Download pdf - Inhibition of Heme Oxygenase-1Increases …clincancerres.aacrjournals.org/content/clincanres/11/10/3790.full.pdfInhibition of Heme Oxygenase-1Increases Responsiveness of Pancreatic

Inhibition of Heme Oxygenase-1Increases Responsiveness ofPancreatic Cancer Cells toAnticancer TreatmentPascal O. Berberat,1,2 Zilvinas Dambrauskas,1Antanas Gulbinas,1 Thomas Giese,2 Nathalia Giese,1

Beat Ku« nzli,1Frank Autschbach,3 StefanMeuer,2 Markus W. Bu« chler,1and Helmut Friess1

Abstract Heme oxygenase-1 (HO-1) is believed to represent a key enzyme for the protection of cellsagainst ‘‘stress.’’ Its overexpression in different types of human cancers supports the notionthat HO-1 provides a growth advantage and contributes to cellular resistance against chemo-therapy and radiotherapy. Given the poor survival rates of patients with pancreatic cancer dueto its aggressive growth behavior and its exceptional resistance to all known forms of antican-cer treatment, we have investigated the expression of HO-1 in human pancreatic cancer cellsgrowth behavior and prognosis. Expression of HO-1was analyzed in human pancreatic cancersamples in comparison with normal pancreas by quantitative PCR,Western blot, and confocalmicroscopy. The influence of radiotherapy and chemotherapy on HO-1expression in pancreaticcancer cell lines was evaluated. Furthermore, HO-1 expression was specifically suppressed bysmall interfering RNA transfection and subsequently the alterations of growth behavior andresistance to anticancer treatment were tested. Human pancreatic cancer showed a 6-foldand 3.5-fold HO-1up-regulation in comparison to normal pancreas based on mRNA and pro-tein level, respectively (P < 0.05). Cancer tissues revealed marked HO-1 immunoreactivity intumor cells and in tumor associated immunocytes. Treatment of the pancreatic cancer cell lineswith gemcitabine or radiation strongly induced HO-1expression. Targeted knockdown of HO-1expression led to pronounced growth inhibition of the pancreatic cancer cells and made tumorcells significantly more sensitive to radiotherapy and chemotherapy. Therefore, specific inhibi-tion of HO-1 expression may be a new option in pancreatic cancer therapy and may be usedas sensitizer to chemotherapy and radiotherapy.

Pancreatic cancer has a very poor prognosis and at presentsurgical resection is still the only chance for cure (1, 2).Unfortunately, most patients are diagnosed at advanced tumorstages with metastasis already in distant organs; therefore, thesepatients are no longer candidates for surgical treatment.Furthermore, even after a so-called ‘‘curative resection,’’ the 5-year survival rate is 5% to 20%. Current chemotherapy andradiotherapy regimens are not very effective (3). Therefore,additional treatment options for advanced disease adjuvant tosurgical therapy are urgently needed.

The heme oxygenase (HO) system catalyzes the degradationof heme to produce equimolar quantities of biliverdin, CO, andfree iron (4). Subsequently, biliverdin is converted to bilirubinby cytosolic biliverdin reductase, and free iron is promptly

sequestered into ferritin (5, 6). To date, three HO isoforms(HO-1, HO-2, and HO-3) that catalyze this reaction have beenidentified (7–9). HO-1 is a 32-kDa inducible heat shockprotein, which is found at low levels in most mammaliantissues but is highly induced by a variety of stress stimuli,including heat shock (10), UV irradiation (11), hydrogenperoxide (12, 13), heavy metals (14, 15), hypoxia (16), andcytokines (17, 18). Recent findings indicate that HO-1 and itsproducts possess anti-inflammatory and antiapoptotic func-tions (19–22). It represents a key biological molecule in theadaptive response to cellular stress. Moreover, new studiessuggest that HO-1 exerts also a role in controlling growth andcell proliferation in a cell-specific manner. Elevated HO-1expression and activity was found in various tumors such ashuman renal cell carcinoma (23), prostate tumors (24), andlymphosarcomas (25). In human gliomas and melanomas,HO-1 is linked to angiogenesis (26–28), and in an experimen-tal mouse model, HO-1 accelerates pancreatic cancer growth bypromoting tumor angiogenesis (28).

These findings suggest that HO-1, with its proangiogenic andgrowth-regulative properties, may also play a crucial role in thedevelopment and progression of pancreatic cancer. Further-more, its anti-inflammatory and antiapoptotic activity impliesthat HO-1 may enhance radioresistance and chemoresistance inpancreatic cancer cells. We, therefore, hypothesized that theinhibition of HO-1 expression and activity may sensitizepancreatic cancer cells to anticancer treatment modalities.

www.aacrjournals.orgClin Cancer Res 2005;11(10) May 15, 2005 3790

Authors’Affiliations: 1Division of Pancreatic Surgery and Molecular PancreaticResearch, Department of General Surgery, 2Institute of Immunology, and3Department of Pathology, University of Heidelberg, Heidelberg, GermanyReceived10/21/04; revised 2/17/05; accepted 2/22/05.The costs of publication of this article were defrayed in part by the payment of pagecharges.This article must therefore be hereby marked advertisement in accordancewith18 U.S.C. Section1734 solely to indicate this fact.Note: P.O. Berberat and Z. Dambrauskas contributed equally to this work.Requests for reprints:HelmutFriess,DepartmentofGeneralSurgery,UniversityofHeidelberg, ImNeuenheimerFeld110,69120Heidelberg,Germany.Phone: 49-6221-564860; Fax: 49-6221-566903; E-mail: [email protected].

F2005 American Association for Cancer Research.

Imaging, Diagnosis, Prognosis

Research. on July 2, 2018. © 2005 American Association for Cancerclincancerres.aacrjournals.org Downloaded from

http://clincancerres.aacrjournals.org/

Materials and Methods

Patients and tissue collection. Twenty-seven ductal pancreatic cancersamples were obtained from 14 male and 13 female patients (medianage, 66 years; range, 36-81 years) who underwent pancreatic resectionsbecause of ductal pancreatic cancer at the University Hospital of Berne(Switzerland) and Heidelberg (Germany). According to the tumor-node-metastasis and histopathologic grading system of the Interna-tional Union Against Cancer, there were four stage I, four stage II, 14stage III, and five stage IV tumors.

Normal pancreatic tissue samples were obtained from 20 individ-uals (7 women and 13 men) who were free of pancreatic diseasethrough an organ donor program. The median age of the organdonors was 46.2 years (range, 20-74 years). All normal tissue sampleswere obtained from the head of the organ donor’s pancreas to ensurecomparability with the tumor samples. In all experiments, tissuesections of normal and cancerous pancreas samples were processedsimultaneously to ensure comparability of the results. Freshly removedtissue samples were immediately fixed in formaldehyde solution for12 to 24 hours and embedded in paraffin for immunohistochemistryand confocal microscopy. Simultaneously, tissues for RNA extractionwere put in RNA Later (Ambion, Huntingdon, United Kingdom)whereas tissue for protein extraction were snap-frozen in liquidnitrogen in the operating room upon surgical removal and maintainedat �80jC until use. The ethical committee of the University of Bernand Heidelberg approved the study and informed consent wasobtained from each patient.

Cell culture. The established human pancreatic cancer cell linesPanc-1, MIaPaCa-2, SU8686, and Colo 357 were used in theseexperiments. All cells were obtained from American Type CellCollection (Manassas, VA).

The PANC-1 and MIA PaCa-2 cells were cultured in Dulbecco’smodified Eagle high-glucose medium with Na pyruvate (Cell Concepts,Umkirch, Germany) supplemented with 10% fetal bovine serum (PANBiotech GmbH, Aidenbach, Germany), 4 mmol/L Glutamine-L (LifeTechnologies, Paisley, United Kingdom), and a penicillin-streptomycinsolution (100 units/mL and 100 Ag/mL, respectively; Life Technolo-gies). The cells were maintained at 37jC in a humidified 5% CO2

atmosphere for at least 72 hours before each experiment and harvestedafter treatment with trypsin-EDTA (Life Technologies).

The SU8686 and Colo 357 cells were cultured in RPMI 1640 with L-glutamine (Life Technologies) supplemented with 10% fetal bovineserum (PAN Biotech) and a penicillin-streptomycin solution (100units/mL and 100 Ag/mL, respectively; Life Technologies). The cellswere maintained at 37jC in a humidified 5% CO2 atmosphere for atleast 72 hours before each experiment and harvested after treatmentwith trypsin-EDTA (X1; Life Technologies).

Heme oxygenase-1 small interfering RNA transfection. To study theeffects of HO-1 inhibition on cell viability, growth rate, induction ofapoptosis, and sensitization to anticancer treatment, HO-1 expressionwas specifically suppressed by introduction of 21-nucleotide duplexsmall interfering RNA (siRNA), which targets nucleotides 612 to 630of the HO-1 mRNA coding sequence (29). The sequences ofthe ribonucleotides used were 5V-rGACUGCGUUCCUGCUCAACdTdT-3V and 5V-rGUUGAGCAGGAACGCAGUCdTdT-3V (Ambion, Inc., Aus-tin, TX; ref. 30). The cells (PANC-1, MIA PaCa-2, SU8686, and Colo357) were each plated on 6-well plates (Nunc, Roskilde, Denmark) indensity of 100,000 cells per well and were preincubated overnight, afterwhich 2 Ag per well of siRNA was introduced into the cells by usingLipofectAMINE 2000 (Invitrogen, Carlsbad, CA) according to themanufacturer’s directions. The Silencer Negative Control siRNA 1(Ambion, Austin, TX) was used as a negative control and wasintroduced into the cells using the same protocol.

3-(4,5-Dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide assay.

The 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromidecell proliferation assay was done to evaluate the effect of transfectionwith siRNA for HO-1 mRNA (siRNA) on cell growth rate and

viability comparing to the transfection with negative control siRNA(control). Cells were seeded in 96-well culture plates (3,000 cells perwell) and after overnight preincubation, cells were transfected with0.3 Ag per well siRNA for HO-1 or negative control siRNA asdescribed above. Measurements were done 24, 48, and 72 hourspost-transfection and the graphical data represent the percentage ofsurviving cells in the siRNA group compared with the vehicle groupat given time points. Control nontransfected cells were plated on thesame experiment plates to assess the normal proliferation rate of thecells.

Fluorescence-activated cell sorting analysis. The cells (Panc-1,

MIaPaCa-2, SU8686, and Colo 357), transfected with siRNA for HO-1

mRNA or negative control siRNA as described above, were exposed to

anticancer treatment, such as g-radiation 20 Gy in ‘‘Gammacell 1000’’

(Atomic Energy of Canada, Ltd., Ottawa, Canada) or gemcitabine HCl

(Eli Lilly and Company, Indianapolis, IN) in cell line–specific EC50

dose (determined on the nontransfected cells; 4 Ag/mL for Colo 357,

8 Ag/mL for SU8686, 14 Ag/mL for MiaPaCa-2, and 150 Ag/mL for

Panc-1) for 48 hours. As control, cells were exposed to oxidative stress,

a known inducer of HO-1. Oxidative stress was induced by 100 mmol/L

1,1V-dimethyl-4,4V-biperidinum dichloride and 100 mmol/L dd-

buthionine-[S ,R]-sulfoximine (Sigma-Aldrich GmbH, Steinheim, Ger-

many) for 48 hours. dd-Buthionine-[S ,R]-sulfoximine is very potent

irreversible inhibitor of y-glutamylcysteine synthetase, which leads to

significant depletion of cellular glutathione levels (31). 1,1V-Dimethyl-4,4V-biperidinum dichloride, a so-called redox-cycling drug,

generates continually superoxide (32). The combination of both

induces pronounced oxidative stress to the cells (33, 34).Some cells were grown for 48 hours after transfection without any

treatment to determine the effect of HO-1 suppression on cell viabilityand to compare these results with the cell proliferation assay.

For analysis, cells were trypsinized, harvested, and added to thecorresponding supernatant in fluorescence-activated cell sorting (FACS)tubes. Then the pooled cells were pelleted, washed with cold PBS (CellConcepts), and resuspended while vortexing in 0.4 mL of cold PBScontaining 2% fetal bovine serum and 2.5 Ag/mL propidium iodide(Sigma-Aldrich, Steinheim, Germany). After incubating the cells at 4jCfor 10 minutes, flow cytometry with the BD LSR ImmunocytometrySystem (BD Biosciences, San Jose, CA) was used to analyze thepropidium iodide staining and to determine the percentage of viableand necrotic cells. The results are presented as a percentage of viablecells in each group.

Western blot analysis. Total proteins were extracted using Laemmlibuffer from treated cultured pancreatic cancer cells and control nativecells. From pancreatic tissues (normal pancreas and pancreatic cancer)total protein was obtained by lyzing the appropriate tissue sample(f500 Ag) in a buffer containing 20 mmol/L Tris (pH 7.5), 100 mmol/LNaCl, 1% NP40, and protease inhibitor cocktail for mammalian tissue50 AL per sample (Sigma, St. Louis, MO).

Total proteins were electrophoresed in NuPage MOPS SDS RunningBuffer, NuPage 10% Bis-Tris Gel (Invitrogen) at 200 V. Using thesemidry transfer at 0.04 A constant for 1 hour, the proteins were blottedon to Invitrolon polyvinylidene difluoride 0.45-Am pore membrane(Invitrogen). After blocking with 8% milk, the membrane wasincubated in a milk solution containing anti–HO-1 mouse monoclo-nal antibody (BD Biosciences Transduction Laboratories, Lexington,KY). The bound antigen-antibody complex was detected by secondaryperoxidase-conjugated affinity purified goat anti-mouse antibody(Jackson ImmunoResearch Laboratories, Inc., West Grove, PA) andthe chemiluminescence kit (Perkin-Elmer Life Sciences, Inc., Boston,MA). The same membranes were used for blotting with anti-g-tubulinor anti-GABDH (Santa Cruz Biotechnology, Inc., Santa Cruz, CA)antibody as an internal loading control.

All the Western blot films were scanned with GS-800 CalibratedDensitometer (Bio-Rad Laboratories, Inc., Hercules, CA) and furtherquantified using the Quantity One Quantification Software accordingto the manufacturer’s guidelines (Bio-Rad Laboratories).

www.aacrjournals.org Clin Cancer Res 2005;11(10) May 15, 20053791

HO-1in Pancreatic Cancer



Immunohistochemistry. Frozen pancreatic tissue sections (5-7 Am)were air-dried and fixed in a mixture of ice-cold acetone and 10%formaldehyde for 5 minutes at �20jC.

In paraffin sections, antigen retrieval was completed by microwavecooking at 800 W for 5 minutes (10 minutes cooling at roomtemperature) and cooking at 400 W 2 � 5 minutes in 10 mmol/Lsodium citrate buffer followed by incubation with 0.05% saponin inPBS for 20 minutes at room temperature.

Further procedures for imunohistochemical detection of HO-1 wereidentical for frozen and paraffin sections. Sections were incubated witha rabbit polyclonal antibody to human HO-1 (Stressgen Biotechnolo-gies Co., Victoria, British Columbia, Canada), diluted 1:250. A mousemonoclonal anti-rabbit immunoglobulin G antibodies, diluted 1:100(DAKO A/S, Glostrup, Denmark), was used as secondary antibody. Fornegative controls, incubation with full rabbit serum in dilutioncorresponding to the concentration of the primary antibody was used.

Fluorescent microscopy. The cells were plated on Teflon-coated slidesand fixed by ice-cold methanol for 5 minutes and acetone for 2minutes, before 0.05% saponin was applied for 20 minutes at roomtemperature. After rinsing in PBS, nonspecific binding was blocked bypreincubating sections with 10% normal goat serum in TBS containing10% g-avenin and 0.1% BSA-c for 1 hour at room temperature. Sectionswere incubated with a mouse monoclonal antibody to human HO-1(BD Biosciences Transduction Laboratories), diluted 1:25 or 1:50 in10% normal goat serum TBS containing 0.05% Tween 20 in 4jC ON.

After rinsing thrice in TBS Tween 0.05% with 1.0% bovine serumalbumin, sections were incubated for 1 hour at room temperaturewithout light exposure with goat anti-mouse Cy3-labeled antibodies,diluted 1:400 (Jackson ImmunoResearch Laboratories). Then the slideswere further incubated with Toto-3 iodide 1:1,000 (Molecular Probes,Eugene, OR). For negative controls, sections were processed asdescribed earlier, except that the incubation with mouse IgG (DAKO)in dilution 1:10 or 1:20 was employed instead of the primary antibody.

Heme oxygenase-1 activity. Benzene was purchased from AppliChemGmbH (Darmstadt, Germany), and barium chloride-2-hydrate fromSigma-Aldrich (Seelze, Germany). Then cells were lyzed by adding 100AL of 10� Triton 100-X to each well for 1 hour and 1.0 mL of eachculture supernatant was collected into corresponding tubes, and 500mg of BaCl2 were added to each sample. After vortexing (10-15seconds), 1.5 mL benzene were added; tubes were vortexed vigorously(50-60 seconds) leading to the formation of a relatively stable milky-white emulsion. After centrifugation (13,000 � g , 30 minutes, withoutcooling), the upper benzene layer was collected and the absorbance at450 nm with reference wavelength at 570 nm was measured using aShimadzu UV-160A spectrophotometer. In a separate tube, 1.0 mL offresh culture medium was processed in the same way and the benzenelayer was collected and used as a blank. The quantity of bilirubinproduced was calculated using a molar extinction coefficient ofbilirubin dissolved in benzene E450 = 27.3 (mmol/L)�1/cm�1.

Statistical analysis. The data are presented as mean F SE or medianand range. P < 0.05 was considered statistically significant. Forcomparison between groups, the Mann-Whitney test was employed.

Results

Heme oxygenase-1 is overexpressed in human pancreaticcancer. To determine whether HO-1 plays a role in humanpancreatic cancer, mRNA and protein levels in humanpancreatic cancer samples (n = 27) were compared withnormal pancreatic tissue (n = 20). Using quantitative reversetranscription-PCR, a significant overexpression of HO-1 mRNAwas detected in all cancer samples (27 of 27). On average, a 6-fold up-regulation of HO-1 mRNA (P < 0.05) in the pancreaticcancer samples compared with the normal pancreatic tissue wasobserved (Fig. 1A). Western blot analysis revealed similarresults on the protein level: pancreatic cancer tissue showed on

average 3.5-fold higher HO-1 levels in comparison with normalpancreatic tissues (P < 0.05; Fig. 1B).

Heme oxygenase-1 is found in tumor-associated immunocytesand pancreatic cancer cells. Next, the localization of HO-1was determined in normal and cancer tissues. Immunohisto-chemistry analysis using an aHO-1 monoclonal antibodyrevealed that HO-1 is located in pancreatic cancer cells as wellas in immunocytes located next to the tumor mass (Fig. 1C).

Pancreatic cancer cell lines showed varying HO-1 mRNAexpression levels: PANC-1 and Colo 357 cell lines had very highHO-1 expression levels, whereas MIaPaCa-2 showed low levels ofHO-1 mRNA expression. In SU8686 cells HO-1 mRNA expres-sion was not detectable (Fig. 2A and B). By confocal microscopystrong HO-1 signals were present in the cytoplasm of all Panc-1and Colo 357 cells, whereas only some MiaPaCa-2 cells and noneof the SU8686 cells showed positive staining (Fig. 2C).

Heme oxygenase-1 can be induced in pancreatic cancer celllines. To determine whether HO-1 is inducible in pancreaticcancer, tumor cells were exposed to exogenous stress in vitroalong with commonly used anticancer treatment modalitiessuch as radiotherapy and chemotherapy.

When exposed to continuous oxidative stress, induced bydd-buthionine-[S,R]-sulfoximine and 1,1V-dimethyl-4,4V-biper-idinum dichloride, all four pancreatic cancer cell linesshowed a transient up-regulation of HO-1 (Fig. 3). Quanti-fication data from the Western blot revealed that oxidativestress increased the expression of HO-1 protein in Colo 357by 1.2 times, in Panc-1 by 1.5 times, in MiaPaCa-2 by 3.5times, and in SU8686 by 8.2 times compared with thenontreated controls. Next, we treated the pancreatic cancercells with an EC50 dose of gemcitabine for 24 hours or a 20Gy bolus dosis of g-radiation in vitro. Both treatmentsresulted in multiple fold increases of HO-1 in all the celllines and to de novo expression in SU8686, which showed nodetectable basal levels of HO-1 (Fig. 3): treatment withgemcitabine increased the expression of HO-1 in Colo 357by 2.1 times, in Panc-1 by 1.3 times, in MiaPaCa-2 by 1.3,and in SU8686 by 9.6 times, compared with the nontreatedcontrols. The 20 Gy bolus dosis of g-radiation resulted insimilar changes in the expression of HO-1, with 2.9-foldincrease in Colo 357, 1.3-fold increase in Panc-1, and 8.8-fold increase in SU8686. There was no significant inductionof HO-1 observed in the MiaPaCa-2 cell line.

Targeted inhibition of heme oxygenase-1 activity results indiminished pancreatic cancer cell proliferation. To determinewhether the induced high expression levels of HO-1 areresponsible for the high proliferation index of pancreaticcancer cells, we suppressed HO-1 expression through siRNAtransfection of a 21-nucleotide duplex HO-1 siRNA andevaluated the growth behavior of the cancer cells. The SilencerNegative Control siRNA 1 was used as a negative control. 48hours post-transfection, Western blot analysis revealed asignificant decrease in HO-1 levels in all tested cancer celllines, compared with the negative control (Fig. 4A): in Colo357, siRNA transfection down-regulated the HO-1 protein levelby 3.5 times compared with the control. In Panc-1, the HO-1level was 3.8 times lower after siRNA transfection. In MiaPaCa-2 and SU8686 HO-1 levels were 2.3 and 1.6 times lower than inthe control group, respectively. Furthermore, the activity of HO-1 was decreased accordingly in all four cell lines: in Colo 357,siRNA transfection decreased bilirubin level by 33 F 2%





compared with the control; in Panc-1, bilirubin level was 32 F6% lower after siRNA transfection, in MiaPaCa-2 and SU8686,it was 21 F 4% and 36 F 2%, respectively, lower than incontrol group (P < 0.037; Fig. 4B).

Subsequently, the proliferation rates of the pancreatic cancercells were evaluated at different time points by 3-(4,5-dime-thylthiazol-2-yl)-2,5-diphenyltetrazolium bromide assay.Down-regulation of HO-1 activity was associated with asignificant inhibition of the proliferation rate in all fourpancreatic cancer cell lines (Fig. 5). The proliferation decreasedin comparison with the controls in Panc-1 by 24 F 2%, inMiaPaCa-2 by 24 F 19, in Colo 357 by 25 F 6%, and in SU8686by 18 F 5%. The evaluation of cell viability at 72 hours revealedthat at least partly this effect is based on a diminished survival

of the cancer cells; decrease of viability in comparison withcontrols: Panc-1 by 13 F 7, MiaPaCa-2 by 12 F 3%, Colo 357by 8 F 5%, and SU8686 by 7 F 6% (all P < 0.05; Fig. 5). Incontrast, there was no statistically significant differencebetween native cells and cells transfected with control siRNA(data not shown).

Targeted inhibition of heme oxygenase-1 sensitizes pancreaticcancer cells to oxidative stress, radiotherapy, and chemotherapy.The anti-inflammatory and antiapoptotic activities of HO-1help cells to adapt to stress and injury. To determine whetherthe radioresistance and chemoresistance of pancreatic cancercells are due to high levels of HO-1 expression, we exposedcancer cells with inhibited HO-1 activity to oxidative stress aswell as radiotherapy and chemotherapy.


Fig. 1. HO-1 is overexpressed in humanpancreatic cancer. By quantitative PCRwedetected a significant (P < 0.03)overexpression of HO-1mRNA in100%(27 of 27) of the cancer samples comparedwith normal pancreas tissue. On average,a 6-fold up-regulation of HO-1mRNA(P < 0.05) in the pancreatic cancer samplescompared with the normal pancreatic tissuewas observed (A).Western blot withaHO-1mouse monoclonal antibody furthersupports the real-time quantitative PCRdata.TheWestern blot quantitative analysisrevealed similar results for the proteincontent: pancreatic cancer tissue showedon average a 3.5-fold (P < 0.05) higherexpression of HO-1in comparisonwithnormal pancreas tissue (B).Immunohistochemistry analysis using anaHO-1mouse monoclonal antibodyrevealed that HO-1is located in pancreaticcancer cells as well as in tumor-associatedimmunocytes and is overexpressed inpancreatic cancer tissue (magnification,�40) compared with the normal pancreastissue (magnification,�20; C). H&Estaining of the frozen sections; HO-1staining with rabbit polyclonal antibodyfor HO-1.




Forty-eight hours after transfection of HO-1 siRNA, thepancreatic cancer cells were exposed to dd-buthionine-[S,R]-sulfoximine and 1,1V-dimethyl-4,4V-biperidinum dichloride(which mediates oxidative stress), gemcitabine (EC50 dose),or g-radiation (20 Gy). Seventy-two hours later, cell viabilitywas evaluated. HO-1 expression and activity was still sup-pressed at this time. The inhibition of HO-1 sensitizespancreatic cancer not only to oxidative stress but also toradiotherapy and chemotherapy. We found that decreasedexpression and activity of HO-1 resulted in increased cellsensitivity to oxidative stress and decreased the numbers ofviable cells by 12% to 20% (P < 0.05; Fig. 6A). A 5% to 9%reduction in viability was observed after HO-1 inhibition and

chemotherapy with gemcitabine (P < 0.05; Fig. 6B). Finally, asignificant sensitization effect of HO-1 inhibition was alsoobserved towards radiotherapy (P < 0.05; Fig. 6C), with anoverall reduction of 13% to 23% in cell viability.

Discussion

A wide range of stress stimuli in all cells induces HO-1. Inseveral experimental systems, it was shown that the inductionof HO-1 is a fundamental self-defense process of cells toassaults from the environment (35). The anti-inflammatoryand antiapoptotic effects of HO-1 and its metabolites protectcells, tissues, and even whole organs. Furthermore, HO-1


Fig. 2. HO-1is overexpressed in pancreatic cancercell lines. Quantitative real-time PCR data showsthat the expression of HO-1mRNA in differentpancreatic cancer cell lines varies considerably.PANC-1and Colo 357 cell lines have the highestHO-1mRNA levels, whereas MIAPaCa-2 andSU.86.86 expressed significantly lower levels ofHO-1mRNA (A). InWesternblot withaHO-1mousemonoclonal antibody, the high levels of HO-1 pro-tein are detected in cell lines PANC-1and Colo 357,whereas MIAPaCa-2 expresses 3- to 4-fold lowerlevels of HO-1, and it is below visual detectionlevel in the SU.86.86 cell line (B).Theimmunofluorescent confocal microscopy withaHO-1mouse monoclonal antibody corresponds tothe above data and shows that HO-1is located incytoplasm of all Colo 357 (aHO-1,1:50,�160) andPanc-1 (aHO-1,1:25,�160) cells, but only some ofMiaPaCa-2 (aHO-1,1:50,�160) cells expressed it,and it was not detectable in SU8686 (aHO-1,1:50,�160; C).




influences cell growth in a specific manner (36). Based on theseobservations, several new experimental treatment options areunder development for various inflammatory disorders, trans-plant rejection, endotoxic shock, artheriosclerosis, and respira-tory diseases (37, 38). HO-1 induction inhibits theinflammatory reactions or even prevents or ameliorates thecourse of the disease (35).

Different human cancers express high levels of HO-1, whichmay provide them a growth advantage (39). Therefore,inhibition of HO-1 might lead to reduced tumor growthin vitro and in vivo (11, 28, 30).

The data presented here show that HO-1 expression issignificantly up-regulated in human pancreatic cancer com-pared with normal pancreatic tissue. A strong overexpressionwas found in cancer cells but also in immunocytes surroundingthe cancer cells.

In human gliomas and melanomas, for example, only tumor-associated macrophages were expressing HO-1 (26, 27, 40, 41),whereas in human prostate, esophageal, tongue, and renalcancer only the tumor cells revealed strong HO-1 signals(23, 24, 42, 43). Expression of HO-1 in macrophages withinthe tumor seems to correlate with increasing vascular densityand therefore ongoing neovascularization, which is known topromote local tumor growth (26, 27, 40, 41). In vitro and in vivostudies in a pancreatic cancer model showed that HO-1stimulates angiogenesis and increases endothelial cell survival(28, 44–47). Furthermore, in melanoma and oligodendro-glioma, expression levels of HO-1 in macrophages correlatewith tumor cell invasiveness and poor prognosis (27, 40).These data imply that HO-1 plays a role in tumor angiogenesismediated by activated macrophages and also mediates anti-apoptotic effects on endothelial cells.

The strong HO-1 positivity in pancreatic tumor cells in situand in several pancreatic cancer cell lines points to yet anotherrole of HO-1 in tumor pathogenesis. The expression level variedamong cancer cells and also in different cell lines. Directcorrelation of the expression level with clinical or histomor-phologic data could, however, not be established in this study.In contrast, others observed some prognostic value of HO-1expression in tumors, although with contradictory results:


Fig. 3. Adjuvant treatment leads to HO-1induction.Western blot shows that24 hours of exposure to oxidative stress (induced by100 mmol/L dd-buthionine-[S,R]-sulfoximine and100 mmol/L1,1V-dimethyl-4,4V-biperidinum dichloride),treatment of the pancreatic cancer cells with EC50 dose of gemcitabine for 24 hours,and 20Gybolus dosis of g-radiation resulted in several fold increase of HO-1in all thecell lines and it was also detectable in SU8686 compared with the control.

Fig. 4. Target siRNA down-regulates HO-1expression in cancer cells.Western blotswith aHO-1mouse monoclonal antibody (1:200) and densitometric quantificationof theseblots showed that the transfectionof siRNAforHO-1mRNA(siRNA) resultsin a several fold decrease of HO-1expression compared with the negative controlsiRNA transfection (A). In Colo 357, siRNA transfection down-regulated HO-1by3.5 times comparedwith the control. In Panc-1, HO-1level was 3.8 times lower aftersiRNA transfection. InMiaPaCa-2 and SU.86.86, it was, respectively, 2.3 and1.6times lower than in the control group. Bilirubin level measurements in cell growthmedium showed that that the transfection of siRNA for HO-1mRNA (siRNA) resultsin a decrease of bilirubin production compared with the negative control siRNAtransfection.This further confirms our notion that the expression andfunctional activity of HO-1canbe successfully down-regulated bymeans of specificsiRNA interference. In Colo 357, siRNA transfection decreased bilirubin levels by33F 2% compared with the control; in Panc-1, bilirubin level was 32F 6% lowerafter siRNA transfection; inMiaPaCa-2 and SU.86.86, it was, respectively, 29F 4%and 36F 2% lower than in the control group (*, P < 0.037; B).




Yanagawa et al. reported that low level HO-1 is associated withan increased risk of developing lymph node metastasis intongue squamous cell carcinoma (42). On the other hand, inprostate and melanoma cancers, increasing HO-1 expressionwas associated with progression (24, 40). Further studies usinghigher numbers of human pancreatic tumor samples andcomparison with inflammatory diseases of the pancreas will benecessary to illustrate the potential prognostic and diagnosticrole of HO-1 in pancreatic carcinoma.

Here we show that HO-1 expression is directly associatedwith the proliferation and/or survival of pancreatic cancer cells.Specific inhibition of HO-1 expression and activity by siRNAtranfection leads to diminished proliferation, which is onlypartly based on cytotoxicity. As the specific inhibition of HO-1has only modest effect on the viability of the pancreatic cancercells, the clearly decreased proliferation rate may be based ondirect effect on the cell cycle.

It seems that HO-1 also acts as an endogenous protectionmechanism in pancreatic cancer cells, which potentially makes


Fig. 5. Down-regulation of HO-1 leads to growth inhibition of pancreatic cancercells. 3-(4,5-Dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide assay showsthat down-regulationof HO-1with siRNA leads to the statistically significant growthinhibition of pancreatic cancer cell lines Colo 357 with P < 0.04, Panc-1withP < 0.03,MiaPaCa-2withP <0.03, and SU.86.86withP <0.04, comparedwith thecontrol group, that was equal to100% at every time point and corresponding valuesof the siRNA groupwere calculated based on this condition. Columns, results ofFACS/propidium iodide analysis 72 hours post-transfection, which show that thenumber of viable cells statistically significantly decreases in siRNA group comparedwith the control group.

Fig. 6. Down-regulation of HO-1expression sensitizes cancer cells to anticancertherapy. Our data shows that the down-regulation of HO-1with siRNA sensitizesthe pancreatic cancer cells to adjuvant treatment. FACS analysis with propidiumiodide staining proves that down-regulation of HO-1lowers the number of viablecells further by12% to 20% (*, P < 0.05) after exposure to oxidative stress incomparisonwith the control group (A). Decreased expression and activity of HO-1also results in increased cell sensitivity to gemcitabine and decreased number ofviable cells by 5% to 8% (**, P < 0.05), as shown by FACS analysis (B). FACSanalysis also confirms that down-regulationof HO-1sensitizes the pancreatic cancercells to radiotherapy, thus resulting in a decreased number of viable cells by13% to23% (***, P < 0.05) after this treatment (C).




cells more resistant to oxidative stress and apoptotic stimuli,thus providing growth advantage. Moreover, the HO-1 systemis known to have direct effects on various components of thecell cycle machinery, which leads to cell-specific promotion orinhibition of cell proliferation depending on the cell type (36).Because tumor growth is based on augmented cell growth andprolonged cell survival (based on resistance to apoptosis), it ispossible that HO-1 develops its tumor-promoting activity byincreasing cell proliferation and protection towards apoptosis.

More importantly, we show that HO-1 may representmechanism for pancreatic cancer cells towards anticancer drugsand radiotherapy. The destructive mechanisms of radiotherapyas well as chemotherapy on cancer cells are mainly based on thegeneration of oxidative stress and/or induction of apoptosis(48). In our present, study we show that radiation as well asgemcitabine (both widely used therapeutic approaches forpancreatic cancer today) lead to the induction of HO-1 inpancreatic cancer cells. This underlines the role of HO-1 as astress response gene, showing pronounced up-regulation afterfirst time treatment, leading to the resistance of further therapy,a generally known clinical problem in cancer treatment. Thatinhibition of HO-1 expression and activity in pancreatic cancercells having high constitutive HO-1 levels enhances theirresponse to radiotherapeutic and chemotherapeutic supportsthis view. Recent studies in other models made similarobservations: induction of HO-1 made leukemia and coloncancer cells resistant to merocyanine and pyrrolidine dithio-carbamate, two experimentally used chemotherapeutic drugs(49, 50). In an experimental murine sarcoma model pretreat-ment with the HO-1 inhibitor PEG-ZnPP augmented thecytotoxic effect of several chemotherapeutics (51).

The molecular basis of these effects on cancers is not yet clear.From other cells and experimental systems, one mighthypothesize that three processes induce resistance of cancercells: (I) HO-1 diminishes the pro-oxidant pool intracellular bydegrading heme, increasing cellular iron efflux and inducingthe synthesis of ferritin, a strong anti-oxidant; (II) HO-1metabolites CO and biliverdin/bilirubin revealed antiapoptoticand inflammatory effects; (III) through direct protein-protein

interaction, HO-1 may have further enzyme-independentprotective functions (35, 52). Furthermore, HO-1 does alsodirectly influences several cell signaling pathway, such asmitogen-activated protein kinases and prostaglandins, whichare known to have crucial roles in pancreatic cancer pathogen-esis (53). Most probably, ROS play a crucial role, as we showthat pure oxidative stress is significantly more effective in killingpancreatic cancer cells after HO-1 inhibition. Irradiationrevealed similar enhancement of killing of pancreatic cancercells, whereas the alteration of the gemcitabine effects was weakto modest. Other HO-1– and ROS-independent pathways maybe in case of gemcitabine of more importance.

In future studies, the known pathways have to be evaluatedin cancer models both in vitro and in vivo. This will help us gaina better understanding of the treatment resistance processes,especially in pancreatic cancer. We believe further that specifictumor-targeted or even systemic inhibition of HO-1 or itsmetabolites in vivo could be of great therapeutic benefit becauseit could sensitize tumor cells to the oxidative burst frominflammatory cells (54, 55). It may also enhance theinflammatory response in tumor tissue and reinforce the hostdefense system by increasing tumor necrosis factor-a produc-tion in macrophages, as some authors suggest that CO derivedfrom heme degradation could potentially inhibit the produc-tion of this proinflammatory cytokines in activated macro-phages (20, 56).

In conclusion, high HO-1 levels in pancreatic cancer cellsmay, at least partly, be responsible for their resistance toanticancer therapy. Furthermore, high HO-1 expression seen inhuman pancreatic cancer may act as endogenous protectionmechanism, which is initiated during carcinogenesis to makepancreatic cancer cells resistant to environmental stress such asimmune attack, hypoxia, and nutrition deprivation. A betterunderstanding of the exact molecular processes of the HO-1system in pancreatic carcinogenesis may lead to new thera-peutic targets. Perhaps, more important, specific inhibition ofHO-1 activity in combination with conventional radiotherapyand chemotherapy may open up new perspectives in thetreatment of pancreatic cancer.


References1. Boring CC, SquiresTS,Tong T, Montgomery S. Can-cer statistics,1994. CACancerJClin1994;44:7^26.

2. Bramhall SR, NeoptolemosJP. Advances in diagnosisand treatment of pancreatic cancer. Gastroenterology1995;3:301^10.

3. Ghaneh P, Greenhalf W, Humphreys M, et al. Ade-novirus-mediated transfer of p53 and p16(INK4a)results in pancreatic cancer regression in vitro andin vivo. Gene Ther 2001;8:199^208.

4. Otterbein LE, Choi AM. Heme oxygenase: colors ofdefense against cellular stress. AmJPhysiol LungCellMol Physiol 2000;279:1029^37.

5. MainesMD.The heme oxygenase system: a regulatorof second messenger gases. Annu Rev PharmacolToxicol1997;37:517^54.

6. Schacter BA. Heme catabolism by heme oxygenase:physiology, regulation, and mechanism of action.Semin Hematol1988;25:349^69.

7. Maines MD. Heme oxygenase: function, multiplicity,regulatory mechanisms, and clinical applications.FASEB J1988;2:2557^68.

8. McCoubrey WK Jr, Ewing JF, Maines MD. Humanheme oxygenase-2: characterization and expressionof a full-length cDNA and evidence suggesting thatthe two HO-2 transcripts may differ by choice of pol-

yadenylation signal. Arch Biochem Biophys 1992;295:13^20.

9. Shibahara S, Yoshizawa M, Suzuki H, Takeda K,Meguro K, Endo K. Functional analysis of cDNAs fortwo types of human heme oxygenase and evidencefor their separate regulation. J Biochem (Tokyo)1993;113:214^8.

10. Stuhlmeier KM. Activation and regulation of Hsp32and Hsp70. EurJBiochem 2000;267:1161^7.

11. Doi K, Akaike T, Fujii S, et al. Induction of haemoxygenase-1 nitric oxide and ischaemia in experi-mental solid tumours and implications for tumourgrowth. Br J Cancer 1999;80:1945^54.

12. Keyse SM,Tyrrell RM. Heme oxygenase is the major32-kDastressproteininducedinhumanskinfibroblastsbyUVAradiation,hydrogenperoxide,andsodiumarse-nite. ProcNatlAcadSciUSA 1989;86:99^103.

13. Lautier D, Luscher P, Tyrrell RM. Endogenousglutathione levels modulate both constitutive andUVA radiation/hydrogen peroxide inducible expres-sionof thehumanhemeoxygenase gene.Carcinogen-esis1992;13:227^32.

14. Elbirt KK,Whitmarsh AJ, Davis RJ, Bonkovsky HL.Mechanism of sodium arsenite-mediated inductionof heme oxygenase-1 in hepatoma cells. Role of

mitogen-activated protein kinases. JBiol Chem1998;273:8922^31.

15. Eyssen-Hernandez R, Ladoux A, Frelin C. Differen-tial regulation of cardiac heme oxygenase-1and vas-cular endothelial growth factor mRNA expressions byhemin, heavy metals, heat shock and anoxia. FEBSLett1996;382:229^33.

16. Motterlini R, Foresti R, Bassi R, Calabrese V, ClarkJE, Green CJ. Endothelial heme oxygenase-1 induc-tion by hypoxia. Modulation by inducible nitric-oxidesynthase and S-nitrosothiols. JBiol Chem 2000;275:13613^20.

17. Rizzardini M, Zappone M,Villa P, et al. Kupffer celldepletion partially prevents hepatic heme oxygenase1messenger RNA accumulation in systemic inflamma-tion in mice: role of interleukin 1h. Hepatology 1998;27:703^10.

18. Terry CM, Clikeman JA, Hoidal JR, Callahan KS.Effect of tumor necrosis factor-a and interleukin-1a on heme oxygenase-1expression in human endo-thelial cells. AmJPhysiol1998;274:883^91.

19. Pae HO, Oh GS, Choi BM, et al. Carbon monoxideproduced by heme oxygenase-1suppressesTcell pro-liferation via inhibition of IL-2 production. J Immunol2004;172:4744^51.





20. Otterbein LE, Bach FH, Alam J, et al. Carbon mon-oxidehas anti-inflammatory effects involving themito-gen-activatedprotein kinase pathway. NatMed 2000;6:422^8.

21. Berberat PO, Katori M, Kaczmarek E, et al. Heavychain ferritin acts as an antiapoptotic gene that pro-tects livers from ischemia reperfusion injury. FASEB J2003;17:1724^6.

22. Liu H, Nowak R, ChaoW, Bloch KD. Nerve growthfactor induces anti-apoptotic heme oxygenase-1in ratpheochromocytoma PC12 cells. J Neurochem 2003;86:1553^63.

23. Goodman AI, Choudhury M, da Silva JL,Schwartzman ML, Abraham NG. Overexpression oftheheme oxygenase gene in renal cell carcinoma.ProcSoc Exp Biol Med1997;214:54^61.

24. Maines MD, Abrahamsson PA. Expression of hemeoxygenase-1 (HSP32) in human prostate: normal, hy-perplastic, and tumor tissue distribution. Urology1996;47:727^33.

25. Schacter BA, Kurz P. Alterations inmicrosomal drugmetabolism and heme oxygenase activity in isolatedhepatic parenchymal and sinusoidal cells in Murphy-Sturm lymphosarcoma-bearing rats. Clin Invest Med1986;9:150^5.

26. Nishie A, Ono M, Shono T, et al. Macrophageinfiltration andheme oxygenase-1expression correlatewith angiogenesis in human gliomas. Clin Cancer Res1999;5:1107^13.

27. Torisu-Itakura H, Furue M, Kuwano M, Ono M. Co-expression of thymidine phosphorylase and hemeoxygenase-1in macrophages in humanmalignant ver-tical growth melanomas. Jpn J Cancer Res 2000;91:906^10.

28. Sunamura M, Duda DG, Ghattas MH, et al.Heme oxygenase-1 accelerates tumor angiogenesisof human pancreatic cancer. Angiogenesis 2003;6:15^24.

29. Elbashir SM, Harborth J, Lendeckel W, Yalcin A,Weber K, Tuschl T. Duplexes of 21-nucleotide RNAsmediate RNA interference in cultured mammaliancells. Nature 2001;411:494^8.

30. Fang J, Sawa T, Akaike T, et al. In vivo antitumoractivity of pegylated zinc protoporphyrin: targeted in-hibition of heme oxygenase in solid tumor. Cancer Res2003;63:3567^74.

31. Griffith OW. Mechanism of action, metabolism, and

toxicity of buthionine sulfoximine and its higherhomo-logs, potent inhibitors of glutathione synthesis. J BiolChem1982;257:13704^12.

32. Rabinowitch HD, Clare DA, Crapo JD, Fridovich I.Positive correlation between superoxide dismutaseand resistance to paraquat toxicity in the green algaChlorella sorokiniana. Arch Biochem Biophys 1983;225:640^8.

33. Allen RG, Balin AK. Effects of oxygen on the antiox-idant responses of normal and transformed cells. ExpCell Res 2003;289:307^16.

34. Nakagawa I, Suzuki M, Imura N, Naganuma A. In-volvement of oxidative stress in paraquat-inducedmetallothionein synthesis under glutathione depletion.Free Radic Biol Med1998;24:1390^5.

35. Otterbein LE, Soares MP, Yamashita K, Bach FH.Heme oxygenase-1: unleashing the protective prop-erties of heme. Trends Immunol 2003;24:449^55.

36. DuranteW. Heme oxygenase-1 in growth controland its clinical application to vascular disease. J CellPhysiol 2003;195:373^82.

37. Katori M, Buelow R, Ke B, et al. Heme oxygenase-1overexpressionprotects rat hearts from cold ischemia/reperfusion injury via an antiapoptotic pathway. Trans-plantation 2002;73:287^92.

38. Ryter SW, Otterbein LE. Carbonmonoxide in biolo-gy and medicine. Bioessays 2004;26:270^80.

39. Fang J, AkaikeT, Maeda H. Antiapoptotic role ofheme oxygenase (HO) and the potential of HO as atarget in anticancer treatment. Apoptosis 2004;9:27^35.

40. Deininger MH, Meyermann R,Trautmann K, et al.Heme oxygenase (HO)-1expressing macrophages/microglial cells accumulate during oligodendrogliomaprogression. Brain Res 2000;882:1^8.

41. Sunamura M, Duda DG, Ghattas MH, et al. Hemeoxygenase-1 accelerates tumor angiogenesis of hu-man pancreatic cancer. Angiogenesis 2003;6:15^24.

42. Yokoyama S, Mita S, Okabe A, Abe M, Ogawa M.Prediction of radiosensitivity in human esophagealsquamous cell carcinomas with heme oxygenase-1: aclinicopathological and immunohistochemical study.Oncol Rep 2001;8:355^8.

43. Yanagawa T, Omura K, Harada H, et al. Hemeoxygenase-1 expression predicts cervical lymphnode metastasis of tongue squamous cell carci-nomas. Oral Oncol 2004;40:21^7.

44. DulakJ, LobodaA, ZagorskaA, Jozkowicz A.Com-plex role of heme oxygenase-1 in angiogenesis. Anti-oxid Redox Signal 2004;6:858^66.

45. Hemmrich K, Suschek CV, Lerzynski G, Kolb-Bachofen V. iNOS activity is essential for endothelialstress gene expression protecting against oxidativedamage. J Appl Physiol 2003;95:1937^46.

46. Jozkowicz A, Huk I, Nigisch A, et al. Heme oxygen-ase and angiogenic activity of endothelial cells: stimu-lation by carbon monoxide and inhibition by tinprotoporphyrin-IX. Antioxid Redox Signal 2003;5:155^62.

47. Suzuki M, Iso-o N,Takeshita S, et al. Facilitated an-giogenesis induced by heme oxygenase-1gene trans-fer in a rat model of hindlimb ischemia. BiochemBiophys Res Commun 2003;302:138^43.

48. SimizuS, TakadaM,UmezawaK, ImotoM.Require-ment of caspase-3(-like) protease-mediatedhydrogenperoxide production for apoptosis induced by variousanticancer drugs. JBiol Chem1998;273:26900^7.

49. Lin F, Girotti AW. Hyperresistance of leukemia cellsto photodynamic inactivation after long-term expo-sure to hemin. Cancer Res1996;56:4636^43.

50. Hellmuth M,Wetzler C, Nold M, et al. Expression ofinterleukin-8, heme oxygenase-1and vascular endo-thelial growth factor in DLD-1colon carcinoma cellsexposed to pyrrolidine dithiocarbamate.Carcinogene-sis 2002;23:1273^9.

51. Fang J, Sawa T, Akaike T, Greish K, Maeda H.Enhancement of chemotherapeutic response of tumorcells by a heme oxygenase inhibitor, pegylated zincprotoporphyrin. IntJCancer 2004;109:1^8.

52. Baranano DE, Wolosker H, Bae BI, Barrow RK,Snyder SH, Ferris CD. A mammalian iron ATPaseinduced by iron. JBiol Chem 2000;275:15166^73.

53. Li D, Xie K,Wolff R, Abbruzzese JL. Pancreaticcancer. Lancet 2004;363:1049^57.

54. Doi K, AkaikeT, Horie H, et al. Excessive produc-tion of nitric oxide in rat solid tumor and its implicationin rapid tumor growth. Cancer1996;77:1598^604.

55.Wu J, AkaikeT, Maeda H. Modulation of enhancedvascular permeability in tumors by a bradykinin antag-onist, a cyclooxygenase inhibitor, and a nitric oxidescavenger. Cancer Res1998;58:159^65.

56. LeeTS, Chau LY. Heme oxygenase-1mediates theanti-inflammatory effect of interleukin-10 in mice. NatMed 2002;8:240^6.




2005;11:3790-3798. Clin Cancer Res Pascal O. Berberat, Zilvinas Dambrauskas, Antanas Gulbinas, et al. of Pancreatic Cancer Cells to Anticancer TreatmentInhibition of Heme Oxygenase-1 Increases Responsiveness

Updated version

http://clincancerres.aacrjournals.org/content/11/10/3790

Access the most recent version of this article at:

Cited articles

http://clincancerres.aacrjournals.org/content/11/10/3790.full#ref-list-1

This article cites 55 articles, 13 of which you can access for free at:

Citing articles

http://clincancerres.aacrjournals.org/content/11/10/3790.full#related-urls

This article has been cited by 22 HighWire-hosted articles. Access the articles at:

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

Subscriptions

Reprints and

[email protected] at

To order reprints of this article or to subscribe to the journal, contact the AACR Publications

Permissions

Rightslink site. (CCC)Click on "Request Permissions" which will take you to the Copyright Clearance Center's

.http://clincancerres.aacrjournals.org/content/11/10/3790To request permission to re-use all or part of this article, use this link



http://clincancerres.aacrjournals.org/content/11/10/3790.full#ref-list-1

http://clincancerres.aacrjournals.org/content/11/10/3790.full#related-urls

http://clincancerres.aacrjournals.org/cgi/alerts

mailto:[email protected]

