arti cleexpression of transcription factor e2f1 and telomerase

12
Journal of the National Cancer Institute, Vol. 97, No. 21, November 2, 2005 ARTICLES 1589 Expression of Transcription Factor E2F1 and Telomerase in Glioblastomas: Mechanistic Linkage and Prognostic Signicance Marta M. Alonso, Juan Fueyo, Jerry W. Shay, Kenneth D. Aldape, Hong Jiang, Ok-Hee Lee, David G. Johnson, Jing Xu, Yasuko Kondo, Takao Kanzawa, Satoru Kyo, B. Nebiyou Bekele, Xian Zhou, Janice Nigro, J. Matthew McDonald, W. K. Alfred Yung, Candelaria Gomez-Manzano Background: Several tumor suppressor pathways have been identied as modulators of telomerase function. We exam- ined the functional role of the retinoblastoma-E2F1 path- way in regulating telomerase activity in malignant gliomas. Methods: Adenovirus vectors were used to transfer cDNAs into human glioblastoma and sarcoma cells. Telomerase ac- tivity was assessed with a telomere repeat amplication pro- tocol. Promoter activity in cancer cells was assessed with promoter–luciferase reporter constructs. Promoter binding was assessed with the chromatin immunoprecipitation (ChIP) assay. We isolated astrocytes from E2F1 transgenic mice and normal mice for in vivo studies. We evaluated the expression of E2F1 and hTERT (the catalytic subunit of human telomerase) mRNAs by reverse transcriptase– polymerase chain reaction and proteins in human glioblas- toma samples by immunoblot analysis. Associations between survival among 61 glioblastoma multiforme patients and expression of E2F1 and hTERT mRNA and protein were examined with Kaplan–Meier analysis, the log-rank test, and Cox proportional hazards regression models. All statis- tical tests were two-sided. Results: Ectopic E2F1 expression increased hTERT promoter activity in cancer cells. We de- tected an interaction between E2F1 protein and the hTERT promoter. Transgenic E2F1 astrocytes contained functional telomerase protein. E2F1 mRNA expression and hTERT mRNA expression were statistically signicantly correlated in human glioblastoma specimens ( R = .8; P<.001). Longer median survival was statistically signicantly associated with lower E2F1 mRNA expression in tumors (103.6 weeks) rather than with higher expression (46.1 weeks) (difference = 57.5 weeks; 95% condence interval [CI] = 14.7 to 159.7; log-rank P = .002). E2F1 mRNA was the only factor that was statistically signicantly associated with overall sur- vival in a multivariable model ( P = .04). Among 27 patients with glioblastoma multiforme samples, the expression of E2F1 protein was statistically signicantly associated with survival (log-rank P<.001). Conclusions: E2F1 may partici- pate in telomerase activity regulation in malignant glioma cells. Its expression appears to be strongly associated with the survival of patients with malignant brain tumors. [J Natl Cancer Inst 2005;97:1589–600] Mechanistic studies of cancer cells at the molecular level may help to elucidate tumor behavior, to predict prognosis, and to develop targeted and efcacious therapies. Transformation of cultured primary cells into tumorigenic variants is a multistep process in which each genetic change confers a proliferative ad- vantage (1). In most cancers, gene mutations disrupt normal cell cycle checkpoints that otherwise limit cell division in response to DNA damage or oncogene activation. The retinoblastoma protein (pRb) is a common downstream target of the signal cascade that regulates cell cycle checkpoints. Particular components of the retinoblastoma (Rb) regulatory network may act as tumor sup- pressors or proto-oncogenes, mutations in which are apparently essential for cancer cell development (2,3). Inactivation of Rb expression in cancer cells leads to the deregulated activity of the transcription factors E2F1, E2F2, and E2F3, which control the expression of genes involved in differentiation, development, proliferation, and apoptosis. E2F1 has oncogenic properties in vitro and in vivo, and its overexpression is sufcient to induce the transformation of rodent primary cells (4) and to promote tumorigenesis in transgenic mice (5). Surprisingly, E2F1 also in- duces apoptosis in vitro and in vivo (6), and E2F1-null mutant mice develop spontaneous malignancies, suggesting that E2F1 may be a tumor suppressor (7,8). Thus far, however, E2F1 inac- tivation or loss-of-function mutations have not been consistently found in human cancer. A current tenet of carcinogenesis is the hypothesis that hu- man cells must overcome at least two barriers before efcient cellular proliferation can occur. The rst barrier, senescence, minimally involves the Rb tumor-suppressor pathway. How- ever, to sustain immortalization, cells must eventually overcome the second proliferative blockade, known as crisis (9,10). To overcome crisis, telomeric DNA must be stabilized through telomerase activation (11) or through activation of the alterna- tive lengthening of telomeres pathway (12,13). Human telo- merase is a reverse transcriptase that contains an RNA moiety, known as hTR, and a catalytic subunit, known as hTERT Afliations of authors: Departments of Neuro-Oncology (MMA, JF, HJ, O-HL, JX, WKAY, CG-M), Pathology (KDA, JMM), Neurosurgery (YK, TK), and Biostatistics (BNB, XZ), University of Texas M. D. Anderson Cancer Center, Houston, TX; Department of Cell Biology, University of Texas Southwestern Medical Center, Dallas, TX (JWS); Department of Carcinogenesis, Univer- sity of Texas M. D. Anderson Cancer Center, Science Park-Research Division, Smithville, TX (DGJ); Department of Obstetrics and Gynecology, Kanazawa University School of Medicine, Ishikawa, Japan (SK); Department of Pathology, University of California School of Medicine, San Francisco, CA (JN). Correspondence to: Candelaria Gomez-Manzano, MD, Department of Neuro- Oncology, Box 1002, University of Texas M.D. Anderson Cancer Center, 1515 Holcombe Blvd., Houston, TX 77030 (e-mail: [email protected]). See “Notes” following “References.” DOI: 10.1093/jnci/dji340 © The Author 2005. Published by Oxford University Press. All rights reserved. For Permissions, please e-mail: [email protected]. Downloaded from https://academic.oup.com/jnci/article-abstract/97/21/1589/2521454 by guest on 23 March 2018

Upload: ngodieu

Post on 15-Jan-2017

231 views

Category:

Documents


3 download

TRANSCRIPT

Page 1: ARTI CLEExpression of Transcription Factor E2F1 and Telomerase

Journal of the National Cancer Institute, Vol. 97, No. 21, November 2, 2005 ARTICLES 1589

ARTICLE Expression of Transcription Factor E2F1 and Telomerase in Glioblastomas: Mechanistic Linkage and Prognostic Signifi cance Marta M. Alonso , Juan Fueyo , Jerry W. Shay , Kenneth D. Aldape , Hong Jiang , Ok-Hee Lee , David G. Johnson , Jing Xu , Yasuko Kondo , Takao Kanzawa , Satoru Kyo , B. Nebiyou Bekele , Xian Zhou , Janice Nigro , J. Matthew McDonald , W. K. Alfred Yung , Candelaria Gomez-Manzano

Background: Several tumor suppressor pathways have been identifi ed as modulators of telomerase function. We exam-ined the functional role of the retinoblastoma-E2F1 path-way in regulating telomerase activity in malignant gliomas. Methods: Adenovirus vectors were used to transfer cDNAs into human glioblastoma and sarcoma cells. Telomerase ac-tivity was assessed with a telomere repeat amplifi cation pro-tocol. Promoter activity in cancer cells was assessed with promoter – luciferase reporter constructs. Promoter binding was assessed with the chromatin immunoprecipitation (ChIP) assay. We isolated astrocytes from E2F1 transgenic mice and normal mice for in vivo studies. We evaluated the expression of E2F1 and hTERT (the catalytic subunit of human telomerase) mRNAs by reverse transcriptase – polymerase chain reaction and proteins in human glioblas-toma samples by immunoblot analysis. Associations between survival among 61 glioblastoma multiforme patients and expression of E2F1 and hTERT mRNA and protein were examined with Kaplan – Meier analysis, the log-rank test, and Cox proportional hazards regression models. All statis-tical tests were two-sided. Results: Ectopic E2F1 expression increased hTERT promoter activity in cancer cells. We de-tected an interaction between E2F1 protein and the hTERT promoter. Transgenic E2F1 astrocytes contained functional telomerase protein. E2F1 mRNA expression and hTERT mRNA expression were statistically signifi cantly correlated in human glioblastoma specimens ( R = .8; P <.001). Longer median survival was statistically signifi cantly associated with lower E2F1 mRNA expression in tumors (103.6 weeks) rather than with higher expression (46.1 weeks) (difference = 57.5 weeks; 95% confi dence interval [CI] = 14.7 to 159.7; log-rank P = .002). E2F1 mRNA was the only factor that was statistically signifi cantly associated with overall sur-vival in a multivariable model ( P = .04). Among 27 patients with glioblastoma multiforme samples, the expression of E2F1 protein was statistically signifi cantly associated with survival (log-rank P <.001). Conclusions: E2F1 may partici-pate in telomerase activity regulation in malignant glioma cells. Its expression appears to be strongly associated with the survival of patients with malignant brain tumors. [J Natl Cancer Inst 2005;97:1589 – 600]

Mechanistic studies of cancer cells at the molecular level may help to elucidate tumor behavior, to predict prognosis, and to develop targeted and effi cacious therapies. Transformation of cultured primary cells into tumorigenic variants is a multistep

process in which each genetic change confers a proliferative ad-vantage ( 1 ) . In most cancers, gene mutations disrupt normal cell cycle checkpoints that otherwise limit cell division in response to DNA damage or oncogene activation. The retinoblastoma protein (pRb) is a common downstream target of the signal cascade that regulates cell cycle checkpoints. Particular components of the retinoblastoma (Rb) regulatory network may act as tumor sup-pressors or proto-oncogenes, mutations in which are apparently essential for cancer cell development ( 2 , 3 ) . Inactivation of Rb expression in cancer cells leads to the deregulated activity of the transcription factors E2F1, E2F2, and E2F3, which control the expression of genes involved in differentiation, development, proliferation, and apoptosis. E2F1 has oncogenic properties in vitro and in vivo, and its overexpression is suffi cient to induce the transformation of rodent primary cells ( 4 ) and to promote tumorigenesis in transgenic mice ( 5 ) . Surprisingly, E2F1 also in-duces apoptosis in vitro and in vivo ( 6 ) , and E2F1-null mutant mice develop spontaneous malignancies, suggesting that E2F1 may be a tumor suppressor ( 7 , 8 ) . Thus far, however, E2F1 inac-tivation or loss-of-function mutations have not been consistently found in human cancer.

A current tenet of carcinogenesis is the hypothesis that hu-man cells must overcome at least two barriers before effi cient cellular proliferation can occur. The fi rst barrier, senescence, minimally involves the Rb tumor-suppressor pathway. How-ever, to sustain immortalization, cells must eventually overcome the second proliferative blockade, known as crisis ( 9 , 10 ) . To overcome crisis, telomeric DNA must be stabilized through telomerase activation ( 11 ) or through activation of the alterna-tive lengthening of telomeres pathway ( 12 , 13 ) . Human telo-merase is a reverse transcriptase that contains an RNA moiety, known as hTR, and a catalytic subunit, known as hTERT

Affi liations of authors: Departments of Neuro-Oncology (MMA, JF, HJ, O-HL, JX, WKAY, CG-M), Pathology (KDA, JMM), Neurosurgery (YK, TK), and Biostatistics (BNB, XZ), University of Texas M. D. Anderson Cancer Center, Houston, TX; Department of Cell Biology, University of Texas Southwestern Medical Center, Dallas, TX (JWS); Department of Carcinogenesis, Univer-sity of Texas M. D. Anderson Cancer Center, Science Park-Research Division, Smithville, TX (DGJ); Department of Obstetrics and Gynecology, Kanazawa University School of Medicine, Ishikawa, Japan (SK); Department of Pathology, University of California School of Medicine, San Francisco, CA (JN).

Correspondence to: Candelaria Gomez-Manzano, MD, Department of Neuro-Oncology, Box 1002, University of Texas M.D. Anderson Cancer Center, 1515 Holcombe Blvd., Houston, TX 77030 (e-mail: [email protected] ).

See “ Notes ” following “ References. ”

DOI: 10.1093/jnci/dji340 © The Author 2005. Published by Oxford University Press. All rights reserved. For Permissions, please e-mail: [email protected] .

Downloaded from https://academic.oup.com/jnci/article-abstract/97/21/1589/2521454by gueston 23 March 2018

Page 2: ARTI CLEExpression of Transcription Factor E2F1 and Telomerase

1590 ARTICLES Journal of the National Cancer Institute, Vol. 97, No. 21, November 2, 2005

( 14 , 15 ) . hTERT adds telomeric repeats to telomeres by using the RNA moiety hTR as a template ( 16 ) . In the absence of telomerase activity, telomeric DNA increasingly erodes with each round of cell division ( 17 ) , so that telomeres progressively shorten, which eventually leads to senescence in normal cells ( 18 ) . The lack of telomerase activity may also result in end- to-end fusion of chromosomes, chromosome instability, and cell death ( 19 , 20 ) .

Several tumor suppressor pathways have been identifi ed as modulators of telomerase function. In addition, recent studies performed in normal human cells found an association between telomerase activity and the cell cycle ( 21 ) . Current research is focusing on the role of E2F1 as a key regulator of telomerase expression in normal cells ( 21 – 23 ) , specifi cally, on the proximal region of the hTERT promoter sequence that contains, at least, three putative E2F1 binding sites ( 22 , 23 ) .

We examined the functional role of E2F1 in regulating telo-merase in malignant brain tumors. Our primary goal was to deter-mine whether there is a direct link between E2F1 and telomerase biology. If successful, our study would indicate that the Rb-E2F1 pathway is one of the critical tumor-suppressor/oncogene path-ways involved in regulating telomerase expression and activity in glioblastoma. An additional goal was to determine whether the overexpression of E2F1 and telomerase might serve as prognos-tic markers in patients with glioblastoma.

P ATIENTS AND M ETHODS

Cell Culture

Human cell lines Saos-2 (osteosarcoma), HeLa (cervical ade-nocarcinoma), and U-251 MG (malignant glioma) were obtained from the American Type Culture Collection (Manassas, VA). Cells were synchronized as described ( 24 ) . Briefl y, confl uent cul-tures of Saos-2 cells were serum-starved for 3 days by culturing in McCoy’s 5a medium (Gibco, Carlsbad, CA) supplemented with 0.1% fetal bovine serum (Gemini, Woodland, CA) and then synchronized by subculturing at a ratio of 1:2 in medium contain-ing 15% fetal bovine serum.

Adenoviral Vectors, Wild-Type Adenovirus, and Infection Conditions

The replication-defi cient adenoviral vectors Ad5CMV-E2F1 (Ad-E2F1, where CMV is cytomegalovirus), Ad5CMV-Rb (Ad-Rb), Ad5CMV-p21 (Ad-p21), and Ad5CMV- β -gal (Ad- β -gal, where β -gal is β -galactosidase) and the wild-type adenovirus serotype 5 (Ad-WT) were as described previously ( 25 – 27 ) . We used a multiplicity of infection of 100 plaque-forming units per cell. As controls, we used the replication-defi cient adenoviral vector Ad5CMV-pA (Ad-CMV) (with an empty expression cas-sette), the wild-type adenovirus inactivated by exposure to seven cycles of 125-J UV radiation (254 nm) (UV-inactivated Ad-WT), and mock infections with culture medium.

Cell Cycle Analysis

The distribution of cells in the cell cycle was analyzed by measuring their DNA content with fl ow cytometry, as described previously ( 28 ) . Cell samples were collected at the indicated

times after infection with Ad-E2F1, Ad-Rb, Ad-p21, Ad-CMV, Ad-WT, or UV-inactivated Ad-WT or treatment with the cyclin-dependent kinase inhibitor olomoucine (1 μ M ).

Luciferase Assays

Cells were cultured at a density of 3 × 10 4 cells per well in 24-well dishes for 24 hours and then transfected with 250 ng of the hTERT promoter – luciferase reporter construct (pGL3-hTERT). To determine transfection effi ciency, cells were trans-fected with 1 ng of the renilla vector pRL-CMV (Promega, Madison, WI) ( 29 ) . All transfections used the FuGENE 6 transfection reagent (Roche Diagnostics Corp., Indianapolis, IN). One hour after transfection, cells were infected with Ad-E2F1, Ad-Rb, Ad-p21, Ad-CMV, Ad- β -gal, Ad-WT, or UV-inactivated Ad-WT at a multiplicity of infections of 100 plaque-forming units, or cells were treated with 1 μ M olomou-cine. Cells were harvested 24 hours after treatment, and hTERT promoter – driven luciferase reporter activity was measured via the Dual Luciferase assay (Promega). Luciferase activity from untreated control cells was used as the background signal, nor-malized by renilla activity, and expressed as the fold of induc-tion relative to mock-treated cells, to which we assigned an arbitrary value of 1 U.

Telomere Repeat Amplifi cation Protocol

We harvested 10 6 U-251 MG or Saos-2 cells by trypsiniza-tion, and the pellet was resuspended in 100 μ L of ice-cold CHAPS lysis buffer {0.5% 3-[(3-cholamidopropyl)dimethylammonio]-1-propanesulfonate [CHAPS], 0.1 m M benzamidine, 10 m M Tris-HCl at pH 7.5, 1 m M MgCl 2 , 1 m M EGTA, 10% glycerol, and 5 m M 2-mercaptoethanol}, and then RNase in-hibitor (Q-Biogene, Irvine, CA) was added to a fi nal concentra-tion of 150 U/mL. Telomerase activity was assayed by using the Chemicon TRAPeze telomerase detection kit according to the manufacturer’s protocol (Chemicon, Temecula, CA). This kit is a modifi cation of the telomere repeat amplifi cation proto-col (TRAP) developed by Kim et al. ( 30 ) . Each analysis in-cluded a sample incubated at 95 °C for 10 minutes before the assay as a negative control and HeLa cells as a positive control. Every sample contained control primers and a telomere sub-strate control for amplifi cation of a 36-base-pair (bp) internal standard (internal control). The products of the TRAP assay were resolved by polyacrylamide gel electrophoresis in nonde-naturing 12% gels in Tris-glycine buffer (25 m M Tris and 192 m M glycine at pH 7.5). The gel was stained with Syber green (Molecular Probes, Eugene, OR) and visualized at a wave-length of 254 nm in a Fluor-S Multimager (Bio-Rad Laborato-ries, Hercules, CA). For every TRAP reaction, the relative signal intensity of the band ladder corresponding to multimers of the telomere repeats was measured with the National Insti-tutes of Health (NIH) Image analysis program (developed at the U.S. NIH and available on the Internet at http://rsb.info.nih.gov/nih-image/ ), and the results were normalized to those from the internal control.

Immunoblot Analysis of TERT and E2F1 Expression

For immunoblot analysis of TERT expression, nuclear ex-tracts were prepared from 80% confl uent cell cultures by use of a

Downloaded from https://academic.oup.com/jnci/article-abstract/97/21/1589/2521454by gueston 23 March 2018

Page 3: ARTI CLEExpression of Transcription Factor E2F1 and Telomerase

Journal of the National Cancer Institute, Vol. 97, No. 21, November 2, 2005 ARTICLES 1591

nuclear extraction kit (Active Motif, Carlsbad, CA) and the man-ufacturer’s instructions. Samples containing identical amounts of protein (20 – 100 μ g, depending on the antibody used) were sub-jected to sodium dodecyl sulfate – Tris-glycine gel electrophoresis and transferred to nitrocellulose membranes, as described else-where ( 28 ) . Blots were incubated with the antibodies against E2F1 (product KH-95) and β -actin (product C-11) from Santa Cruz Biotechnology (Santa Cruz, CA) and hTERT (product Est-21-A) and mouse TERT (mTERT; product Est-23A) from Alpha Diagnostic (San Antonio, TX). The secondary antibodies were horseradish peroxidase – conjugated anti-mouse, anti-rabit, and anti-goat immunoglobulin G (IgG) antibodies (Santa Cruz Biotechnology). Immunoblots were developed according to Amersham’s enhanced chemiluminescence protocol (Amersham Bioscience, Piscataway, NJ). Protein expression was quantifi ed by densitometry analysis with the public domain NIH Image program as explained above.

E2F1 protein levels were analyzed by immunoblot analysis from 27 snap-frozen glioblastoma multiforme specimens (World Health Organization [WHO] grade IV astrocytoma), from pa-tients selected as described below. Signal intensity was normal-ized to that of tubulin (anti-human tubulin monoclonal antibody T5168; Sigma Chemical Corp., St Louis, MO) levels of expres-sion in every sample. Normal human astrocyte cultures were used as the control for E2F1 expression in normal cells.

Reverse Transcriptase – Polymerase Chain Reaction

Analysis of hTERT, mouse (m) TERT, hTR, and glyceralde-hyde-3-phosphate dehydrogenase (GAPDH) mRNA expression was performed by reverse transcriptase (RT) – polymerase chain reaction (PCR) amplifi cation. Total RNA was collected from samples with a kit from Quigen (Valencia, CA) by following the manufacturer’s protocol. hTERT, mTERT, and hTR mRNAs were amplifi ed with primers and PCR conditions as described elsewhere ( 31 – 33 ) . Reaction products were then resolved by electrophoresis on 2% agarose gel. Amplifi cation of GAPDH RNA was used as the loading control to normalize the amount of input RNA.

Chromatin Immunoprecipitation Assays

Cells were infected with Ad-E2F1 or Ad-CMV or were mock-infected for 24 hours. The chromatin immunoprecipitation (ChIP) assay was performed by use of the ChIP assay kit (Upstate Bio-technology, Lake Placid, NJ) by following the manufacturer’s instructions. E2F1 antibodies (product KH-95) or mouse IgG antibodies (Santa Cruz Biotechnology) were used to immuno-precipitate the cross-linked chromatin. The hTERT proximal promoter was amplifi ed by PCR as described elsewhere ( 22 ) .

For the ChIP assay to assess mTERT (GenBank accession number AF121949 ), we designed the following three pairs of primers: 5 ′ -CAAATGCTAGGTCGGTGGAT-3 ′ and 5 ′ -GATGGCAGCTCTGCTAGGTT-3 ′ to generate a product of 273 bp; 5 ′ -CCCTGGCCTCATGTCCTCTAA-3 ′ and 5 ′ -CCCCAAAGAAACCAGCATAG-3 ′ to generate a product of 382 bp; and 5 ′ -CCCTGGCCTCATGTCCTCTAA-3 ′ and 5 ′ -TTTGGCAAACACTGAAATGC-3 ′ to generate a product of 767 bp. PCR conditions were as follows: one cycle at 94 °C for 3 minutes; 33 cycles at 94 °C for 30 seconds, 53 °C for 30 seconds, and 72 °C for 20 seconds; and a 1-minute cycle at 72 °C.

Construction of the Glial Fibrillary Acidic Protein (GFAP)-E2F1 Transgene and Generation of Transgenic GFAP-E2F1 Mouse Lines

The Pst I- Bam HI fragment from pSI (Promega), including the chimeric intron, multiple cloning region, and simian virus 40 late polyadenylation signal, was inserted into the Pst I- Bam HI site of pBluescript II KS+ (Stratagene, La Jolla, CA) to produce pBSI. The blunt-ended full-length human E2F1 cDNA was inserted into the blunted Sma I- Nhe I restriction sites of pBSI, and then the human GFAP promoter ( 34 , 35 ) was inserted into the Hin dIII- Eco RV restriction sites upstream of the chimeric intron. Founder lines were generated by injecting the 4.1-kilobase Spe I- Hin dIII fragment of the GFAP-E2F1 transgene into Friend virus B sus-ceptibility hybrid zygotes. Transgenic founders were developed by the Transgenic Animal Facility at the M. D. Anderson Cancer Center, Science Park-Research Division. Transgenic lines were maintained by breeding founders with a Friend virus B wild-type strain. Animal studies were performed in the veterinary facilities at M. D. Anderson Cancer Center in accordance with institutional guidelines.

Isolation of Cortical Astrocytes

Normal mouse astrocytes and transgenic GFAP-E2F1 mouse astrocytes were obtained from the brain cortex of postnatal day 4 mice as previously described ( 36 , 37 ) . Cultures were maintained in Dulbecco’s modifi ed Eagle medium – F-12 medium (1:1; Gibco) supplemented with 10% fetal bovine serum. To determine the number of cell population doublings, cells were seeded in six-well plates at a density of 10 4 cells per well and counted after 6 days. The number of population doublings was calculated according to the formula: log( N h ) − log( N i ) = x log 2, where N h is the number of cells harvested, N i is the number of cells inocu-lated, and x is the number of population doublings ( 38 ) . The experiment was carried out three times with quadruplicate samples.

Patient Selection

All tumors were glioblastoma multiforme. The two separate groups of fresh-frozen glioblastoma multiforme tissues (WHO grade IV astrocytoma) were obtained from patients enriched for long-term survival (i.e., alive ≥ 2 years after diagnosis; Table 1 ). The fi rst group contained 34 samples from the tissue bank at the Brain Tumor Center in the Department of Neurological Surgery, University of California, San Francisco, that were included in an expression array study (U95Av2; Affymetrix, Santa Clara, CA), for which detailed methodology and validation are published elsewhere ( 39 ) . We studied E2F1 and hTERT mRNA levels in that cohort of patients. Five samples of non-neoplastic brain obtained from patients undergoing surgery for temporal lobe epilepsy were used for comparison. A second group of 27 fresh-frozen glioblastoma multiforme tissues (WHO grade IV astrocy-toma) was selected from the tissue bank at the Brain Tumor Center at M. D. Anderson Cancer Center, Houston. E2F1 protein levels were measured in these samples by immunoblotting. All tumors were from patients with newly diagnosed glioblastoma multiforme who had received no therapy before sample collec-tion. All patients underwent gross total resection of their tumors, as assessed by postoperative imaging studies, and had completed

Downloaded from https://academic.oup.com/jnci/article-abstract/97/21/1589/2521454by gueston 23 March 2018

Page 4: ARTI CLEExpression of Transcription Factor E2F1 and Telomerase

1592 ARTICLES Journal of the National Cancer Institute, Vol. 97, No. 21, November 2, 2005

conventional external beam radiation therapy after surgery. The median Karnofsky performance score of the 61 patients was 90. The glioblastoma multiforme tumor samples contained at least 90% tumor, as assessed in a section cut from the tumor. The local institutional review boards approved the study, and all patients signed consent for specimens to be used for research purposes. Overall survival was determined as the interval from initial sur-gery to death and was obtained from the patient’s medical record or the Social Security Death Index.

Affymetrix Expression Array

E2F1 and hTERT mRNA levels were analyzed from a microarray database containing information on 34 fresh-frozen glioblastoma multiforme tissues (WHO grade IV astrocytoma), as reported previously ( 39 ) . Briefl y, test arrays were used fi rst to determine the quality of the complementary RNA (cRNA), and 15 μ g of cRNA was hybridized to U95Av2 human GeneChip expression arrays, according to the manufacturer’s specifi ca-tions (Affymetrix). Scanning and image analysis of the arrays were performed according to the manufacturer’s protocols. Data were analyzed by use of the PerfectMatch algorithm, as previously described ( 40 ) . Expression data for a small propor-tion of the probe sets were related to the batch in which the samples were processed due to technical differences in pro-cessing, and this set of genes (approximately 5% of the total) was excluded from further analysis. Data were also analyzed by use of the Affymetrix Microarray program suite (MAS ver-sion 5.0). Similar results were obtained with the two methods. Information on validation of the expression array (by real-time RT – PCR and immunohistochemistry) has been published elsewhere ( 39 ) .

Statistical Analyses

For in vitro experiments, statistical analyses were performed with two-tailed Student’s t tests. Data are expressed as mean and 95% confi dence interval (CI). The correlation between E2F1, E2F5, and telomerase expression and activity was assessed with Spearman’s rank test. Survival was assessed by the Kaplan – Meier method, and comparisons were analyzed with the log-rank test. To analyze survival data in the second set of sample patients, patients were reclassifi ed into two groups by E2F1 protein ex-pression as determined from immunoblots (low expression <1 or

high expression ≥ 1) according to the martingale residual plots. Univariate analysis and multivariable models were fi t by use of a Cox proportional hazards regression model. The proportional hazards assumption was evaluated by graphical displays and determined to be adequate for the purpose of this assessment. All statistical tests were two-sided.

R ESULTS

E2F1-Mediated Regulation of hTERT Promoter Activity in Cancer Cells

To investigate the putative role of E2F1 in regulating telome-rase activity, we selected two telomerase-positive human cancer cell lines [U-251 MG and HeLa ( 17 ) ] and one telomerase- negative cell line [Saos-2 ( 13 ) ] ( Fig. 1, A ). Adenoviral-mediated overexpression of exogenous E2F1 increased hTERT promoter activity in U-251 MG cells (9.5-fold, 95% CI = 8.2-fold to 10.7-fold; P <.001), in HeLa cells (11.3-fold, 95% CI = 9.8-fold to 12.7-fold; P <.001), and also in telomerase-negative Saos-2 cells (8.7-fold, 95% CI = 7.8-fold to 9.5-fold; P <.001), all com-pared with basal activity in corresponding mock-infected cells ( Fig. 1, B ). Overexpression of exogenous E2F1 in Saos-2 cells increased levels of hTERT mRNA and protein ( Fig. 1, C and D ) and also reactivated telomerase activity ( Fig. 1, E ), without mod-ifying the mRNA levels of the hTR, the RNA moiety of telome-rase. Thus, overexpression of E2F1 results in hTERT transcription, increased hTERT protein expression, and the reactivation of the enzymatic function of hTERT.

Physical Interactions Between Transcription Factor E2F1 and the hTERT Promoter

The hTERT promoter sequence contains at least three putative E2F1 binding sites in the proximal region of the promoter ( 21 , 41 ) . To determine whether E2F1 can bind to the hTERT promoter, we transferred exogenous E2F1 into Saos-2 cells and examined the interaction between E2F1 and the hTERT promoter with a ChIP assay. We detected an interaction between E2F1 protein and the hTERT promoter within 24 hours of transferring exogenous E2F1 ( Fig. 1, F ). As expected, we detected low levels of interaction between the hTERT promoter and E2F1 in control Ad-CMV – infected or mock-infected cells.

E2F1-Mediated Induction of Telomerase Activity in Saos-2 Cells

To ascertain whether endogenous E2F1 levels were associated with the induction of functional telomerase, we determined the S-phase fraction in Saos-2 cultures synchronized by serum star-vation and measured telomerase activity by the TRAP assay in these cultures. At 36 and 48 hours after serum-starvation release, S-phase cells represented 45.5% (95% CI = 36.8% to 54.1%) and 53.8% (95% CI = 47.3% to 60.2%) of cell populations in the cultures, respectively ( P <.001), compared with that in cells at 0 hour ( Fig. 2, A ). In addition, compared with cultures at 0 hour, corresponding S-phase – enriched cultures contained increased levels of E2F1 protein and increased telomerase activity that increased progressively with time ( Fig. 2, B ).

To confi rm whether the endogenous E2F1 participated in the induction of hTERT expression, we examined the effect of the

Table 1. Clinical profi le and molecular analysis of patients with glioblastoma multiforme *

Characteristic Group 1 Group 2

No. of patients 34 27 Age range, y 25 – 66 25.5 – 80.3 No. censored/observation time range, wk 1/459.4 9/45.3 – 173.6 Long-term survivors, % † 29.4 40.7 hTERT RNA expression Yes NA * E2F1 RNA expression Yes NA E2F1 protein expression NA Yes

* Group 1 contains 34 patients whose tumors were used to assess the expres-sion of E2F1 and hTERT mRNAs with the Affymetrix expression assay. Group 2 contains 27 patients whose tumors were used to assess the expression of E2F1 protein by immunoblot analysis. NA = sample not available; hTERT = catalytic subunit of human telomerase.

† Long-term survivors = patients alive ≥ 2 years after diagnosis.

Downloaded from https://academic.oup.com/jnci/article-abstract/97/21/1589/2521454by gueston 23 March 2018

Page 5: ARTI CLEExpression of Transcription Factor E2F1 and Telomerase

Journal of the National Cancer Institute, Vol. 97, No. 21, November 2, 2005 ARTICLES 1593

Rb-binding adenoviral oncoprotein E1A on the expression of hTERT mRNA and protein. We expected that the E1A- mediated release of E2F1 transcriptional activity would result in in-creased levels of telomerase. In fact, we found that infection of Saos-2 cells with Ad-WT increased both E2F1 transcriptional activity and hTERT promoter activity (11-fold increase, 95% CI = 10.5-fold to 11.2-fold; P <.001), compared with those in mock-infected cells (data no shown). Adenoviral-mediated induction of endogenous E2F1 was also associated with detect-able levels of hTERT mRNA and protein ( Fig. 2, C ) and reacti-

vation of telomerase activity in Saos-2 cultures ( Fig. 2, D ). Thus, increased expression of endogenous E2F1 appears to be associ-ated with increased hTERT transcription and increased telome-rase activity.

Restoration of the Rb Pathway and hTERT Expression

We next investigated whether restoring the Rb pathway in telomerase-positive U-251 MG cells would reduce the tran-scriptional activity of E2F1 and thus reduce hTERT expression

Fig. 1. E2F1 and transcriptional regulation of the expression of hTERT, the catalytic subunit of human telomerase. A ) Analysis of basal telomerase activity in a panel of cancer cell lines, including HeLa (positive control), heat-treated (HT) HeLa (negative control), U-251 MG, and Saos-2 cells with the telomere repeat amplifi cation protocol (TRAP). IC = internal control; bp = base pair(s). B ) hTERT promoter activity after transfer of E2F1 cDNA to cells. Cells were transfected with an hTERT promoter – luciferase reporter construct and 1 hour later were treated with adenovirus carrying E2F1 ( open bars ), adenovirus carrying cytomegalovirus (CMV; solid bars ), or adenovirus carrying β -galactosidase ( β -Gal, hatched bars ) constructs or were mock-treated ( shaded bars ). After 24 hours, cells were harvested, and luciferase activity was measured. All values were normalized to the expression of renilla luciferase; data are expressed as relative to those of mock-infected cells (arbitrary value of 1 U). Data are the means of fi ve experiments. Error bars = 95% confi dence intervals. For U-251 MG cells (*), P <.001; for HeLa cells ( • ), P <.001; and for Saos-2 cells (+), P <.001. C ) Detection of hTERT mRNA expression in Saos-2 cells. Reverse transcriptase – polymerase chain reaction (PCR) was performed 24 or 48 hours after transfer of E2F1 cDNA.

Shown are levels of the RNA component of telomerase hTR and glyceraldehyde-3-phosphate dehydrogenase (GAPDH) as a loading control. D ) hTERT protein expression in Saos-2. Immunoblotting was performed 24 hours transfer of E2F1 cDNA, as indicated. The percentage of cells in S phase, indicated below the lanes, was determined in duplicate samples by fl ow cytometry. Loading control was β -actin. Results are representative of those from three independent experiments, all with similar results. E ) Detection of telomerase activity in Saos-2 cultures. TRAP was performed at the indicated time points after infection with an adenovirus carrying E2F1 cDNA. F ) Interaction between E2F1 protein and the hTERT promoter. The chromatin immunoprecipitation assay was used to investigate this interaction. Cells were collected 24 hours after infection with an adenovirus carrying E2F1 cDNA and then incubated with antibodies ( α ) against E2F1 or immunoglobulin G (IgG). We used a nonspecifi c mouse IgG as negative control for the immunoprecipitation (IP) and a plasmid containing the hTERT promoter as a positive control [(+) C] for the PCR assay. Ten percent of the total cell lysate before immunoprecipitation was subjected to PCR to determine the chromatin input, as shown at the bottom.

Downloaded from https://academic.oup.com/jnci/article-abstract/97/21/1589/2521454by gueston 23 March 2018

Page 6: ARTI CLEExpression of Transcription Factor E2F1 and Telomerase

1594 ARTICLES Journal of the National Cancer Institute, Vol. 97, No. 21, November 2, 2005

and telomerase activity. Within 24 hours of transfer, Rb or p21 cDNAs arrested U-251 MG cells in G 0 – G 1 phase of the cell cycle (G 0 – G 1 /S-phase cell ratio in Ad-Rb – treated cultures = 7.6, 95% CI = 6.1 to 9, P <.001; in Ad-p21 – treated cultures = 13.0, 95% CI = 11.6 to 14.3, P <.001; and in Ad-CMV – infected cultures = 3.5, 95% CI = 2.5 to 4.4). Treating these cells with the cyclin-dependent inhibitor olomoucine ( 42 ) also statistically signifi cantly increased the ratio of G 0 – G 1 /S-phase cells by 48 hours after treatment (9.5, 95% CI = 8.4 to 10.5) compared with that in mock-treated cultures (3.8, 95% CI = 2.4 to 5.1) (differ-

ence = 5.7, 95% CI = 3.9 to 7.7; P <.001). We detected a statisti-cally signifi cant reduction of hTERT promoter activity 24 hours after treatment. The reduction in hTERT promoter activity was statistically signifi cant and strongest in cells treated with Ad-Rb (20% of control luciferase activity, 95% CI = 18.8% to 21.1%; P <.001) compared with that in cells treated with Ad-CMV. Transfer of p21 cDNA (65%, 95% CI = 62.7% to 67.2%; P <.001) or treatment with olomoucine (48%, 95% CI = 46.7% to 49.2%; P <.001) also resulted in moderate and statistically signifi cant inhibition of hTERT promoter activity, compared

Fig. 2. Telomerase activity in synchronized Saos-2 cultures. A ) Progressive accumulation of S-phase cells in synchronized Saos-2 cell cultures. The percentage of cells in S-phase was determined by the DNA content of cells as assessed by fl ow cytometry. Percentages of Saos-2 cells in the S phase of the cell cycle after serum-starvation release are shown below the histograms as the mean of three independent experiments and its corresponding 95% confi dence intervals. Arrowheads indicate the peaks for the 2 n (G 0 /G 1 phase) and 4 n (G 2 /M phase) cell populations. B ) Telomerase activity as assessed by the telomere repeat amplifi cation protocol (TRAP) after cell cycle block release in HeLa (positive control), heat-treated (HT) HeLa (negative control), and Saos-2 cells. Activity of hTERT, the catalytic subunit of human telomerase, was assessed at 0, 12, 24, 36, and 48 hours after releasing Saos-2 cells from serum starvation; it was detected at 36 hours and had increased at 48 hours. E2F1 protein was detected by immunoblot analysis at 24 hours after serum block release, and its level had increased at 36 hours and at 48 hours. Saos-2 cells were collected at

the indicated times after serum stimulation and analyzed by immunoblotting. β -Actin was the loading control for the immunoblot. C ) Detection of hTERT expression in Saos-2 cells. Reverse transcriptase – polymerase chain reaction to detect hTERT mRNA ( upper panel ) and immunoblot analysis to detect hTERT protein ( lower panel ) were performed at the indicated points after wild-type (WT) adenoviral infection. Glyceraldehyde-3-phosphate dehydrogenase (GADPH) was used as the loading control for PCR, and β -actin was the loading control for immunoblotting. Expression of adenoviral E1A protein and increased expression of E2F1 protein are observed. The percentage of cells in the S phase of the cell cycle is indicated below the immunoblot (values were representative of three independent experiments). UVi = UV-inactivated WT adenovirus; mock = mock-infected. D ) Detection of telomerase activity in Saos-2 cultures. TRAP was performed at the indicated times after infection with WT adenovirus or UV-inactivated WT adenovirus or mock infection. HeLa was the positive control, and HT HeLa was the negative control. IC = internal control.

Downloaded from https://academic.oup.com/jnci/article-abstract/97/21/1589/2521454by gueston 23 March 2018

Page 7: ARTI CLEExpression of Transcription Factor E2F1 and Telomerase

Journal of the National Cancer Institute, Vol. 97, No. 21, November 2, 2005 ARTICLES 1595

with that in Ad-CMV – treated cell cultures ( Fig. 3, A ). Expres-sion of hTERT mRNA could not be detected after transferring pRb or p21 cDNAs or treatment with olomoucine ( Fig. 3, B ). The reductions in hTERT promoter activity and mRNA ex-pression were also associated with reductions in the hTERT protein levels ( Fig. 3, C ). Finally in U-251 MG cells, when the

Rb pathway was restored, telomerase activity could not be detected ( Fig. 3, D ).

E2F1-Mediated Regulation of Telomerase in Normal Murine Glial Cells

Telomerase activity in the mouse brain decreases shortly after birth ( 43 ) . To investigate the role of E2F1 in the regulation of telomerase in glial cells, we isolated astrocytes from E2F1 trans-genic mice and wild-type mice. E2F1 protein levels were higher in astrocytes derived from E2F1 transgenic mice than in those derived from their counterpart wild-type mice (i.e., normal mouse astrocytes) ( Fig. 4, A ). The modifi cation in E2F1 expression was related to the proliferative phenotype of the cultures. That is, dur-ing a 6-day period, the population size of normal mouse astro-cytes doubled three times (95% CI = 1.4 to 4.5 times) and that of transgenic E2F1 astrocytes doubled fi ve times (95% CI = 3.5 to 6.4 times).

Transgenic E2F1 astrocytes contained increased levels of mTERT protein ( Fig. 4, A ) and of promoter activity (7.0-fold increase, 95% CI = 6.5-fold to 7.4-fold; P <.001), compared with the low level of basal transcriptional activity observed in normal mouse astrocytes ( Fig. 4, B ). Moreover, expression of exogenous E2F1 in normal mouse astrocytes resulted in in-creased hTERT promoter activity (8.5-fold, 95% CI = 7.8-fold to 9.1-fold; P <.001), compared with mock-, Ad-CMV – , or Ad- β -gal – treated normal mouse astrocytes. Furthermore, transfer of Rb cDNA to transgenic E2F1 astrocytes reduced hTERT transcriptional activity ( P = .005), compared with the level observed in mock- or Ad-CMV – treated transgenic E2F1 astrocytes, to a level similar to the level observed in normal mouse astrocytes ( Fig. 4, B ). In agreement with these data, the mTERT transcript was present in transgenic E2F1 but was dramatically reduced in transgenic E2F1 astrocytes transfected with pRb cDNA ( Fig. 4, C ). Finally, we observed telomerase activity in astrocytes from postnatal day 4 transgenic E2F1 mice but not in control normal mouse astrocytes ( 41 ) . How-ever, telomerase activity was clearly noticeable in normal mouse astrocytes expressing ectopic E2F1. Transfer of pRb to transgenicE2F1 astrocytes reduced mTERT expression and telomerase activity ( Fig. 4, D ).

Analyses of the mTERT promoter sequence identifi ed four possible binding sites for E2F1 ( Fig. 4, E ). Experiments with dif-ferent combinations of primers showed that the putative E2F-binding site was located between positions – 1682 and – 1313 of the mTERT promoter. This direct interaction between E2F1 and the mTERT promoter was detected in astrocytes isolated from wild-type and E2F-transgenic animals, suggesting that E2F1 directly interacts with the mTERT promoter in normal mouse glial cells ( Fig. 4, E ).

Correlation of E2F1 and hTERT Expression Levels in Glioblastoma

We next examined the mRNA expression levels of E2F1 and hTERT in a series of human glioblastoma multiforme specimens ( Table 1 ). We found a correlation between E2F1 mRNA expression levels and hTERT mRNA expression levels ( R = .8; P <.001) ( Fig. 5, A ). To demonstrate the specifi city of this correlation, we found that hTERT mRNA expression was not correlated ( R = – .04) with the expression of E2F5, another E2F

Fig. 3. Restoration of the retinoblastoma (Rb) pathway and expression of hTERT, the catalytic subunit of human telomerase. A ) hTERT promoter activity assessed by a luciferase reporter assay (as described in Fig. 1, B ). hTERT promoter activity was assayed in U-251 MG cells 24 hours after the transfer of pRb or p21 cDNA or treatment with 1 μ M olomoucine (Olo). All values were normalized to the expression of renilla luciferase; data are expressed relative to those of mock-treated cells (mock; arbitrary value of 1 U). Cultures infected with Ad- β -Gal ( β -gal) were used as control for specifi city. * P <.001, compared with cells infected with adenovirus carrying cytomegalovirus (CMV). Data are the mean of fi ve independent experiments. Error bars = 95% confi dence intervals. B and C ) Modulation of the Rb-E2F1 pathway and the expression of hTERT mRNA ( B ) and protein ( C ) in U-251 MG cells. hTERT mRNA was assessed by reverse transcriptase – polymerase chain reaction, and hTERT protein was assessed by immunoblotting at the indicated times after mock infection; infection with adenovirus carrying CMV, pRb cDNA, or p21 cDNA; or treatment with 1 μ M olomoucine (Olo). Glyceraldehyde-3-phosphate dehydrogenase (GADPH) was the loading control. D ) Telomerase activity after the transfer of pRb or p21 cDNAs or treatment with 1 μ M olomoucine in U-251 MG cells. Telomerase activity was assessed by the telomere repeat amplifi cation protocol in lysates of U-251 MG cells at the indicated times after treatment. Histone 3 was the loading control. IC = internal control.

Downloaded from https://academic.oup.com/jnci/article-abstract/97/21/1589/2521454by gueston 23 March 2018

Page 8: ARTI CLEExpression of Transcription Factor E2F1 and Telomerase

1596 ARTICLES Journal of the National Cancer Institute, Vol. 97, No. 21, November 2, 2005

family member whose activity is not involved in the positive transcriptional activation of cell cycle progression. Thus, E2F1 may be involved in triggering or maintaining hTERT expression in malignant gliomas and, therefore, favoring the perpetuation of the neoplastic phenotype in these tumors.

E2F1 and hTERT as Prognostic Factors in Malignant Gliomas

Because we observed a strong correlation between hTERT expression and E2F1 expression in glioblastoma multiforme

Fig. 4. Telomerase activity in transgenic E2F1 astrocytes. A ) Expression of mTERT, the catalytic subunit of mouse telomerase, and E2F1 protein in GFAP-E2F1 astrocytes from transgenic E2F1 (tgE2F1) mice and normal mouse astrocytes (NMAs). Proteins were assessed by immunoblot analyses, with β -actin as the loading control. B ) hTERT promoter activity was assayed by use of hTERT-driven luciferase activity in NMA and tgE2F1 astrocytes. Higher basal hTERT promoter activity was detected in tgE2F1 astrocytes than in NMAs ( P <.001). Transferring E2F1 (* P <.001) or retinoblastoma (Rb) ( • , P = .005) cDNAs modulated this activity, relative to that in mock-infected cells. Data are the mean of fi ve independent experiments. Error bars = 95% confi dence intervals. CMV = cytomegalovirus. C ) mTERT mRNA in astrocytes and the transfer of Rb cDNA. mRNA was assessed by reverse transcriptase – polymerase chain reaction(PCR) in normal and transgenic astrocytes and 24 hours after the transfer of E2F1 or Rb cDNAs. Glyceraldehyde-3-phosphate dehydrogenase (GAPDH) was used as the loading control. D ) Telomerase activity in astrocytes from normal and tgE2F1 mice and after the transfer of E2F1 and Rb cDNAs and

the percentage of cells in S phase. The telomere repeat amplifi cation protocol was used to measure telomerase activity. The DNA content of cells as assessed by fl ow cytometry was used to determine the percentage of cells in S phase. HeLa was the positive control, and heat-treated (HT) HeLa was the negative control. IC = internal control. Percentages of cells in S phase in parallel astrocyte cultures are shown below the lanes. E ) Chromatin immunoprecipitation (ChIP) assay and the interaction between E2F1 protein and the mTERT promoter in NMA and tgE2F1 astrocytes. Putative E2F-responsive elements are shown by boxes that contain sequences of primers used; predicted PCR product sizes in base pairs (bp) are indicated as A , B , and C . Cultures were collected and incubated with antibodies ( α ) against E2F1 or immunoglobulin G (IgG). The negative control [( − ) C] for the immunoprecipitation (IP) was a nonspecifi c mouse IgG. The positive control [(+) C] for the PCR assay was a plasmid containing the mTERT promoter. Ten percent of the total cell lysates was subjected to PCR before IP to determine the chromatin input, as shown at the bottom of the panel.

Downloaded from https://academic.oup.com/jnci/article-abstract/97/21/1589/2521454by gueston 23 March 2018

Page 9: ARTI CLEExpression of Transcription Factor E2F1 and Telomerase

Journal of the National Cancer Institute, Vol. 97, No. 21, November 2, 2005 ARTICLES 1597

Fig. 5. Expression of E2F1 and hTERT, the catalytic subunit of human telomerase, in human glioblastoma multiforme (GBM). A ) E2F1 and hTERT mRNA expression in a set of 34 human GBM samples. Scatter plot and loess smoothed line illustrate the correlation between hTERT and E2F1 expression levels. B ) Kaplan – Meier analysis of the association between survival and the expression of hTERT and E2F1 mRNA. The differences in overall survival were statistically signifi cant among the four groups (log-rank test, P = .015). The 4-year overall survival rates were 20% (95% confi dence interval [CI] = 7 to 55) for the low-E2F1/low-hTERT group, 50% (95% CI = 13 to 100) for the low-E2F1/high-hTERT group, and 7% (95% CI = 1 to 44) for the high-E2F1/high-hTERT group. C and D ) Kaplan – Meier survival curves stratifi ed by hTERT ( C ) or E2F1 ( D ) mRNA levels, with prognostic signifi cance by the

log-rank test using the median values as cut points. The probability of survival (%) and corresponding 95% confi dence levels are shown below the graphs. E ) E2F1 protein expression in two GBM samples (GBM A and GBM B = low and high E2F1 expression, respectively) from the second set of 27 GBM samples. Representative immunoblots are shown; E2F1 expression level is expressed as relative to the expression level of tubulin. Controls were lysates from U-251 MG cells infected with adenovirus carrying an E2F1 cDNA [(+) C] and from normal human astrocytes (NHA). Patient survival data are shown at the bottom. F) Kaplan – Meier survival curves stratifi ed by E2F1 protein expression levels in the second set of 27 GBM samples, with prognostic signifi cance by the two-sided log-rank test. All statistical tests were two-sided.

Downloaded from https://academic.oup.com/jnci/article-abstract/97/21/1589/2521454by gueston 23 March 2018

Page 10: ARTI CLEExpression of Transcription Factor E2F1 and Telomerase

1598 ARTICLES Journal of the National Cancer Institute, Vol. 97, No. 21, November 2, 2005

samples and because hTERT expression has been associated with reduced survival in glioma patients ( 44 , 45 ) , we investigated whether E2F1 and hTERT mRNA levels in the glioblastoma mul-tiforme samples were associated with patient survival. We used the median values for E2F1 expression (i.e., 234.1 arbitrary fl uo-rescent units) and hTERT expression (i.e., 347.61 arbitrary fl uo-rescent units) as cut points to divide patients into the following four groups: those with low expression of E2F1 and low expres-sion of hTERT ( n = 15); those with high expression of E2F1 and low expression of hTERT ( n = 2); those with low expression of E2F1 and high expression of hTERT ( n = 2); and those with high expression of E2F1 and high expression of hTERT ( n = 15). The differences in overall survival were statistically signifi cant among the four groups (log-rank test, P = .015) ( Fig. 5, B ). We evaluated potential differences among these four groups by age and sex with the adjusted Cox proportional hazards regression model. The analyses showed that high expression of E2F1 plus high or low expression of hTERT was negatively associated with sur-vival ( P = .009 and P = .05, respectively), compared with low expression of both genes. However, by conditioning on low- expressing E2F1 tumors, hTERT expression levels were not sta-tistically signifi cantly associated with patient survival ( P = .93).

Low hTERT mRNA expression in tumors was associated with longer median survival than high hTERT mRNA expression in tumors. The median overall survival was 97.9 weeks (95% CI = 60.9 to 206 weeks) for patients with low tumor hTERT expres-sion compared with 46.1 weeks (95% CI = 31.1 to 83 weeks) for patients with high tumor hTERT expression (difference = 51.8 weeks, 95% CI = 2.5 to 146; log-rank test, P = .038) ( Fig. 5, C ). In addition, low E2F1 mRNA expression in tumors was associ-ated with a longer median survival (103.6 weeks, 95% CI = 82 to 227) than was high E2F1 mRNA expression in tumors (46.1 weeks, 95% CI = 31.1 to 80) (difference = 57.5 weeks, 95% CI = 14.7 to 159.7; log-rank test, P = .002) ( Fig. 5, D ). The one study subject who was still alive at the end of this study was a member of the group of patients with tumors with low E2F1 and low hTERT expression.

Next, we assessed whether the association between overall survival and the expression levels of E2F1 and hTERT was mod-ifi ed by age (44.5 years, 95% CI = 44.2 to 47.7 years) and sex (male to female ratio = 1.13/1). The results were consistent with the log-rank analysis; i.e., the expression of E2F1 and hTERT was statistically signifi cantly associated with overall survival of patients with glioblastoma multiforme ( P = .003 and P = .042, respectively), but age and sex were not associated with overall survival ( P = .38 and P = .71, respectively).

Prognostic Value of E2F1 in Human Glioblastomas

Transcription factor E2F1 regulates the expression of genes involved in many cell functions in addition to telomerase ( 46 ) . To determine whether overexpression of E2F1 mRNA could be used as a prognostic factor that was independent of telomerase expres-sion, we investigated whether E2F1 protein expression was associated with overall survival, by use of an adjusted Cox proportional-hazards regression model by age and sex. We found that E2F1 protein expression was the only factor that statistically signifi cantly affected overall survival ( P = .04). Although hTERT was statistically signifi cantly associated with survival in the uni-variate Cox proportional hazards regression analysis, the adjusted analysis for age and sex showed that, in the presence of E2F1

protein expression, hTERT protein expression was not statisti-cally signifi cantly associated with survival ( P = .62). Further-more, tumors with a high level of E2F1 mRNA expression were associated with a 3.94-fold higher risk of death (95% CI = 1.07-fold to 14.52-fold) than that associated with tumors with a low level of E2F1 mRNA expression.

Although protein extracts from tumors were not available from the same sample cohort that was examined by the expres-sion array methodology, we used specimens from an independent group of 27 glioblastoma multiforme patients to examine the expression of E2F1 protein ( Table 1 ). We found that E2F1 protein levels were higher in protein extracts from these tumor speci-mens than that in protein extracts from normal human astrocytes ( Fig. 5, E ), which is consistent with the higher levels of E2F1 mRNA observed in the expression array of glioblastoma multi-forme compared with that in normal brain ( P = .018, Wilcoxon rank sum test). Moreover, the level of E2F1 protein expression was statistically signifi cantly associated with overall survival ( P <.001, log-rank test) ( Fig. 5, F ). In addition, consistent with the expression array data, low E2F1 protein expression in tumors was associated with longer survival (median survival time = 220 weeks, 95% CI = 126 to [not available]) than was high E2F1 protein expression in tumors (22.7 weeks, 95% CI = 11.6 to [not available]). Results from the univariate Cox proportional hazards regression model analysis were consistent with the log-rank anal-ysis, showing that the expression of E2F1 protein was associated with overall survival of patients with glioblastoma multiforme ( P <.001). Although age (57.2 years, 95% CI = 51.4 to 62.97) was also associated with overall survival ( P = .011) in the multivari-able analysis, only E2F1 protein expression was associated with the survival rate, increasing the risk of death 25.7-fold (95% CI = 4.97-fold to 133.81-fold).

D ISCUSSION

We found that the Rb-E2F pathway was involved in the regu-lation of telomerase expression and activity in cancer and normal cells, and we also found a statistically signifi cant correlation be-tween the levels of E2F1 and hTERT mRNA and protein in tissue samples of malignant brain tumors. Specifi cally, hTERT mRNA and protein levels, as well as telomerase activity, increased in cells overexpressing E2F1 protein. Moreover, the increased expression of hTERT mRNA and protein in cancer cells was sup-pressed by restoration of Rb pathway function. Isogenic astro-cytes that differed only in their levels of E2F1expression had markedly different levels of hTERT expression and activity, sug-gesting that the expression of E2F1 protein in normal cells may be suffi cient to trigger telomerase activity. Moreover, E2F1 and hTERT levels were strongly associated with survival in patients with glioblastoma multiforme, with E2F1 being the more strongly associated factor in multivariable analysis.

Our fi nding that E2F1 directly binds the hTERT promoter pro-vides a mechanistic link that underlies the regulation of telome-rase activity during cell cycle progression. In this model, free E2F1 activates telomerase in normal and cancer cells, and the Rb-E2F1 complex functions by repressing telomerase expression and activity. Our data support previous observations ( 47 – 49 ) about the role of Rb protein in repressing telomerase activity, and Rb protein may be at least partially responsible for the block of cell cycle progression and the decreased telomerase activity that is observed in quiescent cells cultured in the absence of serum

Downloaded from https://academic.oup.com/jnci/article-abstract/97/21/1589/2521454by gueston 23 March 2018

Page 11: ARTI CLEExpression of Transcription Factor E2F1 and Telomerase

Journal of the National Cancer Institute, Vol. 97, No. 21, November 2, 2005 ARTICLES 1599

( 50 , 51 ) . Furthermore, in normal cells, Rb protein is part of the E2F complexes that bind the hTERT promoter during G 0 /G 1 phases, but not during S phase ( 23 ) . Intriguingly, Rb and Rb- related proteins directly regulate telomere length, and they do so independently of the regulation of telomerase activity and the roles of Rb protein in the cell cycle ( 52 ) . Fibroblasts that do not express the Rb pocket proteins (i.e., pRb, p107, and p130) have the ability to rapidly elongate their telomeres, bolstering the pos-sibility that Rb family members have a direct role in regulating telomere length or telomere structure. Thus, multiple complex relationships between Rb protein and telomerase activity appear not only to govern telomerase function during the cell cycle but also to control chromosome structure and genomic stability beyond the strict limits of cell cycle regulation. The Rb pathway may also be involved in regulating hTR, the mRNA component of the telomerase complex ( 53 , 54 ) .

In normal cells evading Rb-mediated regulation of cell cycle progression, the resultant free E2F1 protein can induce hTERT promoter activity ( 21 ) . We found that mouse astrocytes engi-neered to constitutively express E2F1 had increased mTERT pro-moter transcriptional activity, mTERT protein levels, and telomerase activity. In addition, telomerase activity was not de-tected in tumor tissue from E2F-1 − / – mice but was detected in tissue from wild-type mice ( 55 ) . Thus, the combined results from transgenic and E2F1-null mouse models support the theory that E2F1 has a physiologic role in the regulation of telomerase.

The statistically signifi cant correlation between E2F1 and hTERT mRNA expression in a subset of patients with glioblas-toma multiforme confers clinical relevance to the mechanistic association between the expression of E2F1 and hTERT. Lower levels of E2F1 or hTERT mRNA expression were associated with longer median survival in patients with glioblastomas, compared with a higher level of either. Previous reports showed that telomerase activity in tumor cells is associated with progres-sion of malignant gliomas ( 56 ) , that hTERT expression is associ-ated with reduced survival in patients with malignant gliomas ( 45 ) , and that the absence of telomerase activity and the presence of the mechanism of alternative lengthening of telomeres are associated with longer survival in patients with glioblastoma multiforme ( 57 ) .

E2F1 was the factor most strongly associated with the overall survival of patients with high-grade gliomas. A previous study ( 44 ) reported that the expression of E2F1 protein was not corre-lated with the survival of patients with gliomas. Chakravarti et al. ( 44 ) also found underexpression of E2F1 in some glioblas-toma multiforme tumors. However, the expression of E2F1 in our glioblastoma multiforme samples was consistently the same or higher than that in our control samples (i.e., normal brain tis-sue and normal human astrocytes). Our results were confi rmed at protein and RNA levels in two independent sample cohorts, and a high level of E2F1 expression in both analyses was associated with poor prognosis for patients with glioblastoma multiforme. The discrepancy between these two studies can, at least in part, be explained by differences in the sample cohorts, because we analyzed a cohort enriched for patients with long-term survival.

Although our results are encouraging, our study had some limitations. First, because it made use of archival material, it had the potential biases associated with a retrospective study. In this regard, a prospective study may offer the possibility for yielding defi nitive conclusions. Moreover, because we were examining two cohorts of patients who were relatively homogeneous in

terms of prolonged survival, we cannot rule out the possibility that our results would be muted in a more heterogeneous group of patients. In addition, it would be premature to consider E2F1 expression as a reliable prognostic marker for clinical use, because more standard tests, such as immunohistochemistry or quantitative PCR, must be done to validate our conclusions. In addition, our conclusions pertain only to glioblastomas, and stud-ies in other types of solid tumors are required to broaden the role of E2F1 in the telomerase biology of these other tumors.

In conclusion, we have presented evidence that the level of E2F1 expression is strongly associated with survival of glioblas-toma patients. This association deserves further analysis in prospective studies. Because the treatment of glioblastoma multiforme is multimodal and involves several levels of inter-vention, the identifi cation of molecular markers that are active early in the development of this disease, such as E2F1 expression and telomerase activity, should facilitate the screening and selec-tion of those patients who require an intense treatment regimen as soon as the disease is diagnosed.

REFERENCES

(1) Hahn WC, Weinberg RA. Modelling the molecular circuitry of cancer. Nat Rev Cancer 2002 ; 2 : 331 – 41.

(2) Sherr CJ. Cancer cell cycles. Science 1996 ; 274 : 1672 – 7. (3) Classon M, Harlow E. The retinoblastoma tumour suppressor in develop-

ment and cancer. Nat Rev Cancer 2002 ; 2 : 910 – 7. (4) Johnson DG, Cress WD, Jakoi L, Nevins JR. Oncogenic capacity of the

E2F1 gene. Proc Natl Acad Sci U S A 1994 ; 91 : 12823 – 7. (5) Pierce AM, Schneider-Broussard R, Gimenez-Conti IB, Russell JL,

Conti CJ, Johnson DG. E2F1 has both oncogenic and tumor-suppressive properties in a transgenic model. Mol Cell Biol 1999 ; 19 : 6408 – 14.

(6) DeGregori J, Leone G, Miron A, Jakoi L, Nevins JR. Distinct roles for E2F proteins in cell growth control and apoptosis. Proc Natl Acad Sci U S A 1997 ; 94 : 7245 – 50.

(7) Field SJ, Tsai FY, Kuo F, Zubiaga AM, Kaelin WG Jr, Livingston DM, et al. E2F-1 functions in mice to promote apoptosis and suppress prolifera-tion. Cell 1996 ; 85 : 549 – 61.

(8) Yamasaki L, Jacks T, Bronson R, Goillot E, Harlow E, Dyson NJ. Tumor induction and tissue atrophy in mice lacking E2F-1. Cell 1996 ; 85 : 537 – 48.

(9) Hahn WC, Stewart SA, Brooks MW, York SG, Eaton E, Kurachi A, et al. Inhibition of telomerase limits the growth of human cancer cells. Nat Med 1999 ; 5 : 1164 – 70.

(10) Stewart SA, Weinberg RA. Telomerase and human tumorigenesis. Semin Cancer Biol 2000 ; 10 : 399 – 406.

(11) Masutomi K, Hahn WC. Telomerase and tumorigenesis. Cancer Lett 2003 ; 194 : 163 – 72.

(12) Reddel RR. Alternative lengthening of telomeres, telomerase, and cancer. Cancer Lett 2003 ; 194 : 155 – 62.

(13) Bryan TM, Englezou A, Dalla-Pozza L, Dunham MA, Reddel RR. Evidence for an alternative mechanism for maintaining telomere length in human tumors and tumor-derived cell lines. Nat Med 1997 ; 3 : 1271 – 4.

(14) Nakamura TM, Cech TR. Reversing time: origin of telomerase. Cell 1998 ; 92 : 587 – 90.

(15) Feng J, Funk WD, Wang SS, Weinrich SL, Avilion AA, Chiu CP, et al. The RNA component of human telomerase. Science 1995 ; 269 : 1236 – 41.

(16) Greider CW, Blackburn EH. A telomeric sequence in the RNA of Tetrahy-mena telomerase required for telomere repeat synthesis. Nature 1989 ; 337 : 331 – 7.

(17) Harley CB, Futcher AB, Greider CW. Telomeres shorten during ageing of human fi broblasts. Nature 1990 ; 345 : 458 – 60.

(18) Wei W, Sedivy JM. Differentiation between senescence (M1) and crisis (M2) in human fi broblast cultures. Exp Cell Res 1999 ; 253 : 519 – 22.

(19) Blasco MA, Lee HW, Hande MP, Samper E, Lansdorp PM, DePinho RA, et al. Telomere shortening and tumor formation by mouse cells lacking telomerase RNA. Cell 1997 ; 91 : 25 – 34.

Downloaded from https://academic.oup.com/jnci/article-abstract/97/21/1589/2521454by gueston 23 March 2018

Page 12: ARTI CLEExpression of Transcription Factor E2F1 and Telomerase

1600 ARTICLES Journal of the National Cancer Institute, Vol. 97, No. 21, November 2, 2005

(20) Counter CM, Avilion AA, LeFeuvre CE, Stewart NG, Greider CW, Harley CB, et al. Telomere shortening associated with chromosome insta-bility is arrested in immortal cells which express telomerase activity. EMBO J 1992 ; 11 : 1921 – 9.

(21) Won J, Yim J, Kim TK. Opposing regulatory roles of E2F in human telom-erase reverse transcriptase (hTERT) gene expression in human tumor and normal somatic cells. FASEB J 2002 ; 16 : 1943 – 5.

(22) Masutomi K, Yu EY, Khurts S, Ben-Porath I, Currier JL, Metz GB, et al. Telomerase maintains telomere structure in normal human cells. Cell 2003 ; 114 : 241 – 53.

(23) Won J, Chang S, Oh S, Kim TK. Small-molecule-based identifi cation of dynamic assembly of E2F-pocket protein-histone deacetylase com-plex for telomerase regulation in human cells. Proc Natl Acad Sci U S A 2004 ; 101 : 11328 – 33.

(24) Claudio PP, De Luca A, Howard CM, Baldi A, Firpo EJ, Koff A, et al. Functional analysis of pRb2/p130 interaction with cyclins. Cancer Res 1996 ; 56 : 2003 – 8.

(25) Jones N, Shenk T. Isolation of adenovirus type 5 host range deletion mutants defective for transformation of rat embryo cells. Cell 1979 ; 17 : 683 – 9.

(26) Fueyo J, Gomez-Manzano C, Yung WK, Liu TJ, Alemany R, McDonnell TJ, et al. Overexpression of E2F-1 in glioma triggers apoptosis and suppresses tumor growth in vitro and in vivo. Nat Med 1998 ; 4 : 685 – 90.

(27) Fueyo J, Gomez-Manzano C, Yung WK, Liu TJ, Alemany R, Bruner JM, et al. Suppression of human glioma growth by adenovirus-mediated Rb gene transfer. Neurology 1998 ; 50 : 1307 – 15.

(28) Gomez-Manzano C, Fueyo J, Kyritsis AP, McDonnell TJ, Steck PA, Levin VA, et al. Characterization of p53 and p21 functional interactions in glioma cells en route to apoptosis. J Natl Cancer Inst 1997 ; 89 : 1036 – 44.

(29) Takakura M, Kyo S, Kanaya T, Hirano H, Takeda J, Yutsudo M, et al. Cloning of human telomerase catalytic subunit (hTERT) gene promoter and identifi cation of proximal core promoter sequences essential for transcriptional activation in immortalized and cancer cells. Cancer Res 1999 ; 59 : 551 – 7.

(30) Kim NW, Piatyszek MA, Prowse KR, Harley CB, West MD, Ho PL, et al. Specifi c association of human telomerase activity with immortal cells and cancer. Science 1994 ; 266 : 2011 – 5.

(31) Ulaner GA, Hu JF, Vu TH, Giudice LC, Hoffman AR. Telomerase activity in human development is regulated by human telomerase reverse transcriptase (hTERT) transcription and by alternate splicing of hTERT transcripts. Cancer Res 1998 ; 58 : 4168 – 72.

(32) Yi X, Tesmer VM, Savre-Train I, Shay JW, Wright WE. Both transcriptional and posttranscriptional mechanisms regulate human telomerase template RNA levels. Mol Cell Biol 1999 ; 19 : 3989 – 97.

(33) Martâin-Rivera L, Herrera E, Albar JP, Blasco MA. Expression of mouse telomerase catalytic subunit in embryos and adult tissues. Proc Natl Acad Sci U S A 1998 ; 95 : 10471 – 6.

(34) Ding H, Roncari L, Shannon P, Wu X, Lau N, Karaskova J, et al. Astrocyte-specifi c expression of activated p21-ras results in malignant astrocytoma formation in a transgenic mouse model of human gliomas. Cancer Res 2001 ; 61 : 3826 – 36.

(35) Radany EH, Brenner M, Besnard F, Bigornia V, Bishop JM, Deschepper CF. Directed establishment of rat brain cell lines with the phenotypic character-istics of type 1 astrocytes. Proc Natl Acad Sci U S A 1992 ; 89 : 6467 – 71.

(36) Noble M, Murray K. Purifi ed astrocytes promote the in vitro division of a bipotential glial progenitor cell. EMBO J 1984 ; 3 : 2243 – 7.

(37) McCarthy KD, de Vellis J. Preparation of separate astroglial and oligoden-droglial cell cultures from rat cerebral tissue. J Cell Biol 1980 ; 85 : 890 – 902.

(38) Phillips PD, Cristofalo VJ. Cell culture of human diploid fi broblast in serum-containing and serum-free media. In: Baserga R, editor. Cell growth and divi-sion: a practical approach. Washington, DC: IRL Press; 1989 , p. 121 – 31.

(39) Nigro JM, Misra A, Zhang L, Smirnov I, Colman H, Griffi n C, et al. In-tegrated array-comparative genomic hybridization and expression array

profi les identify clinically relevant molecular subtypes of glioblastoma. Cancer Res 2005 ; 65 : 1678 – 86.

(40) Zhang L, Miles MF, Aldape KD. A model of molecular interactions on short oligonucleotide microarrays. Nat Biotechnol 2003 ; 21 : 818 – 21.

(41) Crowe DL, Nguyen DC, Tsang KJ, Kyo S. E2F-1 represses transcription of the human telomerase reverse transcriptase gene. Nucleic Acids Res 2001 ; 29 : 2789 – 94.

(42) Alessi F, Quarta S, Savio M, Riva F, Rossi L, Stivala LA, et al. The cyclin-dependent kinase inhibitors olomoucine and roscovitine arrest human fi bro-blasts in G1 phase by specifi c inhibition of CDK2 kinase activity. Exp Cell Res 1998 ; 245 : 8 – 18.

(43) Klapper W, Shin T, Mattson MP. Differential regulation of telomerase activ-ity and TERT expression during brain development in mice. J Neurosci Res 2001 ; 64 : 252 – 60.

(44) Chakravarti A, Delaney MA, Noll E, Black PM, Loeffl er JS, Muzikansky A, et al. Prognostic and pathologic signifi cance of quantitative protein expres-sion profi ling in human gliomas. Clin Cancer Res 2001 ; 7 : 2387 – 95.

(45) Tchirkov A, Rolhion C, Kâemâeny JL, Irthum B, Puget S, Khalil T, et al. Clinical implications of quantitative real-time RT-PCR analysis of hTERT gene expression in human gliomas. Br J Cancer 2003 ; 88 : 516 – 20.

(46) Mèuller H, Bracken AP, Vernell R, Moroni MC, Christians F, Grassilli E, et al. E2Fs regulate the expression of genes involved in differentiation, development, proliferation, and apoptosis. Genes Dev 2001 ; 15 : 267 – 85.

(47) Xiang H, Wang J, Mao Y, Liu M, Reddy VN, Li DW. Human telomerase accelerates growth of lens epithelial cells through regulation of the genes mediating RB/E2F pathway. Oncogene 2002 ; 21 : 3784 – 91.

(48) Xu HJ, Zhou Y, Ji W, Perng GS, Kruzelock R, Kong CT, et al. Reexpression of the retinoblastoma protein in tumor cells induces senescence and telom-erase inhibition. Oncogene 1997 ; 15 : 2589 – 96.

(49) Nguyen DC, Crowe DL. Intact functional domains of the retinoblastoma gene product (pRb) are required for downregulation of telomerase activity. Biochim Biophys Acta 1999 ; 1445 : 207 – 15.

(50) Holt SE, Aisner DL, Shay JW, Wright WE. Lack of cell cycle regulation of telomerase activity in human cells. Proc Natl Acad Sci U S A 1997 ; 94 : 10687 – 92.

(51) Holt SE, Wright WE, Shay JW. Regulation of telomerase activity in immor-tal cell lines. Mol Cell Biol 1996 ; 16 : 2932 – 9.

(52) Garcâia-Cao M, Gonzalo S, Dean D, Blasco MA. A role for the Rb family of proteins in controlling telomere length. Nat Genet 2002 ; 32 : 415 – 9.

(53) Zhao J, Bilsland A, Hoare SF, Keith WN. Involvement of NF-Y and Sp1 binding sequences in basal transcription of the human telomerase RNA gene. FEBS Lett 2003 ; 536 : 111 – 9.

(54) Zhao JQ, Glasspool RM, Hoare SF, Bilsland A, Szatmari I, Keith WN. Activation of telomerase rna gene promoter activity by NF-Y, Sp1, and the retinoblastoma protein and repression by Sp3. Neoplasia 2000 ; 2 : 531 – 9.

(55) Crowe DL, Nguyen DC. Rb and E2F-1 regulate telomerase activity in human cancer cells. Biochim Biophys Acta 2001 ; 1518 : 1 – 6.

(56) Langford LA, Piatyszek MA, Xu R, Schold SC Jr, Shay JW. Telomerase activity in human brain tumours. Lancet 1995 ; 346 : 1267 – 8.

(57) Hakin-Smith V, Jellinek DA, Levy D, Carroll T, Teo M, Timperley WR, et al. Alternative lengthening of telomeres and survival in patients with glioblastoma multiforme. Lancet 2003 ; 361 : 836 – 8.

NOTES

Supported by grant RO1-CA80748 from the National Cancer Institute. We thank Joann Aaron, Department of Neuro-Oncology, University of Texas

M.D. Anderson Cancer Center, Houston, TX, and Vicky J. Williams, Department of Scientifi c Publication, University of Texas M.D. Anderson Cancer Center, Houston, TX, for editorial assistance.

Manuscript received February 22, 2005; revised August 24, 2005; accepted September 9, 2005.

Downloaded from https://academic.oup.com/jnci/article-abstract/97/21/1589/2521454by gueston 23 March 2018