Targeting heat shock protein 90 with non-quinone inhibitors: A novel chemotherapeutic approach in human hepatocellular carcinoma

Authors


  • Potential conflict of interest: Nothing to report.

Abstract

The inhibition of heat shock protein 90 (Hsp90) has emerged as a promising antineoplastic strategy in diverse human malignancies. Hsp90 has been predicted to be involved in hepatocellular carcinoma (HCC) development; however, its role in hepatocarcinogenesis remains elusive. Using chemically distinctive Hsp90 inhibitors, we show that Hsp90 capacitates the aberrant expression and activity of crucial hepatocarcinogenesis-driving factors (e.g., insulin-like growth factor receptor 1, hepatocyte growth factor receptor, protein kinase B, v-raf-1 murine leukemia viral oncogene homolog 1, and cyclin-dependent kinase 4). In vitro, Hsp90 inhibition with both geldanamycin analogs (17-allylamino-17-desmethoxygeldanamycin (17-AAG) and 17-dimethylaminoethylamino-17-desmethoxygeldanamycin (17-DMAG)) and the non-quinone compound 8-(6-iodobenzo[d][1,3]dioxol-5-ylthio)-9-(3-(isopropylamino)propyl)-9H-purin-6-amine (PU-H71) reduced the viability of various HCC cell lines, induced the simultaneous degradation of numerous hepatocarcinogenic factors, and caused substantial cell cycle arrest and apoptosis. In contrast, nontumorigenic hepatocytes were less susceptible to Hsp90 inhibition. Because conventional geldanamycin-derivate Hsp90 inhibitors induce dose-limiting liver toxicity, we tested whether novel Hsp90 inhibitors lacking the benzoquinone moiety, which has been deemed responsible for hepatotoxicity, can elicit antineoplastic activity without causing significant liver damage. In HCC xenograft mouse models, PU-H71 was retained in tumors at pharmacologically relevant concentrations while being rapidly cleared from nontumorous liver. PU-H71 showed potent and prolonged in vivo Hsp90 inhibitory activity and reduced tumor growth without causing toxicity. Conclusion: Hsp90 constitutes a promising therapeutic target in HCC. Non-quinone Hsp90 inhibitors exhibit tumor-specific accumulation and exert potent antineoplastic activity without causing significant hepatotoxicity. (HEPATOLOGY 2009.)

The molecular chaperone heat shock protein 90 (Hsp90) constitutes a relevant therapeutic target in various types of cancer.1 In general, Hsp90 is involved in controlling protein homeostasis by regulating the stabilization and activation of so-called client proteins. Its function is linked to adenosine triphosphate–driven conformational changes and a dynamic interplay with cochaperones.2 Cancer cells, however, co-opt Hsp90 in order to establish their malignant phenotype. Hsp90 controls the folding and activity of numerous bona fide oncoproteins to which malignant cells become addicted and hence safeguards the dysregulated expression of these proteins and their mutational status.1, 3 This transformation-specific function is reflected by findings that demonstrate a higher affinity for adenosine triphosphate and also intriguingly for Hsp90 inhibitors in Hsp90 found in cancer cells in comparison with Hsp90 in their nontransformed counterparts.4, 5

Several chemically distinctive and highly specific Hsp90 inhibitors have shown compelling anticancer activity because they induce the simultaneous proteasomal degradation of Hsp90 clients, thereby depriving cancer cells of tumor-promoting proteins.1, 2 Hsp90 inhibition is therefore anticipated to surpass targeted therapies that exclusively attack one oncoprotein, the function of which might be compensated via collateral pathways.6 Moreover, Hsp90 inhibition renders tumor cells susceptible toward chemotherapy that otherwise affords only limited benefit.1

Hsp90 up-regulation in combination with cyclin-dependent kinase 4 (CDK4) activity has been predicted to contribute to hepatocellular carcinoma (HCC) development.7 Because HCC is frequently associated with liver injury,8 Hsp90 inhibition with geldanamycin analogs is likely to be an inadequate option in this malignancy, as these compounds provoke dose-limiting hepatotoxicity.1 Novel Hsp90 inhibitors, lacking geldanamycin's benzoquinone moiety, which is presumably responsible for the hepatotoxicity,1 may offer new promise for Hsp90 inhibition in HCC.

HCC is the sixth most common cancer worldwide, with over 500,000 new cases annually. It is the fastest growing cause of cancer-related mortality in men in the United States, and its incidence is increasing globally.9, 10 The etiology of HCC primarily involves hepatitis B virus (HBV) and hepatitis C virus (HCV) infections, although virtually all cirrhosis-inducing conditions, such as alcohol abuse, can contribute to hepatocarcinogenesis.11 Because of late detection and the high frequency of tumor recurrence, HCC prognosis remains poor.10 Alarming increases in HCV infection rates portend a looming HCC pandemic, and this emphasizes the urgent need for efficient treatments.9

Here we show that Hsp90 inhibition induced the simultaneous degradation of various hepatocarcinogenesis-driving factors, blocked HCC cell growth, and induced apoptosis. In vivo, the non-quinone Hsp90 inhibitor (PU-H71) showed tumor-specific accumulation, efficiently reduced HCC tumor growth, and was well tolerated, showing a lack of significant hepatotoxicity. Hsp90 inhibition with non-quinone compounds therefore represents a valid therapeutic option in HCC.

Abbreviations

17-AAG, 17-allylamino-17-desmethoxygeldanamycin; 17-DMAG, 17-dimethylaminoethylamino-17-desmethoxygeldanamycin; Akt, protein kinase B; ALT, alanine aminotransferase; CDK4, cyclin-dependent kinase 4; DMSO, dimethyl sulfoxide; Erk, extracellular signal-regulated kinase; FACS, fluorescence-activated cell sorting; HBV, hepatitis B virus; HCC, hepatocellular carcinoma; HCV, hepatitis C virus; HSF1, heat shock transcription factor 1; Hsp70, heat shock protein 70; Hsp90, heat shock protein 90; IGF-1R, insulin-like growth factor receptor 1; KNK-437, N-Formyl-3,4-methylenedioxy-benzylidine-g-butyrolactam; Met, Hepatocyte growth factor (HGF)-receptor; NQO1, NAD(P)H dehydrogenase quinone 1; pAkt, phosphorylated protein kinase B; PARP, nuclear poly(ADP-ribose) polymerase; PBS, phosphate-buffered saline; pErk, phosphorylated extracellular signal-regulated kinase; PU-H71, 8-(6-iodobenzo[d][1,3]dioxol-5-ylthio)-9-(3-(isopropylamino)propyl)-9H-purin-6-amine; RAF-1, v-raf-1 murine leukemia viral oncogene homolog 1; siRNA, small interfering RNA.

Materials and Methods

Note that detailed methods are included as supplementary material.

Reagents

PU-H71 (8-(6-iodobenzo[d][1,3]dioxol-5-ylthio)-9-(3-isopropylamino)propyl)-9H-purin-6-amine) was synthesized as previously reported.12 17-Allylamino-17-desmethoxygeldanamycin (17-AAG) and 17-dimethylaminoethylamino-17-desmethoxygeldana- mycin (17-DMAG) were obtained from LC Laboratories (Woburn, MA). KNK-437 and the caspase-9 inhibitor Z-LEHD-FMK were purchased from Calbiochem/Merck (Darmstadt, Germany). The broad-spectrum caspase inhibitor Z-VAD-FMK was obtained from BioVision (Mountain View, CA). For in vitro use, compounds were dissolved in dimethyl sulfoxide, and stock solutions were stored at −20°C.

Cell Culture

The human liver tumor cell lines HepG2 (wild-type p53) and Hep3B (p53-deficient and pRb-deficient and HBV-positive; both were obtained from the Deutsche Sammlung von Mikroorganismen und Zellkulturen, Braunschweig, Germany), HuH7 [mutated p53 (Tyr220Cys); p21-deficient] and PLC/PRF/5 [mutated p53 (Arg249Ser); p21-deficient and HBV-positive; both were obtained from JHSF, Osaka, Japan], and SkHep1 (obtained from the American Type Culture Collection, Wesel, Germany) were used.13, 14 Additionally, HuH6, FLC4, and the nontumorigenic human liver epithelial cell line HACL-115 were employed (kindly provided by R. Bartenschlager).

HepG2 and HuH6 cells were cultured in Roswell Park Memorial Institute 1640 medium, Hep3B and SkHep1 cells were cultured in minimum essential medium, and HuH7, FLC4, and PLC/PRF/5 cells were cultured in Dulbecco's modified Eagle's medium (all media were obtained from PAA Laboratories, Cölbe, Germany) supplemented with 10% fetal calf serum and 1% penicillin/streptomycin (Sigma, St. Louis, MO) in an atmosphere containing 5% CO2. HACL-1 cells were cultured as previously described and became senescent after prolonged passaging.15 HACL-1 cells in passages 18 to 20 were used for the experiments.

Cell Viability Measurement

The number of viable cells was determined with MTT (3-[4,5-dimethylthiazol-2-yl]-2,5-diphenyl-tetrazolium- bromide) assays (EZ4U, Biomedica, Vienna, Austria) as previously described.16

Cell Cycle Analysis and Detection of Apoptosis

The DNA content was analyzed by flow cytometry of propidium iodide–stained nuclei with a FACSCalibur flow cytometer (Becton Dickinson, Franklin Lakes, NJ) as previously described.16 Data were analyzed with CellQuest and ModFit LT 3.00 software. A hypodiploid peak (sub-G1 phase) indicated apoptosis.

Animal Studies

All animal experiments were performed according to the guidelines and with approval of the local institutional animal care committee. Four- to eight-week-old nu/nu athymic female mice (Charles River Laboratories, Wilmington, DE, and Taconic Farms, New York, NY) were used. Animals were kept under pathogen-free conditions in a 12-hour light/dark cycle at 21.5°C and 60% relative humidity and received an autoclaved standard diet and water ad libitum. Xenografts were established by the subcutaneous injection of 5 × 106 HuH7 cells or 1 × 107 Hep3B cells per 150 μL of phosphate-buffered saline (PBS) into the flanks of mice with 20-gauge needles. For in vivo use, PU-H71 was dissolved in sterile PBS (pH 7.4) and stored at room temperature. 17-DMAG was dissolved in sterile PBS prior to each application.

Pharmacokinetic and Pharmacodynamic Studies.

For pharmacokinetic and pharmacodynamic assays, HuH7 and Hep3B tumors were allowed to reach a volume of ∼500 mm3. Mice were administered a single intraperitoneal dose of PU-H71 (75 mg/kg). Animals were sacrificed by cervical dislocation 12, 24, and 48 hours post-administration (n = 2 per time point). A control group (n = 2) was included that received one intraperitoneal application of PBS. At sacrifice, tumors and livers were collected. The pharmacokinetics of PU-H71 were investigated with high-performance liquid chromatography/tandem mass spectrometry at the Analytical Core Facility of the Memorial Sloan-Kettering Cancer Center as previously described.12 The pharmacodynamic profile of PU-H71 was analyzed by an evaluation of the expression of select Hsp90 clients and heat shock protein 70 (Hsp70) with western immunoblotting.

Efficacy Studies.

In order to investigate the in vivo response toward Hsp90 inhibition in HCC, mice with established tumors (volume of 50-100 mm3) were randomly distributed into either one control group receiving PBS or treatment groups receiving 17-DMAG or PU-H71 (n = 5 each). Because xenograft establishment of Hep3B cells was less efficient, the efficacy studies were performed with HuH7 cells. Mice were treated five times per week (daily/2 days off schedule) with intraperitoneal injections of PU-H71 (75 mg/kg). Because of the potential toxicity of daily intraperitoneal administration of geldanamycin analogs,17, 18 we chose a schedule employing 25 mg/kg 17-DMAG administered intraperitoneally four times per week (2 days/2 days off schedule). Animal weight was determined every day. Tumor progression was monitored by the daily measurement of tumor dimensions with Vernier calipers. The formula length × width2 × 0.5 was used to calculate tumor volumes. Treatment continued for 9 days, at which time tumors in the control group (PBS-treated) reached a volume of ∼1000 mm3, and mice were sacrificed by cervical dislocation within 24 hours after the last application of PBS, 17-DMAG, or PU-H71. A gross necropsy was performed on all mice. Tumors and livers were excised, halved, and either formalin-fixed for immunohistochemistry or snap-frozen for protein extraction. Blood was collected for an analysis of transaminase levels. Measurements of serum transaminase concentrations were performed according to the standard procedure guidelines for clinical diagnosis at the Central Laboratory of Diagnostic Analysis, University Hospital of Heidelberg.

Statistical Analysis

Statistical analyses were performed with the Student t, Wilcoxon, or Kruskal-Wallis test with the SAS statistical software system (SAS Institute, Cary, NC). All statistical analyses were two-sided. For statistical analyses of pharmacokinetic data, a linear model was applied, and the difference in concentration between tumor tissue and liver tissue in a mouse was chosen as the dependent variable. Time was the only factor in the fixed-effects model (Proc MIXED in SAS). Two-sided t tests with 3 degrees of freedom for the null hypothesis that the true mean differences between the tissues were equal to zero were performed. Additionally, Bonferroni correction was applied.

The asterisks in each graph indicate statistically significant changes given as P values: *P < 0.05, **P ≤ 0.01, and ***P ≤ 0.001. P values less than 0.05 were considered statistically significant.

Results

Hsp90 Inhibition Selectively Inhibits the Growth of HCC Cells

The effects of chemically distinct Hsp90 inhibitors (Fig. 1A) on cell viability were investigated with a panel of HCC cell lines and the nontumorigenic liver epithelial cell line HACL-1,15 all of which expressed Hsp90 and Hsp70 (Fig. 1B). Hsp90 inhibition significantly reduced HCC cell viability in a concentration-dependent and time-dependent manner (Fig. 1C). Susceptibility differed among the cell lines and Hsp90 inhibitors used. The calculated half-maximal growth inhibitory concentrations after 48 hours did not exceed 3 μM for all HCC cell lines and Hsp90 inhibitors employed. HACL-1 cells were markedly less susceptible toward Hsp90 inhibition, with calculated half-maximal growth inhibitory concentrations either beyond the tested concentrations of 17-AAG and PU-H71 or at ∼10 μM for 17-DMAG.

Figure 1.

Hsp90 inhibition reduces HCC cell growth. (A) Chemical structures of the Hsp90 inhibitors 17-DMAG and 17-AAG and the non-quinone Hsp90 inhibitor PU-H71. (B) Expression of Hsp90 and Hsp70 in the cell lines employed as determined by western immunoblotting. Representative results are shown (n = 3). (C) HCC cell viability reduction. HCC cell lines and the nontumorigenic hepatocyte cell line HACL-1 were treated with increasing concentrations of Hsp90 inhibitors, and cell viability was analyzed after 24 and 48 hours with MTT assays. Assays were performed in sextuplicate, and data are expressed as the mean ± standard deviation (n ≥ 3). Abbreviations: Hsp70, heat shock protein 70; Hsp90, heat shock protein 90.

Hsp90 Inhibition Induces the Simultaneous Degradation and Inactivation of Hepatocarcinogenesis-Driving Factors

Hsp90 inhibition, independent of the compound employed, reduced the expression of insulin-like growth factor receptor 1, Met receptor, CDK4, protein kinase B (Akt), and v-raf-1 murine leukemia viral oncogene homolog 1 (RAF-1) or inhibited the activation of Akt and extracellular signal-regulated kinase 1/2 (Erk1/2) in a concentration-dependent and time-dependent manner, regardless of the HCC cell line investigated (Fig. 2 and Supporting Fig. 1). Densitometrical analyses (n ≥ 3) revealed the significance of the concentration-dependent and time-dependent down-regulation of the detected proteins, except for the time-dependent effect on Akt in 17-AAG–treated Hep3B cells (Fig. 2B). In contrast, Hsp90 inhibition, independent of the compound employed, did not induce a significant degradation of Akt and CDK4 in the nontumorigenic liver epithelial cell line HACL-1. The other factors investigated were, however, degraded in a time-dependent manner (Supporting Fig. 2). Generally, Erk1/2 expression remained unaltered because Erk1/2 itself is not an Hsp90 client. The expression of additional Hsp90 clients (survivin, epidermal growth factor receptor, and mutated p53) was not found to be markedly altered at the time points and concentrations investigated (data not shown); however, the variable sensitivity of different Hsp90 clients to Hsp90 inhibitors is well acknowledged.6 Significantly increased expression of the inducible chaperone Hsp70 and of Hsp90 itself was noted after Hsp90 inhibition.

Figure 2.

Hsp90 inhibition induces the simultaneous degradation and inactivation of hepatocarcinogenesis-driving factors. (A) Concentration dependence. Hep3B and HuH7 cells were treated with increasing concentrations of Hsp90 inhibitors, and the expression of the indicated proteins was investigated after 24 hours with western immunoblotting. (B) Time dependence. Hep3B and HuH7 cells were treated with vehicle (co) or 1 μM of each Hsp90 inhibitor, and the expression of the indicated proteins was investigated after different time points with western immunoblotting. Representative results are shown (n = 3). Actin served as a loading control. Densitometrical data are given as the mean ± standard deviation (n ≥ 3). Comparable results were obtained with HepG2 and PLC cells (see Supporting Fig. 1). Abbreviations: Akt, protein kinase B; CDK4, cyclin-dependent kinase 4; Erk, extracellular signal-regulated kinase; Hsp70, heat shock protein 70; Hsp90, heat shock protein 90; IGF-1R, insulin-like growth factor receptor 1; Met, Hepatocyte growth factor (HGF)-receptor; pAkt, phosphorylated protein kinase B; pErk, phosphorylated extracellular signal-regulated kinase; RAF-1, v-raf-1 murine leukemia viral oncogene homolog 1.

Hsp90 Inhibition Induces Cell Cycle Arrest and Apoptosis in HCC Cells

After 24 hours, Hsp90 inhibition induced a concentration-dependent decrease in the percentage of cells in the S phase on the basis of G1 and G2/M arrest. The extent and predominance of G1 or G2/M arrest were cell line–dependent and Hsp90 inhibitor–dependent (Fig. 3A and Supporting Table 1).

Figure 3.

Hsp90 inhibition induces cell cycle arrest and apoptosis in HCC cells. (A) Cell cycle arrest. HCC cell lines were treated with increasing concentrations of Hsp90 inhibitors, and the cell cycle distribution was analyzed after 24 hours with FACS measurements. Assays were performed in triplicate. Representative results are shown (n = 3). (For detailed information, see Supporting Table 1.) (B) Apoptosis induction. HCC cell lines were treated with increasing concentrations of Hsp90 inhibitors, and the percentage of hypodiploid cells (sub-G1 phase) was analyzed after 24 hours with FACS measurements. Assays were performed in triplicate. Data are the mean ± standard deviation (n = 3).

Furthermore, Hsp90 inhibition manifested a concentration-dependent increase in the amount of hypodiploid cells (sub-G1 phase). This effect was pronounced in HepG2 and Hep3B cells and moderate but significant (P ≤ 0.05) in HuH7 and PLC/PRF/5 cells (Fig. 3B). Compared to 17-DMAG and PU-H71, 17-AAG was less effective in stimulating cell cycle arrest and hypodiploidy in Hep3B cells.

In order to confirm that the increase of cells in the sub-G1 phase was due to apoptosis, analyses of nuclear poly(ADP-ribose) polymerase (PARP) cleavage and caspase activation were performed, and they substantiated the finding that Hsp90 inhibition induced apoptosis in HCC cells (Supporting Fig. 3).

Hsp90 Inhibition–Induced Apoptosis Is Selective for HCC Cells and Is Mediated by Caspases

Having defined the induction of apoptosis as a result of Hsp90 inhibition with chemically distinctive compounds in HCC cells, we performed further analyses with PU-H71 in order to rule out potential Hsp90-independent effects of geldanamycin analogs associated with their benzoquinone moiety.19, 20

In order to evaluate the relevance of apoptosis induction following Hsp90 inhibition in HCC, we examined whether apoptosis was restricted to HCC cells. In contrast to the HCC cell lines investigated, no PARP cleavage was detected in nontumorigenic HACL-1 cells after PU-H71 treatment (Fig. 4A). Sub-G1 phase analyses confirmed this finding (data not shown).

Figure 4.

Hsp90 inhibition–induced apoptosis is selective for HCC cells and caspase-dependent. (A) Selectivity of Hsp90 inhibition for HCC cells. HCC cell lines and the nontumorigenic hepatocyte cell line HACL-1 were exposed to PU-H71 (1 μM) for 24 and 48 hours, and PARP cleavage was determined by western immunoblotting. Representative results are shown (n = 3). Actin served as a loading control. (B) Apoptosis induction and caspase activation. HCC cell lines were treated with vehicle (control) or PU-H71 (1 μM), and the percentage of cells in the sub-G1 phase (left) and the activities of select caspases (right) were analyzed. Assays were performed in triplicate. Data are the mean ± standard deviation (n = 3). (C) Caspase dependence. HCC cell lines were treated with vehicle (control), PU-H71 (1 μM), or PU-H71 (1 μM) in combination with the broad-spectrum caspase inhibitor zVAD (5 μM). Apoptosis was investigated after 24 and 48 hours with fluorescence-activated cell sorting analysis. Assays were performed in duplicate. Data are the mean ± standard deviation (n = 3). Abbreviations: cl., cleaved; fl., full length; PARP, nuclear poly(ADP-ribose) polymerase.

We then analyzed apoptosis following PU-H71 treatment for up to 48 hours. In Hep3B and HepG2 cells, a remarkable increase in the sub-G1 phase was noted. PU-H71 also induced a significant increase of hypodiploidy in HuH7 and PLC/PRF/5 cells (P ≤ 0.001; Fig. 4B, left, and Supporting Fig. 4). After 48 hours of PU-H71 treatment, caspase-3–like activities were significantly increased in all HCC cell lines (P ≤ 0.01), although time kinetics varied. No significant increase in caspase-8 activity was observed. Caspase-9 activity was significantly increased in Hep3B and HepG2 cells (Fig. 4B, right).

Next, we evaluated the biological relevance of caspase activation in Hsp90 inhibition–induced apoptosis (Fig. 4C). Broad-spectrum caspase inhibition significantly reduced PU-H71–induced apoptosis (P ≤ 0.05). These results were essentially confirmed by the analysis of PARP cleavage (Supporting Fig. 5). Therefore, Hsp90 inhibition–induced apoptosis is mediated by caspases, although a contribution of caspase-independent mechanisms cannot be ruled out.

Because Hsp90 inhibition–induced caspase-9 activation suggested the involvement of mitochondria-mediated apoptosis,21 we examined intrinsic apoptosis mechanisms (Supporting Fig. 6). These analyses revealed that Hsp90 inhibition–induced apoptosis is channeled through intrinsic pathways in HepG2 and Hep3B cells (Supporting Fig. 6A-D) and that factors previously reported to determine cell fate following Hsp90 inhibition22, 23 are not relevant in HCC cells (Supporting Fig. 6E).

Hsp70 Up-Regulation Does Not Limit Hsp90 Inhibition in HCC Cells

Hsp90 inhibition increased the expression of Hsp70 in HCC cells (Fig. 2). Hsp70 has been described as an antiapoptotic factor, and its up-regulation is anticipated to limit the effects of Hsp90 inhibition.24 In HepG2 and Hep3B cells, Hsp90 inhibition–induced apoptosis was marked despite Hsp70 up-regulation, but it was less intense in HuH7 and PLC cells (Fig. 4B). Because Hsp70 depletion significantly reduces cell viability in HuH7 cells (Nylandsted et al.25 and data not shown), we tested whether Hsp70 up-regulation reduced the effects of Hsp90 inhibition in the latter cell lines.

Therefore, we employed a specific approach to block Hsp70 induction using RNA interference to knock down heat shock transcription factor 1 (HSF1), the transcription factor responsible for Hsp90 inhibition–induced Hsp70 up-regulation.26 Successful HSF1 knockdown sufficiently abrogated PU-H71–induced Hsp70 and Hsp90 up-regulation but did not render HuH7 cells more susceptible to Hsp90 inhibition (Fig. 5). These results were confirmed and extended with a pharmacological approach to blocking Hsp70 up-regulation in HuH7 (Fig. 5B) and PLC cells (Supporting Fig. 7A) with KNK-437, a compound previously reported to repress Hsp70 induction.27 Additionally, Hsp70 overexpression prior to Hsp90 inhibition did not confer resistance to Hsp90 inhibition in HuH7 cells (Supporting Fig. 7B).

Figure 5.

Hsp70 up-regulation does not limit Hsp90 inhibition in HCC cells. (A) HSF1 siRNA–mediated abrogation of Hsp70 up-regulation does not affect susceptibility to Hsp90 inhibition. HuH7 cells were transfected with either HSF1 siRNA (20 nM) or nonsense siRNA (control) for 48 hours following PU-H71 (1 μM) treatment for an additional 8 and 24 hours, and Hsp70 and Hsp90 expression was investigated with western immunoblotting (left). HSF1 knockdown efficiency was determined in parallel. Representative results are shown (n = 2). Actin served as a loading control. Cells were transfected with either HSF1 siRNA or nonsense siRNA (control) for 48 hours following an additional 24 hours of treatment with increasing concentrations of PU-H71. Cell viability was analyzed with MTT assays (right). Assays were performed in sextuplicate. Data are the mean ± standard deviation (n = 2). (B) KNK-437–mediated abrogation of Hsp70 up-regulation does not affect susceptibility to Hsp90 inhibition. HuH7 cells were simultaneously treated with PU-H71 (1 μM) and KNK-437 (50 μM), and Hsp70 and Hsp90 expression was investigated after 8 and 24 hours with western immunoblotting (left). Cell viability after simultaneous treatment with either vehicle (dimethyl sulfoxide) or KNK-437 and increasing concentrations of PU-H71 was analyzed after 24 hours (right). Data are the mean ± standard deviation (n = 3). Abbreviations: DMSO, dimethyl sulfoxide; HSF1, heat shock transcription factor 1; Hsp70, heat shock protein 70; Hsp90, heat shock protein 90; siRNA, small interfering RNA.

PU-H71 Exhibits Potent and Prolonged In Vivo Activity, Is Preferentially Retained in HCC Tumors, Significantly Reduces HCC Tumor Growth, and Is Well Tolerated in Mice

Having defined the antihepatocarcinogenic potential of Hsp90 inhibitors in vitro, we investigated the biological consequences of in vivo Hsp90 inhibition in HCC. Regarding the pharmacodynamics of PU-H71 in HuH7 and Hep3B tumor xenografts, a single intraperitoneal application of PU-H71 resulted in concurrent and sustained degradation of RAF-1, Akt, its active phosphorylated form (pAkt), and CDK4. As a marker of Hsp90 inhibition, Hsp70 up-regulation was detected for all time points post-application. Densitometrical analyses confirmed the statistical significance of the down-regulation of respective proteins (n = 2). In contrast, no significant down-regulation of the active phosphorylated form of Erk, Akt, pAkt, or CDK4 was observed in mouse livers (Fig. 6A).

Figure 6.

PU-H71 exhibits potent and prolonged in vivo activity and preferential intratumoral retention and significantly reduces HCC tumor growth without inducing toxicity. (A) Pharmacodynamics of PU-H71 in mice with HCC xenografts. HuH7 and Hep3B tumor–bearing mice were treated with a single intraperitoneal application of PU-H71 (75 mg/kg) or PBS (co), and tumors and livers were harvested at the time points indicated. Degradation or inactivation of indicated proteins, as well as Hsp70 expression, was determined by western immunoblotting. Representative results are shown (n ≥ 2). Actin served as a loading control. Densitometrical data are given as the mean ± standard deviation (n ≥ 2). (B) Pharmacokinetics of PU-H71 in mice with HCC xenografts. HuH7 tumors and livers were collected at the indicated time points post-application, and PU-H71 levels were measured with high-performance liquid chromatography/tandem mass spectrometry. PU-H71 levels of 2 μg/g correspond to a concentration of ∼3.9 μM. Data are given as the mean ± standard deviation (n = 2). For all time points investigated, the differences in PU-H71 levels in tumors compared to livers were statistically significant (P ≤ 0.001). (C) Tumor growth reduction and lack of visible toxicity. HuH7 tumor–bearing mice were randomized into three groups (n = 5 each), and an intraperitoneal application with PBS (control), 17-DMAG, or PU-H71 was performed as outlined in the Materials and Methods section. Left: Tumor volume was determined daily. An arrow indicates the start of treatment. Data are expressed as the mean ± standard deviation (n = 5). Right: Photograph of three representative mice of each group at the end of the experiment. Tumors were excised and photographed separately. Scale bar: 1 cm. (D) Decreased tumor cell proliferation was paralleled by reduced CDK4 expression. Left: The number of Ki-67–positive tumor cells was evaluated after immunohistochemistry staining. Data are the mean ± standard deviation (n = 5). Representative staining patterns of Ki-67 are depicted. Magnification: ×200. Right: CDK4 immunohistochemistry of representative tumors from each group. Magnification: ×50; insert: ×400.

Regarding the pharmacokinetics of PU-H71 after a single intraperitoneal application in HuH7 tumor xenografted mice, calculated PU-H71 concentrations in liver tissues were below ∼0.4 μM for all time points examined, whereas intratumoral concentrations remained in a pharmacologically relevant range with a calculated concentration of ∼4 μM (a corresponding PU-H71 level of ∼2 μg/g) after 48 hours (Fig. 6B). For all time points investigated, the differences in PU-H71 levels in tumors compared to livers were statistically significant (P ≤ 0.001).

Significant tumor growth reduction was observed with both Hsp90 inhibitors, the geldanamycin analog 17-DMAG and the non-quinone compound PU-H71 (P ≤ 0.05; Fig. 6C). Hsp90 inhibition significantly reduced the number of actively duplicating, Ki-67–positive tumor cells (Fig. 6D, left), and this was paralleled by reduced levels of the cell cycle regulator CDK4 (Fig. 6D, right).

Regarding toxicity, histological analyses of liver sections did not reveal any hepatocellular damage following Hsp90 inhibition with the compounds and application schedules employed (Fig. 7A), and no other internal organ damage was detected (data not shown). Serum transaminase (alanine aminotransferase) levels displayed no statistically significant differences between the treatment groups, and no significant weight loss occurred (Fig. 7B,C). 17-DMAG–treated mice started suffering from diarrhea at day 7 after the onset of medication.

Figure 7.

Nontoxicity of Hsp90 inhibition with the non-quinone Hsp90 inhibitor PU-H71 in mice. (A) No traceable hepatic alterations. Mice were treated with PBS (control), 17-DMAG (25 mg/kg three times per week), or PU-H71 (75 mg/kg five times per week), as outlined in the Material and Methods section. Select liver sections were stained with hemalum and eosin following a histological examination. Representative results are shown. Magnification: ×100. (B) No changes in serum transaminase levels. Serum alanine aminotransferase levels were measured as described. Data are the mean ± standard deviation (n = 5). (C) No differences in final body weight. Data are the mean ± standard deviation (n = 5). Abbreviations: ALT, alanine aminotransferase.

Discussion

HCC employs a variety of protumorigenic molecular mechanisms.28, 29 Thus, specific molecular therapeutic efforts are expected to be obstructed by a high degree of primary or secondary resistance.30 Recent progress using the multikinase inhibitor sorafenib for HCC treatment31 suggests that the simultaneous inhibition of various molecular pathways could be advantageous over strategies that solely affect a single oncoprotein.8

We have demonstrated that Hsp90 modulates the expression and function of many known hepatocarcinogenic factors.32–34 Our data therefore indicate that targeting Hsp90 leads to broad-spectrum inhibition of numerous independent hepatocarcinogenic mechanisms. Consequently, Hsp90 inhibition induced marked cell cycle arrest and apoptosis, thereby efficiently reducing the viability of all tested HCC cell lines, despite their heterogeneous etiological backgrounds and mechanisms of oncogenic transformation (e.g., mutated p53 or p53 deficiency). On the basis of these findings, it seems rational to predict that Hsp90 inhibition could render HCC cells more susceptible toward chemotherapeutic strategies that otherwise show only inadequate efficacy. Further experiments are warranted to substantiate this assumption. Furthermore, our results support previous findings that demonstrated a selectivity of Hsp90 inhibitors for transformed cells.12 It remains, however, elusive whether the lack of apoptosis induction following Hsp90 inhibition in the investigated nontumorigenic liver epithelial cell line is based on the lack of degradation of Akt and CDK4, which was exclusively observed in this cell line. Additional studies are necessary to elucidate the transformation specificity of Hsp90 inhibition in HCC cells.

We also have shown that several of the recently discussed limitations of Hsp90 inhibition in cancer treatment are either irrelevant in HCC or overcome by non-quinone Hsp90 inhibitors. First, the DT-diaphorase status is a major source of variability in response to Hsp90 inhibition with 17-AAG and constitutes a potential resistance mechanism.35 Correspondingly, the poor performance of 17-AAG in Hep3B cells is well explained by the inactivating hypermethylation of the NAD(P)H dehydrogenase quinone 1 (NQO1) gene.36 Because no such restrictions were observed for PU-H71, our data corroborate findings demonstrating that the antitumorigenic response toward novel Hsp90 inhibitors is independent of the NQO1 status.37, 38

Second, Hsp90 inhibition up-regulates the expression of chaperones, such as Hsp70, via HSF1 induction.26 Because of its potent antiapoptotic capacity, Hsp70 up-regulation is believed to limit the antineoplastic activity of Hsp90 inhibitors.24, 27 Nevertheless, our data demonstrate that Hsp70 up-regulation did not compensate the antitumorigenic effects of Hsp90 inhibition in HCC cells, in analogy to results obtained in small-cell lung carcinoma cells.12 Therefore, Hsp70 induction is apparently no constraint for Hsp90 inhibition in HCC. In the same line, the up-regulation of Hsp90 itself did not alter the response toward Hsp90 inhibition.

A major drawback for the use of Hsp90 inhibition in HCC is the dose-limiting hepatotoxicity elicited by geldanamycin analogs. Clinical findings have reported hepatotoxicity due to Hsp90 inhibition with 17-AAG or 17-DMAG,39, 40 and the chronic intraperitoneal application of geldanamycin analogs has been proved toxic in rodents, causing weight loss, liver injury, and death.17, 18 Several studies suggest that the benzoquinone moiety shared by these Hsp90 inhibitors is responsible for hepatotoxicity1 because it has potential for cytochrome P450–associated redox metabolism and glutathione adduct formation, both of which contribute to Hsp90-independent toxicity.19, 20 We have shown that the non-quinone Hsp90 inhibitor PU-H71 did not impair liver function in our in vivo experiments. Importantly, no toxicity was observed in PU-H71–treated mice, and this further extends recent results demonstrating that non-quinone Hsp90 inhibitors are well tolerated in vivo.12, 37, 38 Long-term treatment periods, in which mice received up to 45 doses of 75 mg/kg PU-H71 on an alternate-day schedule, and a more extensive toxicity study performed in Balb/c mice treated with 75 mg/kg/day PU-H71 (n = 5 each) consecutively for 10 days and in B6D2F1 mice treated with 100 mg/kg on a 3 times per week schedule for 21 days (13 males and 13 females/group) again showed no evidence of macroscopic (e.g., weight and posture) or microscopic toxicity upon histological examination of all vital tissues compared to vehicle-treated control mice. There was also no evidence of hematologic, renal, or hepatic toxicity, as determined by the performance of complete blood counts, blood chemistry, and liver function tests. No abnormalities in liver enzyme levels and no gastrointestinal mucositis were observed in PU-H71–treated mice (G. Chiosis, unpublished data). Contrarily, gastrointestinal complications, side effects also observed in humans,39, 40 were noticed in mice treated with the geldanamycin analog 17-DMAG, despite the moderate dosing schedule employed in our experiments.

Besides their advantageous toxicological profile, non-quinone Hsp90 inhibitors also have improved pharmacological features, such as insensitivity toward multidrug resistance and favorable water solubility.41 Notably, the intratumoral retention of PU-H71, paralleled by its rapid clearance from the liver demonstrated herein, adds to the favorable therapeutic profile of non-quinone Hsp90 inhibitors in HCC. This is furthermore corroborated by the fact that PU-H71 application did not induce a significant degradation or inactivation of Erk, Akt, or CDK4 in mouse livers; this is in line with and extends previous findings with a chemically distinctive non-quinone Hsp90 inhibitor.37

It may also be noteworthy to mention that Hsp90 inhibition blocks HBV and HCV replication.42, 43 Given that HBV and HCV infections are major risk factors for HCC development,9 these observations possibly cast an additional positive light on Hsp90 inhibition as an anti-HCC strategy.

Collectively, our results are the first to identify Hsp90 as a valuable target structure in HCC and to provide a solid basis for prospective clinical trial investigations using non-quinone Hsp90 inhibitors in HCC treatment.

Acknowledgements

The authors thank Dr. Ralf Bartenschlager (Department of Molecular Virology, University of Heidelberg) for providing the cell lines and Dr. Marja Jäättelä (Apoptosis Department and Centre for Genotoxic Stress Response, Institute for Cancer Biology, Danish Cancer Society) for the Hsp70 expression vector. They are indebted to Dr. Michael Aulmann (Central Laboratory of Diagnostic Analysis, University Hospital of Heidelberg) for his support in serum enzyme measurements. They thank Dr. Volker Ehemann and Hieris Aléz Vuer-Dicaz for advice regarding fluorescence-activated cell sorting analysis. They thank Eva Maria Eiteneuer and Sarah Simone Messnard for their expert technical assistance. They are highly indebted to Dr. Josef Högel (University of Ulm) for his expert assistance with statistical analyses.

Ancillary