Efficacy of Onalespib, a Long-Acting Second-Generation HSP90 Inhibitor, as a Single Agent and in Combination with Temozolomide against Malignant Gliomas

Glioblastomas (GBM) are malignant primary brain tumors characterized by notable resistance to conventional chemotherapy and radiotherapy, resulting in tumor recurrence and an overall median survival of about 16 to 18 months despite aggressive therapy (1, 2). GBM exhibits a marked heterogeneity of genetic and epigenetic alterations which influences tumor resistance and clinical outcome (3, 4). In addition, the presence of chemotherapy and radiation-resistant glioma-initiating cells (GSC) within GBM further contributes to the heterogeneity and highly malignant nature of these tumors (5). Significantly, both inter- and intratumoral driver events converge on a few key pathways involving EGFR, AKT, and ERK, indicating that these pathways are central to tumorigenesis in GBM (6, 7). However, targeting individual pathways in GBM has largely failed to yield the expected improvement in outcome (8), suggesting that strategies to target GBMs may require an alternative approach that simultaneously blocks multiple pathways critical to their growth and survival. One such strategy involves targeting mechanisms of protein processing that are highly active in tumors compared with normal cells and simultaneously disabling several key components of the signaling pathways essential for GBM growth (9).

Heat shock protein 90 (HSP90) is an evolutionarily highly conserved molecular chaperone that functions predominantly as a cytoplasmic protein responsible for ATP-dependent folding and activation of client proteins and is overexpressed in cancer at levels up to 2- to 10-fold higher than in normal cells (10). A large number of receptor and protein kinases and DNA damage repair/response proteins depend on HSP90 for their function (11–13). Tumor cells respond to chemotherapy, radiotherapy, or targeted therapy by adaptively activating DNA repair pathways and by oncogenic reprograming that switches tumor dependence to alternative kinase pathways, by a process facilitated by HSP90 (13, 14). This suggests that targeting HSP90 could potentially overcome these resistance mechanisms and improve sensitivity to therapies (15, 16). Currently, several HSP90 inhibitors are being evaluated in trials against solid tumors and are in different stages of clinical development (17–19). They bind to the N-terminal ATPase-pocket, affecting the chaperone activity and leading to the depletion of HSP90 substrates by subsequent proteasomal degradation (11–13). Recent studies in several solid tumor models have shown sensitization to currently used chemotherapeutic agents when combined with new generation HSP90 inhibitors (20, 21). Onalespib, a potent second-generation HSP90 inhibitor, differs from other HSP90 inhibitors in its longer duration of target inhibition (22) and its favorable toxicity profile in phase I studies in patients with refractory solid tumors (17). However, there is no significant preclinical data regarding the effects of onalespib against glioblastoma; in addition, the potential for chemosensitization of onalespib in combination with alkylating agents such as temozolomide (TMZ) has not been previously studied.

In this study, we characterized the antitumor activity of onalespib in glioma cell lines and GSCs and provide evidence that its use in combination with TMZ causes a synergistic antiproliferative effect in malignant glioma. The study also examines the ability of the drug to cross the blood–brain barrier (BBB) and to inhibit in vivo tumor progression in combination with TMZ.

The human glioma cell lines, LN229 and A172, were obtained from the ATCC. U251HF glioma cells were kindly provided by Dr. W. K. Alfred Yung (M. D. Anderson Cancer Center, Houston, TX); U251HF-Luc were generated by stably transfecting a firefly luciferase gene under a CMV promoter into U251HF cells and maintained as described with the addition of Geneticin (G418-800 μg/mL) for U251HF-Luc cells (23). GSC2, GSC20, GSC11, GSC23, and GSC811 patient-derived GSC lines (24) were kindly provided by Dr. Frederick Lang (M. D. Anderson Cancer Center, Houston, TX) and cultured as neurospheres passaged every 5 to 7 days in serum-free DMEM/F12 medium containing B-27 Supplement (Life Technologies), bFGF (FGF2 Gold BioTechnology), and EGF (Gold Bio Technology). Cell lines were authenticated at the University of Arizona Genetics Core (http://uagc.arl.arizona.edu/services/cell-line-authentication-human). Onalespib (AT13387) was purchased from Medkoo Biosciences.

Glioblastoma cells were seeded into 96-well plates and exposed to vehicle (PBS), onalespib, TMZ, or a combination of onalespib and TMZ for 24 to 96 hours. Glioma and GSC lines were exposed to a 0.1–0.8 μmol/L onalespib for varying times. Cell viability was measured using the WST-1 assay as per the manufacturer's instructions (Roche).

LN229, U251HF, and A172 cells were exposed to increasing concentrations of onalespib following which Annexin assays were conducted by incubating cells with Annexin V–fluorescein isothiocyanate and propidium iodide (BD Biosciences) for 15 minutes and accumulating the fluorescence of at least 10,000 cells on a BD Biosciences FACSCalibur flow cytometer to determine the percentage of apoptotic and viable cells.

LN229, U251HF, and A172 cell lines were treated with onalespib or vehicle (PBS) for 24 hours and seeded at 1,000 cells/well in triplicate. After 7 to 10 days, colonies were fixed, stained with crystal violet, and counted using Gelcount (Oxford Optronix). The Compusyn software (ComboSyn) was used to determine the combinatorial index (CI).

Immunoblotting was performed with antibodies against HSP90, HSP70, p-AKT (Ser 308), AKT, pERK1/2 (Thr 202/Tyr 204), ERK 1/2, pS6 (Ser 240/244), S6, GAPDH (Cell Signaling Technology), pEGFR, EGFR, and anti-mouse and anti-rabbit IgG-HRP as secondary antibodies (GE Healthcare).

LN229, U251HF, and A172 cells were seeded in triplicate on 6-well dishes. Upon reaching 80% confluence, they were exposed to vehicle or onalespib for 24 hours after which a scratch was made and images taken at various time points using an AxioScope A1 (Zeiss). The degree of migration was determined using the Image J software (https://imagej.nih.gov/ij/). In vitro angiogenesis assays were performed as described (25), and the formation of capillary-like structures was captured as images using a Nikon microscope attached to a charge-coupled device camera.

Animal experiments were conducted under a protocol approved by the Institutional Animal Safety Committee in compliance with the Humane Care and Use of Laboratory Animals Policy. Nude mouse intracranial glioma xenografts were generated by implanting U251HF-Luc cells (10⁶ cells/5 μL) into the forebrain as described (26), after which luciferin solution (100 mg/kg) was injected and tumor growth was determined by measuring the bioluminescence emitted using a ZFOV-24 zoom lens-installed IVIS Lumina Series III Pre-clinical In Vivo Imaging System (Perkin Elmer).

Glioma cells isolated from normal brain or GBM xenografts were seeded at 4 × 10⁵ on chamber slides, fixed in ethanol, and blocked with 1%(w/v) BSA before being incubated with HSP90 antibody (Cell Signaling Technology), and subsequently incubated with fluorochrome (Cy3) tagged anti-rabbit secondary antibodies. Nuclei were stained with DAPI, following which slides were washed and optical sections collected using an Olympus FV1000 confocal microscope equipped with three lasers and 100X oil immersion objective. Image processing was conducted using Olympus Fluoview software.

Formalin-fixed, paraffin-embedded tissue from normal brain tissue or GBM xenografts were cut into 15-μm-thick coronal sections for immunohistochemical staining. Sections were deparaffinized in xylene, dehydrated in ethanol, and rinsed with distilled water before incubation with antibodies against HSP90 (Cell Signaling Technology) or HSP70 (Enzo Life Science) which were visualized using the DAB staining Kit (Roche-Ventana) according to the manufacturer's instructions followed by counterstaining with hematoxylin. High-resolution images were acquired and analyzed as detailed in the Supplementary Methods.

Institute for Cancer Research (ICR) wild-type mice were injected with a single intravenous tail vein bolus dose of onalespib (30 mg/kg) following which brain tissue and plasma were harvested at various time points after injection (5 minutes, 10 minutes, 20 minutes, 30 minutes, 1 hours, 2 hours, 4 hours, 6 hours, 8 hours, and 24 hours), lysed, and analyzed by LC-MS-MS (TSQ Quantum Discovery Max mass spectrometer, Thermo Scientific) in the Ohio State University Pharmacoanalytic Shared Resource.

U251HF-GFP cells, grown to 70%–80% confluence, were transplanted (25–50 cells/animal) in the vicinity of the midbrain–hindbrain boundary of 36 hours postfertilization tricaine-anesthetized zebrafish embryos (Sigma-Aldrich) and larvae screened at 24 hours after transplantation to determine implantation (27). Five days after transplantation [5 days post transplant (dpt)], animals (24 animals/group) were randomized to vehicle (DMSO), onalespib at 0.5 μmol/L, TMZ at 10 μmol/L, or a combination of onalespib and TMZ in fish water for 5 days (from 5 to 10 dpt). After 5 days of treatment, tumor burden was assessed by imaging live fish on an Andor spinning disc confocal microscope as described previously (27).

Human patient-derived GS811 neurospheres were dissociated and stereotactically implanted (0.8 × 10⁶ cells in 5 μL) into the right frontal lobe of 7–8-week-old female athymic nude mice (Charles River). Fifteen days after implantation, mice were randomized to vehicle (15% Captosil) or treatment cohorts of TMZ (4 mg/kg), onalespib (30 mg/kg), or a combination with TMZ and onalespib. TMZ was administered by oral gavage daily for 5 consecutive days and onalespib by tail vein injection twice (days 16 and 19). Animals were observed daily and euthanized when they showed signs of morbidity (hunched posture and weight loss).

All in vitro experiments were conducted at least 3 times independently and expressed as mean ± standard error (SEM). Comparisons between groups in vitro and in vivo were performed using two-tailed t tests, and comparisons between multiple groups were performed using one-way ANOVA. In vivo experiments with zebrafish and mouse were analyzed for survival with the Kaplan–Meier method and the significance evaluated by ANOVA.

HSP90 is known to be overexpressed in several solid and hematologic malignancies (10). To assess the expression levels of HSP90 in glioma versus nontumor tissue, U251HF-luc cells were implanted into the forebrain of nude mice to generate bioluminescent intracranial glioma xenografts (Fig. 1A). Hematoxylin–eosin (H&E) staining (Fig. 1B, top) was used to classify tissue sections as non–tumor-bearing (NTB) or tumor-bearing (GBM) areas. Both nontumor and glioma-containing sections were assessed for HSP90 expression by immunocytochemistry and immunofluorescence. HSP90 was robustly expressed in tumor tissue compared with nontumor brain (Fig. 1B, bottom). In addition, using confocal microscopy, HSP90 expression was seen to be predominantly cytoplasmic in location in the tumor cells (Fig. 1C). These results confirm the in vivo cytoplasmic overexpression of HSP90 in human glioma xenografts compared with adjacent nontumor mouse brain.

Onalespib inhibits HSP90 activity by blocking its chaperone function (28), which is associated with a reciprocal increase in the levels of HSP70, which appears to be a class effect of all HSP90 inhibitors and has been used as a biomarker of HSP90 inhibition (13). Exposure of LN229, U251HF, and A172 cells to increasing concentrations of onalespib resulted in HSP90 inhibition as measured by the characteristic increase in HSP70 levels in cells in a dose-dependent manner (Fig. 1D).

Gliomas are characterized by enhanced proliferation, survival, and cellular migration, which contribute to their poor prognosis (29–31). We hence examined the effects of onalespib on these characteristics in LN229, U251HF, and A172 cells and found a dose- and time-dependent decrease in proliferation as measured by the WST assay in all cell lines tested, with an IC₅₀ ≤ 0.25 μmol/L and with maximal inhibition seen at 72 hours (Fig. 2A). Inhibition of growth was paralleled by a dose-dependent activation of caspase-3 and caspase-7 (Supplementary Fig. S1A) and increase in the percentage of Annexin-positive cells (Fig. 2B) in response to onalespib. Correspondingly, colony-forming assays demonstrated that exposure of LN229, U251, and A172 cells to 0.4 μmol/L onalespib for 24 hours significantly inhibited the proliferative capacity of glioma cell lines (Fig. 2C). These results show a broad and sustained in vitro antitumor activity by onalespib in several different glioma cell lines.

GBMs are also characterized by neoangiogenesis and invasion, which is a hallmark of these tumors and is associated with poorer prognosis (31). Quantitation of the microtubules formed by endothelial cells after exposure to media conditioned (CM) by glioma cells was used as a surrogate of tumor-mediated angiogenesis (25, 31) and showed that exposure to 0.4 μmol/L onalespib for 24 hours strongly inhibited the number of microtubules formed in response to CM in LN229, U251HF, and A172 cells (Fig. 2D). Further, when we examined the effect of onalespib on glioma cell migration in vitro using a wound-healing assay, U251HF, LN229, and A172 cells treated with 0.4 μmol/L onalespib showed reduced ability to migrate compared with untreated controls (Fig. 2E; Supplementary Fig. S1B). Taken together, these data provided the evidence that onalespib-induced HSP90 inhibition also inhibits the proliferation, migration, angiogenic potential, and survival of glioma cells.

Recent evidence indicated that the cell death response to HSP90 inhibitors does not fully correlate with degradation of individual oncogenic clients but instead may depend on the inhibition of entire signaling networks driven by key client proteins (32). In this context, amplification and mutations of EGFR (e.g., EGFRvIII), leading to its constitutive activation, is a signature abnormality in GBM that facilitates cell survival and chemoresistance by activating the AKT, ERK, and S6 signaling (33, 34). Hence, we examined the impact of onalespib on the EFGR signaling pathway in gliomas.

Exposure of U251HF, LN229, and A172 cells to onalespib for 48 hours resulted in a robust dose-dependent decrease in the levels of EGFR and p-EGFR and its downstream signaling intermediaries, phospho- and total AKT (also a HSP90 client), phospho- and total ERK1/2, and phospho- and total S6 (Fig. 3A). Of particular note, HSP90 inhibition was equally effective at depleting mutant EGFRvIII and its downstream intermediates (Fig. 3B). In a parallel time-course experiment, exposure to 0.4 μmol/L onalespib resulted in a time-dependent decline in the HSP90 client EGFR, with accompanying losses in p-EGFR as well as an inhibition of their downstream signaling intermediaries phospho- and total AKT, phospho- and total ERK1/2, and phospho- and total S6 by 24 hours in each cell line tested (Fig. 3C). These results point to a profound impact on the entire EGFR signaling pathway in glioma cells upon onalespib treatment.

GSCs represent a subpopulation of glioma cells postulated to contribute to treatment resistance and to have the ability to repopulate the tumor causing disease recurrence (35). GSC lines established from primary tumors recapitulate patient-specific disease phenotypes and thus are valuable as preclinical models (36). Cell lines derived from GBMs (GSC11 and GSC23) when exposed to onalespib exhibited a characteristic increase in HSP70 levels (Fig. 4A). Given the relevance of the EGFR signaling to GSC (37), we determined the effects of onalespib on the EGFR pathway in GSC cells. Exposure to onalespib for 48 hours led to steep declines in the levels of total EGFR, pEGFR, and an inhibition of its signaling via the AKT–ERK and S6 axis, suggesting a potent effect against this key signaling pathway in GSCs (Fig. 4A). In parallel, inhibition of HSP90 in GSC11 and GSC23 cells was also associated with a dose-dependent decrease in the expression of the stem cell markers CD133 (Supplementary Fig. S2A) and Nestin (Supplementary Fig. S2B) with accompanying decreases in the levels of Olig2 which has been reported to be highly expressed in GSC (ref. 38; Supplementary Fig. S2B).

Gene expression analysis has revealed that GBMs can be classified into classical, mesenchymal (MES), or proneural (PN) subtypes; of these, PN tumors exhibit the most favorable prognosis, whereas MES subtype exhibits are more resistant to treatment and have the most aggressive phenotype (3). To examine if there were differences in response of these subtypes to HSP90 inhibition, we tested the efficacy of onalespib against MES or PN GSCs (39). Exposure of MES (GSC2 and GSC20) and PN (GSC11 and GSC23) GSC lines (Fig. 4B) to onalespib showed a dose- and time-dependent decrease in survival (Fig. 4C) irrespective of molecular subtype. However, at the highest concentrations of onalespib, both MES-GSC lines lost viability to 50%, whereas the response of PN-GSCs was variable with loss of viability from 50% in GSC23 to over 90% in GSC11 cells.

Although EGFR activation is necessary for the survival of both bulk tumor cells and GSCs, phosphorylation and activation of STAT3 appear to occur preferentially in GSCs and increase the tumorigenic potential of the GSC population (40). p-STAT3 was clearly evident in MES-GSC and less apparent in PN-GSCs; however, exposure to onalespib for 48 hours inhibited p-STAT3 in both GSCs (Fig. 4D). Finally, exposure to onalespib markedly decreased the invasive ability of each of these cell lines when measured using the transwell migration assay (Supplementary Fig. S2C) Taken together, these results confirm the inhibitory effects of onalespib against glioma tumor-initiating cells. They also support the likelihood that onalespib-induced antitumor activity is not dependent to gene expression–based subtypes of GBM.

A major requirement for agents targeting gliomas is their ability to cross the BBB and achieve target inhibition; the lack of transport of most chemotherapeutic and targeted agents across the BBB is a known reason for their lack of activity against gliomas (41). Hence, we assessed the ability of onalespib to cross the BBB in non–tumor-bearing nude mice (with intact BBB) by intravenously injecting a single dose of onalespib at a dose of 30 mg/kg following which brain tissue and plasma samples were collected at various time points and analyzed by LC-MS/MS to determine onalespib concentration. As expected, the C_max of onalespib was higher in plasma compared with brain tissue in the early time points after intravenous injection. However, levels of onalespib were found to be higher in brain tissue compared with the plasma at each time point analyzed from 2 up to 24 hours, indicating the ability of the drug to cross the BBB and achieve higher than plasma levels (Fig. 5A).

In parallel, normal brain tissue sections obtained from mice intravenously injected with 30 mg/kg of onalespib or from untreated controls were evaluated for evidence of HSP90 inhibition by measuring the levels of HSP70 by immunohistochemistry (IHC). The IHC image was then scanned and digitized to enable an unbiased quantitation of the intensity as well as the percentage of DAB-stained HSP70-positive light brown pixels within the 100 × 100 region of interest. These data were also converted into a color-coded population maps (heat maps) which were used to visually locate positive areas on whole-slide images (Fig. 5B). Our data indicate that onalespib exposure resulted in successful HSP90 inhibition within brain tissues in vivo; however, the HSP70 staining also demonstrated heterogeneity across tissue sections with some areas showing dense HSP70 pixels and others a variable intensity (Fig. 5B). Quantitation of the pixel intensity demonstrated the increase in the percentage and intensity of HSP70 pixels over time in onalespib-treated mice in comparison with the vehicle-treated mice (Fig. 5C).

The standard-of-care first-line treatment for glioblastoma consists of radiotherapy with concurrent and adjuvant administration of TMZ, an alkylating agent (2). Resistance mechanisms that lead to tumor progression during alkylator therapy include the repair of TMZ-induced monofunctional alkylation and single-strand DNA lesion by methyl-guanyl methyltransferase (MGMT) and repair of double-stranded DNA breaks (DSB) by specific DNA repair enzymes in glioma cells and in GSCs (42). Consequently, therapeutic strategies that can potentiate the activity of TMZ are of high clinical relevance. To determine whether exposure to onalespib would sensitize glioma cells to TMZ, we treated U251HF, LN229, and A172 glioma cells with various concentrations of onalespib alone, TMZ alone, or a combination of onalespib and TMZ (Supplementary Fig. S3A) and determined the CI of the treatments (43); the combination of onalespib and TMZ was synergistic in the U251 and LN229 cells with a CI of 0.11 ± 0.06 and 0.7 ± 0.1, respectively, and was additive in the A172 cell line (CI = 1.0 ± 0.16). The IC₅₀ for the combination was achieved by 48 hours of drug exposure in LN229 and U251 but between 48 and 72 hours in A172 cells (Supplementary Fig. S3B). These results demonstrate that the combination of onalespib and TMZ is synergistic or additive in vitro in glioma cells.

To examine whether the combination of onalespib and TMZ was synergistic in vivo, we employed two distinct glioma xenograft models. The first was an orthotopic intracranial xenograft glioma model in zebrafish which we had previously utilized to demonstrate the efficacy of TMZ in reducing tumor burden and improving survival (27). To investigate whether HSP90 inhibition would potentiate the cytotoxic effects of TMZ, we transplanted U251HF cells expressing green fluorescent protein (U251HF-GFP) into the midbrain of zebrafish embryos to generate gliomas, which were evident at 5 dpt (Fig. 6A) and grew robustly by 10 days dpt. Zebrafish were randomized to the following four groups: 1% DMSO, TMZ (10 μmol/L), onalespib (0.5 μmol/L) or TMZ (10 μmol/L) plus onalespib (0.5 μmol/L), and treated between 5 and 10 dpt (Fig. 6A). Fish treated with DMSO showed an exponential increase in intracranial tumor growth during this period (Fig. 6A, top), whereas those exposed to TMZ or onalespib showed moderate reductions in tumor burden (Fig. 6A and B). However, fish exposed to the combination of onalespib and TMZ showed the greatest reduction in tumor burden (Fig. 6B). Correspondingly, onalespib in combination with TMZ significantly extended survival to a greater extent than each individual treatment alone, indicating that the combination of onalespib with TMZ synergizes to reduce tumor growth and extends survival in this in vivo vertebrate model of glioma (Fig. 6C).

Given that conventional cell lines do not fully represent human tumors, we further evaluated the efficacy of this combination in a more clinically relevant patient-derived xenograft mouse model. GSC811 cells represent aggressive GSCs and were implanted intracranially into NOD/SCID mice. Fifteen days after implant, mice bearing GSC811 xenografts were randomized to vehicle, TMZ, onalespib, or a combination of onalespib with TMZ treatment cohorts. Although treatment with TMZ or onalespib alone did not extend survival compared with vehicle-treated mice, the combination of onalespib and TMZ significantly improved survival (P < 0.01, log rank test) in comparison with mice treated with vehicle or each single agent (Fig. 6D).

Increased expression and activity of HSP90 has been implicated in the malignant process in several human tumors including gliomas. Hsp90 inhibitors have been shown to induce a variety of biological effects against gliomas in preclinical evaluations. For instance, geldanamycin, a benzoquinone ansamycin antibiotic, or its derivatives 17-AAG or 1-DMAG have been shown to inhibit glioma growth in vitro through their potent inhibition of HSP90 and consequent biological antitumor effects that included inhibition of survival, invasion, and migration (30), and effects against GSCs (9). Similar results have been seen with newer nongeldanamycin HSP90 inhibitors such as NVP-AUY922 and NVP-BEP800 (44) and IPI-504 (45). Of note, upregulation of HSP70 levels upon inhibition of HSP90 is considered a clinically identifiable class-effect of HSP90 inhibitors across tumor types and can serve as a marker of target inhibition in tumor tissue. This effect is mediated by HSF1, which is normally bound by HSP90 and maintained in an inactive state in the cytoplasm but which, upon HSP90 inhibition, translocates in its active trimeric form to the nucleus and binds to HSP70 promoter, transcriptionally inducing this target gene resulting in increased HSP70 levels (46).

Although 80% of the kinome in the cell and 30% of transcription factors are strong or weak clients of HSP90 (11), strength of binding of client proteins to HSP90 alone does not appear to determine sensitivity to HSP90 inhibition. Rather, the degree to which the tumor cell depends on the HSP90 client signaling network and the extent to which the signaling networks are inhibited as a whole as opposed to individual oncogenic clients appear critical in determining the cellular response to HSP90 inhibitors (32). EGFR amplification occurs in approximately 40% to 50 % of GBMs and results in an increase in prosurvival and proliferative signals. In response to onalespib, there was rapid reduction of EGFR which was accompanied by an inhibition of the key downstream signaling intermediaries, AKT, ERK, and S6 signaling indicating an onalespib-mediated shutdown of the signaling network activated by EGFR. Over half of the tumors that show EGFR amplification also express EGFRvIII mutant, which is recognized as a strong driver of tumor progression in this subset of GBMs and correlates with poor prognosis (33). The potent inhibition of downstream signaling in EGFRvIII-expressing cells demonstrates that onalespib can exhibit a profound effect on this pathway even in the setting of constitutive ligand-independent signaling.

Onalespib treatment was also effective against glioma stem–like cells which are considered the treatment-resistant subset of tumor-initiating clones. Tumorigenicity is preferentially augmented in GSCs by the phosphorylation and activation of STAT3, whereas activation of EGFR signaling is key to survival in both GSCs and bulk tumor cells. The finding that onalespib exposure inhibited p-STAT3 and EGFR signaling to compromise survival in GSCs suggests that onalespib could be active against GBM regardless of EGFR mutation status or stem cell subtype. Further, the activity of onalespib was independent of GBM subtypes in patient-derived GSC lines that recapitulated the relatively favorable PN or more aggressive MES signature. These findings taken in conjunction with the reduction of stem cell markers in GSCs and the observed inhibition of migration, angiogenesis, and proliferation in glioma lines exposed to onalespib suggest a potential for this agent to effectively target the diverse cell populations and pathways that drive heterogeneity and resistance in gliomas.

Although preclinical reports have shown broad activity of HSP90 inhibitors across various tumor types, single-agent activity of HSP90 inhibitors against most malignancies including gliomas has been modest likely due to insufficient target inhibition, short duration of action, or toxicities (47). Unlike other HSP90 inhibitors, Onalespib has a favorable pharmacokinetic and pharmacodynamic profile exerting a prolonged inhibition of HSP90 resulting in a sustained reduction in levels of its client oncoproteins and their downstream targets (22). This may be particularly relevant clinically where sustained inhibition of oncogenic client proteins (12) could result in a more pronounced clinical effect. Onalespib also has the unique advantage of being able to traverse the BBB. Although the brain C_max was 4.7-fold lower than that in plasma, the overall brain exposure of onalespib was greater than of plasma. This was reflected with a consistently higher level of HSP70 staining (marker of HSP90 inhibition) in the nontumor brain tissue of mice exposed to onalespib compared with vehicle. We also observed variations in the intensity of HSP70 signal (marker of HSP90 inhibition) across the IHC-stained section which could be of relevance to the variability of the drug's effects across various portions of the tumor, a finding that could be further examined in clinical specimens as a measure of heterogeneity in target and client protein inhibition across the tumor.

Single-agent HSP90 inhibitor therapy has shown only modest activity in clinical studies, suggesting that although these agents induce changes in a variety of cellular processes, they may not be able to induce a robust anticancer activity alone; instead, several studies suggest that they could potentially serve to prime cancer cells by depleting HSP90 client oncoproteins and sensitize them to cytotoxic signals from other anticancer agents (20, 48, 49). Previous studies of combining shorter acting HSP90 inhibitors with TMZ, a monofunctional alkylating agent that is the most commonly used cytotoxic chemotherapeutic agent against GBM, did not reveal any significant additive or synergistic activity (9, 48). Our results however show that although both onalespib or TMZ could inhibit glioma growth individually in vivo, onalespib was able to potentiate the cytotoxic effects of TMZ in several glioma cells as well as improve survival in two independent glioma xenograft models, one generated using a glioma cell line in zebrafish and the second using a patient-derived GSC xenograft in mice suggesting a potential for efficacy of the combination in the treatment of bulk tumor as well as the GSC populations in GBM. We attribute this finding to the prolonged inhibition of HSP90 by onalespib and to its broad effects on key signaling pathways potentiating the effects of TMZ.

This study is the first to determine the preclinical activity of onalespib against gliomas and assess its potential as an antitumor agent both as a single agent and in combination with TMZ. In addition, onalespib was seen to cross the intact BBB and show target inhibition, which are key requirements for agents targeting infiltrative gliomas. The prolonged pharmacodynamic effect of onalespib differentiates it from other HSP90 inhibitors, and its effects on diverse glioma cells and GSC suggest a potential for targeting glioma in a manner that can overcome tumor resistance; this was shown in its activity in combination with TMZ; these results provide a strong rationale for development of onalespib in combination with TMZ as a therapeutic approach against gliomas.

M.N. Gurcan is a consultant/advisory board member for Inspirata, Inc. V.K. Puduvalli is a consultant/advisory board member for Nektar Pharma, Novocure, and Orbus Therapeutics. No potential conflicts of interest were disclosed by the other authors.

Conception and design: A.M. Welker, D. Sampath, V.K. Puduvalli

Development of methodology: A.M. Welker, J.Y. Yoo, P. Nagarajan, M.N. Gurcan, D. Sampath, V.K. Puduvalli

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): A.M. Welker, J.Y. Yoo, D. Kesanakurti, C.E. Beattie, J. Liu, F.F. Lang

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): A.M. Welker, F.S. Abas, C.E. Beattie, M.N. Gurcan, D. Sampath, V.K. Puduvalli

Writing, review, and/or revision of the manuscript: A.M. Welker, E.P. Sulman, M.N. Gurcan, B. Kaur, D. Sampath, V.K. Puduvalli

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): J. Xu, E.P. Sulman, J. Gumin, B. Kaur

Study supervision: M.N. Gurcan, D. Sampath, V.K. Puduvalli

This project was supported by the Ohio State University Comprehensive Cancer Center core grant (NCIP30CA016058) through the Target Validation Shared Resource (TVSR) core facility (providing athymic nude mice), the Pharmacoanalytic Shared Resource and the Comparative Pathology and Mouse Phenotyping (assistance with tissue processing and IHC), and NINDS P30NS045758 for the zebrafish studies. This study was also supported by NCI grant K24CA160777, the Salvino Family & Accenture Brain Cancer Research Fund, Lisa B. Landes Brain Cancer Fund, and the Roc on Research Fund for Neuro-oncology (V.K. Puduvalli), the Ohio State University Cancer Center Support grant NCI-CA16058, and the MD Anderson Brain Cancer SPORE P50 CA127001 (F.F. Lang and E.P. Sulman).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.