Elevated expression of c-myc has been detected in a broad range of human cancers, indicating a key role for this oncogene in tumor development. Recently, an interaction between FUSE-binding protein–interacting repressor (FIR) and TFIIH/p89/XPB helicase was found to repress c-myc transcription and might be important for suppressing tumor formation. In this study, we showed that enforced expression of FIR induced apoptosis. Deletion of the NH2-terminal repression domain of FIR rescued the cells from apoptosis as did coexpression of c-Myc with FIR; thus, repression of Myc mediates FIR-driven apoptosis. Surprisingly, a splicing variant of FIR unable to repress c-myc or to drive apoptosis was frequently discovered in human primary colorectal cancers but not in the adjacent normal tissues. Coexpression of this splicing variant with repressor-competent FIR, either in HeLa cells or in the colon cancer cell line SW480, not only abrogated c-Myc suppression but also inhibited apoptosis. These results strongly suggest the expression of this splicing variant promotes tumor development by disabling FIR repression and sustaining high levels of c-Myc and opposing apoptosis in colorectal cancer. (Cancer Res 2006; 66(3): 1409-17)
- c-myc suppressor
- FBP-interacting repressor
- alternative splicing
- colorectal cancer
- gastrointestinal cancers: colorectal
- gastrointestinal cancers: colorectal
c-Myc plays a critical role in cell proliferation and tumorigenesis. The c-myc proto-oncogene is activated in a variety of tumors, and its ectopic expression induces transformation. The recent genetic construction of mice in which the expression of c-myc can be switched on or off in vivo has emphasized the significance of c-Myc expression for tumorigenesis. Ectopic c-myc expression in hematopoietic cells using the tetracycline regulatory system caused malignant T-cell lymphomas and acute myeloid leukemia; the subsequent inactivation of the transgene caused regression of established tumors ( 1). c-myc activation was also shown to be required for skin epidermal and pancreatic β-cell tumor maintenance in c-MYC-ERTAM transgenic mice ( 2, 3). These observations have provided encouragement for the future development of cancer therapies based on targeting individual oncogenes, such as c-myc.
c-Myc expression is tightly regulated, and in turn, c-Myc modifies the expression of a large and diverse set of target genes. The molecules and mechanisms regulating c-myc expression have not been fully enumerated, elucidated, or understood. The far upstream element (FUSE) is a sequence required for proper expression of the human c-myc gene ( 4). The FUSE is located 1.5 kb upstream of c-myc promoter P1 and binds the FUSE-binding protein (FBP), a transcription factor stimulating c-myc expression in a FUSE-dependent manner ( 5– 7). Yeast two-hybrid analysis revealed that FBP binds to a protein that has transcriptional inhibitory activity termed the FBP-interacting repressor (FIR). FIR interacts with the central DNA-binding domain of FBP ( 8). Recently, FIR was found to engage the TFIIH/p89/XPB helicase and repress c-myc transcription by delaying promoter escape. Cells from XPB and XPD patients are defective for FIR repression, suggesting that mutations in TFIIH impair the regulation of c-myc by FIR, perhaps contributing to tumor development ( 9).
Up to 60% of all human genes present at least one alternative splice variant ( 10), and alternative splicing has been documented to play a significant role in human disease, including cancer ( 11). Although alternative splicing has been observed in many cancer-associated genes, the functional significance of the splice variants related to carcinogenesis is unknown. In this study, we show that a splice variant of the c-myc repressor FIR plays an important role in the pathogenesis of human colorectal cancer. A FIR splicing variant lacking exon 2, existing only in tumors but not in the adjacent normal tissue, failed to repress c-Myc and inhibited FIR-induced apoptosis, suggesting an important role for this splicing variant of FIR in the tumorigenesis of human colorectal cancer.
Materials and Methods
Plasmids. Full-length FIR cDNA [hemagglutinin (HA)-FIR] and the FIR deleted of its first 77 amino acids (HA-FIRΔN77) and the alternatively spliced form of FIR lacking exon 2 (HA-FIRΔexon2), obtained from cancer tissues, were cloned into the pCGNM2 vector plasmid ( 9), respectively, to introduce the HA-tag at their NH2 termini. The human c-Myc expression vector (pcDNA3.1-c-myc; GeneStorm Expression-Ready Clones; Invitrogen, Huntsville, AL) was purchased, and c-Myc expression was confirmed by Western blot using anti-c-Myc (Upstate Biotechnology, Lake Placid, NY). Full-length FIR cDNA was cloned into pcDNA3.1. Plasmids were prepared by CsCl ultracentrifugation or Endofree Plasmid Maxi kit (Qiagen, Germantown, MD), and the DNA sequences were verified.
Human tissue samples. Tissues from 15 cases of primary colorectal cancer were surgically excised. Written informed consent was obtained from each patient before surgery. The tumor samples were obtained from tumor epithelium immediately after operative excision; tissues and the corresponding nontumor epithelial samples were 5 to 10 cm from the tumor. Two pathologists microscopically confirmed all tissue samples as adenocarcinomas. All excised tissues were immediately placed into liquid nitrogen and stored at −80°C until analysis.
Immunocytochemistry and flow cytometry analysis. HeLa cells were grown on coverslips overnight and then transfected with plasmids using LipofectAMINE Plus reagents (Life Technologies, Carlsbad, CA). At the indicated time after plasmid transfection, cells were treated for immunocytochemistry as described previously ( 12). The primary antibodies, mouse monoclonal anti-HA (Santa Cruz Biotechnology, Santa Cruz, CA) and rabbit polyclonal anti-c-Myc (Upstate Biotechnology), were diluted 1:500 and 1:200 in the blocking buffer, respectively. The primary rabbit polyclonal antibody against FIR was prepared using two synthetic peptides GDKWKPPQGTDSIKME (30-45) and EVYDQERFDNSDLSA (528-542) simultaneously immunized to enhance the possibility of antibody production (Japan Bio Services Co. Ltd., Saitama, Japan) and was diluted 1:200 in the blocking buffer. The coverslips were incubated at room temperature for 1 hour. After washing with PBS, the secondary antibodies Alexa Fluor 488–conjugated goat anti-rabbit or Alexa Fluor 594–conjugated goat anti-mouse IgG secondary antibody (Molecular Probes, Eugene, OR) were used at 1:1,000 dilution. DNA was counterstained with 4′,6-diamidino-2-phenylindole (DAPI) III (Vysis, Abbott Park, IL) and cells were observed under immunofluorescence microscopy (Leica QFISH, Leica Microsystems, Tokyo, Japan).
Cells were processed for two-color FACScan analysis ( 12) to quantify c-Myc suppression by FIR. Briefly, at the indicated times after transfection, cells were fixed at least 1 hour with −20°C ethanol and then treated with mouse anti-HA and rabbit anti-c-Myc as primary antibodies (1:200 dilution). After washing with PBS, the secondary antibodies FITC-conjugated anti-rabbit IgG (Sigma) and R-phycoerythrin (PE)–conjugated anti-mouse IgG (PharMingen, Mississauga, Ontario, Canada) were used at 1:100 dilution, respectively. Ten thousand cells of each sample were analyzed with flow cytometry setting c-Myc-FITC into FL1 and HA-PE into FL2. The transfected cells were identified as PE-positive cells shown on the X axis.
Apoptosis detection. Apoptotic cells were detected by terminal deoxynucleotidyl transferase (TdT)–mediated dUTP nick-end labeling (TUNEL) assay according to the manufacturer's instructions (Apoptosis Detection System, Fluorescein, Promega, Madison, WI). Briefly, HeLa cells cultured on cover glasses were fixed with paraformaldehyde at 4°C for 10 minutes on ice and permeabilized with 0.5% Triton X-100 solution in PBS for 5 minutes. After washing with PBS twice, apoptotic cells were visualized through detection of internucleosomal fragmentation of DNA using in situ nick-end labeling with TdT and FITC-labeled dUTP (MEBSTAIN Apoptosis kit; Medical & Biological Laboratories, Nagoya, Japan). HeLa cells were treated with DNase I (MessageClean kit, GenHunter Corp., Nashville, TN) for positive control. Briefly, cells were trypsinized following 4% paraformaldehyde fixation for 15 minutes on ice. Washing with PBS twice, cells were refixed with 70% ethanol at −20°C for 1 hour. After washing with PBS twice, cells were treated with 1 unit/mL DNase I for 5 minutes at room temperature following TdT treatment with FITC-labeled dUTP. DNA was counterstained with either DAPI III for microscopy or 5 μg/mL propidium iodide (PI) with 250 μg/mL RNase in PBS solution for two-color FACScan analysis.
Protein extraction and immunoblotting. Frozen tissue samples were dissolved in lysis buffer [7 mol/L urea, 2 mol/L thiourea, 2% CHAPS, 0.1 mol/L DTT, 2% IPG buffer (Amersham Pharmacia Biotech, Buckinghamshire, United Kingdom), 40 mmol/L Tris] using a Polytron homogenizer (Kinematica, Lucerne, Switzerland) following centrifugation (100,000 × g) for 1 hour at 4°C. The amount of protein in the supernatant was measured by protein assay (Bio-Rad, Hercules, CA). The proteins were separated by electrophoresis on polyacrylamide gels of suitable concentration and transferred to a polyvinylidene fluoride membrane (Millipore, Bedford, MA) in a tank transfer apparatus (Bio-Rad). The membrane was blocked with 5% skim milk in PBS for 1 hour. Rabbit polyclonal anti-FIR antibody, goat polyclonal anti-β-actin antibody (Santa Cruz Biotechnology), rabbit polyclonal anti-Myc (Upstate Biotechnology), and mouse monoclonal anti-caspase-9 (Upstate Biotechnology), diluted 1:1,000, 1:500, 1:500, and 1:1,000, respectively, in blocking buffer were used as primary antibodies. Goat anti-rabbit IgG horseradish peroxidase (HRP) conjugate (Jackson, West Grove, PA) diluted 1:3,000 and rabbit anti-goat IgG HRP (Cappel, West Chester, PA) diluted 1:500 were used as secondary antibodies. Antigens on the membrane were detected by enhanced chemiluminescence detection reagents (Amersham Pharmacia Biotech). The intensity of each band was measured by NIH Image.
Reverse transcription-PCR and real-time quantitative PCR. Total RNA and genomic DNA were extracted from tumor and nontumor epithelial tissues with the RNeasy Mini kit and DNeasy Tissues kit (Qiagen). cDNA was synthesized from total RNA with the First-Strand cDNA Synthesis Kit for reverse transcription-PCR (RT-PCR; Roche, Mannheim, Germany). Using the cDNA as a template, FIR cDNA was amplified with suitable primers by RT-PCR: forward 5′-GGCCCCATCAAGAGCATC-3′ and reverse 5′-GGGGCTGGGCCAGGGTCAG-3′. For control, glyceraldehyde-3-phosphate dehydrogenase (GAPDH) cDNA was amplified. The NH2-terminal region of FIR was amplified by RT-PCR with primers: forward 5′-AGACAGCGGAAGGAGCAAGAGTGG-3′ and reverse 5′-CTGTGCAGCTTCGGGGACCTCATA-3′. The PCR product was loaded on a 2.5% agarose gel (Promega), purified by gel extraction kit (Qiagen), and cloned with the pGEM-T Easy vector system (Promega) for DNA sequencing.
Real-time quantitative PCR of FIR cDNA using the LightCycler instrument (Roche) was carried out in 20 μL reaction mixture that consisted of a master mixture (LightCycler FastStart DNA Master SYBR Green I) that contained FastStart Taq DNA polymerase, deoxynucleotide triphosphate mixture, and buffer (LightCycler DNA Master hybridization probes; Roche), 3.0 mmol/L MgCl2, 0.5 μmol/L each of sense and antisense primers, and 1 μL template cDNA in a LightCycler capillary. LightCycler software version 3.3 (Roche) was used for the analysis of quantitative RT-PCR. Primers for FIR cDNA and c-myc cDNA are as follows: forward 5′-GCACCTGGAGTCATCACA-3′ and reverse 5′-CGCAGAACCATCACTGTAG-3′ (FIR) and forward 5′-GCCTCAGAGTGCATCGAC-3′ and reverse 5′-TCCACAGAAACAACATCG-3′ (c-myc).
The NH2-terminal domain of FIR is required to repress endogenous c-Myc expression. Previous studies revealed that the NH2 terminus of FIR was necessary to repress transcription from the c-myc promoter of a transfected reporter plasmid ( 8). To test the effect of FIR and its NH2-terminal deletion mutant (HA-FIRΔN77) on the endogenous c-Myc expression, HA-tagged, full-length FIR (HA-FIR) and the deletion mutant (HA-FIRΔN77) were expressed in HeLa cells and c-Myc expression was examined. HA-FIR strongly suppressed c-myc in both mRNA and protein levels compared with HA-FIRΔN77 ( Fig. 1B ). Note that both plasmids were confirmed to express the equal amounts of proteins detected by Western blot ( Fig. 1A). The suppression of c-Myc was also visualized by immunostaining with anti-c-Myc ( Fig. 1C, top). c-Myc levels were greatly diminished in HA-FIR-expressing cells (arrowheads) but were unperturbed in HA-negative cells, showing that FIR represses endogenous c-Myc expression through a cell autonomous mechanism. In great contrast to the full-length protein, deleting its NH2 terminus enfeebled the repressor activity of FIR (arrows). To quantify the repression of endogenous c-Myc, HA-FIR and HA-FIRΔN77 were transfected into HeLa cells. Analysis of c-Myc and HA-FIR levels using two-color FACScan (red for HA-FIR and green for c-Myc) confirmed that FIR, but not HA-FIRΔN77, effectively reduced c-Myc in a large population of cells ( Fig. 1D). Within the HA-FIR-positive population, c-Myc levels were distinctly bimodal ( Fig. 1D, left); c-Myc levels were most sharply depressed in the HA-FIR-transfected cells found in region 2 ( Fig. 1D, R2). In contrast, HA-FIRΔN77-transfected or HA-tag only–transfected cells were found only in region 1 ( Fig. 1D, R1), with c-Myc levels uniform and indistinguishable between transfected and nontransfected cells ( Fig. 1D, middle and right).
FIR-induced apoptosis is prevented by enforced expression of c-Myc. Inopportune up-regulation or down-regulation of c-myc may each be associated with apoptosis. In the absence of survival factors, c-myc is a potentiator of apoptosis, whereas in other instances, such as glucocorticoid or nitric oxide–driven apoptosis, c-Myc is required to protect cells from death ( 13). Because FIR depresses native c-myc expression, we tested the influence of enforced FIR expression on apoptosis. HA-FIR and HA-FIRΔN77 were transfected into HeLa cells and apoptosis was examined by TUNEL assay ( Fig. 1E). HA-FIR induced apoptosis with DNA fragmentation ( Fig. 1E, top, arrows; Supplementary Fig. S1), whereas little apoptosis occurred in cells transfected with HA-FIRΔN77 or the control vector. The extent of apoptosis driven by full-length HA-FIR was 16.5% as assessed with flow cytometry but dropped to only 6.6% with transfected HA-FIRΔN77 ( Fig. 1F).
If FIR suppression of c-myc drives apoptosis, then bypassing this repression should rescue cells from death. Expression from an exogenous promoter enabled the elevation of c-Myc levels even when cotransfected with FIR ( Fig. 2A, top right, arrowheads ). Augmented c-Myc expression protected HeLa cells from the FIR-induced nuclear swelling and degradation ( Fig. 2A, bottom right, arrowheads). Coexpression of c-Myc along with FIR reduced the TUNEL-positive cells ( Fig. 2B; Supplementary Fig. S1). The extent of apoptosis driven by FIR declined from 21.1% to 4.2% when c-Myc expression was enforced ( Fig. 2B). The prevention of apoptosis by c-Myc was also confirmed by gradual decrease of the cleavage product of caspase-9 along with the increase expression of c-Myc ( Fig. 2C, arrow). These results indicate that increasing FIR levels trigger apoptosis most likely due to c-myc suppression.
FIR is paradoxically up-regulated in colorectal cancer correlating with increased c-Myc. It is well known that c-Myc is overexpressed in the majority ( 14, 15) of colorectal cancers due to deregulation of c-myc expression ( 16). Thus, we hypothesized that cancer cells must escape FIR repression of c-myc. FIR itself might be down-regulated in cancer or the effector actions of FIR in transcription and apoptosis might be disabled. To test if FIR is down-regulated in colorectal cancer, FIR protein levels in tumors and normal tissue from 10 cases were examined by immunoblot. Surprisingly, FIR levels were actually increased in most colorectal cancer tissue compared with the corresponding nontumor epithelium ( Fig. 3A ). To determine if the increased FIR levels resulted from increased levels of FIR mRNA, RNA from colorectal cancer cells and normal colonic epithelium were examined by RT-PCR ( Fig. 3B). All cases, except case 5, showed higher FIR mRNA levels in tumors compared with nontumor tissue. FIR mRNA and protein levels paralleled each other in all cases. The increased FIR mRNA level in colorectal cancer cells was further confirmed by real-time, quantitative RT-PCR ( Fig. 3C). FIR gene amplification was also evaluated but was not observed (data not shown). Hence, the increase in FIR protein reflects either increased transcription or increased stability of FIR mRNA in colorectal cancers. The high levels of FIR in tumors were clearly ineffective at repressing c-myc; in fact, the level of FIR mRNA expression was positively correlated with the level of c-myc ( Fig. 3D). The overexpression of FIR in colorectal cancer, therefore, might represent a futile attempt at negative feedback.
An alternatively spliced form of FIR is expressed in tumors but not in the adjacent normal tissue. The results above suggest that the deregulation of c-myc in colorectal cancer is not due to down-regulation of FIR. How then does c-myc evade FIR repression in tumors? Either FIR must be defective or the c-myc promoter must somehow be made resistant to FIR. Although currently there is no evidence to suggest inactivating mutations in FIR, such as those occurring in tumor suppressor genes, alternative splicing would be another way to alter protein function. Although multiple splice variants of FIR exist (Ensembl NM_014281), the most common RNA variants, FIR and PUF60 ( 17), differ in that FIR lacks the 17 amino acids of exon 5. Another NH2-terminal variation removes exon 2, including a stretch of alanines ( 18); such stretches have been associated with transcriptional repression ( 19). RT-PCR of full-length FIR cDNAs isolated from HeLa cells or colorectal cancer tissues using primers to amplify the NH2-terminal region revealed the four variants (FIR and PUF60, FIRΔexon2, and PUF60Δexon2) expected from the alternative utilization of the two optional exons. DNA sequencing confirmed that the four products reflected all combinations of inclusion/exclusion of exons 2 and 5 ( Fig. 4A ). Although FIR protein is abundant in HeLa cells, PUF60 is at best scarce, below the threshold necessary for detection by immunoblot using anti–exon 5 antibody, and likely to be down-regulated at the protein level (anti–exon 5 antibody easily detects transiently expressed PUF60, which represses transcription as well as transfected FIR). 5 To examine if the alternative splicing of FIR is a tumor-specific event, FIR cDNA isolated from tumors and normal tissues of several cases were examined by RT-PCR. The splicing variant FIRΔexon 2 was only observed in tumor tissues and not in adjacent normal tissues or in blood cells from most of the matched cases ( Fig. 4B). The existence of FIRΔexon 2 could explain the high c-Myc expression in tumor cells due to its dominant-interfering effects. In normal cells where the c-myc is not activated, there would be little call to express any variant of FIR at high levels, and in fact, the level of both full-length FIR and FIRΔexon2 in most of the normal tissues is very low.
FIRΔexon2 is impaired for c-Myc suppression and apoptosis induction. The NH2 terminus of FIR is indispensable for the suppression of c-Myc and induction of apoptosis. Therefore, the deletion of the alanine-rich segment from FIR, as occurred in colorectal cancer, was likely to have some influence on these functions. To examine this possibility, expression plasmids of HA-FIRΔexon2 were introduced into HeLa cells and c-Myc expression was examined by RT-PCR and Western blot ( Fig. 5A ). Whereas HA-FIR reduced c-myc expression at both mRNA and protein levels, HA-FIRΔexon2 did not ( Fig. 5A), although HA-FIR and HA-FIRΔexon2 plasmids express the equal amounts of proteins (data not shown). This failure to suppress c-Myc was also visualized by immunostaining. FIRΔexon2 not only failed to suppress but also actually increased c-Myc expression ( Fig. 5B, arrows versus arrowheads) as confirmed by fluorescence-activated cell sorting analysis ( Fig. 5C, 3, arrow). This observation was most likely due to a dominant-interfering effect of FIRΔexon2 on the c-Myc suppression activity of endogenous FIR.
To ask if the removal of exon 2 also impaired the induction of apoptosis, as expected if occurring through depressed levels of c-Myc, FIRΔexon2 was transfected into HeLa cells. As hypothesized, FIRΔexon2 failed to induce apoptosis ( Fig. 6A and B, 1-3 ). Next, we asked if coexpression of FIRΔexon2 along with full-length FIR diminished apoptosis induced by the latter. Again as predicted, expression of FIRΔexon2 along with FIR not only reversed the c-Myc suppression but also protected cells from the apoptotic activity of FIR ( Figs. 5D and 6B, 4 and 5). These dominant-interfering effects of FIRΔexon2 on the c-Myc suppression and apoptotic induction of endogenous FIR were also observed in a colon cancer cell line SW480 (Supplementary Fig. S2). Taken together, the above results strongly indicate that up-regulation of c-Myc frequently observed in colorectal cancer is indeed enabled by the dominant interference of an alternatively spliced variant of FIR. The higher levels of c-Myc and reduced apoptosis supported by FIRΔexon2 would enhance the growth advantage of cancer cells over their normal neighbors.
Deregulated expression of c-Myc is detected in many tumor cell types and it has been proposed that increased c-Myc expression is instrumental in the initiation of the neoplastic phenotype in many, if not most, human tumors. However, the mechanisms that normally regulate c-myc expression, the defects that deregulate it in tumors, and how deregulated c-Myc expression contributes to tumorigenesis have not been fully elucidated ( 20). This study showed that FIR strongly repressed endogenous c-myc transcription and induced apoptosis. Most importantly, a splicing variant of FIR found frequently in human primary colorectal cancer tissues not only lacked the c-Myc-suppressing and apoptosis-inducing action of FIR but also prevented normal FIR from performing these activities. The alternative splicing of FIR may contribute to tumor progression by enabling higher levels of c-myc expression and greater resistance to apoptosis in tumors than in normal cell.
A large number of cancer-related genes that exhibit alternative splicing have been characterized, including CD44, WT1, BRCA1, MDM2, FGFR, and kallikrein family members ( 11). For example, WT1, a zinc finger transcription factor that is inactivated in the germ line of children with genetic predisposition to Wilms' tumor, has two major splicing variants, −KTS and +KTS ( 21, 22). The −KTS isoform acts as sequence-specific DNA binding factor, whereas the +KTS isoform plays a role in RNA processing ( 23– 26). A recent report showed that several splicing variants of HDM2, the human homologue of MDM2, were found together with full-length transcript; one of these variants inhibited the interaction of HDM2 with p53, thus increasing p53 levels and enhancing p53 transcriptional activity ( 27). Although these splice variants can function as dominant-negative inhibitors and interfere with the wild type, their selective occurrence in tumors has not been proven. This study has provided the first evidence of the selective expression and function of such alternative splicing for carcinogenesis. To show how the alternative splicing of FIR is differentially regulated between tumors and the surrounding normal tissue promises to expose further links between the earliest events in carcinogenesis with tumor progression.
Reducing c-myc mRNA causes apoptosis in a variety of transformed cell types ( 12). In addition, regression of established tumors due to inactivation of c-myc transgene in hematopoietic cells was associated with rapid proliferative arrest, differentiation, and apoptosis of tumor cells ( 1). These experiments indicate the FIR is an important player in these processes by directly regulating c-myc. Evidence for this scheme includes the following: (a) the NH2 terminus of FIR, essential for c-myc suppression, is also necessary for induction of apoptosis; (b) a splicing variant of FIR lacking exon 2 in the NH2 terminus failed not only to suppress c-Myc expression but also to induce apoptosis; and (c) enforced expression of c-Myc rescued cells from FIR-induced apoptosis. Given the central role of c-Myc in the development of many cancers and that cell death in tumors is increased on c-Myc suppression, one route to the development of cancer therapies directed against c-Myc may go through FIR and its variants.
Grant support: Grant-in-Aid 13214016 for Priority Areas in Cancer Research (T. Tomonaga) and 21st Century Center of Excellence Programs from the Ministry of Education, Science, Sports and Culture of Japan (T. Ochiai).
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 18 U.S.C. Section 1734 solely to indicate this fact.
We thank Dr. Chi Van Dang for his helpful suggestions.
Note: Supplementary data for this article are available at Cancer Research Online (http://cancerres.aacrjournals.org/).
↵5 H-J. Chung and D. Levens, unpublished data.
- Received December 20, 2004.
- Revision received November 14, 2005.
- Accepted November 21, 2005.
- ©2006 American Association for Cancer Research.