The CLN3 Gene Is a Novel Molecular Target for Cancer Drug Discovery1

[CANCER RESEARCH 62, 801–808, February 1, 2002] The CLN3 Gene is a Novel Molecular Target for Cancer Drug Discovery1

Svetlana N. Rylova, Andrea Amalfitano, Dixie-Ann Persaud-Sawin, Wei-Xing Guo, Jerry Chang, Paul J. Jansen, Alan D. Proia, and Rose-Mary Boustany2 Departments of Pediatrics and Neurobiology [S. N. R., D-A. P-S., W-X. G., J. C., P. J. J., R-M. B.], Genetics [A. A.], and Pathology [A. D. P.], Duke University Medical Center, Durham, North Carolina, 27710

ABSTRACT growth inhibition and also rescues these cells from death caused by treatment with vincristine, etoposide, and staurosporine (6). We have Juvenile Batten disease is a neurodegenerative disease caused by accel- also shown that CLN3 up-regulation decreases the level of the lipid erated apoptotic death of photoreceptors and neurons attributable to second messenger, ceramide, in these cells and also attenuates vinc- defects in the CLN3 gene. CLN3 is antiapoptotic when overexpressed in NT2 neuronal precursor cells. CLN3 negatively modulates endogenous ristine-induced activation of ceramide (6). However, overexpression ceramide levels in NT2 cells and acts upstream of ceramide generation. of CLN3 fails to protect NT2 cells from exogenous ceramide-induced Because defects in regulation of apoptosis are involved in the development killing. These facts place CLN3 upstream of ceramide in apoptosis of cancer, we evaluated the expression of CLN3 on both mRNA and signaling and suggest that CLN3 plays an important role in mecha- protein levels in a variety of cancer cell lines and solid colon cancer tissue. nisms of cell death and survival. The fact that CLN3 is highly We also observed the effect of the blocking of CLN3 protein expression on conserved across species from human to yeast and also in Caenorh- cancer cell growth, survival, ceramide production, and apoptosis by using abditis elegans and Drosophila underscores its importance for cell an adenovirus-bearing antisense CLN3 construct. We show that CLN3 function. CLN3 has also been found to be developmentally regulated mRNA and protein are overexpressed in glioblastoma (U-373G and T98g), in differentiating hNT neurons and in neonatal rat brain. Peaks of neuroblastoma (IMR-32 and SK-N-MC), prostate (Du145, PC-3, and expression are noted just after hNT cells exit the cell cycle, and on day LNCaP), ovarian (SK-OV-3, SW626, and PA-1), breast (BT-20, BT-549, and BT-474), and colon (SW1116, SW480, and HCT 116) cancer cell lines Po in neonatal rat brain, which corresponds to the period of maximum but not in pancreatic (CAPAN and As-PC-1) or lung (A-549 and NCI- neuronal growth (7). This suggests that CLN3 is an oncofetal, anti- H520) cancer cell lines. CLN3 is also up-regulated in mouse melanoma apoptotic gene. These facts led us to investigate whether CLN3 could and breast carcinoma cancer cell lines. We found CLN3 expression is be differentially expressed in some cancers. 22–330% higher than in corresponding normal colon control tissue in 8 of Defects in proapoptotic events can contribute to cancer formation 10 solid colon tumors. An adenovirus-expressing antisense CLN3 (Ad-AS- by allowing cells to survive and to proliferate beyond their normal life CLN3) blocks CLN3 protein expression in DU-145, BT-20, SW1116, and span (8–12). Regulation of apoptosis is involved in the development T98g cancer cell lines as seen by Western blot. Blocking of CLN3 expres- of tumors and plays an essential role in their treatment. A variety of sion using Ad-AS-CLN3 inhibits growth and viability of cancer cells. It chemotherapeutic agents and radiation kill tumor cells by inducing also causes elevation in endogenous ceramide production through de novo apoptosis (10, 13, 14). It has been established that many tumor types ceramide synthesis and results in increased apoptosis as shown by propidium iodide and JC-1 staining. This suggests that Ad-AS-CLN3 may be are resistant to chemotherapy-induced apoptosis. This can occur either an option for therapy in some cancers. More importantly these results because of inactivation of tumor suppressor genes, such as p53 or suggest that CLN3 is a novel molecular target for cancer drug discovery. retinoblastoma, or because of overexpression of antiapoptotic onco- genes, such as Bcl-2 (B-cell lymphoma), Bcr-Abl (myelogenous leukemia), BUG-1 (breast cancer), and survivin in a number of cancers INTRODUCTION and lymphomas (15–18). Identification of novel antiapoptotic genes

3 expressed in cancer cells provides a basis for a better understanding of Mutations in the CLN3 gene are responsible for the JNCL (1). JNCL the biology of these tumors and may lead to discovery of new targets is a recessively inherited neurodegenerative disorder of childhood (1–3). for anticancer drug development (9, 19–21). Direct inactivation of The clinical hallmarks are progressive loss of vision, seizures, and mental antiapoptotic gene expression may promote cancer cell death. Sup- deterioration. These symptoms are attributable to massive cortical pression of antiapoptotic genes can be achieved by antisense strate- neuronal death and gradual loss of photoreceptor cells (2, 3). Apoptosis gies, which in some instances may improve the efficacy of conven- has been shown to be the mechanism of neurodegeneration in the brain tional chemotherapy (9). Antisense-based therapies have already been of patients with the juvenile form of Batten disease (4). Moreover, developed to block Bcl-2 overexpression in non-Hodgkin’s lym- up-regulation of Bcl-2 and elevation of endogenous ceramide levels in phoma (22). the brain from affected individuals provides mechanistic evidence for In this study we establish that CLN3 mRNA and protein are apoptotic death of neurons in this disorder (5). overexpressed in a number of cancer cell lines including breast, colon, We have demonstrated previously that stable CLN3 overexpression malignant melanoma, prostate, ovarian, neuroblastoma, and glioblas- protects NT2 neuronal precursor cells from serum starvation-induced toma multiforme but not lung or pancreatic cancer cell lines. We also show that CLN3 is overexpressed in 8 of 10 solid human colon cancer Received 4/18/01; accepted 12/4/01. cases. Additionally, we demonstrate that blocking CLN3 protein The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with expression in these cancer cell lines, using adenovirus-mediated an- 18 U.S.C. Section 1734 solely to indicate this fact. tisense CLN3 methodology, inhibits cancer cell growth and viability. 1 Supported in part by NIH Grant RO-2NS30170 (to R-M. B.) and in part by NIH Treatment of cancer cells with Ad-AS CLN3 virus also affects the de Grant RO-IDK52925 (to A. A.). The work was also supported by a grant from the Batten Disease Support and Research Association foundation and by a grant from the Battin’ for novo ceramide synthetic pathway and results in ceramide elevation Betsy Charities Inc. and cancer cell death by apoptosis. 2 To whom requests for reprints should be addressed, at Duke University Medical Center, MSRB Box 2604, Durham, NC, 27710. Phone: (919) 681-6220; Fax: (919) 681- 8090; E-mail: [email protected]. MATERIALS AND METHODS 3 The abbreviations used are: JNCL, juvenile form of neuronal ceroid lipofuscinosis; FBS, fetal bovine serum; BSS, balanced salt solution; Ad-AS-CLN3, adenovirus antisense-CLN3 vector; MOI, multiplicity of infection; RT-PCR, reverse transcription-PCR; Cell Culture. Cancer cell lines were obtained from American Type Culture PI, propidium iodide. Collection (Manassas, VA) and maintained at 37°C and 5% CO2 except for 801

SW1116 cells, which are cultured under CO2-free conditions by sealing the RT-PCR for CLN3. Total RNA was isolated from cancer cells using the flasks with parafilm. Human cancer cell lines were maintained in medium as RNeasy Mini kit (Qiagen Inc., Valencia, CA). Then mRNA was isolated from total described below. Normal fibroblasts and A 375 (melanoma) cells were cul- RNA by the Oligotex mRNA spin-column Mini kit (Qiagen Inc.). First-strand tured in DMEM with 10% FBS; SW 626 (ovarian adenocarcinoma) cells were cDNA was then synthesized from mRNA by reverse transcription using Omnis- cultured in DMEM with 1 mM sodium pyruvate, 10% FBS; PC3 (prostate cript Reverse Transcription kit (Qiagen Inc.). PCR reaction was set up with adenocarcinoma) cells were propagated in Ham’s F12K medium with 7% FBS; first-strand cDNA, 2.5 units of Ampli Taq Gold (PE Applied Biosystems, Foster 32 A549 (lung carcinoma) cells were cultured in Ham’s F12K medium with 2 mM City, CA), and 5 ␮Ci of [␣- P]dCTP in each reaction. The reaction was per- ␮ glutamine adjusted to contain 1.5 g/liter sodium bicarbonate, 10% FBS; formed in PCR buffer (Applied Biosysteme) containing 2.5 mM MgCl2,50 M of SW1116 (colon adenocarcinoma) cells were cultured in Leibovitz’s L-15 dCTP, and 100 ␮M each of dATP, dTTP, and dGTP. The primers used for medium with 10% FBS; SW-480 (colon adenocarcinoma) cells were cultured amplification of human CLN3 were 5Ј primer, 5Ј-GGTGGACAGTAT- in Leibovitz’s L-15 medium with L-glutamine, 10% FBS; SK-OV-3 (ovary TCAAGGG-3Ј (958–976), and 3Ј primer, 5Ј-CTTGGCAGAAAGACGAAC-3Ј adenocarcinoma) and HCT116 (colon carcinoma) cells were cultured in Mc- (1229–1246). Cyclophilin was used as an internal control, and primers used for Coy’s 5a medium with 10% FBS; DU145 (prostate carcinoma) cells were amplification of cyclophilin were the 5Ј primer, 5Ј-AAATGCTGGACCCAA- cultured in MEM with 10% FBS; C-32 (melanoma) and PA 1 (ovary terato- CAC-3Ј (317–334), and 3Ј primer, 5Ј-AAACACCACATGCTTGCC-3Ј (384– Ј Ј carcinoma) cells were cultured in MEM with Earle’s BSS, 0.1 mM nonessential 401). The primers used for amplification of mouse CLN3 were 5 primer, 5 - amino acids, and 10% FBS (heat inactivated for PA-1 cells); T98 G and U-373 AGGTGGACAGTGTTCAAGGGTC-3Ј (961–982), and 3Ј primer, 5Ј- Ј MG (glioblastomas) cells and IMR-32 and SK-N-MC (neuroblastoma) cells GGAAGGTATTCACGTAAGCGGC-3 (1300–1321). The primers used for Ј Ј were cultured in MEM with L-glutamine and Earle’s BSS adjusted to contain amplification of mouse cyclophiline were 5 primer, 5 -ACAGCAAGTTC- Ј Ј Ј 1.5 g/liter sodium bicarbonate, 0.1 mM nonessential amino acids, 1 mM sodium CATCGTGTCATC-3 (285–307), and 3 primer, 5 -TGCTCTTTCCTCCTGT- Ј pyruvate, 10% FBS; LNCaP (prostate carcinoma) and BT 549 (breast carci- GCCATC-3 (347–368). The reaction conditions used for human cDNA were 10 noma) were cultured in RPMI 1640 with 10% FBS; CAPAN-1 and As-PC-1 min at 95°C, and then 1 min at 94°C, 1 min at 50°C, and 2 min at 72°C, for 20 (pancreas adenocarcinomas) cells were cultured in RPMI 1640 with 15% and cycles. The reaction conditions used for mouse cDNA were 10 min at 95°C, and 20% FBS, respectively; and NCI-H520 (lung carcinoma) and BT 474 (breast then 1 min at 94°C, 1 min at 56°C, and 2 min at 72°C, for 20 cycles. PCR amplified products were separated on 8% nondenaturing polyacrylamide gel. The carcinoma) cells were cultured in RPMI 1640 with 2 mM glutamine adjusted to bands of interest were quantitated using a PhosphoImager (Molecular Dynamics contain 1.5 g/liter sodium bicarbonate, 4.5 g/liter glucose, 10 mM HEPES, 1 Inc., Sunnyvale, CA). The results are expressed as the ratio of CLN3 signal to that mM sodium pyruvate, and 10% FBS. Penicillin-streptomycin (1%) were added in all of the mediums. Mouse B16 (melanoma) cells and 4T1 (breast carci- of the internal control, cyclophilin. noma) cells were cultured in DMEM high glucose medium with 10% heat Immunocytochemistry. Cells were grown on glass coverslips until confluent, fixed in 4% paraformaldehyde and 4% sucrose in PBS at 4°C, and inactivated FBS, 1% penicillin-streptomycin-fungezon; mouse embryo and permeabilized in 0.2% Triton X-100 in PBS at room temperature. After contact sensitive C3H10T1/2 cells were used as a control and were cultured in blocking in 3% BSA in PBS, cells were incubated with human CLN3 anti- BME medium with Earle’s BSS, 10% heat inactivated FBS, and 1% penicillin- serum. The CLN3 antibody used in this study is a polyclonal antibody raised streptomycin-fungezon. against the peptide sequence AAHDILSHDRTSGNQSHVDP corresponding Human Colon Cancer and Normal Colon Tissues. Samples of human to amino acids 58–77 of the CLN3 protein (Research Genetics, Huntsville, colon cancer and normal colon tissue from the same patient were obtained AL). An additional glass coverslip of each cell type was exposed to rabbit IgG using an Institutional Review Board-approved procedure for collecting left- at the same concentration under the same conditions Cells were washed and over, “deidentified,” or “unlinked” specimens from surgical pathology. In then incubated with 1:500 dilution of biotin-conjugated goat antirabbit IgG brief, at the time of frozen section analysis of the patient tumor, samples of (Zymed, San Francisco, CA) for 1 h. This was followed with exposure to tumor and normal colon tissue not needed for diagnostic purposes were snap streptavidin-conjugated horseradish peroxidase (Zymed) 1:300 for 30 min. frozen in embedding medium and stored at Ϫ80°C. The samples were deiden- Cells were stained with 3,3Ј-diaminobenzidine in 0.2 M Tris-HCl, counter- tified by removing all of the patient information from the samples except for stained with hematoxylin blue, dehydrated in ethanol, washed in xylene, and a pathological diagnosis. Frozen sections of each specimen were stained with mounted on glass slides with Permount. H&E and examined by a pathologist to insure that the frozen samples con- Western Blot. Western Blot for CLN3 protein detection was performed as tained representative normal or neoplastic tissue. In all of the cases, the colon described previously (6). Cells were harvested on day 3 after infection and cancer specimens consisted mostly of neoplastic glands invading into the lysed in buffer containing 250 mM NaCl, 0.1% NP40, 50 mM HEPES (pH 7.0), muscularis externa. The normal colon specimens contained mucosa, submu- 5 mm EDTA, 1 mM DTT, and 10% protease inhibitor mixture (Sigma Chem- cosa, and muscularis externa. ical Co.). After 40 min of incubation at 4°C, lysates were centrifuged at Construction of Ad-AS-CLN3. Ad-AS-CLN3 virus was constructed using 12,000 ϫ g, and the supernatants were quantitated for total proteins using the Ad-Easy method (23). CLN3 cDNA was cloned in the antisense direction into Bio-Rad protein assay. Equal amounts of protein (200 ␮g) were separated by the multiple cloning site of pShuttle-CMV vector. The Ad-AS-CLN3 vector SDS-PAGE using 4.5% and 12% acrylamide for stacking and resolving gels, was constructed by homologous recombination using cotransformation of respectively. Proteins were transferred to nitrocellulose membrane and probed linearized AS-CLN3 pShuttle-CMV plasmid and pAd-Easy (adenoviral back- with a polyclonal CLN3 antibody raised against the peptide sequence AAH- bone) into Escherichia coli BJ5183-competent cells. The Ad-AS-CLN3 vector DILSHKRTSGNQSHVDP corresponding to amino acids 58–77 of the CLN3 was isolated, propagated, and purified exactly as described previously (24). protein (Research Genetics). CLN3 primary antibody complexes were detected Infection units were determined by limiting dilution assay in 293 cells. using goat antirabbit IgG conjugated with horseradish peroxidase and visual- 3 Limiting Dilution Assay. Serial dilutions of viral stock from 1 ϫ 10 to ized by SuperWest Pico Chemiluminescent Substrate (Pierce, Rockford, IL). 15 1 ϫ 10 were made. B6 cells were plated to 80% confluency in two 24-well [3H]Thymidine Incorporation Cell Proliferation Assay. Cells were plates. Serial dilutions of virus were then added in each row of the 24-well plated in 12-well plates at a density of 7 ϫ 104 cells/well and were incubated plate and allowed to infect for 1 h. Additional medium was added, and the with 0.5 ␮Ci/ml [3H]thymidine (DuPont) in medium for 2 h before harvesting. plates were incubated in a 37°C incubator at 5% CO2 for 2 weeks. Limiting For viral transduction experiments, cells were infected with 5, 10, 15, 20, and dilution was determined by the wells with the lowest viral titer that showed 40 MOI of control or Ad-AS-CLN3 virus as described above. On day 2 after cytopathic effect. infection cells were incubated with 0.5 ␮Ci/ml [3H]thymidine (DuPont) in Transduction of Cancer Cells with Control Vector or Ad-AS-CLN3 medium for 2 h. The cells were washed twice with ice-cold PBS and the DNA Virus. Transduction of cancer cells was carried out in 60-mm dishes for precipitated with 5% trichloracetic acid. The DNA precipitate was dissolved in Western blot and in 12-well plates for growth curves and proliferation assays. 0.4 ml of 0.25 M NaOH, and incorporated [3H]thymidine was determined by At 70% confluency, the number of cells was determined. Cells were infected liquid scintillation counting. Each time point was determined in triplicate. with the MOI (number of viral particles used per cell) of virus indicated in the Measurement of Ceramide Levels. DU145 prostate cancer cells were figures by incubation with virus in a minimum amount of medium for1hand plated in 100-mm dishes and at 50% confluency were transduced with 40 MOI addition of medium to a final volume after that. of Ad-AS CLN3 virus or control virus. Three days later cells were washed with 802

PBS, and lipids were extracted using the Bligh and Dyer method. Fumonisin, coverslips, and transduced with 40 MOI of Ad-AS CLN3 virus or control virus. an inhibitor of de novo ceramide biosynthesis (Biomol, Plymouth, PA) was After 72 h, cells were stained with 1 ␮g/␮l JC-1 in medium for 15 min at 37°C. added at a concentration of 25 ␮M to two dishes of DU 145 cells 12 h before After that, cells were washed twice with PBS and mounted on glass slides with transduction with virus. Ceramide levels in the cells were measured using PBS/glycerol (1:1). At the onset of apoptosis, a drop in potential was indicated by diacylglycerol kinase assay, as described previously (6). The results are ex- a fluorescence emission shift from green to red (525 to 590 nm) attributable to pressed as pmols of ceramide per nmol of total phospholipids and represent an formation of J-aggregates. These appear as bright dots in Fig. 9. J-aggregates in average from three experiments. Du145 prostate cancer cells are visualized using ϫ400 magnification. Percentage Measurement of Sphingomyelin Content. Lipids were extracted from of apoptotic cells is calculated from two fields (60–150 cells/field), and the data 10 ϫ 107 cells using the method of Bligh and Dyer. Lipid extracts in 1 ml of are presented as the average of both counts. The statistical significance of the chloroform each were subjected to hydrolysis by adding 0.1 ml of 0.2 N NaOH difference in the number of apoptotic cells after transduction with control or in methanol. Aliquots were taken for quantitation of total phospholipids before Ad-AS CLN3 virus is determined by the two-tailed Student t test. hydrolysis. After neutralization with an equal amount of 2 N HCl, lipids were reextracted by adding 1 ml of chloroform and 1 ml of water and vortexing. The lower organic phase was then evaporated under the nitrogen gas, resuspended RESULTS in 75 ␮l of chloroform-methanol (2:1), and 50 ␮l out of it was spotted onto a CLN3 Is Overexpressed in Cancer Cell Lines Both at the RNA TLC plate (Whatman Inc., Clifton, NJ). The plate was developed in CHCl3- and Protein Levels. We had shown previously that CLN3 is in- CH3OH-acetic acid-H2O (50:30:8:5). Lipids were visualized using iodine vapor, and bands corresponding to sphingomyelin standard were scraped. The volved in antiapoptotic pathways. Because defects in apoptosis amount of phosphor was determined using phosphomolybdate phosphate as- play an important role in carcinogenesis, we compared CLN3 say. The content of sphingomyelin was expressed in nmol per nmol of total expression at the RNA level in neuroblastoma, glioblastoma, pros- phospholipids, and percentage change in sphingomyelin in patient cells in tate, ovarian, colon, breast, melanoma, pancreas, and lung cancer comparison to control fibroblasts was calculated. cell lines to CLN3 expression in normal fibroblasts. Mouse CLN3 PI Staining. DU145 prostate cancer cells were grown on glass coverslips expression at the RNA level is also measured in cells derived from and transduced with 40 MOI of Ad-AS CLN3 virus or control virus. After naturally occurring mouse melanoma and mouse breast carcinoma 72 h, cells were stained with PI in PBS at a concentration of 5 ␮g/␮l for 5 min at 4°C. After that, cells were washed with PBS and mounted in PBS/glycerol (Figs. 1 and 2). The results demonstrate that expression of CLN3 (1:1). PI-positive apoptotic cells were observed under fluorescence (excitation is increased 3.5-fold in glioblastoma, 1– 4-fold in neuroblastoma, wavelength: 525 nm, emission wavelength: 600 nm) at ϫ100 magnification. 2– 4-fold in prostate cancer cell lines, 1.5–2-fold in ovarian cancer JC-1 Staining. JC-1 is a cationic lypophilic dye and exhibits potential- cell lines, and 2– 4-fold in breast carcinomas in comparison to dependent accumulation in mitochondria (25, 26). Cell were grown on glass normal fibroblasts. CLN3 mRNA expression in C 32 (melanoma)

Fig. 1. CLN3 expression in human cancer cell lines by RT-PCR. mRNA was isolated from different cancer cell lines, cDNA synthesized, and PCR was performed in the presence of CLN3 5Ј and 3Ј primers as described in “Ma- terials and Methods.” Cyclophilin was used as an internal control. Results are expressed as the ratio of the CLN3 signal to that of cyclophilin. a, neuroblastoma (SK-N-MC, IMR-32), glioblastoma (T98g, U373); b, prostate cancer cell lines; c, ovarian cancer cell lines; d, breast carcinomas; e, melanoma (C32, A375) and colon (SW1116) cancer cell lines; f, pancreas (Capan, As-PC-1) and lung (A-549, NCI- H520) cancer cell lines. Average of four separate experi- ;P Ͻ 0.05 versus the controlء ;ments; bars, Ϯ SD .P Ͻ 0.01 versus the control; au, arbitrary unitsءء

803

demonstrate in 7 tumors that CLN3 expression is 50–330% higher than in the corresponding normal colon tissue. In an eighth case CLN3 expression was increased by 22% (Fig. 4). Ad-AS-CLN3 Virus Blocks CLN3 Protein Expression and Decreases Viability of Cancer Cells. To demonstrate the impact of CLN3 overexpression on cancer cells we engineered an adenoviral vector with an antisense CLN3 cDNA construct capable of blocking CLN3 overexpression and studied its effect on cancer cell growth and apoptosis. Infection of prostate DU 145, neuroblastoma SK-N-MC, and breast BT-20 cancer cells with 40 MOI Ad-AS-CLN3 virus and colon SW1116 cells with 5 MOI appreciably block expression of the Fig. 2. CLN3 expression in mouse melanoma, breast carcinoma cell lines, and CLN3 protein in comparison to cells infected with the same amount of C3H101/2 control mouse fibroblasts. mRNA was isolated from different cell lines, cDNA synthesized, and PCR was performed in the presence of CLN3 5Ј and 3Ј primers as control virus (Fig. 5). described in “Materials and Methods.” Cyclophilin was used as an internal control. The We then analyzed the effect of different MOIs of Ad-AS-CLN3 results are expressed as the ratio of CLN3 signal to that of cyclophilin. Average of four virus on cancer cell growth. Incorporation of [3H]thymidine was .P Ͻ 0.05 versus the control; au, arbitrary unitsء ;separate experiments; bars, Ϯ SD measured in DU 145, SK-N-MC, BT-20, and SW1116 cells infected with 5, 10, 15, 20, and 40 MOI of Ad-As CLN3 virus or control virus. Incorporation of [3H]thymidine into cancer cells was inversely related to an increase in MOI or the number of Ad-AS-CLN3 viral particles infecting each cell (Fig. 6, a, c, e, and g). Ad-AS-CLN3 at a dose of

Fig. 4. CLN3 mRNA expression in human colon tumors compared with normal colon tissue. mRNA is isolated from 10 solid colon cancers, and expression of CLN3 is Fig. 3. CLN3 protein expression by immunocytochemistry. Confluent cells were fixed, compared with normal colon tissue obtained from the same patient using RT-PCR. permeabilized, and incubated with human CLN3 antiserum for 12 h. After that cells were Cyclophilin is used as an internal control. Results are expressed as percentage change of exposed to goat antirabbit biotin-conjugated antibody for 1.5 h and to horseradish CLN3 expression in colon cancer samples compared with corresponding normal colon peroxidase conjugated streptavidin for 30 min. Next, cells were stained with diaminoben- controls. Each experiment was repeated twice. zidine, dehydrated, and mounted in Permount. Brown staining correlates with CLN3 protein expression. Note increased amount of staining in Du145 prostate, SK-N-MC neuroblastoma, BT-549 breast, PA-1 ovarian, and T98g glioblastoma cancer cells compared with normal human fibroblasts. and SW1116 colon cancer cell lines is 2-fold higher than in the normal control. Pancreas and lung cancer cells only show slight changes in CLN3 expression (Fig. 1). CLN3 expression in mouse melanoma and breast carcinoma cell lines is 25–38-fold higher in comparison to CLN3 expression in control mouse C3H10T1/2 cells (Fig. 2). To compare the CLN3 protein levels in cancer cells to control fibroblasts we performed immunocytochemistry using polyclonal CLN3 antibody. Prostate DU145, ovarian PA1, glioblastoma T 98G, neuroblastoma SK-N-MC and breast carcinoma BT-549 cancer cells show intense staining for CLN3 as opposed to light staining in normal fibroblasts (Fig. 3). Therefore, increase in CLN3 protein levels cor- Fig. 5. Western Blot. Cancer cells were transduced with 10 or 40 MOI of Ad-AS-CLN3 relates with CLN3 mRNA overexpression in cancer cells. virus or control virus. Lane 1, control cells; Lane 2, control virus 10 MOI; Lane 3, Expression of CLN3 Is Increased in 8 of 10 Solid Colon Ad-AS-CLN3 virus 10 MOI; Lane 4, control virus 40 MOI; Lane 5, Ad-AS-CLN3 virus 40 MOI; Lane 6, Ad-AS-CLN3 virus 2 MOI; Lane 7, Ad-AS-CLN3 virus 5 MOI. Protein Tumors. To corroborate the results obtained in cancer cells in solid was isolated 3 days after infection and quantitated. Equal amounts of protein were human tumors, we analyzed CLN3 expression in 10 solid colon separated by SDS-PAGE then transferred to nitrocellulose membrane and probed with CLN3 antibody. To confirm equal protein loading, the membrane was stained with tumors. Normal colon tissue obtained from the same patient at the Ponceau red dye. a, prostate carcinoma DU145 cells; b, colon adenocarcinoma SW1116 time of surgery and assessed pathologically is used as a control. We cells; c, neuroblastoma SK-N-MC cells; d, breast carcinoma BT-20 cells. 804

Fig. 6. Left panels, Ad-AS-CLN3 virus dose- response curves for (a) prostate DU145, (c) breast BT-20, (e) colon SW1116, (g) neuroblastoma SK-N-MC cancer cell lines. Cells were seeded in 24-well plates, infected with 5, 10, 15, 20, 40 MOI of Ad-AS-CLN3 virus or control virus. On day 2 proliferation rates were determined by measuring [3H]thymidine incorporation. Right panels, growth curves of cancer cells after infection with Ad-AS-CLN3 virus or control virus. Cells were plated at equal density and infected with 15 MOI of Ad-AS-CLN3 virus or control virus. Number of cells was counted using trypan blue dye exclusion method at the respec- tive time points over a 4-day period; (b) prostate DU145 cancer cells; (d) breast carcinoma BT-20 cells; (f) colon SW1116 cancer cells; (h) neuroblastoma SK-N-MC cells. Note that there is an inverse correlation between amount of viral particles used per cell and cancer cell thymidine uptake (a, c, e, g). Also note decreased cancer cell number after transduction with Ad-AS CLN3 (b, d, f, h).

15 MOI was sufficient and effective in the killing of cancer cells as myelin content. In fact, it was 26.9% higher than in normal cells seen from the dose-response curves in Fig. 6, a, c, e, and g. (Fig. 7b). This suggests that the increase in ceramide levels seen with The growth profile of DU145, BT-20, SW1116, and SK-N-MC diminished expression of CLN3 is not attributable to sphingomyelin cancer cells infected with control virus practically matches that of breakdown or the sphingomyelin cycle but more likely suggests an untreated cells (Fig. 6, b, d, f, and h). In contrast, infection of the cells impact of CLN3 on the de novo ceramide pathway. with Ad-AS-CLN3 halts cancer cell growth. This effect was notice- Blocking of CLN3 Expression in Prostate Cancer Cells Results able on day 2 after transduction with virus. At the end of the treatment in Increased Ceramide Levels. We have shown previously that the amount of live cells drastically dropped compared with control, CLN3 negatively modulates ceramide generation in NT2 cells. Cer- which indirectly indicates an increase in the number of dying cells. amide levels in DU145 prostate cancer cells were increased by 91.4% Increase in Ceramide Levels in CLN3 Deficient Fibroblasts Is after transduction with Ad-AS CLN3 virus compared with cells trans- Not Sphingomyelin-derived. We also measured ceramide levels in duced with control virus (Fig. 7a). Moreover, pretreatment of these fibroblasts obtained from patients with juvenile Batten disease that are cells with fumonisin, an inhibitor of ceramide synthase, abrogated deficient in CLN3. There was a 78.5% increase in endogenous cer- ceramide elevation. Fumonisin equally blocks ceramide elevation in amide in CLN3-deficient fibroblasts compared with normal fibro- DU145 cells after transduction with Ad-AS CLN3 virus and control blasts. At the same time there was no simultaneous drop in sphingo- virus by 93.3% and 91.4%, respectively. Therefore, blocking of de 805

Fig. 7. a, ceramide levels in DU145 cells after transduction with Ad-AS CLN3 virus. DU145 cells were transduced with 30 MOI of Ad-AS CLN3 virus or control virus. Fumonisin was added 12 h before transduction with virus. Three days after transduction, lipids were harvested and ceramide levels were measured using the diacyl glycerol kinase assay. Results are an average of three experiments and are expressed in pmol of ceramide per nmol of total phospholipids. a, bar 1, DU145 cells ϩ control virus; bar 2, DU145 cells ϩ control virus ϩ fumonisin; bar 3, DU145 cells ϩ Ad-AS CLN3 virus; bar 4, DU145 cells ϩ Ad-AS CLN3 virus ϩ fumonisin. b, ceramide and sphingomyelin levels in CLN3 deficient fibroblasts. Ceramide and sphingomyelin were measured as described in “Materials and Methods” and expressed as percentage change compared with levels in normal fibroblasts. Note that the increase in ceramide levels is not accompanied by a decrease in sphingomyelin content suggesting that ceramide is not derived from activation of the sphingomyelin cycle; bars, Ϯ SD.

Fig. 8. a, growth curves for CLN3-deficient lymphoblasts. Lymphoblasts (5 ϫ 104) are plated in 12-well plates and, after 24 h, transfected with pGEM ϩ 7f-CLN3 or empty vector using LipofectAMINE 2000 kit (Stratagene, La Jolla, CA). The number of cells was counted using trypan blue dye exclusion assay starting 24 h after transfection. Note that JNCL patient cells (Œ) show diminished growth when compared with normal lymphoblasts (ࡗ). Transfection of JNCL patient cells with the full CLN3 gene (gray triangles), as opposed to transfection with an empty vector, restores growth to normal levels. b, thymidine incorporation in CLN3-deficient lymphoblasts. Thymidine uptake by CLN3- deficient lymphoblasts was measured 24 h after transfection with pGEM ϩ 7f-CLN3 (gray triangles) or empty plasmid (Œ). Note that CLN3 deficient lymphoblasts incorporate thymidine at a rate similar to lymphoblasts transfected with full CLN3 gene. novo ceramide synthesis suppresses ceramide level increases induced to the empty vector, restores growth almost to normal levels but has by Ad-AS CLN3 virus, and this suggests that CLN3 impacts the de no effect on thymidine incorporation (Fig. 8b). This indicates that the novo ceramide pathway (Fig. 7a). CLN3 gene impacts apoptotic pathways and not cell proliferation. Transfection of CLN3-deficient Lymphoblasts with CLN3 Blocking of CLN3 Expression in Prostate Cancer Cells Containing Plasmid Restores Growth but Has No Effect on Increases Apoptosis. Transduction of DU145 prostate cancer cells Thymidine Incorporation. CLN3-deficient lymphoblasts from pa- with Ad-AS CLN3 results in increased apoptosis. There is a higher tients manifest slowed growth compared with normal lymphoblasts number of apoptotic cells in DU145 prostate cancer cells after trans- (Fig. 8a). Reintroduction of CLN3 into JNCL patient cells, as opposed duction with Ad-AS CLN3 virus compared with transduction with

Fig. 9. PI (a, b) and JC-1 staining (c, d)of Du145 cells. On day 3 after transduction with control virus (a) or Ad-AS CLN3 virus (b), cells were washed, stained with PI (5 ␮g/ml) for 5 min, and observed under fluorescence (excitation wavelength: 525 nm, emission wavelength: 600 nm). Note an increase in the number of white apoptotic Du145 cells after transduction with Ad-AS CLN3 virus. On day 3 after transduction with control virus (c) or Ad-AS CLN3 virus (d), Du145 cells were washed, stained with JC-1 in medium (1 ␮g/ml) for 15 min at 37°C, and observed under fluorescence. At the onset of apoptosis there was a drop of potential indicated by a fluorescence emission shift from green to red (525 to 590 nm) because of formation of red fluorescent J-aggregates that appear as bright dots. The JC-1-stained cells are visualized using ϫ400 magnification. Ap- optotic cells demonstrate formation of red J-aggregates. Note decrease in the total number of cells and an increase in those with red J-aggregate formation.

806

Downloaded from cancerres.aacrjournals.org on September 27, 2021. © 2002 American Association for Cancer Research. CLN3: A NOVEL MOLECULAR TARGET FOR CANCER control virus shown by PI staining (Fig. 9, a and b). Also, the number show a 78.5% increase in endogenous ceramide compared with nor- of cells with J-aggregate formation is increased from 8.2 Ϯ 3.7% to mal fibroblasts. Different apoptotic stimuli induce ceramide genera- 34.5 Ϯ 9% (P Ͻ 0.05) after transduction with Ad-AS CLN3 virus as tion via two different routes. One route involves activation of the shown by JC-1 staining. (Fig. 9, c and d). sphingomyelin cycle, which results in cleavage of sphingomyelin and release of ceramide, the other is via activation of enzymes in the de DISCUSSION novo ceramide synthetic pathway (35–43). Pretreatment of DU145 cells with fumonisin, an inhibitor of ceramide synthase, before trans- Deregulation of antiapoptotic genes and/or mutated proapoptotic duction with the Ad-AS CLN3 virus completely blocks elevation in genes may contribute to carcinogenesis by allowing abnormal cells to ceramide. This indicates that blocking of CLN3 protein expression survive and thereby become cancerous (21, 27, 28). Bcl-2 was one of induces an increase in ceramide levels not through sphingomyelin the first antiapoptotic genes found to be up-regulated in cancer (29). breakdown but via the de novo ceramide synthetic pathway. Addi- Other genes have been discovered that also are overexpressed in tional evidence is the fact that increases in ceramide levels in CLN3- cancer and that inhibit apoptosis. Examples include Bcr-Abl in chronic deficient brain and CLN3-deficient fibroblasts were not accompanied myeloid leukemia, BUG-1 in breast tumors, and Survivin in a number by a change in sphingomyelin levels. of cancers and lymphomas (10, 15, 16, 30–34). Suppressing levels of CLN3 protein in cancer cells enhances cer- Previous work has shown that CLN3 is antiapoptotic (6). Ceramide amide production and results in death of cancer cells. There are a levels are elevated in the brain of patients with the juvenile form number of anticancer drugs of which the effects include boosting of of Batten disease compared with normal brain (5). NT2 cells stably ceramide levels in cancer cells (44). or transiently overexpressing CLN3 protein have lower ceramide In conclusion, CLN3 overexpression may be developed as yet levels then control cells and are unable to increase ceramide levels another marker for some but not all cancers. Moreover, blocking in response to vincristine. CLN3 overexpression protects NT2 cells CLN3 overexpression using an antisense adenoviral CLN3 construct from apoptosis induced by the chemotherapeutic agents vincris- inhibits cancer growth, increases ceramide levels, and promotes death tine, staurosporine, etoposide, and serum starvation. Therefore, of cancer cells. The implications are 2-fold. The use of antisense CLN3 negatively modulates endogenous ceramide levels. Alterna- CLN3 methodology may be valid for treatment of some cancers. tively, loss of functional CLN3 protein in juvenile Batten disease Additionally, the CLN3 gene provides a novel molecular target to results in increased ceramide production and accelerated apoptosis. screen for new drugs effective in combating cancer. Up-regulation of CLN3 failed to prevent apoptosis caused by exogenous C2-orC6-ceramide, suggesting that CLN3 impacts ceramide generation upstream. REFERENCES We demonstrate that CLN3 is overexpressed at the mRNA and 1. The International Batten Disease Consortium. Isolation of a novel gene underlying protein levels in a number of human cancer cell lines including Batten disease, CLN3. Cell, 82: 949–957, 1995. prostate, glioblastoma, neuroblastoma, ovarian, breast, colon, and 2. Zeman, W. Studies in neuronal ceroid lipofuscinosis. J. Neuropathol. Exp. Neurol., malignant melanoma. Pancreas and lung cancer cell lines do not 33: 1–12, 1974. 3. Boustany, R-M., Alroy, J., and Kolodny, E. H. Clinical classification of neuronal show increased CLN3. CLN3 was also found to be up-regulated in ceroid lipofuscinosis subtypes. Am. J. Med. Genet. 5 (Suppl. 5): 47–58, 1988. mouse melanoma and breast carcinoma cell lines. On the basis of 4. Lane, S. C., Jolly, R. D., Schmechel, D. E., Alroy, J., and Boustany R-M. Apoptosis these observations, it may become possible to develop semiquan- as the mechanism of neurodegeneration in Batten disease. J. Neurochem., 67: 677– 683, 1996. titative assays of CLN3 expression for screening or to monitor 5. Puranam, K., Qian, W-H., Nikbaht, K., Venable, M., Obeid, L., Hannun, Y., and effectiveness of therapy in some cancers. Overexpression of CLN3 Boustany, R-M. Upregulation of Bcl-2 and elevation of ceramide in Batten disease. in 80% of solid colon cancers tested corroborates data obtained Neuropediatrics, 28: 37–41, 1997. 6. Puranam, K. L., Guo, W-X., Qian, W-H., Nikbakht, K., and Boustany, R-M. CLN3 from human cancer cell lines. Discovery of novel antiapoptotic defines a novel antiapoptotic pathway operative in neurodegeneration and mediated genes involved in cancer may lead to a better understanding of by ceramide. Molecular Genetics and Metabolism, 66: 294–308, 1999. 7. Pane, M. A., Puranam, K. L., and Boustany, R-M. Expression of CLN3 in human NT2 cancer biology. Also, these novel genes may prove to be effective neuronal precursor cells and neonatal rat brain. Pediatr. Res., 46: 367–374, 1999. targets for cancer drug development. Potential strategies for induc- 8. Villunger, A., and Strasser, A. Does “death receptor” signaling play a role in tion of apoptosis in cancer cells include inactivation of anti- tumorigenesis and cancer therapy? Oncol. Res., 10: 541–550, 1998. 9. Favrot, M., Coll, J-L., Louis, N., and Negoescu, A. Cell death and cancer: replace- apoptotic genes by blocking their expression (9, 18). Reduction in ment of apoptotic genes and inactivation of death suppressor genes in therapy. Gene Bcl-2 expression by antisense methods increases the susceptibility Ther., 5: 728–739, 1998. of cancer cells to induction of apoptosis by multiple chemothera- 10. Noteborn, M-H. M., Zhang, Y-H., and van der Eb, A. J. Apoptin specifically causes apoptosis in tumor cells and after UV-treatment in untransformed cells from cancer- peutic agents (19, 22). Similarly, down-regulation of CLN3 ex- prone individuals: a review. Mutat. Res., 400: 447–455, 1998. pression with a resultant increase in ceramide generation may be 11. Green, D. R., Bissonette, R. P., and Cotter, T. G. Apoptosis and cancer. In: V. DeVita, explored as a potential cancer therapeutic strategy. We demon- S. Hellman, and S. A. Rosenberg (eds.), Important Advances in Oncology, pp. 37–52. Philadelphia: J. B. Lippincott Co., 1994. strate that expression of the CLN3 protein drops significantly after 12. Kerr, J. F., Winterford, C. M., and Harmon, B. V. Apoptosis. Its significance in cancer transduction with 40 MOI of Ad-AS-CLN3 virus. Blocking of and cancer therapy. Cancer (Phila.), 73: 2013–2026, 1994. CLN3 expression in cancer cells by Ad-AS-CLN3 virus suppresses 13. Liu, X., Fuks, Z., and Kolesnick, R. Ceramide mediates radiation-induced death of endothelium. Crit. Care Med., 28: 87–93, 2000. their growth and results in an increase in the number of apoptotic 14. Chiu, S. M., Davis, T. W., Meyers, M., Ahmad, N., Mukhtar, H., and Separovic, D. cells. CLN3-deficient lymphoblasts show no difference in thymi- Phtalocyanine 4-photodinamic therapy induces ceramide generation and apoptosis in dine incorporation when transfected with empty plasmid or plas- acid sphingomyelinase-deficient mouse embryonic fibroblasts. Int. J. Oncol., 16: 423–427, 2000. mid carrying full-length CLN3 cDNA. This suggests that CLN3 15. Yang, X., Hao, Y., Ding, Z., Pater, A., and Tang, S-C. Differential expression of impacts apoptotic pathways and not proliferative pathways. antiapoptotic gene BAG-1 in human breast normal and cancer cell lines and tissues. Clin. Cancer Res., 1816–1822, 1999. Transduction of DU145 prostate cancer cells with Ad-AS CLN3 16. Ambrosini, G., Adida, C., and Altieri, D. C. A novel anti-apoptotic gene, survivin, virus elevates ceramide levels in these cells compared with those expressed in cancer and lymphoma. Nat. Med., 3: 917–921, 1997. transduced with control virus. The fact that less CLN3 expression 17. Steller, H. Mechanisms and genes of cellular suicide. Science (Wash. DC), 267: 1445–1449, 1995. results in increased ceramide was confirmed in CLN3-deficient fibro- 18. McDonnell, T. J., Meyn, R. E., and Robertson. Implication of apoptotic cell death blasts derived from patients with juvenile Batten disease. These cells regulation in cancer therapy. Semin. Cancer Biol., 6: 53–60, 1995. 807

19. Reed, J. C. Mechanisms of apoptosis avoidance in cancer. Curr. Opin. Oncol., 11: 33. Takayama, S., Sato, T., Krajewski, S., Kochel, K., Irie, S., Millan., J. A., and Reed, 68–75, 1999. J. Cloning and functional analysis of BAG-1: a novel bcl-2-binding protein with 20. Williams, G. T Programmed cell death: apoptosis and oncogenesis. Cell, 65: 1097– anti-cell death activity. Cell, 80: 279–284, 1995. 1098, 1991. 34. Zapata, J. M., Krajewski, M., Krajewski, S., Huang, R. P., Takayama, S., Wang, 21. Thompson, C. B. Apoptosis in the pathogenesis and treatment of disease. Science H. G., Adamson, E., and Reed, J. C. Expression of multiple apoptosis-regulatory (Wash. DC), 267: 1456–1462, 1995. genes in human breast cancer cell lines and primary tumors. Breast Cancer Res. 22. Webb, A., Cunningham, D., Cotter, F., di Stefano, F., and Ross, P. BCL-2 antisense Treat., 47: 129–140, 1998. therapy in patients with non-Hodgkin lymphoma. Lancet, 349: 1137–1141, 1997. 35. Kolesnick, R. N., and Kronke, M. Regulation of ceramide production and apoptosis. 23. He, T-C., Zhou, S., Da Costa, L. T., Yu, J., Kinzler, K. W., and Vogelstein, B. A Annu. Rev. Physiol., 60: 643–665, 1998. simplified system for generation recombinant adenoviruses. Proc. Natl. Acad. Sci. 36. Hannun, Y. A., and Obeid, L. M. Mechanisms of ceramide-mediated apoptosis. Adv. USA, 95: 2509–2514, 1998. Exp. Med. Biol., 407: 145–149, 1997. 24. Amalfitano, A., Hauser, M. A., Hu, H., Serra, D., Begy, C. R., and Chamberlain, J. S. 37. Haimovitz-Friedman, A., Kolesnick, R. N., and Fuks, Z. Ceramide signaling in Production and characterization of improved vectors with the E1, E2b, and E3 genes apoptosis. Br. Med. Bull., 53: 539–553, 1997. deleted. J. Virol., 72: 926–933, 1998. 38. Kroesen, B-J., Pettus, B., Luberto, C., Busman, M., Sietsma, H., Leij, L., and Hannun, Y. A. Induction of apoptosis through B-cell receptor cross-linking occurs via de novo 25. Dubinsky, J., and Levi, Y. Calcium-induced activation of the mitochondrilal perme- generated C -ceramide and involves mitochondria. J. Biol Chem, 276: 13606– ability transition in hippocampal neurons. J. Neurosci. Res., 53: 728–741, 1998. 16 13614, 2001. 26. White, R., and Reynolds, I. Mitochondrial depolarization in glutamate-stimulated 39. Hannun, Y. A., and Chiara, L. Ceramide in the eukaryotic stress response. Trends Cell neurons: an early signal specific to excitotoxin exposure. J. Neurosci., 16: 5688– Biol., 10: 73–80, 2000. 5697, 1996. 40. Mathias, S., Pena, L. A., and Kolesnick, R. N. Signal transduction of stress via 27. Savitz, S. I., Daniel, B. A., and Rosenbaum, M. Apoptosis in neurological disease. ceramide. Biochem. J., 335: 465–480, 1998. Neurosurgery, 42: 555–567, 1998. 41. Xu, J., Yeh, C. H., Chen, S., He, L., Sensi, S. L., Canzoniero, L. M., Choi, D. W., and 28. Orrenius, S. Apoptosis: molecular mechanisms and implications for human disease. Hsu, C. Y. Involvement of de novo ceramide biosynthesis in tumor necrosis factor- J. Intern. Med., 237: 529–536, 1995. ␣/cycloheximide-induced cerebral endothelial cell death. J. Biol. Chem., 273: 16521– 29. Tsujimoto, Y., and Croce, C. A. Analysis of the structure, transcripts, and protein 16526, 1998. products of bcl-2, the gene involved in human follicular lymphoma. Proc. Natl. Acad. 42. Farrell, A. M., Uchida, Y., Nagiec, M. M., Harris, I. R., Dickson, R. C., Elias, P. M., Sci. USA, 83: 5214–5218, 1986. and Holleran, W. M. UVB irradiation up-regulates serine palmitoyltransferase in 30. Hockenberry, D. M. Bcl-2 in cancer, development and apoptosis. J. Cell Sci. Suppl., cultured human keratinocytes. J. Lipid Res., 39: 2031–2038, 1998. 18: 51–55, 1994. 43. Suzuki, A., Iwasaki, M., Kato, M., and Wagai, N. Sequential operation of ceramide 31. Sachs, L., and Lotem, J. Control of programmed cell death in normal and leukemia synthesis and ICE cascade in CPT-11-initiated apoptotic death signaling. Exp. Cell cells: new implication for therapy. Blood, 82: 15–21, 1993. Res., 233: 41–47, 1997. 32. Bedi, A. Inhibition of apoptosis by Bcr-Abl in myeloid leukemia. Blood, 83: 2038– 44. Radin, N. S. Killing cancer cells by poly-drug elevation of ceramide levels: a 2044, 1994. hypothesis whose time has come. Eur. J. Biochem., 268: 193–204, 2001.

808

Svetlana N. Rylova, Andrea Amalfitano, Dixie-Ann Persaud-Sawin, et al.

Cancer Res 2002;62:801-808.

Updated version Access the most recent version of this article at: http://cancerres.aacrjournals.org/content/62/3/801

Cited articles This article cites 39 articles, 11 of which you can access for free at: http://cancerres.aacrjournals.org/content/62/3/801.full#ref-list-1

Citing articles This article has been cited by 4 HighWire-hosted articles. Access the articles at: http://cancerres.aacrjournals.org/content/62/3/801.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://cancerres.aacrjournals.org/content/62/3/801. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.