Blocking Heat Shock Protein-90 Inhibits the Invasive Properties and Hepatic Growth of Human Colon Cancer Cells and Improves

2868

Blocking heat shock protein-90 inhibits the invasive properties and hepatic growth of human colon cancer cells and improves the efficacy of oxaliplatin in p53-deficient colon cancer tumors in vivo

Christian Moser,1 Sven A. Lang,1 Silvia Kainz,2 significantly reduced tumor growth and vascularization. Andreas Gaumann,3 Stefan Fichtner-Feigl,1,2 Furthermore, blocking Hsp90 reduced hepatic tumor Gudrun E. Koehl,2 Hans J. Schlitt,1 burden and metastatic nodules in an experimental model Edward K. Geissler,2 and Oliver Stoeltzing1,2 of hepatic colon cancer growth. Importantly, combining oxaliplatin with 17-DMAG in vivo significantly improved 1Departments of Surgical Oncology and 2Experimental growth inhibitory and proapoptotic effects on p53- Surgery, 3Institute of Pathology, University of Regensburg deficient cells, compared with either substance alone. In Medical Center, Regensburg, Germany conclusion, inhibition of Hsp90 abrogates the invasive properties of colon cancer cells and modulates the Abstract expression of the antimetastatic factor activating transcription factor-3. Hence, targeting Hsp90 could prove We recently showed that inhibition of heat shock protein valuable for treatment of advanced colorectal cancer 90 (Hsp90) decreases tumor growth and angiogenesis in by effectively inhibiting colon cancer growth and hepa- gastric cancer through interference with oncogenic sig- tic metastasis and improving the efficacy of oxaliplatin. naling pathways. However, controversy still exists about [Mol Cancer Ther 2007;6(11):2868–78] the antimetastatic potential of Hsp90 inhibitors. More- over, in vitro studies suggested that blocking Hsp90 could overcome p53-mediated resistance of cancer cells Introduction to oxaliplatin. We therefore hypothesized that blocking Invasion into adjacent tissues and metastasis to distant sites oncogenic signaling with a Hsp90 inhibitor would impair are major features of malignant cancer cells, which are metastatic behavior of colon cancer cells and also improve complex processes that require coordinated actions of a the efficacy of oxaliplatin in vivo. Human colon cancer large assortment of growth factors and their receptors, as cells (HCT116, HT29, and SW620) and the Hsp90 well as downstream signaling intermediates (1). With inhibitor 17-(dimethylaminoethylamino)-17-demethoxy- respect to colorectal cancer metastasis, the epidermal geldanamycin (17-DMAG) were used for experiments. growth factor (EGF) receptor (EGFR) system and c-Met, In vitro, 17-DMAG substantially inhibited phosphorylation the primary receptor for hepatocyte growth factor/scatter of epidermal growth factor receptor, c-Met, and focal factor (HGF), have been shown to play an important role adhesion kinase, overall resulting in a significant decrease (2–5). Both receptor systems are overexpressed in colorec- in cancer cell invasiveness. Importantly, 17-DMAG led to tal cancer and colorectal cancer liver metastases. In an up-regulation of the transcription factor activating addition, both kinase systems are involved in carcinogenic transcription factor-3, a tumor suppressor and antimeta- processes such as cell proliferation, motility, survival, static factor, on mRNA and protein levels. In a cell death apoptosis, and tumor angiogenesis (3, 6–8). Hence, EGFR ELISA, 17-DMAG markedly induced apoptosis in both and c-Met systems represent interesting molecular targets p53-wt and p53-deficient cells. In vivo, 17-DMAG for reducing metastasis and angiogenesis of colorectal cancer (4, 9). However, development of c-Met inhibitors has been challenging with studies currently ongoing. Interestingly, inhibition of heat shock protein 90 (Hsp90) may be a novel approach to achieve effective interference Received 6/18/07; revised 9/12/07; accepted 10/1/07. with the function of both receptor systems and respective Grant support: AACR-Littlefield Grant for Metastatic Colon Cancer Research (O. Stoeltzing) and the German Cancer Society (Deutsche downstream signaling intermediates. Hsp90 is a molecular Krebshilfe, Max-Eder Program, Bonn, Germany; O. Stoeltzing). chaperone that is ubiquitously and abundantly expressed. The costs of publication of this article were defrayed in part by the There are numerous proteins involved in the control of payment of page charges. This article must therefore be hereby marked advertisement in accordance with18 U.S.C. Section 1734 solely to physiologic and pathophysiologic processes that require indicate this fact. Hsp90 for their biogenesis, regulation, and function (10, 11). Requests for reprints: Oliver Stoeltzing, Departments of Surgery and In cancer, the expression of Hsp90 is increased at 2- to 10- Surgical Oncology, University of Regensburg Medical Center, Franz-Josef- fold higher levels when compared with that of normal Strauss-Allee 11, 93042 Regensburg, Germany. Phone: 49-941-944- 6801; Fax: 49-941-944-6802. E-mail: oliver.stoeltzing@ tissues (12). Indeed, compounds that inhibit Hsp90 action, klinik.uni-regensburg.de such as geldanamycin or its derivates 17-allylamino-17- Copyright C 2007 American Association for Cancer Research. demethoxygeldanamycin (17-AAG) and 17-dimethylami- doi:10.1158/1535-7163.MCT-07-0410 noethylamino-17-demethoxygeldanamycin (17-DMAG),

Mol Cancer Ther 2007;6(11). November 2007

have been shown to effectively inhibit tumor growth in chased from Invivogen (Cayla-Invivogen). Antibodies several preclinical xenograft models (13–16), and the against MAPK/Erk kinase 1/2, phospho-Ser217/221 MAPK/ efficacy of 17-AAG is currently being investigated in cancer Erk kinase 1/2, Akt, phospho-Thr308 Akt, p44/42 MAPK, patients in clinical phase I/II trials (17–20). phospho-Tyr202/204 p44/42 MAPK, signal transducer Importantly, a number of cancer-associated proteins have and activator of transcription-3, phospho-Tyr705 signal been identified as Hsp90 clients, including EGFR family transducer and activator of transcription 3, focal adhesion members, c-MET, AKT, mutant p53, mitogen-activated kinase (FAK), phospho-Tyr925 FAK, phospho-Tyr1349 c-Met, protein kinase (MAPK)/extracellular signal–regulated ki- and EGFR were purchased from Cell Signaling Technolo- nase (Erk), and the transcription factor hypoxia-inducible gies. Anti–h-actin and anti–activating transcription factor-3 factor-1a (8, 21–23). Because these are important partic- (ATF3) antibodies were purchased from Santa Cruz Bio- h ipants in pathways driving tumor cell survival, prolifera- technology. -Actin serves as a loading control in Western tion, and progression, Hsp90 seems to be an excellent target blot analyses. Recombinant human EGFand HGFwere in cancer therapy (22, 24, 25). In a previous study, we obtained from R&D Systems. showed that blocking Hsp90 significantly interferes with 3-(4,5-Dimethylthiazol-2-yl)-2,5-Diphenyltetrazo- EGFR signaling in gastric cancer, which, in part, is lium Bromide Analyses mediated through down-regulation of EGFR itself (26). To evaluate cytotoxic effects of 17-DMAG on tumor cells, Moreover, we also showed that targeting multiple path- HCT116, HT29, and SW620 cells were seeded into 96-well Â 3 ways does not require therapy with Hsp90 inhibitors at plates (1 10 per well) and exposed to various j maximum tolerated doses to achieve potent inhibition of concentrations of 17-DMAG for 24 or 48 h at 37 C. We tumor growth and angiogenesis. Nevertheless, controver- used the 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazo- sies still exist about the antimetastatic potential of Hsp90 lium bromide assay to assess cell viability, as previously inhibitors. In one report, Price et al. (27) have shown an described (30). enhanced incidence of bone metastasis and osteolytic Immunoblot Analysis of Signaling Intermediates lesions with 17-AAG treatment in a human breast cancer To determine the effects of Hsp90 inhibition on signaling mouse model. This crucial aspect has not been investigated intermediates, cancer cells were incubated for 16 h with for colon cancer and associated hepatic metastases. In favor FCS-RPMI containing 17-DMAG (250 nmol/L) before sti- of a Hsp90-based approach is evidence that Hsp90 mulation with either recombinant human EGF(100 ng/mL, inhibition may overcome resistance of p53-deficient colon 10 and 30 min) or HGF(50 ng/mL, 10 and 30 min) under cancer cells to oxaliplatin through prevention of S-phase serum-reduced conditions (1% FCS-RPMI). Pretreatment of cells for a minimum of 16 h is required to down-regulate/ accumulation, thus leading to additive or synergistic inhibit Hsp90 client proteins, as recently reported (8, 16, 26). apoptotic effects in vitro (28, 29). However, this hypothesis Whole-cell lysates were prepared, as described elsewhere has not yet been validated in in vivo models. (30). Protein samples (75 Ag) were subjected to Western Because these are very important aspects to highlight blotting on a denaturating 10% SDS-PAGE. before initiating clinical trials with Hsp90 inhibitors for ELISA for Vascular Endothelial Growth Factor-A therapy of colorectal cancer (metastatic to liver), we Secretion particularly focused in this study on determining the To determine changes in vascular endothelial growth potential of the Hsp90 inhibitor 17-DMAG to inhibit the factor (VEGF)-A secretion by cancer cells, we used an migratory and invasive properties of p53-wt and p53- ELISA kit specific for human VEGF-A (BioSource Europe), deficient colon cancer cells, and on investigating the as previously described (30). Colorectal cancer cells were antineoplastic efficacy of oxaliplatin with concomitantly plated at 40% to 50% density and incubated with or administered 17-DMAG in p53-deficient cancer cells without 17-DMAG (250 nmol/L) for 24 or 48 h. Analyses in vivo. of culture supernatants were done according to the manu- facturers’ protocol. After harvesting the cells by trypsi- Materials and Methods nization, they were counted. Detected VEGF-A levels Cells and Culture Conditions were calculated as picograms per milliliter per 1,000 viable The human colorectal cancer cell lines HCT116, SW620, cells. and HT29 were obtained from American Type Culture Real-time PCR analyses Collection. Cells were cultured in RPMI 1640 or DMEM For real-time PCR, total RNA was isolated using Trizol (Life Technologies, Inc.) supplemented with 20% FCS Reagent (Invitrogen) and subsequently purified by ethanol (HCT116 and SW620) or 10% FCS (HT29) and were precipitation. For each RNA sample, a 1-Ag aliquot was j maintained in 5% CO2 at 37 C. All in vitro experiments reverse transcribed into cDNA using the Superscript II Kit were done at 60% to 70% cell density to avoid effects of (Qiagen). Primer pairs were as follows: VEGF165, 5¶-GCAC- confluence. For in vivo experiments, trypsinized cells were CCATGGCAGAAGGAGGAG and 3¶-AGCCCCCGCATCG- resuspended in HBSS. CATCAG; ATF3, 5¶-CTGCAGAAAGAGTCGGAG and 3¶- Reagents and Antibodies TGAGCCCGGACAATACAC. Primers were optimized for Water-soluble Hsp90 inhibitor 17-(dimethylaminoethyla- MgCl2 and annealing, and PCR products were confirmed mino)-17-demethoxygeldanamycin (17-DMAG) was pur- by gel electrophoresis. Real-time PCR was done using the

Mol Cancer Ther 2007;6(11). November 2007

LightCycler system and Roche Fast-Start Light Cycler- calculated. For experiments, two different doses of 17- Master Hybridization Probes master mix (Roche Diagnos- DMAG (100 and 250 nmol/L) were investigated to mini- tics). Changes in VEGF-A expression were determined after mize the bias through cytotoxic effects in this particular 24 and 48 h of exposure to 17-DMAG (250 nmol/L). assay. Migration and Invasion Assays Scratch Assay To determine the effect of 17-DMAG treatment on cancer To validate experiments on cell migration, scratch assays cell motility in vitro, migration assays were done using were done. In brief, cancer cells were plated into six-well modified Boyden chambers, as described elsewhere (2). cell culture plates and grown to 90% density in 10% FCS– Briefly, 1 Â 105 cells were resuspended in 1% FCS-RPMI containing RPMI medium within 24 h. Plates were then and seeded into inserts with 8-Am filter pores, which were rinsed with serum-free medium and cells subsequently either uncoated (migration assay) or Matrigel coated (inva- serum starved overnight. One 2-mm-wide scratch was sion assay; Becton Dickinson Bioscience). 1% FCS-RPMI, drawn across the cell layer using a pipette tip. After rin- with or without EGF(100 ng/mL) or HGF(50 ng/mL) sing with serum-free medium, RPMI medium containing added, served as a chemoattractant. Cells were fixed and 250 nmol/L 17-DMAG was added to the cells. Plates were stained (Diff-Quick reagent, Dade Behring) after 48 h, photographed after 6, 24, and 48 h and scratch width was counted in four random fields, and average numbers were measured.

Figure 1. Effect of Hsp90 inhibition on signaling in colorectal cancer cells. Western blot analyses for activated signaling intermediates. A, HCT116 cells were treated with17-DMAG (16 h)and subsequently stimulated withEGF for indicated time points. 17-DMAG down-regulated EGFR and inhibited downstream signaling. B, HCT116 and SW620 (bottom) cells were treated with 17-DMAG (16 h) and subsequently stimulated for the indicated time points with HGF. Blocking Hsp90 reduced HGF-mediated activation of c-Met. In addition, phospho-MAPK/Erk kinase 1/2 (Mek), phospho-p44/42 mitogen- activated protein kinase (MAPK), phospho-Akt, phospho-signal transducer and activator of transcription 3 (STAT3), and phospho-FAK were diminished. h-Actin served as loading control. All results were confirmed in a second colorectal cancer cell line (SW620). C, HCT116 were exposed to 17-DMAG for indicated time. By Western blotting, blocking Hsp90 led to an increase in ATF3 protein expression. D, real-time PCR showed that treatment with 17-DMAG also induced ATF3 mRNA expression in colon cancer cells (HCT116 and SW620). ATF3 mRNA expression is normalized to h-actin.

Mol Cancer Ther 2007;6(11). November 2007

Cell Death Detection ELISA Assay Immunohistochemical Analyses of Tumor Vasculari- The Cell Death Detection ELISAPlus kit (Roche) was used zation,CancerCellProliferation,andApoptosis to measure DNA fragmentation as a marker for apoptosis. Multiple cryosections and paraffin-embedded sections Cells were seeded into 96-well plates at 10,000 per well. After were obtained from tumors for all immunohistochemical 24 h, cells were treated with 17-DMAG (250 nmol/L) and/or analyses. CD31-positive vessel area was assessed with rat oxaliplatin (1 Amol/L) and allowed to grow for additional anti-mouse CD31/platelet-endothelial cell adhesion mole- 24 and 48 h in RPMI 1640 containing 20% FCS. Cytoplas- cule 1 antibody (PharMingen) and peroxidase-conjugated matic fractions of control and treated cells were transferred goat anti-rat IgG (The Jackson Research Laboratory) on into streptavidin-coated 96-well plates and incubated with frozen sections, as previously described (30). Antibody biotinylated mouse antihistone antibody and peroxidase- binding was visualized with stable diaminobenzidine. conjugated mouse anti-DNA antibody at room temperature Images were obtained in four different quadrants of each for 2 h. Absorbance was determined at 405 to 490 nm. To tumor section (2 mm inside the tumor-normal tissue calculate the specific enrichment of mononucleosomes and interface) at Â40 magnification. Measurement of vessel oligonucleosomes released into the cytoplasm (enrichment area of CD31-stained vessels was done by converting factor) of the treated cells, the following formula was used: images to grayscale and setting a consistent threshold for milliunits of the sample (dying/dead cells)/milliunits of all slides using ImageJ software (version 1.33, NIH). Vessel corresponding negative control (28). areas were expressed as pixels per high-power field (26). To Animal Models determine the amount of proliferating tumor cells, mice Eight-week-old male athymic nude mice (BALB/c nu/nu; received i.p. injections of bromodeoxyuridine (BrdUrd; Charles River) were used for experiments, as approved by Sigma-Aldrich; 1 mg/mouse) 2 h before termination of the Institutional Animal Care and Use Committee of the animal studies. A commercially available BrdUrd detection University of Regensburg and the regional authorities. In kit (Becton Dickinson) was used to visualize BrdUrd addition, experiments were conducted according to Guide- uptake of cells in sections of tumors. Briefly, sections were lines for the Welfare of Animals in Experimental Neoplasia incubated with anti-BrdUrd antibody solution, followed by published by The United Kingdom Coordinating Commit- streptavidin-conjugated horseradish peroxidase–linked tee on Cancer Research. goat anti-mouse IgG2. Antibody binding was visualized S.c. Model. The effects of Hsp90 inhibition on the growth by incubating slides in diaminobenzidine, with the aid of of human colorectal cancer cells were evaluated in a s.c. hematoxylin counterstaining. BrdUrd-positive tumor cells xenograft model, as described (26). The water-soluble were counted in four fields per tumor section at Â20 Hsp90 inhibitor 17-DMAG was used for these experiments. magnification and averages were calculated (26). For In brief, 1 Â 106 human colorectal cancer cells (HCT116) analyses of tumor cell apoptosis, a commercially available were injected into the subcutis (right flank) of nude mice. terminal deoxyribonucleotidyl transferase–mediated dUTP After implantation, tumors were allowed to grow 7 days nick end labeling detection kit was used (Promega Corp.) (volume f50 mm3) before treatment was initiated. Mice according to the manufacturer’s instructions. Briefly, the were randomized into one of two groups (n = 9–10 per 3¶OH DNA ends were labeled with a biotinylated nucleo- group) receiving either vehicle (saline) or 17-DMAG tide, coupled with horseradish peroxidase–labeled strep- (15 mg/kg/d). Tumor diameters were measured every tavidin, and then detected hydrogen peroxide and other day, and tumor volumes were calculated (width2 Â diaminobenzidine. Apoptotic nuclei were stained dark length Â 0.5). For immunohistochemical analyses, tumors brown. Four fields at Â40 magnification were selected in were either paraffin embedded or snap frozen. each tumor at the proliferation front, and the terminal Hepatic Tumor Growth. In a following orthotopic experi- deoxyribonucleotidyl transferase–mediated dUTP nick ment to determine the effects of Hsp90 inhibition on end labeling–positive cells were counted. Data are the growth of colorectal liver metastases, HCT116 cells expressed as number of apoptotic cells per high-power (1 Â 106) were injected into the liver of athymic BALB/c field. Using ImageJ software (NIH), necrotic areas of the nude mice. After randomization into one of two groups, tumor sections were measured. Results are expressed as treatment with 17-DMAG 15 mg/kg/d or vehicle (saline) percentage of necrosis in relation to total tumor area. was started on day 7. On day 28, following necropsy, liver Statistical Analyses and Densitometry weights were measured and nodules counted. Results of in vivo experiments were analyzed for Chemotherapy Model. SW620 cells (2 Â 106)were significant outliers using the Grubb’s test for detecting implanted into the subcutis (right flank) of BALB/c nude outliers.4 Tumor-associated variables in in vivo experiments mice to evaluate the effects of Hsp90 inhibition in were tested for statistical significance using the Mann- combination with chemotherapy on the growth of colorec- Whitney U test for nonparametric data. The two-sided tal cancer cells. Treatment with 17-DMAG and/or oxali- Student t test was applied for analysis of in vitro data. All platin versus vehicle (saline) was started on day 9 after results are expressed as the mean F SE. tumors were established. Mice were treated with low- dose 17-DMAG (15 mg/kg/d; ref. 26) and/or oxaliplatin (10 mg/kg; twice per week), respectively. Tumor volumes were determined every other day as described above. 4 http://www.graphpad.com

Mol Cancer Ther 2007;6(11). November 2007

Figure 2. Effect of Hsp90 inhibition on cancer cell migration and invasion. Changes in cancer cell migration were evaluated using Modified Boyden chambers in the presence or absence of 17-DMAG. A, in migration assays (SW620), 17-DMAG (250 nmol/L) effectively reduced cancer cell motility and blunted the response to EGF (*, P < 0.01). B, for determining the effects on cancer cell invasiveness, Matrigel-coated inserts were used. Inhibition of Hsp90 with 17-DMAG (250 nmol/L) significantly abrogated the EGF- and HGF-mediated invasive properties of colorectal cancer cells (HCT116) compared withcontrol (*, P < 0.01, for all). Experi- ments were done in triplicates.

Results cell growth and invasiveness (tumor suppressor function, Effects of Hsp90 Inhibition on EGFR and c-Met Sig- i.e., ATF3; refs. 31–33). Treatment of colon cancer cells with naling in Human Colon Cancer Cells 17-DMAG led to a time-dependent induction of ATF3 Because the EGFand c-Met receptor systems both protein and mRNA (Fig. 1C and D), yet provided evidence represent important mediators of colon cancer growth for another mechanism of Hsp90-regulated cancer cell and metastasis (2, 4, 9), we focused on investigating the invasiveness or growth. However, the time point of effects of 17-DMAG on EGF- and HGF-induced signaling maximal induction differed among cell lines. This novel pathways in human colon cancer cells (HCT116 and finding was confirmed in other colon cancer cells (HT29 SW620). Treatment with 17-DMAG substantially disrupted and SW620) and also in gastric cancer cells (TMK-1; data EGFsignaling in terms of EGFRdown-regulation, inhibi- not shown). tion of MAPK/Erk kinase activation, and diminishing As a functional consequence of multiple signaling downstream phosphorylation of the substrates Erk1/2 and pathway and receptor inhibition and up-regulation of Akt (Fig. 1A). Moreover, constitutive signal transducer and ATF3, 17-DMAG significantly inhibited EGF- and HGF- activator of transcription-3 phosphorylation was dimin- induced cancer cell migration and invasiveness, as deter- ished on Hsp90 blockade. Importantly, inhibition of Hsp90 mined in modified Boyden chambers (Fig. 2A and B). In also markedly reduced HGF-mediated activation of the scratch assays, treatment of HCT116 cell with 17-DMAG c-Met receptor and its downstream effectors (Fig. 1B). In markedly impaired cell migration, compared with controls, addition to the c-Met system, EGF- and HGF-induced which had completely filled the gap within 24 h (data not phosphorylation of FAK, yet another essential mediator for shown). Moreover, as another consequence, blocking cancer cell invasiveness, was dramatically reduced by 17- Hsp90 also significantly reduced constitutive VEGFsecre- DMAG (Fig. 1A and B). Similar effects have been observed tion in colon cancer cells after 24 and 48 h of treatment 5 in other cancer cell lines (26). Interestingly, besides down- (Supplementary Fig. S1). These ELISA results were regulating Hsp90 client proteins, we detected that 17-

DMAG leads to a selective up-regulation of a transcription 5 Supplementary material for this article is available at Molecular Cancer factor, which has been shown to be a negative regulator of Therapeutics Online (http://mct.aacrjournals.org/).

Mol Cancer Ther 2007;6(11). November 2007

Figure 3. Effect of Hsp90 blockade on growthof colorectal cancer cells in vivo. HCT116 cells were implanted s.c. and mice received either 17-DMAG (15 mg/kg/d) or diluent. A, treatment with17-DMAG (n = 9) led to a significant growthinhibition of xenografted colorectal tumors compared withcontrols (*, P < 0.01; n =8).B, immunohistochemical analysis of tumor vascularization. Densitometric analysis of images from CD31-stained tissue sections of all tumors showed that vessel area was significantly reduced in 17-DMAG – treated tumors (*, P < 0.01). C, 17-DMAG significantly decreased the number of proliferating (BrdUrd-positive) tumor cells in tissue sections (*, P < 0.01). For bothvessel area and cell proliferation data, representative images are given to the right of each graph. Columns, mean; bars, SE.

additionally confirmed by real-time PCR (data not shown). (data not shown). Importantly, as determined by immuno- Together, these experiments provide evidence that blocking histochemical staining, therapy with 17-DMAG significant- Hsp90 in colon cancer cells not only affects tumor ly reduced tumor vascularization (CD31-positive vessels; angiogenesis but also harbors the potential to reduce colon Fig. 3B) and tumor cell proliferation (BrdUrd-positive cells; cancer metastasis through direct interference with promi- Fig. 3C). gratory signaling components. Next, we addressed the question on whether Hsp90 Effect of Hsp90 Inhibition on s.c. and Hepatic Tumor inhibition would also be effective in reducing the growth of Growth In vivo colon cancer cells in the liver because the microenviron- To estimate the effects of Hsp90 blockade on colon cancer ment plays a critical role for the efficacy of targeted growth and vascularization in vivo, we first used a s.c. therapies. For this purpose, human colon cancer cells were xenograft model of colon cancer (HCT116). Treatment with directly implanted into the liver of mice. Indeed, treatment 17-DMAG (15 mg/kg/d, i.p.) significantly inhibited tumor with 17-DMAG significantly reduced hepatic tumor bur- growth of p53-wt colon cancer cells (Fig. 3A). This was also den, compared with controls, as reflected by liver weights reflected by measuring final tumor weights, which also at the end of the experiment (Fig. 4A). Normal non–tumor- were significantly reduced in 17-DMAG–treated mice bearing mice served as an additional control, showing that

Mol Cancer Ther 2007;6(11). November 2007

17-DMAG (15 mg/kg/d) itself did not significantly alter in vitro results indeed translate into improved antineoplas- the liver weight (data not shown). In addition to measuring tic efficacy when combining oxaliplatin with Hsp90 liver weights of tumor bearing mice, hepatic tumor nodules inhibitors for treating colon cancer growth in vivo,as were counted, revealing that blocking Hsp90 with 17- suggested by these authors. DMAG significantly reduced the number of hepatic In preliminary experiments, we observed in 3-(4,5- metastases (Fig. 4B). Mouse body weights did not statisti- dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide cally differ among treatment groups. We conclude from analyses that 17-DMAG itself leads to dose-dependent these experiments that targeting Hsp90 inhibits both s.c. growth inhibition (IC50, 250 nmol/L; 24 h) in all three colon (‘‘primary’’) and hepatic growth of colon cancer. Our data cancer cell lines (HT29, HCT116, and SW620) in vitro (data suggest that this inhibition is directly on the tumor cells not shown). To quantify cell death (apoptosis/DNA and indirectly through inhibition of angiogenesis. fragmentation) associated with Hsp90 inhibition, and also Effect of Hsp90 Inhibition on Susceptibility of Colon in combination with oxaliplatin, we used a cell death Cancer Cells to Chemotherapy In vitro and In vivo detection ELISA kit (28). Analyses show that 17-DMAG The potential of Hsp90 inhibitors to improve susceptibil- (250 nmol/L) led to time-dependent apoptotic events in all ity to chemotherapy has been suggested by studies in other three cell lines regardless of the p53 status (Fig. 5). cancer entities; however, the effects of 17-DMAG on the Interestingly, in HCT116 cells (p53-wt), both single agent chemosensitivity of colon cancer cells to oxaliplatin have therapies showed apoptotic efficacy, but at 24 h, combina- not been fully investigated to date. It has been reported that tion therapy seemed to be more potent (Fig. 5A). However, Hsp90 inhibitors promote oxaliplatin-dependent caspase we did not see substantial increases in apoptosis by such activation and cytotoxicity by down-regulating antiapop- combination of 17-DMAG with oxaliplatin in p53-deficient totic signaling through the transcription factor nuclear cells (HT29 and SW620) at 48 h (Fig. 5B and C). factor nB in colon cancer cell lines (28). Moreover, a recent Nevertheless, a mild additive effect was detectable at 24 study showed that Hsp90 inhibitors harbor the potential to h in HT29 cells (Fig. 5C). Overall, we rather observed that inhibit G1-S transition in p53-deficient colon cancer cells, the rates of apoptosis occurring in the combination therapy thereby enhancing cell cycle effects of oxaliplatin (29). groups equaled the apoptotic rates seen with 17-DMAG Hence, we sought to investigate whether these solely monotherapy, suggesting that therapy with Hsp90 inhibitors would be superior to oxaliplatin in p53-deficient colon cancer cells. Furthermore, these results show that the issue of a Hsp90 inhibition–mediated improvement of chemosensitivity cannot be answered by in vitro assays. To validate the hypothesis that Hsp90 inhibition indeed can improve the efficacy of oxaliplatin in p53-deficient colon cancer cells in vivo, we used a s.c. xenografted model and SW620 cancer cells. The results show that combination therapy with oxaliplatin and 17-DMAG clearly led to improved growth inhibitory effects compared with single- agent therapy in vivo (Fig. 6A). Moreover, significant reduction in tumor vascularization (CD31) was observed in 17-DMAG therapy arms as a consequence of an indirect antiangiogenic effect. Moreover, the number of proliferating tumor cells (BrdUrd) was significantly reduced by oxaliplatin alone and in all 17-DMAG therapy arms (Fig. 6B and C). However, we did not detect a significant difference in numbers of proliferating tumor cells of the 17-DMAG therapy arm and the combination therapy group. In contrast, by terminal deoxyribonucleotidyl transferase– mediated dUTP nick end labeling analyses, we detected that combination therapy indeed significantly increased apoptotic rates in tumor tissues compared with either single therapy arm or the control group (Fig. 6D). This increase in terminal deoxyribonucleotidyl transferase– Figure 4. Effect of Hsp90 inhibition of hepatic growth of colon cancer mediated dUTP nick end labeling–positive tumor cells cells. HCT116 cells were injected into the livers of BALB/c nude mice (1 Â 106 per mouse; n = 7 – 10 per group). Treatment with17-DMAG was also reflected by results from analyses of necrotic areas 15 mg/kg/d (i.p. injection) was started on day 7. On day 28, necropsy liver within tumors. Tumors in the combination therapy arm weights were measured and nodules counted. A, treatment with17- elicited a significantly higher percentage of necrotic areas DMAG significantly inhibited hepatic tumor burden, as reflected by final compared with other groups (Fig. 6E). Mouse body weights liver weights (*, P < 0.05). Columns, mean; bars, SE. B, therapy with 17-DMAG significantly reduced hepatic tumor nodules compared with did statistically not differ among treatment groups. In controls (bar, median; *, P < 0.05). summary, these studies provide evidence that Hsp90

Mol Cancer Ther 2007;6(11). November 2007

Figure 5. Effect of Hsp90 inhibition on induction of apoptosis in colorectal cancer cells. Cell Death Detection ELISA kit was used to measure DNA fragmentation as a marker for apoptosis. After 24 h, cells were treated with17-DMAG (250 nmol/L) and/or oxaliplatin (1 Amol/L) and allowed to grow for an additional 24 and 48 h. The specific enrichment of mononucleosomes and oligonucleosomes released into the cytoplasm of the treated cells was additionally calculated (enrichment factor). A, in HCT116 cells (p53-wt), combination therapy led to improved apoptotic effects compared witheithersubstance alone at 24 h. In p53-deficient cells, suchas HT29 colon cancer cells (B) and SW620 cells (C), 17-DMAG seemed to be superior to oxaliplatin. Columns, fold induction of DNA fragmentation.

inhibitors can be used for inhibiting colon cancer growth The fact that Hsp90 is an interesting molecular target for and angiogenesis and for improving the antineoplastic cancer therapy has been reported and validated in various efficacy of oxaliplatin on p53-mutated tumors in vivo. preclinical studies. Indeed, the antineoplastic efficacy of geldanamycin derivates and other synthetic inhibitors is still under active development. However, controversy Discussion continues about the antimetastatic potential of Hsp90 In this study, we show that inhibition of Hsp90 harbors the inhibitors. We and others have shown that certain signaling potential to substantially decrease the migratory and components, such as EGFR, c-Met, and FAK, are inhibited invasive properties of colon cancer cells, which, in part, is by 17-AAG/17-DMAG treatment, hence leading to reduced mediated through inhibition of EGFR, c-Met, and FAK. cancer cell migration in vitro. However, our study is the Moreover, we provide the first evidence that Hsp90 first to show that blocking Hsp90 leads to a marked up- inhibitors are effective in reducing the hepatic growth of regulation of a transcription factor that has been considered colon cancer and in improving the efficacy of oxaliplatin in to function as a tumor suppressor and antimetastatic factor p53-deficient colon cancer tumors in vivo. ATF3. ATF3 may act both as an inducer and as a repressor

Mol Cancer Ther 2007;6(11). November 2007

of transcription for genes such as gadd153/Chop10, which is Hsp90 blockade on the metastatic potential of cancer cells is associated with cell growth (34), and matrix metalloprotei- highly dependent on tumor entity and tumor cell line. The nase-2, which is associated with invasion (32, 35). Impor- role of Hsp90 inhibitors for reducing the hepatic growth of tantly, ATF3 is repressed in colon cancer, indicating that it colon cancers has not been investigated to date. Formation may be involved in the tumorigenic process. Bottone et al. of colorectal liver metastases represents a relevant clinical (31) recently showed that the anti-invasive activity of problem, and only few patients qualify for surgery. With cyclooxygenase-2 inhibitors or nonsteroidal anti-inflamma- results from this study, we provide evidence that blocking tory drugs involves up-regulation of ATF3. RNA interfer- Hsp90 could indeed decrease hepatic ‘‘metastatic’’ tumor ence of ATF3 led to a significant increase in the invasive growth of colon cancer, thus offering a novel approach for properties of cancer cells, suggesting that ATF3 acts as a targeted therapy concepts for downsizing of colorectal repressor of metastasis. Nevertheless, to date, in vivo effects cancer liver metastases. are not quite clear because Price et al. (27) have elegantly Interestingly, the presence of p53-deficiency has been shown that Hsp90 inhibitors promote the formation of bone associated with an increased incidence of liver metastasis metastases through activation of osteoclasts in an experi- and progression in patients with colorectal cancer (36). In mental model of breast cancer. On the other hand, because addition, p53 status has been shown to be relevant for formationofbonemetastasesisnotapredominant susceptibility of colon cancer cells to chemotherapy with phenomenon of colon cancer, we suggest that the effect of oxaliplatin. Because oxaliplatin is now widely used for

Figure 6. Effect of Hsp90 blockade in combination with chemotherapy on growth of colorectal cancer cells in vivo. SW620 human colorectal cancer cells were implanted into the subcutis of nude mice. Mice (n = 9 – 10 per group) were randomized into one of four treatment groups and received vehicle (saline), oxaliplatin (Ox; 10 mg/kg; biweekly), 17-DMAG (15 mg/kg/d), or the combination of both agents. A, combination therapy elicited a significantly reduced tumor burden, as reflected by tumor volumes, compared withcontrols (*, P < 0.05). B, immunohistochemical analysis of tumor vascularization. Densitometric analysis of images from CD31-stained tissue section shows that vessel area in 17-DMAG – treated tumors is significantly reduced compared withcontrol (*, P < 0.01). C, immunohistochemical analysis tumor cell proliferation in tissues. Both treatment with either 17-DMAG or oxaliplatin alone and combination therapy significantly decreased the number of proliferating (BrdUrd-positive) tumor cells in tissue sections compared with control (*, P < 0.01). D, analyses of tumor cell apoptosis by terminal deoxyribonucleotidyl transferase – mediated dUTP nick end labeling (TUNEL). Treatment witheitheroxaliplatin or 17-DMAG significantly increased apoptotic rates in tumor sections (*, P < 0.01, versus control). Combination therapy enhanced apoptotic rates compared witheithersubstance alone ( #, P < 0.01, versus 17-DMAG or oxaliplatin). E, measurement of necrotic area within tumor sections. Both17-DMAG and oxaliplatin induced tumor necrosis (*, P < 0.01, versus control), with the highest necrotic area observed in the combination therapy arm (#, P < 0.01, versus 17-DMAG or oxaliplatin). Columns, mean; bars, SE.

Mol Cancer Ther 2007;6(11). November 2007

multimodality regimens in treating advanced or metastatic morphology or permeability, suggesting that Hsp90 inhib- (liver) colorectal cancer, we focused on investigating the itors, administered below maximum tolerated dose, might effects of 17-DMAG in combination with oxaliplatin on not have a biologically significant direct effect on vascula- p53-wt and p53-deficient colon cancer cells. This aspect has, ture in vivo. in part, been addressed in a recent study by Rakitina et al. In conclusion, this present study shows that inhibition of (29), showing that the geldanamycin derivate 17-AAG Hsp90 abrogates the invasive properties of colon cancer could prevent S-phase accumulation (G1-S transition), cells in vitro and effectively inhibits growth, angiogenesis, which is induced by oxaliplatin in p53-deficient colon and hepatic tumor growth in vivo. Because p53 status in cancer cells (HT29), thereby improving apoptotic effects by colon cancer cells is a relevant determinant of susceptibility 17-AAG in vitro. In an earlier study, authors similarly used to oxaliplatin, and Hsp90 inhibition leads to the improved a cell death detection ELISA to detect effects of Hsp90 efficacy of oxaliplatin in p53-deficient cancer cells in vivo, inhibition with or without addition of oxaliplatin on colon our data suggest Hsp90 antagonists to be valuable for cancer cell apoptosis. In concordance with our results, improving therapy of advanced colorectal cancer and for Rakitina et al. (28) found that combination therapy may the (neo-adjuvant) therapy of colorectal liver metastases. lead to additive/synergistic apoptotic effects in HCT116 Further studies will be required to define the risk of cells (p53-wt). However, in our hands, p53-deficient cells surgery when treating with novel Hsp90 inhibitors peri- solely responded to 17-DMAG whereas only a modest operatively. effect of oxaliplatin alone was detectable. A critical determinant for interpreting results seemed to be the time Acknowledgments point when evaluating apoptosis (24 versus 48 h). The best We thank Christine Wagner and Kathrin Stengel for excellent technical apoptotic effect achievable in p53-deficient cells was in our assistance. experiments observed with 17-DMAG after 48 h, but we References did not detect clear synergistic effects by combination 1. Fidler IJ. Cancer metastasis. Br Med Bull 1991;47:157 – 77. therapy in such p53-deficient cells in vitro. However, 2. Bauer TW, Fan F, Liu W, et al. Insulinlike growthfactor-I-mediated Rakitina et al. (28) also showed that interference with the migration and invasion of human colon carcinoma cells requires activation nuclear factor-nB pathway through Hsp90 inhibitors of c-Met and urokinase plasminogen activator receptor. Ann Surg 2005; improves susceptibility to oxaliplatin, suggesting that these 241:748 – 56; discussion 56 – 8. potential additive or synergistic effects might be more 3. Di Renzo MF, Olivero M, Giacomini A, et al. Overexpression and amplification of the met/HGF receptor gene during the progression of complex and difficult to detect by in vitro analyses. colorectal cancer. Clin Cancer Res 1995;1:147 – 54. However, the hypothesis that in vivo the efficacy of 4. Herynk MH, Stoeltzing O, ReinmuthN, et al. Down-regulation of c-Met oxaliplatin could be substantially improved by adding inhibits growth in the liver of human colorectal carcinoma cells. Cancer Hsp90 inhibitors to the regimen had not been tested to date. Res 2003;63:2990 – 6. In our first model, we showed that 17-DMAG could 5. Shaheen RM, Ahmad SA, Liu W, et al. Inhibited growth of colon cancer carcinomatosis by antibodies to vascular endothelial and epidermal growth decrease hepatic growth of colon cancer cells in vivo; factor receptors. Br J Cancer 2001;85:584 – 9. hence, we subsequently addressed the question of whether 6. Singh RK, Tsan R, Radinsky R. Influence of the host microenvironment combination therapy indeed would be more effective on the clonal selection of human colon carcinoma cells during primary in vivo. For this purpose, we again used a s.c. tumor model tumor growthand metastasis. Clin Exp Metastasis 1997;15:140 – 50. because continuous evaluation of tumor growth in the liver 7. Vigna E, Naldini L, Tamagnone L, et al. Hepatocyte growthfactor and its receptor, the tyrosine kinase encoded by the c-MET proto-oncogene. is challenging. With this study, we were able to show for Cell Mol Biol (Noisy-le-grand) 1994;40:597 – 604. the first time that 17-DMAG, administered below the 8. Zhang H, Burrows F. Targeting multiple signal transduction pathways maximum tolerated dose (26), improves the antineoplastic through inhibition of Hsp90. J Mol Med 2004;82:488 – 99. efficacy of oxaliplatin in p53-deficient colon cancer cells 9. Yokoi K, Thaker PH, Yazici S, et al. Dual inhibition of epidermal growth in vivo. Moreover, the (indirect) antiangiogenic potential of factor receptor and vascular endothelial growth factor receptor phosphorylation by AEE788 reduces growthand metastasis of humancolon Hsp90 inhibition was confirmed by CD31 staining in the carcinoma in an orthotopic nude mouse model. Cancer Res 2005;65: 17-DMAG therapy groups, suggesting that blocking Hsp90 3716 – 25. can improve antineoplastic effects of oxaliplatin and lead to 10. Pratt WB, Toft DO. Regulation of signaling protein function and trafficking by the hsp90/hsp70-based chaperone machinery. Exp Biol Med inhibition of tumor angiogenesis in colon cancer. Impor- (Maywood) 2003;228:111 – 33. tantly, by analyses for tumor cell apoptosis and by 11. Young JC, Barral JM, UlrichHartl F. More thanfolding: localized measuring necrotic areas in tumors, we confirmed that functions of cytosolic chaperones. Trends Biochem Sci 2003;28:541 – 7. combining oxaliplatin with a Hsp90 inhibitor indeed leads 12. GooljarsinghLT, Fernandes C, Yan K, et al. A biochemicalrationale for to improved antineoplastic efficacy in vivo. In contrast to the anticancer effects of Hsp90 inhibitors: slow, tight binding inhibition by geldanamycin and its analogues. Proc Natl Acad Sci U S A 2006;103: this finding, the antiangiogenic effect of Hsp90 inhibitors 7625 – 30. has been suggested by a few studies (24, 26, 37, 38). 13. Basso AD, Solit DB, Munster PN, Rosen N. Ansamycin antibiotics Notably, in one study, Sanderson et al. (39) showed that 17- inhibit Akt activation and cyclin D expression in breast cancer cells that AAG, administered at maximum tolerated dose, signifi- overexpress HER2. Oncogene 2002;21:1159 – 66. cantly reduces the expression of three VEGFreceptors 14. Kelland LR, Sharp SY, Rogers PM, Myers TG, Workman P. DT- Diaphorase expression and tumor cell sensitivity to 17-allylamino, 17- (VEGF-R1, VEGF-R2, and VEGF-R3) in mouse vasculature demethoxygeldanamycin, an inhibitor of heat shock protein 90. J Natl in vivo. However, we did not detect changes in vascular Cancer Inst 1999;91:1940 – 9.

Mol Cancer Ther 2007;6(11). November 2007

15. Solit DB, Basso AD, Olshen AB, Scher HI, Rosen N. Inhibition of heat cell lines results from inhibition of nuclear factor nB signaling. Cancer Res shock protein 90 function down-regulates Akt kinase and sensitizes 2003;63:8600 – 5. tumors to Taxol. Cancer Res 2003;63:2139 – 44. 29. Rakitina TV, Vasilevskaya IA, O’Dwyer PJ. Inhibition of G1/S 16. Solit DB, Zheng FF, Drobnjak M, et al. 17-Allylamino-17-demethox- transition potentiates oxaliplatin-induced cell deathin colon cancer cell ygeldanamycin induces the degradation of androgen receptor and HER-2/ lines. Biochem Pharmacol 2007;73:1715 – 26. neu and inhibits the growth of prostate cancer xenografts. Clin Cancer Res 30. Stoeltzing O, Liu W, ReinmuthN, et al. Regulation of hypoxia- 2002;8:986 – 93. inducible factor-1a, vascular endothelial growth factor, and angiogenesis 17. Banerji U, O’Donnell A, Scurr M, et al. Phase I pharmacokinetic and by an insulin-like growthfactor-I receptor autocrine loop in human pharmacodynamic study of 17-allylamino, 17-demethoxygeldanamycin in pancreatic cancer. Am J Pathol 2003;163:1001 – 11. patients withadvanced malignancies. J Clin Oncol 2005;23:4152 – 61. 31. Bottone FG, Jr., Moon Y, Kim JS, et al. The anti-invasive activity of 18. Grem JL, Morrison G, Guo XD, et al. Phase I and pharmacologic study cyclooxygenase inhibitors is regulated by the transcription factor ATF3 of 17-(allylamino)-17-demethoxygeldanamycin in adult patients with solid (activating transcription factor 3). Mol Cancer Ther 2005;4:693 – 703. tumors. J Clin Oncol 2005;23:1885 – 93. 32. Chen HH, Wang DL. Nitric oxide inhibits matrix metalloproteinase-2 19. HeathEI, Gaskins M, Pitot HC, et al. A phaseII trial of 17-allylamino- expression via the induction of activating transcription factor 3 in 17-demethoxygeldanamycin in patients with hormone-refractory meta- endothelial cells. Mol Pharmacol 2004;65:1130 – 40. static prostate cancer. Clin Prostate Cancer 2005;4:138 – 41. 33. Fan F, Jin S, Amundson SA, et al. ATF3 induction following DNA 20. Sausville EA, Tomaszewski JE, Ivy P. Clinical development of 17- damage is regulated by distinct signaling pathways and over-expression of allylamino, 17-demethoxygeldanamycin. Curr Cancer Drug Targets 2003; ATF3 protein suppresses cells growth. Oncogene 2002;21:7488 – 96. 3:377 – 83. 34. Fawcett TW, Martindale JL, Guyton KZ, Hai T, Holbrook NJ. 21. Maloney A, Workman P. HSP90 as a new therapeutic target for Complexes containing activating transcription factor (ATF)/cAMP-respon- cancer therapy: the story unfolds. Expert Opin Biol Ther 2002;2:3 – 24. sive-element-binding protein (CREB) interact with the CCAAT/enhancer- binding protein (C/EBP)-ATF composite site to regulate Gadd153 expres- 22. Neckers L, Ivy SP. Heat shock protein 90. Curr Opin Oncol 2003;15: sion during the stress response. Biochem J 1999;339:135 – 41. 419 – 24. 35. Stearns ME, Kim G, Garcia F, Wang M. Interleukin-10 induced 23. Shimamura T, Lowell AM, Engelman JA, Shapiro GI. Epidermal activating transcription factor 3 transcriptional suppression of matrix growthfactor receptors harboring kinase domain mutations associate with metalloproteinase-2 gene expression in human prostate CPTX-1532 Cells. the heat shock protein 90 chaperone and are destabilized following Mol Cancer Res 2004;2:403 – 16. exposure to geldanamycins. Cancer Res 2005;65:6401 – 8. 36. Kastrinakis WV, Ramchurren N, Rieger KM, et al. Increased incidence 24. Isaacs JS, Xu W, Neckers L. Heat shock protein 90 as a molecular of p53 mutations is associated withhepaticmetastasis in colorectal target for cancer therapeutics. Cancer Cell 2003;3:213 – 7. neoplastic progression. Oncogene 1995;11:647 – 52. 25. Workman P. Combinatorial attack on multistep oncogenesis by 37. Georgakis GV, Li Y, Rassidakis GZ, et al. Inhibition of heat shock inhibiting the Hsp90 molecular chaperone. Cancer Lett 2004;206:149 – 57. protein 90 function by 17-allylamino-17-demethoxy-geldanamycin in 26. Lang SA, Klein D, Moser C, et al. Inhibition of heat shock protein 90 Hodgkin’s lymphoma cells down-regulates Akt kinase, dephosphorylates impairs epidermal growthfactor-mediated signaling in gastric cancer cells extracellular signal-regulated kinase, and induces cell cycle arrest and cell and reduces tumor growthand vascularization in vivo. Mol Cancer Ther death. Clin Cancer Res 2006;12:584 – 90. 2007;6:1123 – 32. 38. Park JW, Yeh MW, Wong MG, et al. The heat shock protein 90- 27. Price JT, Quinn JM, Sims NA, et al. The heat shock protein 90 binding geldanamycin inhibits cancer cell proliferation, down-regulates inhibitor, 17-allylamino-17-demethoxygeldanamycin, enhances osteoclast oncoproteins, and inhibits epidermal growth factor-induced invasion in formation and potentiates bone metastasis of a human breast cancer cell thyroid cancer cell lines. J Clin Endocrinol Metab 2003;88:3346 – 53. line. Cancer Res 2005;65:4929 – 38. 39. Sanderson S, Valenti M, Gowan S, et al. Benzoquinone ansamycin 28. Rakitina TV, Vasilevskaya IA, O’Dwyer PJ. Additive interaction of heat shock protein 90 inhibitors modulate multiple functions required for oxaliplatin and 17-allylamino-17-demethoxygeldanamycin in colon cancer tumor angiogenesis. Mol Cancer Ther 2006;5:522 – 32.

Mol Cancer Ther 2007;6(11). November 2007

Downloaded from mct.aacrjournals.org on September 30, 2021. © 2007 American Association for Cancer Research. Blocking heat shock protein-90 inhibits the invasive properties and hepatic growth of human colon cancer cells and improves the efficacy of oxaliplatin in p53-deficient colon cancer tumors in vivo

Christian Moser, Sven A. Lang, Silvia Kainz, et al.

Mol Cancer Ther 2007;6:2868-2878.

Updated version Access the most recent version of this article at: http://mct.aacrjournals.org/content/6/11/2868

Cited articles This article cites 39 articles, 18 of which you can access for free at: http://mct.aacrjournals.org/content/6/11/2868.full#ref-list-1

Citing articles This article has been cited by 4 HighWire-hosted articles. Access the articles at: http://mct.aacrjournals.org/content/6/11/2868.full#related-urls

E-mail alerts Sign up to receive free email-alerts related to this article or journal.

Reprints and To order reprints of this article or to subscribe to the journal, contact the AACR Publications Subscriptions Department at [email protected].

Permissions To request permission to re-use all or part of this article, use this link http://mct.aacrjournals.org/content/6/11/2868. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.