and fixation times. These controls never resulted in MFV shifts exceeding 15% in either direction between NVP-AUY922- and DMSO-treated cells. This was thus assumed to represent the range of noise for this protocol in myeloma cells, and additional controls with primary cells were forgone in order to maximize the scope of experiments with this strictly limited and unique resource.

experiment (primary patient cells), by non-linear regression (variable slope dose response curve) analysis using GraphPad Prism 3.0. (GraphPad Software, CA, USA).

Results

NVP-AUY922 efficiently induces apoptosis in myeloma cell lines In order to appraise the efficacy of the novel Hsp90 inhibitor NVP-AUY922 on myeloma cells, we established kill curves for eight myeloma cell lines (Figure 1). A 3-day treatment led to pronounced loss of viability in all MM lines tested, with halfmaximally effective concentrations (EC50) in the low nanomolar range (10–25 nM). Five MM lines (AMO-1, INA-6, MOLP-8, OPM-2, MM.1s) displayed steep kill curves and virtually complete cell death at NVP-AUY922 concentrations below 30 nM, whereas for three cell lines (KMS-12-BM, RPMI-8226, U266) EC90 values were not reached at concentrations up to 100 nM (Figure 1). Pretreatment of MM cell lines with the pan-caspase inhibitor z-VAD-fmk showed that NVP-AUY922-mediated cell death was predominantly executed through apoptosis. Exposure to 75 mM caspase inhibitor prior to incubation with maximally effective concentrations of Hsp90 inhibitor (50 nM) was sufficient to largely prevent the appearance of annexin V-positive cells in all MM cell lines tested (Figure 2a). In accordance with processing of apoptotic caspases, decreases in the full-length forms of caspases 3, 8 and 9 were observed in western blots for MM cells treated for 18 h with NVP-AUY922. Reduction of full-length caspases appeared more pronounced in strongly sensitive cell lines (INA-6 and AMO-1) than in those displaying greater resilience to drug treatment, such as KMS-12-BM and RPMI8226. The latter actually did not even show noticeable decreases in caspase levels in this time frame (Figure 2b).

Pharmacologic effect of NVP-AUY922 on primary myeloma cells Data analysis Dose response curves were calculated from at least three independent experiments (MM cell lines), or from a single

To assess the effects of NVP-AUY922 on primary myeloma cells (n ¼ 47 different patients), these were exposed to drug concentrations within the range established with MM cell lines. No

Figure 1 Sensitivity of multiple myeloma (MM) cell lines to NVP-AUY922. (a) Exemplary kill curves for three different MM cell lines. Values reflect the survival of NVP-AUY922-treated cells with respect to that of dimethyl sulphoxide (DMSO)-treated control cells as measured by annexin V-FITC/PI staining and fluorescence-activated cell sorting (FACS) analysis after 3-day exposure to the drug. Three independent experiments were conducted for each cell line. Whereas AMO-1 and INA-6 cells are very susceptible to the drug, KMS-12-BM cells are more resilient. (b) EC50 and EC90 values for all MM cell lines tested. A dash denotes that the applicable value had not been reached in this setting at up to 100 nM NVP-AUY922. Leukemia


Molecular signature of Hsp90 inhibition in MM cell lines

Figure 2 NVP-AUY922-induced myeloma cell death is predominantly mediated through apoptosis. (a) Prior incubation with pan-caspase inhibitor z-VAD-fmk (30 min, 75 mM) blocks apoptosis in NVP-AUY922-sensitive (AMO-1) and resilient (U266) myeloma cell lines. Annexin V-FITC/PI staining was performed after 36 h exposure to z-VAD-fmk (left panel), 50 nM NVP-AUY922 (middle panel) or a combination of both (right panel). (b) Western analyses of cells incubated for 18 h with 50 nM NVP-AUY922 (A) or dimethyl sulphoxide (DMSO; D) show notable decreases in the levels of fulllength caspases 3, 8 and 9 in cell lines efficiently driven into apoptosis by the Hsp90 inhibitor (INA-6, AMO-1). Caspase activation appears much less pronounced in multiple myeloma (MM) cell lines with slower and incomplete response to the drug.

difference in the extent of apoptosis was observed between myeloma cells cultured in medium alone or in co-culture with BMSCs (n ¼ 5; see Figure 3b for a typical example), which may at least in part be a consequence of effects of the drug on BMSCs (see below). However, because co-culture with BMSCs tends to provide better overall survival and is a more rigorous assay for the in vitro assessment of drug effects on primary myeloma cells, experiments were routinely conducted in this setting. The majority of primary MM samples was effectively killed by treatment with low nanomolar concentrations of NVP-AUY922 for 3 days (Figure 3). Again, two types of response could be distinguished: most primary MM samples showed a steep decrease in relative survival (that is, compared to DMSO-treated controls) over a very small concentration range (5–20 nM) and generally reached EC90 values at drug concentrations below 20 nM. The remainder displayed flatter dose dependency and consequently did not attain EC90 values at up to 50 nM (see Figure 3a for examples of both types of response and Figure 3c for a listing of the EC50/90 values of all samples (n ¼ 20) for which kill curves were determined). The high sensitivity of primary myeloma cells against NVP-AUY922 is also evident from Figure 3d, where the percentage of surviving cells at 50 nM of the drug is plotted for all samples tested at this specific concentration (n ¼ 39). The large majority responded efficiently (480% apoptosis) whereas a smaller group (8/39 samples tested) illustrates resilience to Hsp90 inhibition. PBMCs were much less affected by drug treatment than MM cells (Figure 3d).

1607

Inhibition of Hsp90 activity through geldanamycin-derivatives is known to affect a multitude of client proteins that are components of important proliferative and anti-apoptotic signalling pathways. Because such pathways are often found activated in myeloma cells, we analysed the effects of NVP-AUY922 treatment on the expression of client proteins in MM cells by western blotting. Treatment with 50 nM NVP-AUY922 for 18 h (that is, prior to the onset of apoptosis as revealed by annexin V staining) negatively affected IL-6R/ STAT3, Ras/MAPK and the Akt pathway (Figure 4). Notably, this resulted in sizeable decreases in the amounts of activated (phosphorylated) pathway components where these were constitutively present, for example, phospho-ERK1/2 in AMO-1 and RPMI-8226 cells, phospho-STAT3 in INA-6 and AMO-1 cells, and phospho-glycogen synthase kinase-3b (GSK-3b) in all four cell lines tested (the latter as a plausible surrogate marker for signalling through activated Akt, which we were unable to display in western blots with unstimulated myeloma cells; Figure 4). Treatment with NVP-AUY922 led to a marked decrease in the levels of Raf-1 and Akt, both established clients of Hsp90, whereas ERK1/2 and STAT3 remained unaffected. This was also true for Hsp90 itself, whereas robust increases in the amounts of Hsp70, a characteristic response to Hsp90 inhibition, were obvious in all samples (Figure 4). A fourth pathway known to be involved in the survival of MM cells is the one leading to activation of the anti-apoptotic transcription factor NF-kB. Components of this pathway are also known to be clients of Hsp90. Decreases in the levels of inhibitor of kB kinase a (IKKa) and of the IKK-activating RIP in the exquisitely NVP-AUY922-sensitive MM cell lines AMO-1 and INA-6 suggest that impairment of NF-kB activity adds to apoptosis induction in MM cells after Hsp90 inhibition. Interestingly, two MM cell lines more resilient to NVPAUY922-induced apoptosis (RPMI-8226 and KMS-12-BM), displayed noticeably smaller relative declines in both proteins, albeit from a low base in the case of RIP in RPMI-8226 (Figure 4).

Analysis of Hsp90-dependent survival pathways in primary myeloma cells To assess the effects of NVP-AUY922-mediated Hsp90 blockade on the above-mentioned signalling and survival pathways in primary myeloma cells, we employed intracellular antibody staining of pathway components followed by FACS analysis. Staining conditions were established to measure ERK1/2, phospho-ERK1/2, STAT3, phospho-STAT3, Akt and phosphoAkt, thus representing a means to monitor changes in the relative levels and activation status of one central component of Ras/MAPK, IL-6R/STAT3 and the Akt pathway, respectively. Additionally, staining for Hsp70 was performed, which represents a good functional control because it is upregulated due to Hsp90 inhibition. Primary myeloma cells were left overnight to interact with BMSCs and were treated for 18 h with either 50 nM NVP-AUY922 or DMSO. Due to limited cell numbers not all targets could be tested for every primary sample, and when choices had to be made, staining for the phosphorylated (activated) forms was always preferred. In total, 18 different tumour samples were analysed (Figure 5). Every primary sample tested for Hsp70 (n ¼ 11) displayed a very pronounced increase in signal intensity after treatment with NVP-AUY922. MFV shifts ranged from a minimum of about 200% to the two extremes depicted in Figure 5a, which surpassed 800% (sample a) and Leukemia


1608

Figure 3 NVP-AUY922-induced apoptosis in primary myeloma cells. (a) Exemplary kill curves for three very responsive and one resilient primary sample. (b) Exemplary kill curves for a primary tumour sample cultured in medium supplemented with 10 ng/ml interleukin-6 (IL-6; .) or cocultured with bone marrow stromal cells (BMSCs; J). No difference in NVP-AUY922-mediated apoptosis was ever observed for any sample between these two settings (6 multiple myeloma (MM) cell lines, 5 primary samples tested). (c) EC50 and EC90 values for NVP-AUY922-mediated cell death (3-day drug exposure) for all 20 primary MM samples where dose/response relationships could be deduced (all established in co-culture with BMSCs). Both values were reached for 15/20 samples in the low nanomolar range, attesting to the rather steep slope of the dose/response relationship in these samples. EC90 was not attained (n.a.) at up to 50 nM NVP-AUY922 in five samples tested. (d) Survival as determined by annexin V-FITC/PI staining and fluorescence-activated cell sorting (FACS) analysis after 3-day exposure to 50 nM NVP-AUY922 relative to dimethyl sulphoxide (DMSO) controls for all primary tumour samples (K) tested at this concentration (n ¼ 39) or for PBMCs (J; n ¼ 11). The drug exerts a strong apoptotic effect on the large majority of primary MM cells (median at 8% survival) but is less toxic for PBMCs, indicating the scope for establishment of a suitable therapeutic window.

1800% (sample b), respectively. The majority of MFV shifts was in the range of 200 to 400% (Figure 5b). MFV shifts for (phosphorylated) signalling pathway components were generally smaller, which in part reflects the way decreases in signal intensity were defined (a fourfold decrease represents 25% of the control value, a fourfold increase 400% of the control value). Based on the range of background noise, as established on myeloma cell lines (see Materials and methods), MFV shifts representing less than 80% of control values were considered to unambiguously represent drug-induced decreases. Intracellular epitope staining in primary tumour cells pretty faithfully reflected the effects seen for ERK1/2 and phospho-ERK1/2 in western analyses with MM cell lines. Treatment with NVPAUY922 had virtually no effect on the levels of ERK1/2 (less than 20% change in 7/8 samples tested) but resulted in clearly diminished levels of phospho-ERK1/2 in 12/15 samples (Figure 5b). The median values for NVP-AUY922-induced Leukemia

MFV shifts were 94% of control for ERK1/2 and 66% of control for phospho-ERK1/2, indicating strong and selective decline of the active state of these proteins on drug treatment. Similarly, STAT3 protein was largely unchanged in all primary samples evaluated (less than 20% deviation from the control value in 8/8 samples, median ¼ 91% of control), whereas downregulation was observed for phospho-STAT3 (14/17 samples, median ¼ 58% of control; Figure 5b). The signal for Akt fell below the 80% of control threshold in 5/7 samples (median ¼ 75%), which again largely confirms the effect seen with NVP-AUY922 in MM cell lines. MFV shifts in phospho-Akt, however, did not necessarily reflect this situation because an equal number of samples (7/14) either displayed or failed to display downregulation after drug treatment (median ¼ 81%; Figure 5b). All of these measurements solely reflect changes with respect to DMSO-treated control cells. Absence of an MFV shift might therefore either indicate true failure of pathway downregulation


1609 effects of NVP-AUY922 on strongly proliferating cultures of primary BMSCs (Figure 6). A 3-day exposure to concentrations ranging from 25 to 400 nM led to a virtually complete halt in proliferation at all concentrations tested. Additionally, with increasing concentrations of the drug BMSCs became progressively easier to detach from the culture wells by mild physical force (gentle pipetting; TS personal observations). However, complete removal of the drug after 3 days followed by continued cell culture showed that even at the highest concentration tested, the remaining BMSCs had not only survived but even retained their capacity to strongly proliferate. These results indicate that acute treatment with the Hsp90 inhibitor may exert beneficial antitumour effects not just through direct targeting of the cancer cells, but also through the temporary withdrawal of support from cells from the BM microenvironment. However, they also attest to the resilience against Hsp90 inhibition of non-malignant cells, which would be important for the establishment of a therapeutic window.

Discussion

Figure 4 Regulation of Hsp90 client proteins after treatment with NVP-AUY922 in multiple myeloma (MM) cell lines. Cells were treated for 18 h with 50 nM NVP-AUY922 (A) or an applicable amount of dimethyl sulphoxide (DMSO; D) and processed for western blotting. Downregulation was recorded for all major signalling and survival pathways analysed (interleukin-6 (IL-6)R/STAT3, Ras/MAPK, Akt and nuclear factor (NF)-kB). This was visible either through decreases in the levels of major pathway components (Raf-1, Akt, RIP, IKKa) or through decreases in the amounts of the activated states of pathway components (phospho-STAT3, phospho-ERK1/2, phospho-GSK-3b). Also shown is the strong upregulation of Hsp70, and lack of effect on Hsp90b levels. For this picture three gels were loaded to permit independent staining of proteins of similar size and of the phosphorylated versus the unphosphorylated forms of the same protein. Loading controls with either actin or tubulin were performed for every gel. For clarity, only one tubulin control (corresponding to the gel also stained for IKKa and RIP) is shown, but signal intensity was the same between drug treated and corresponding control samples for all cases presented.

or, more likely, that the pathway was not activated in these tumour cells in the first place, and thus indicate heterogeneity between oncogenic networks in primary tumour samples. The prominent NVP-AUY922-mediated MFV shifts seen for phospho-ERK1/2, phospho-STAT3 and especially for Hsp70 in the majority of primary MM cells would highlight these targets as candidates for development/use as pharmacodynamic biomarkers for future clinical trials.

Effects of NVP-AUY922 on BMSCs It is known that at least some of the antimyeloma activity of 17-AAG is mediated through effects that impair the tumoursupporting function of stromal cells.9 We therefore tested the

Inhibition of Hsp90 has recently emerged as a promising therapeutic strategy to produce antitumour effects in a wide range of neoplastic malignancies. These effects are believed to reside in cancer cells’ exquisite requirement to maintain a host of deregulated signalling pathways that allow them to survive, propagate and spread even under the adverse conditions that transformation, growth deregulation and host defences normally impose. As many of these pathways rely on proteins whose stability is dependent on Hsp90, blockade of this chaperone has a particularly debilitating effect on the networks that maintain cancer cell survival.5,6 In addition, and for reasons not entirely understood, Hsp90 inhibitors of different chemical composition appear to be preferentially retained in tumour tissue.21 The novel Hsp90 inhibitor NVP-AUY922 is a synthetic diarylisoxazol resorcinol compound, which binds to the ATPbinding pocket of Hsp90. In a competitive fluorescent polarization binding assay, NVP-AUY922 has excellent potency against Hsp90 and cellular potency against a panel of tumour cell lines.18 In addition, NVP-AUY922 has been shown to exhibit good antitumour efficacy in human xenograft tumour models in nude mice at well tolerated doses and schedules indicating that this compound offers a therapeutic opportunity.18 The promising preclinical data obtained with NVP-AUY922 supported the initiation of clinical Phase I trials in patients with solid tumours (www.clinicaltrials.gov). In this report, we evaluated NVP-AUY922 in a large cohort of primary myeloma samples and in MM-derived cell lines to assess its pharmacodynamic profile and antitumour activity. Both, myeloma cell lines and every primary sample thus tested displayed hallmarks of the molecular signature associated with Hsp90-inhibition after 18 h exposure to 50 nM NVP-AUY922. A ubiquitous and very pronounced effect was the strong upregulation of Hsp70, which is a well-established cellular response to inhibition of Hsp90. This increase was easily monitored in primary MM cells and confirms Hsp70 as a prime candidate to serve as pharmacodynamic biomarker. NVP-AUY922-mediated interference with survival pathways was evident from strong decreases for Hsp90 clients such as Akt, Raf-1, RIP and IKKa. The effect on pathway activity was generally a clear decrease, as shown either by western analysis (MM cell lines) or by intracellular antibody staining and FACS analysis (primary MM samples). Thus, pronounced downward shifts of the MFV were observed for phosphorylated ERK1/2 and phosphorylated STAT3 Leukemia


1610

Figure 5 Regulation of Hsp90 client proteins in primary myeloma cells and possible application as pharmacodynamic markers. (a) Intracellular epitope staining followed by fluorescence-activated cell sorting (FACS) analysis for the STAT3 and phospho-STAT3, ERK1/2 and phospho-ERK1/2, Akt and phospho-Akt pairs, and for Hsp70 in two primary multiple myeloma (MM) samples. Note the strong median fluorescence value (MFV) shifts of Hsp90 client proteins Hsp70 and Akt in both samples, MFV shifts for phospho-STAT3 and phospho-ERK in sample a, and for phosphoSTAT3 and phospho-Akt in sample b. Also note absence of MFV shifts for phospho-Akt in sample a and phospho-ERK in sample b, possibly indicating that these pathways had not been active when the cells were harvested after 2-day co-culture with bone marrow stromal cells (BMSCs). (b) Range of MFV shifts for the abovementioned targets for all primary samples investigated. Open circles mark samples where due to limitations in cell numbers only the phosphorylated form was analysed, closed circles mark samples where the respective phosphorylated/unphosphorylated pair was measured.

in NVP-AUY922-treated primary MM cells. Conversely, signal intensity was largely unchanged for ERK1/2 or STAT3 protein. Although it is well established that ERK1/2 do not belong among Hsp90 clients, STAT3 has previously been reported as client of Hsp9022Fan assessment, that in this study was not reflected for either MM cell lines or primary samples. However, the stability of upstream components of STAT3 activation, namely the Janus kinases, has also been reported to be compromised by Hsp90 inhibition.23 Whereas NVP-AUY922-treatment strongly diminished Akt protein in all MM cell lines and in the majority of primary tumour cells, MFV shifts for phospho-Akt failed to materialize in about half of the primary samples tested, indicating that this pathway was inactive in these cells. Taken together, the NVP-AUY922-induced MFV shifts in phosphoLeukemia

STAT3, phospho-ERK1/2 and Akt indicated that in the large majority of primary MM samples the pharmacodynamic consequences of Hsp90 inhibition were faithfully represented and that these targets, too, might serve to monitor target inhibition in primary tumour cells. The novel Hsp90 inhibitor was very effective at inducing apoptotic cell death of MM cells from established lines and of primary tumour samples alike, with the majority of EC50 values below 20 nM and EC90 values below 30 nM. This was true even though primary MM cells were co-cultured with BMSCs, which can protect them against many pharmacologic interventions.24 However, inhibition of Hsp90 with its multiple effects on signalling pathways may be too broadbased for BMSC coculture to make a difference, and, additionally, debilitating


1611

Figure 6 Effect of NVP-AUY922 on the proliferation of primary bone marrow stromal cells. Freshly seeded bone marrow stromal cells (BMSCs) were treated for 3 days with NVP-AUY922 (range: 25– 400 nM) and after complete drug removal further cultured to monitor their future development (the day numbers refer to the time after drug removal). The day 0 row shows proliferation of BMSCs in the control well (clearly indicated by the appearance of the typical ‘whirls’ of cells) but virtually complete absence of growth in the Hsp90-inhibitortreated samples. However, strong proliferation was resumed after drug removal for all NVP-AUY922 concentrations tested. The pictures show one of two primary BMSC samples from different patients tested with virtually identical results. Scale bars: 200 mm.

effects of Hsp90 inhibitors on the BMSCs themselves might be conducive for MM cell death in co-culture. Some MM lines and a fraction of primary tumour samples were clearly more resilient against NVP-AUY922-mediated apoptosis. Neither in MM cell lines nor in primary samples this partial resistance seemed to bear any connection to the mutation status of p53, because INA-6, OPM-2 and primary plasma cell leukaemia cellsFall defective in TP53Fwere easily driven into apoptosis. Furthermore, the extent of decreases in survival pathway activity, as assessed by MFV shift analyses in primary MM cells, was not obviously correlated to their aptitude to execute apoptosis. Likewise, upregulation of Hsp70, which has been touted as a possible mechanism of anti-apoptotic protection,7 was not consistently stronger in resilient tumour samples. Pharmacologic blockade of Hsp90 focuses on a wellcharacterized target, but the effects are immediately transmitted to such a large group of client proteins and biochemical pathways that on a cellular level it can be viewed as a multitargeted therapy rather than a pathway-specific approach. Our analysis of NVP-AUY922-mediated Hsp90 inhibition on survival pathway activity confirms the validity of this concept for primary myeloma cells. Multi-targeted approaches have the potential to be effective in a large number of myeloma patients even after previous (different) chemo- and systemic therapies, and to provide room for improvements by gainful combination with other drugs that work through different mechanisms. The strength of such concepts has recently been proven by the successful introduction of the proteasome inhibitor bortezomib, which also leads to deregulation of multiple pathways and whose effects are much more damaging to myeloma cells even though no genetic lesions predisposes them as far as the drugs’ primary target is concerned.25 Inhibition of Hsp90 with ansamycin-based compounds has previously been shown to induce myeloma cell death,9–12 and 17-AAG and IPI-504 have been evaluated in clinical Phase I/II trials with myeloma patients. Although both drugs have proven tolerable and displayed pharmacodynamics in line with Hsp90 inhibition,

only 17-AAG is currently being further evaluated in clinical Phase II/III and Phase III trials, where different formulations and combinations with the proteasome inhibitor bortezomib are being tested for their efficacy in relapsed/refractory myeloma patients or as therapeutic option after first relapse (www. clinicaltrials.gov). Nevertheless, formulation and toxicity issues, but also possible resistance mechanisms through biochemical modification, make the development and implementation of structurally different inhibitors of Hsp90 highly desirable.26,27 NVP-AUY922 is the result of the structure-guided optimization of 4,5-diarylisoxazole compounds to block the ATP-binding pocket of Hsp90. Its preclinical effectivity in primary MM cells, the signature changes of Hsp90 client proteins and the drugs’ subduingFbut not fatalFeffects on non-malignant cells from the BM microenvironment are all important points that strengthen the case of Hsp90 inhibition as a promising novel (that is, not yet utilized by any finally approved drug) antimyeloma approach. Our results extend the biological and molecular evidence for this beyond the realm of ansamycin derivatives. The existence of fairly resilient primary myeloma samples in our in vitro analysis shows that resistance against NVP-AUY922, and likely against other Hsp90 inhibitors, will have to be anticipated. Extending the structural range of Hsp90 inhibitors could provide a means to circumvent or attenuate at least some potential resistance mechanisms. This could serve to produce longer lasting benefits of Hsp90-directed mono- or combination therapies in MM. In summary, the preclinical antitumour activity of NVP-AUY922 in myeloma cell lines and in a large panel of primary MM cells suggests that the drug could be effective in a majority of myeloma patients whereas the lower impact of the drug on the viability of healthy primary cells (PBMCs, BMSCs) indicates that in analogy to the situation with ansamycin-class Hsp90 inhibitors a suitable therapeutic window may be established. These results therefore warrant initiation of clinical testing.

Acknowledgements This work was supported in parts by grants from the Deutsche Forschungsgemeinschaft (SFB-TR17/C5), the Deutsche Krebshilfe (grant no. 107715) and the Deutsche Jose´ Carreras Leuka¨mieStiftung e. V. (grant no. R06/17).

References 1 Kuehl WM, Bergsagel PL. Early genetic events provide the basis for a clinical classification of multiple myeloma. Hematology 2005, 346–352. Available at http:\\asheducationbook.hematologylibrary. org/content/vol2005/issue1/. 2 Kyle RA, Rajkumar SV. Multiple myeloma. Blood 2008; 111: 2962–2972. 3 http://www.picard.ch/downloads/downloads.htm; accessed on 2 April 2008. 4 Whitesell L, Lindquist SL. HSP90 and the chaperoning of cancer. Nat Rev Cancer 2005; 5: 761–772. 5 Zhang H, Burrows F. Targeting multiple signal transduction pathways through inhibition of Hsp90. J Mol Med 2004; 82: 488–499. 6 Cullinan SB, Whitesell L. Heat shock protein 90: a unique chemotherapeutic target. Semin Oncol 2006; 33: 457–465. 7 Beere HM. The stress of dying: the role of heat shock proteins in the regulation of apoptosis. J Cell Sci 2004; 117: 2641–2651. 8 Sanderson S, Valenti M, Gowan S, Patterson L, Ahmad Z, Workman P et al. Benzoquinone ansamycin heat shock protein 90 inhibitors modulate multiple functions required for tumor angiogenesis. Mol Cancer Ther 2006; 5: 522–533. Leukemia


1612 9 Mitsiades CS, Mitsiades NS, McMullan CJ, Poulaki V, Kung AL, Davies FE et al. Antimyeloma activity of heat shock protein-90 inhibition. Blood 2006; 107: 1092–1100. 10 Chatterjee M, Jain S, Stu¨hmer T, Andrulis M, Ungethu¨m U, Kuban R-J et al. STAT3 and MAPK signaling maintain overexpression of heat shock proteins 90a and b in multiple myeloma cells, which critically contribute to tumor cell survival. Blood 2007; 109: 720–728. 11 Sydor JR, Normant E, Pien CS, Porter JR, Ge J, Grenier L et al. Development of 17-allylamino-17-demethoxygeldanamycin hydroquinone hydrochloride (IPI-504), an anti-cancer agent directed against Hsp90. Proc Natl Acad Sci USA 2006; 103: 17408–17413. 12 Duus J, Bahar HI, Venkataraman G, Ozpuyan F, Izban KF, Al-Masri H et al. Analysis of expression of heat shock protein-90 (HSP90) and the effects of HSP90 inhibitor (17-AAG) in multiple myeloma. Leuk Lymphoma 2006; 47: 1369–1378. 13 Whitesell L, Mimnaugh EG, Costa BD, Myers CE, Neckers LM. Inhibition of heat shock protein HSP90-pp60v-src heteroprotein complex formation by benzoquinone ansamycins: essential role for stress proteins in oncogenic transformation. Proc Natl Acad Sci USA 1994; 91: 8324–8328. 14 Supko JG, Hickman RL, Grever MR, Malspeis L. Preclinical pharmacologic evaluation of geldanamycin as an antitumor agent. Cancer Chemother Pharmacol 1995; 36: 305–315. 15 Egorin MJ, Zuhowski EG, Rosen DM, Sentz DL, Covey JM, Eiseman JL et al. Plasma pharmacokinetics and tissue distribution of 17-(allylamino)-17-demethoxygeldanamycin (NSC 330507) in CD2F1 mice1. Cancer Chemother Pharmacol 2001; 47: 291–302. 16 Glaze ER, Lambert AL, Smith AC, Page JG, Johnson WD, McCormick DL et al. Preclinical toxicity of a geldanamycin analog, 17-(dimethylaminoethylamino)-17-demethoxygeldanamycin (17-DMAG), in rats and dogs: potential clinical relevance. Cancer Chemother Pharmacol 2005; 56: 637–647. 17 Neckers L, Ivy SP. Heat shock protein 90. Curr Opin Oncol 2003; 15: 419–424.

Leukemia

18 Brough PA, Aherne W, Barril X, Borgognoni J, Boxall K, Cansfield JE et al. 4,5-diarylisoxazole Hsp90 chaperone inhibitors: potential therapeutic agents for the treatment of cancer. J Med Chem 2008; 51: 196–218. 19 Janz M, Stu¨hmer T, Vassilev LT, Bargou RC. Pharmacologic activation of p53-dependent and p53-independent apoptotic pathways in Hodgkin/Reed-Sternberg cells. Leukemia 2007; 21: 772–779. 20 Siegmund D, Wicovsky A, Schmitz I, Schulze-Osthoff K, Kreuz S, Leverkus M et al. Death receptor-induced signaling pathways are differentially regulated by gamma interferon upstream of caspase 8 processing. Mol Cell Biol 2005; 25: 6363–6379. 21 Chiosis G, Neckers L. Tumor selectivity of Hsp90 inhibitors: the explanation remains elusive. ACS Chem Biol 2006; 1: 279–284. 22 Sato N, Yamamoto T, Sekine Y, Yumioka T, Junicho A, Fuse H et al. Involvement of heat-shock protein 90 in the interleukin-6mediated signaling pathway through STAT3. Biochem Biophys Res Commun 2003; 300: 847–852. 23 Shang L, Tomasi TB. The heat shock protein 90-CDC37 chaperone complex is required for signaling by types I and II interferons. J Biol Chem 2006; 281: 1876–1884. 24 Mitsiades CS, Mitsiades NS, Munshi NC, Richardson PG, Anderson KC. The role of the bone microenvironment in the pathophysiology and therapeutic management of multiple myeloma: interplay of growth factors, their receptors and stromal interactions. Eur J Cancer 2006; 42: 1564–1573. 25 Chauhan D, Hideshima T, Mitsiades C, Richardson P, Anderson KC. Proteasome inhibitor therapy in multiple myeloma. Mol Cancer Ther 2005; 4: 686–692. 26 Sharp S, Workman P. Inhibitors of the HSP90 molecular chaperone: Current status. Adv Cancer Res 2006; 95: 323–348. 27 Sharp SY, Boxall K, Rowlands M, Prodromou C, Roe SM, Maloney A et al. In vitro biological characterization of a novel, synthetic diaryl pyrazole resorcinol class of heat shock protein 90 inhibitors. Cancer Res 2007; 67: 2206–2216.