Amplification and Overexpression of the Dual-Specificity Tyrosine-(Y)- Phosphorylation Regulated Kinase 2 (DYRK2) Gene in Esophageal and Lung Adenocarcinomas1

[CANCER RESEARCH 63, 4136–4143, July 15, 2003] Amplification and Overexpression of the Dual-Specificity Tyrosine-(Y)- Phosphorylation Regulated Kinase 2 (DYRK2) Gene in Esophageal and Lung Adenocarcinomas1

Charles T. Miller, Sanjeev Aggarwal, Theodore K. Lin, Susan L. Dagenais, Jorge I. Contreras, Mark B. Orringer, Thomas W. Glover, David G. Beer, and Lin Lin2 Departments of Surgery, Section of Thoracic Surgery [C. T. M., S. A., T. K. L., J. I. C., M. B. O., D. G. B., L. L.] and Human Genetics [S. L. D., T. W. G.], University of Michigan Medical School, Ann Arbor, Michigan 48109

ABSTRACT fied previously a novel amplicon at 8p22–23 in gastroesophageal adenocarcinomas (8). Using the STS3-amplification mapping ap- Genomic amplification can lead to the activation of cellular proto- proach, the core-amplified domain was narrowed to include the cys- oncogenes during tumorigenesis, and is observed in most, if not all, human teine protease gene CTSB and the transcription factor GATA-4 (8, 9). malignancies, including adenocarcinomas of lung and esophagus. Using a two-dimensional restriction landmark genomic scanning technique, we Similarly, we identified and characterized an amplicon at 19q12 and identified five NotI/HinfI fragments with increased genomic dosage in an revealed cyclin E as the best candidate gene selected by the 19q12 adenocarcinoma of the gastroesophageal junction. Four of these amplified amplification in gastroesophageal adenocarcinomas (10). In addition fragments were matched within three contigs of chromosome 12 using the to oncogene activation via gene amplification, inactivation of the bioinformatics tool, Virtual Genome Scan. All three of the contigs map to tumor suppressor gene p53 occurs frequently in esophageal adenocar- the 12q13-q14 region, and the regional amplification in the tumor was cinoma (Ͼ87%) and early in low- and high-grade Barrett’s dysplasia verified using comparative genomic hybridization analysis. The 12q14 (11, 12). Amplification and overexpression of the MDM2 oncogene amplicon was characterized using sequence tagged site-amplification were reported in sarcomas lacking detectable p53 alterations, indicat- mapping with DNA from paired normal-tumor tissues of 75 gastroesoph- ing the role of coordinately abrogating the suppression of cell growth ageal and 37 lung adenocarcinomas. The amplicon spans a region of >12 regulated by the p53 pathway (13, 14). Whereas involvement of p53 Mb between genes DGKA and BLOV1. The core-amplified domain was determined to be <0.5 Mb between marker WI-12457 and gene IFNG. was observed frequently in esophageal adenocarcinomas, amplifica- However, MDM2, a well-documented oncogene of the region, is outside the tion of the MDM2 gene was not detected in one study (15) and was core-domain. Eleven genes and expressed sequence tags within the ampli- reported infrequently (4%) in another study in this type of tumor (16). con were selected for quantitative reverse transcription-PCR, and DYRK2, Using CGH, Moskaluk et al. (17) and van Dekken et al. (18) reported a member of the dual-specificity kinase family, was overexpressed in all of increased 12q DNA copy number in 27% and 39% of esophageal the tumors showing gene amplification. Among the sequence tagged site/ adenocarcinomas. However, these regions of gain were mapped to expressed sequence tag/gene markers tested, DYRK2 demonstrated the 12q21-q24, suggesting an additional 12q amplicon because MDM2 is highest DNA copy number and the highest level of mRNA overexpression mapped to the 12q13–15 region (13, 19). Tumors with amplified in the tumors. Moreover, DYRK2 mRNA overexpression (>2.5-fold of genes closely linked to MDM2 but without MDM2 amplification or normal mean) was found in 18.6% of additional 86 lung adenocarcinomas with discontinuous amplicons including MDM2 were also reported in an assay using oligonucleotide microarrays. DYRK2 mRNA overexpression occurs more frequently than gene amplification in both esophageal (20–22). To clarify the amplification status within the 12q region, the and lung adenocarcinomas. This is the first report of amplification and current investigation was undertaken using RLGS (23, 24) combined overexpression of DYRK2 in any tumor type. with a recently developed bioinformatics tool, VGS (25). Use of these tools allows individual genomic fragments amplified in a tumor genome to be identified. DYRK2, a dual-specificity tyrosine-(Y)- INTRODUCTION phosphorylation regulated kinase gene, was identified as the most The incidence of esophageal adenocarcinoma has rapidly increased frequently amplified and overexpressed gene among the STS/EST/ in many western countries including the United States, the United gene markers examined, and also demonstrated the highest mRNA Kingdom, the Netherlands, Denmark, and Australia over the past 3 overexpression level among the genes tested in this series of gastroe- decades (1, 2). A poor overall 5-year survival rate of Ͻ10% portends sophageal and lung adenocarcinomas. Both lung and esophageal ad- the future clinical importance of this disease (3). Invasive adenocar- enocarcinomas were examined in this study to determine whether the cinoma is accompanied frequently with Barrett’s metaplasia, an in- amplicon is tumor type-specific or whether it occurs in multiple tumor testinalized columnar epithelium that replaces the normal squamous types. The present study represents the first complete characterization epithelium at the distal esophagus as a result of chronic gastroesoph- of the 12q14 amplification in both lung and esophageal adenocarci- ageal reflux (4, 5). Smoking and obesity have been implicated as nomas, and is the first to report amplification and overexpression of additional risk factors for the disease (6, 7). DYRK2 in any human cancer. Amplification of dominant-acting oncogenes plays a significant role in the development and/or progression of many types of human MATERIALS AND METHODS cancer, including esophageal and lung adenocarcinomas. We identi- Tissue Collection. After obtaining written consent, 75 esophageal and Received 12/4/02; accepted 5/7/03. gastric cardia adenocarcinomas, the corresponding normal squamous esopha- The costs of publication of this article were defrayed in part by the payment of page gus or gastric mucosa, and 123 lung adenocarcinomas were obtained from charges. This article must therefore be hereby marked advertisement in accordance with patients undergoing esophagectomy or pulmonary resection at the University 18 U.S.C. Section 1734 solely to indicate this fact. 1 Supported by National Cancer Institute Grant CA71606 and the Weber Research Fund. Supported also by NIH Training Grant CA009672-08 (to S. A.). 3 The abbreviations used are: STS, sequence tagged site; RLGS, restriction landmark 2 To whom requests for reprints should be addressed, at MS RB II B560, Box 0686, genome scanning; VGS, Virtual Genome Scan; CGH, comparative genomic hybridiza- Department of Surgery, Section of General Thoracic Surgery, University of Michigan tion; QG-PCR, quantitative genomic-PCR; EST, expressed sequence tag; NCBI, National Medical School, Ann Arbor, MI 48109. Phone: (734) 647-2787; Fax: (734) 763-0323; Center for Biotechnology Information; RT-PCR, reverse transcription-PCR; GAPDH, E-mail: [email protected]. glyceraldehyde-3-phosphate dehydrogenase. 4136

Fig. 1. Two-dimensional RLGS analysis of NotI/HinfI DNA fragments from the normal and tumor tissues of patient E13, and a VGS image of virtual NotI/HinfI fragments of chromosome 12. DNA fragments from a gastroesophageal adenocarcinoma of patient E13 (B) and normal tissue from the same patient (A) were resolved on two-dimensional gels. As shown, the comparison of relative intensity of each fragment (spot) between normal (A) and tumor (B) reveals five such fragments, B1 through B5 (arrows in B), that are darker in tumor as compared with corresponding spots A1 through A5 (arrows in A) in normal, suggesting increased DNA copy number in each of the five fragments in the tumor. C, matched virtual fragments C1 through C4 (black dots indicated by arrows in C) are shown on a VGS image with NotI/HinfI fragments from chromosome 12 (white dots in C). Fragment C5 was not found with any chromosomes using the VGS tool analysis, indicating the possible unavailability of sequences for this fragment in the GenBank databases.

of Michigan Medical Center between 1992 and 2000. Patients in this study had for quantitative genomic PCR as described previously (9). PCR primers were no preoperative radiotherapy or chemotherapy. A small portion of each tissue designed to assure that the melting temperature (Tm) of each primer set was was embedded in OCT compound (Miles Scientific, Naperville, IL) and stored compatible with that of the internal control, herein GAPDH, which resides on Ϫ Ϫ at 70°C; the remainder was stored at 70°C until use. the same chromosome 12 so the ratio of tumorsample/GAPDH: normalsample/GAPDH DNA Isolation and RLGS-combined VGS Analyses. High molecular will reflect the actual DNA copy numbers increased in tumor. The quantity of weight DNA was extracted as described previously (26) from tumors with the normal and tumor genomic DNA was measured by fluorometry (TKO100; Ͼ70% tumor cellularity. Two-dimensional RLGS was performed as described Hoefer Scientific Instruments, San Francisco, CA) to ensure equity of the previously (27). In brief, DNA from normal and tumor tissues was double starting DNA. Forward primers for GAPDH and each STS/gene marker were digested using the restriction enzymes NotI and EcoRV (New England Biolabs end-labeled with [␥-32P]ATP (NEN Life Science Products) using T4 polynu- Inc., Beverly, MA), and NotI ends were filled with [␣-32P]dCTP and cleotide kinase (New England Biolabs). PCR was performed using Taq po- [␣-32P]dGTP (NEN Life Science Products, Boston, MA). First-dimensional lymerase (Promega, Madison, WI), and the PCR products were resolved in 8% size-fractioning was performed in a 0.9% disk-agarose gel. The resulting DNA denaturing polyacrylamide gels. PCR product signal ratios for both the tumor fragments were then in-gel digested with HinfI and separated in the second (tumor STS fragment:tumor GAPDH fragment) and normal DNA samples dimension on a 5.25% polyacrylamide gel. Gels were then dried and autora- (normal STS fragment:normal GAPDH fragment) were quantified using Im- diographic images obtained (Molecular Dynamics, Sunnyvale, CA). Digital ageQuant software (Molecular Dynamics). Values Ն2.0 were considered in- images were analyzed using ImageQuant v1.2 software (Molecular Dynamics). dicative of DNA amplification. All of the results were verified by repeating Amplified DNA fragments in tumors were quantified by densitometry using the two-dimensional images from the corresponding normal tissue DNA as standards. The VGS bioinformatics tools4 were used to match the positive Table 1 Primer sequences of gene/STS markers used for QG-PCR analysis of the fragments identified by actual two-dimensional RLGS analysis as described 12q14 amplicon previously (25). The chromosome 12 VGS image was captured and used to Gene/STS Forward primer (5Ј-3Ј) Reverse primer (5Ј-3Ј) C create Fig. 1 . DGKAa ggactctactcccgaggct ttcagatgaggcactttcaat CGH. The CGH procedure was based on the protocol described previously CDK2a tgccgtaccaatctctgaatgcc gactcctcccagtggttttgttttac with modifications (28). Briefly, 200 ng of DNA samples isolated from tumor ERBB3a ggtttcctccctttcctcttcttg gctttcccagtcccattcctcag a and the corresponding normal tissues was labeled by nick translation with STAT6 agctctgtccagcgagttc gcagccactgcttacactgc D12S329b aagcaatcagccagccct tgtcagaacctaacaacccagaaag fluorescein-isothiocyanate-dUTP (green) and Texas-Red-dUTP (red), respec- HMGA2a ttcaacaagcaagcgattc tctaccatttctgcaagttagg tively. The probes were mixed with human Cot-1 DNA (10 ␮g), denatured, MAN1c ggagtgacatcagcagaataggaa acagctttgatacaggtggttttg reannealed, and hybridized to methotrexate-synchronized male metaphase IRAK-Ma cccgtgcaagcaagctaaaccaa gacttcagccaaacaaaacaatcaaaa a,b chromosomes on glass slides, and incubated at 37°C for 3 days under humid WI-12457 cttactgttacaggagcaccaca tctgtgctggggaacaaag DYRK2a cagatgacagatgccaatgggaata ttgagcaaataaaaacaggtctgagc conditions. The preparations were washed to remove nonspecific bound DNA SHGC-57328a,b ctcccacttcccagcaaaag tggcaacagacccagagaga and counterstained with 4Ј,6-diamidoino-2-phenylindol for chromosome iden- D12S335 ctcccaacctcgcccagaga attatggtggcaaggacagacacaat tification. Images were captured using a Zeiss Axioskop fluorescent micro- D12S313 ctcagtcttgccttccacctc taattgattccaaactcattgatgtct c scope equipped with filters for CGH and analyzed using QUIPS software IFNG gacattcacaattgattttattcttaca cccccaatggtacaggtttc IL26(AK155)a,b aagcagtcagtaaaccaaagcca tttgaaaacacagaaaatgtgcaa (Vysis, Downers Grove, IL). Data from 20 metaphase spreads were used to MDM1a,b ttccccaagttttcattaccc aagtagcctggaggtaggttct generate a composite CGH profile. Threshold values were 0.8 for loss and 1.2 MDM2a gagggctttgatgttcctga gctactagaagttgatggc for gain. BLOV1c ctcgggggcggaagcactggag acagatcaaccccgggcgagtagcac STS-Amplification Mapping Using QG-PCR. STS markers located in D12S80 caccaccacccggctaatttttctat caggctgggtgcggtgctc the region identified as having high-level amplification by CGH were selected a Primers that prime PCR reactions for both genomic DNA and cDNA. b The sequences of these primers are retrieved from the NCBI databases from various resources. 4 Internet address: http://dot.ped.med.umich.edu:2000/VGS/index.html. c These primers can only be utilized for genomic PCR amplification. 4137

Table 2 The matched virtual NotI/HinfI fragments C1, C2, C3, and C4 (Fig. 1) using A260:280 ratio were used to assess RNA quality. RNA samples were treated with VGS tools (VGS 1.01) DNase I (Promega) before reverse transcription. Two ␮g of total RNA were One-dimensional, incubated with reverse transcriptase (Invitrogen Life Technologies, Inc.) and two-dimensional Matched contig primed by both (dT) and random hexamers in a total of 25 ␮l of reaction volume. Fragment (bp)a IDb Match sitec Locationd 18 cDNA (1.0–1.5 ␮l) underwent RT-PCR with GAPDH coamplified as an internal C1 1373, 732 NT_029419 2533943–2534671 12q14.2-q14.3 control. PCR products were then resolved on 8% PAGE gels, and gel data analyses C2 1260, 517 NT_009509 295882–296400 12q13.13-13.2 C3 887, 175 NT_009458 456700–456880 12q13.13 were performed using ImageQuant software as in QG-PCR. C4 1805, 312 NT_009458 1026030–1026339 Immunohistochemical Analysis. Cryostat sections (5 ␮m) were placed on a One-dimensional number indicates the NotI/EcoRV fragment size, in base pairs, 0.1% poly-L-lysine-coated slides and fixed with 100% acetone at Ϫ20°C for 10 migrated in the first-dimensional gel and two-dimensional number provides the size (bp) min. Endogenous peroxidase activity was quenched with three changes of of NotI/HinfI fragment resolved in the second-dimensional gel. b Contig ID is the ID name used in the NCBI databases. 0.3% hydrogen peroxide for 30 min each. Horse serum, at a dilution of 1:20 in c Match site denotes where the matched virtual NotI/HinfI fragments aligned within the PBS-1% BSA, was used to block nonspecific binding. The MDM2 protein was identified contigs. detected using an anti-MDM2 antibody (Santa Cruz Biotechnology, Santa d Chromosome map site is derived from the NCBI databases and maps (http://www. ncbi.nlm.nih.gov/cgi-bin/Entrez/map_search). Cruz, CA) at a 1:500 dilution in PBS-1% BSA. Negative controls were incubated without primary antibody. Immunoreactivity was detected using the Vectastain avidin/biotin complex kit (Vector Laboratories, Burlingame, CA) three times. The primers for the STS/EST/gene markers analyzed in the present and slides counterstained with Harris-modified hematoxylin. study are listed in Table 1. DYRK2 Expression Analysis in Lung Adenocarcinomas Using Oligonu- RNA Isolation and RT-PCR. Total RNA was isolated using TRIzol reagent cleotide (Affymetrix) Microarray Data. Total cellular RNA was isolated (Life Technologies, Inc., Gaithersburg, MD). Agarose gel electrophoresis and the from 10 normal and 86 lung adenocarcinoma samples using TRIzol reagent

Fig. 2. CGH analysis of tumor E13. The digitized fluorescent image of an arranged metaphase spread and the chromosomal ideograms from tumor E13 are shown. A, high-intensity green bands, indicative of increased regional DNA dosage, are present on the long arm of chromosome 12 near the centromere (arrow). B, the composite CGH profile from 20 metaphase spreads demonstrates that the regional genomic amplification is located at the 12q14 band in tumor E13.

4138

(Life Technologies, Inc.) and RNeasy Mini kit (Qiagen, Valencia, CA). Prep- aration of cRNA and hybridization of the HuGeneFL Arrays were performed according to the manufacturer’s protocol (Affymetrix, Santa Clara, CA). The arrays were scanned using the GeneArray scanner (Affymetrix), and data analysis was performed using GeneChip 4.0 software as described previously (29, 30). The level of DYRK2 expression was assessed by analyzing the DYRK2 data from our Affymetrix microarray database (30).

RESULTS Two-Dimensional RLGS and VGS of Gastroesophageal Adeno- carcinomas. High molecular weight DNA from 44 primary esophageal and gastric cardia adenocarcinomas was analyzed using the two-dimensional RLGS technique. Approximately 3000 individual NotI/HinfI fragments can be visualized in a two-dimensional gel allowing global comparison of two genomes from normal and tumor tissues (Fig. 1). Comparison of the DNA from a gastroesophageal adenocarcinoma (tumor E13, Fig. 1B) with the DNA from its corresponding normal tissue (Fig. 1A) revealed 5 DNA fragments with increased intensity in the tumor tissue (Fig. 1B). VGS 1.01, a bioinformatics tool for comparison of sample-derived RLGS patterns with patterns predicted from the human genome sequence, was used. The amplified NotI/HinfI fragment B1 that was ϳ1359 bp in the first dimensional gel and 739 bp in the second dimensional gel (1359/739), B2 (1239/530), B3 (896/178), and B4 (1786/312) were matched to their respective virtual fragments C1 through C4, all located in chromosome 12 (Figs. 1 and 4). The sequences of matched virtual NotI/ HinfI fragments were aligned within three chromosome 12 contigs, NT_029419, NT_009509, and NT_009458 (Table 2). A virtual fragment of similar RLGS size for restriction fragment B5/C5 (1878/300) was not found using VGS tools (Fig. 1), indicating possible unavailability of this specific genomic sequence in the current databases. All three contigs were confirmed in the region of chromosomal band 12q13-q14 after a comprehensive analysis using NCBI GenBank databases and tools5 (Table 2; Fig. 4B). Many of the very intensive spots visible in the two-dimensional gels represent ribosomal DNA, which is present in multiple copies (Fig. 1). Changes in the patterns of Fig. 3. QG-PCR applied to characterize the 12q14 amplicon in 75 gastroesophageal and 37 lung adenocarcinomas. The housekeeping gene GAPDH was used as an internal these two-dimensional fragments as well as other CpG island-con- control and coamplified with selected STS/EST/gene markers located in the chromosome taining fragments are attributable to alterations of DNA methylation 12q13-14 region. The intensity ratio (Ts/c:Ns/c) between tumor (Tsample/GAPDH) and and are often observed during tumor development (31, 32). normal (Nsample/GAPDH) samples from QG-PCR were quantified as described in “Mate- rials and Methods,” and the ratio numbers are listed under each pair of samples. Five Validation of Amplified Two-dimensional Fragments Using STS/EST/gene markers are shown representing the 19 markers investigated (Table 1). As CGH Analysis. DNA from tumor tissue of patient E13 and normal shown, tumor E13 is amplified at all markers (ϳ12 Mb) suggesting that a large amplicon is contained in tumor E13. The highest increased copy number was observed in the reference DNA were labeled by nick-translation with FITC-dUTP DYRK2 gene (6.4-fold) and a closely linked EST marker, WI-12457 (5.1-fold), in tumor (tumor) and Texas Red-dUTP (normal) then hybridized to normal G12. Tumor B81 contains CDK2 amplification but no other markers tested, indicating an metaphase chromosomes. High-level regional amplification was ob- existence of additional amplicon in the 12q13 region. MDM2 was not amplified in tumor G12, the tumor showing the highest DNA copy number of DYRK2. served in tumor E13 at chromosome 12q (Fig. 2A) with the peak at 12q14 based on a composite CGH profile of 20 metaphase spreads (Fig. 2B). A genomic gain was also observed at chromosome 12p and (Fig. 4). In tumor G12, QG-PCR analysis indicated a possible low 18p in the composite profile of the tumor (Fig. 2B). gain at the MDM2 locus. However, much higher values of increased Characterization of the 12q14 Amplicon Using STS-Amplifica- DNA dosage (Ͼ5-fold) were consistently observed at DYRK2 and tion Mapping. To determine the core-amplified domain and bound- WI-12457 in the same tumor G12 (Fig. 3; Fig. 4A). WI-12457 is an aries of the 12q14 amplicon, Ͼ19 STS/EST/gene markers, spanning EST fragment with high homology to TATA box binding protein- from the DGKA gene to D12S80 and covering ϳ15 Mb in the region, interacting protein (TIP120A) and is closely linked to DYRK2 (ϳ300 were selected from NCBI databases5 for QG-PCR analysis. The DNA Kbp). The core-amplified domain of the 12q14 amplicon is a contin- from paired tumor and corresponding normal tissues of 75 gastroe- uous and common region between both gastroesophageal and lung sophageal and 37 lung adenocarcinoma patients was screened with adenocarcinomas (Fig. 4A). A separate regional amplification was each of the markers using the QG-PCR assay (Fig. 3). QG-PCR found in the 12q13 region at the CDK2 locus in one tumor (B81, Fig. results were verified at least two times. The 12q14 amplicon is Ͼ12 3; Fig. 4A), suggesting an additional amplicon in the 12q region. We Mb with the centromeric cutoff at the DGKA gene and the telomeric also observed the centromeric boundary of the 12q14 amplicon in border around the D12S80 locus (Fig. 4). The core-amplified domain tumor E13 to include CDK2 (Fig. 3; Fig. 4A). is confined to a region Ͻ0.5 Mb between WI-12457 and IFNG DYRK2 Is a Potential Candidate Gene Selected by the 12q14 Regional Amplification. Eleven genes, including those located 5 Internet address: http://www.ncbi.nlm.nih.gov. within the core region of the 12q14 amplicon, were selected to 4139

Fig. 4. Diagrammatic map of the 12q13-14 amplicon in gastroesophageal and lung tumors determined using QG-PCR analysis, and a map of the matched virtual NotI/HinfI fragments aligned to the actual amplification map. A, F indicate that genomic amplification was detected in the individual tumors at the markers tested. E indicate that DNA amplification was not observed at the loci tested. The map order is based on QG-PCR analysis of 75 gastroesophageal and 37 lung adenocarcinoma tumors, but closely follows the December 2002 NCBI databases from various sources. As shown, the core-amplified domain (excluding tumor B81 of which the amplicon is centered on CDK2) is between the marker WI-12457 and gene IFNG, leaving MDM2 outside of the core-amplified region. Overexpression status of genes and ESTs in this region was analyzed, and DYRK2 was found to be the only gene overexpressed in all tumors amplified. B, the physical map of the matched virtual NotI/HinfI fragments as related to the actual amplicon map shown at left.

determine RNA expression levels in the amplified tumors using quantitative RT-PCR (Fig. 5). DYRK2 was overexpressed in all of the amplified tumors and also in tumor B81, which was not amplified at the DYRK2 locus (Figs. 3 and 5). mRNA overexpression of the remaining 10 genes/ESTs tested, including MDM2, was also detected but none of them were overexpressed in all of the amplified tumors (Fig. 5; Table 3). Moreover, in the tumors, the level of DYRK2 mRNA overexpression was much greater than all of the other genes/ESTs located within the 12q14 amplicon including MDM2 (Fig. 5). Gene amplification and overexpression seemed to be confined only to DYRK2 when the other DYRK family members, DYRK1A, DYRK1B, DYRK3, and DYRK4, were examined and amplification and overexpression were not detected in this series of tumors (data not shown). Although MDM2 is not included in the core-amplified domain in our series of adenocarcinomas (Fig. 3; Fig. 4A), we used an immunohistochemical assay to determine MDM2 protein expression because of the well-documented involvement of this gene in the p53 pathway in tumors. Tumors E13, G12, H38, and S01, the four tumors containing the 12q14 amplicon, were examined (Fig. 6, A–D). The 0SAS sarcoma cell line, which overexpresses MDM2, was used as a positive control (Fig. 6A). All of the negative controls showed no staining (data not shown). Lung adenocarcinoma H38 shows intense nuclear staining of MDM2 in a majority of tumor cells (Fig. 6B) compared with no staining in H38 normal lung (Fig. 6C). Tumor G12, which demonstrated only low gain in MDM2 copy number (Fig. 3), shows barely detectable nuclear staining in Ͻ5% of tumor nuclei (Fig. 6D). MDM2 staining was light and detected in Ͻ10% of tumor cell nuclei of tumor E13, and no nuclear staining was observed in lung tumor S01 (data not shown). DYRK2-specific antibodies are currently unavailable. Frequent Overexpression of DYRK2 mRNA in Lung Adenocar- cinomas. DYRK2 gene expression was analyzed using our oligonucleotide microarray database of 86 lung adenocarcinomas and 10 normal lungs (30). The gene expression data demonstrated a significant increase of DYRK2 expression in stage I lung tumors Fig. 5. Quantitative RT-PCR analysis applied to 11 genes within the 12q14 amplicon; (P ϭ 0.00088) and stage III tumors (P ϭ 0.00073) as compared with 5 of them are shown. GAPDH was coamplified as a control in each reaction. As shown, the expression level in normal tissues. The difference of DYRK2 all 5 tumors demonstrate DYRK2 mRNA overexpression, including tumor B81 that did not contain DYRK2 gene amplification (Fig. 3). MDM2 mRNA overexpression is not detected expression between stage I and III lung tumors was not found in two tumors, B81 and S01, and the latter (tumor S01) contains low-level MDM2 (P ϭ 0.20). Array analysis of the expression patterns of the genes amplification (Fig. 3). 4140

Table 3 Candidate search results using quantitative RT-PCR for amplified genes/ESTs was observed to be frequently amplified at high-levels (Ͼ30-fold) in a located within the 12q13-14 amplicon sarcomas (33.3%) and gliomas (8–10%; Refs. 14, 33), our findings Gene/EST B81 E13 G12 H38 S01 are consistent with two previous publications in which amplification CDK2 ϩϩϪϪϪof MDM2 was observed in only 0–4% of esophageal adenocarcino- ERBB3 ϪϪϪϪϪ MAN1 ϪϪϩϩnt mas (15, 16). Infrequent or absent amplification of MDM2 was also IRAK-M ϪϩϩϩϪdocumented in many other human cancers, including esophageal WI-12457 ϪϪϩϩϪsquamous carcinomas (0 of 24 tumors; Ref. 34), primary Ewing DYRK2 ϩϩϩϩϩ SHGC-57328 ϪϪϩϩϩtumors (0 of 37 tumors; Ref. 35), primary cervical carcinomas (0 of 35 IFNG ϪϪϩϪϪtumors; Ref. 36), breast cancers (1 of 60 tumors; Ref. 37), and MDM1 ϪϪϪϩϪ MDM2 ϪϩϩϩϪhepatoblastomas (0 of 19 tumors; Ref. 38). The present study is the a Ն first to delineate the genes selected by the 12q14 amplification with The ratio of tumorsample/GAPDH:normalsample/GAPDH 2.0 is defined as gene overexpression. “ϩ”, overexpressed; “Ϫ”, not overexpressed. “nt”, not tested. close proximity to MDM2 in esophageal and lung adenocarcinomas. The DYRK family includes dual-specificity tyrosine-(Y)-phosphorylation regulated kinases, which are distinguished from other protein located within the 12q14 amplicon indicated that overexpression of kinase families by their ability to phosphorylate both Ser/Thr and Tyr DYRK2 in tumors acts independently from the rest of the genes substrates (dual-specificity), and includes at least seven mammalian closely linked (Fig. 6E), suggesting that DYRK2 is less likely a members, DYRK1A, DYRK1B, DYRK1C, DYRK2, DYRK3, DYRK4A, passenger of a closely linked gene, which is the driving force of the and DYRK4B (41). In addition, DYRK1A, DYRK2, DYRK3, MNB, and 12q14 amplicon. The frequency of DYRK2 mRNA overexpression YakA can autophosphorylate their tyrosine residues in vitro (39, 41). (Ͼ2.5-fold of the normal mean expression) in this series of 86 lung YakA may be the homologue of DYRK2 or DYRK3, and all three are adenocarcinomas is 18.6% (16 of 86) and is higher than the gene predominantly cytoplasmic, whereas DYRK1A shows nuclear local- amplification frequency (5.4%) in the series of the 37 lung adenocar- ization (40, 41). DYRK1A is the homologue of Drosophila MNB, cinomas. Amplification analysis for these 86 lung tumors was not which has been mapped to human Down syndrome critical region assessed because of finite amounts of tissue material. DYRK2 mRNA overexpression was also found in one esophageal tumor, B81, which 21q22.2 (42, 43). Presently, the cellular function of all of the DYRK did not contain DYRK2 gene amplification (Figs. 3 and 5). These data family members remains unclear. However, DYRK1A has been im- indicate that additional mechanism(s) other than gene amplification plicated in normal embryo neurogenesis (44) and as a potential can also elevate DYRK2 expression in esophageal and lung adeno- candidate gene responsible for the Down syndrome (45). The DYRK carcinomas. family may also be involved in early heart development (46) and might play a role in cell cycle regulation (47). DYRK1A has been reported to phosphorylate the transcription factor FKHR at Ser329 in DISCUSSION vitro (48). Both DYRK1A and DYRK2 have been shown to phos- In the present study, DYRK2 demonstrated the highest level and phorylate STAT3 in vitro (49). DYRK1A might be directly involved frequency of amplification (Fig. 4A) and RNA overexpression in regulating the transcriptional activity of Gli1, a transcription factor (Table 3) among 19 STS/EST/gene markers examined in the12q13-14 that is a target of the Sonic hedgehog pathway involved in basal cell region. MDM2, on the other hand, was not included in the core- carcinoma development (50–52). Interestingly, mirk, a recently amplified domain in this series of 75 gastroesophageal and 37 lung cloned minibrain related kinase of which the cellular localization is adenocarcinomas (Fig. 3). Although the well-documented oncogene similar to DYRK2 and DYRK3, is overexpressed in primary colon

Fig. 6. Immunohistochemical analysis of MDM2 protein expression in esophageal and lung tumors (A–D), and gene expression of DYRK2 in lung tumors using Affymetrix microarray analysis (E). A, the sarcoma cell line 0SAS overexpresses MDM2 and was used as a positive control for nuclear staining (arrow). B, lung tumor H38, which contains the 12q14 amplicon, shows strong nuclear staining of MDM2 in most tumor cells (arrows). C, normal lung section from patient H38 displays no MDM2 staining. D, esophageal tumor from patient G12, which contains highly increased DYRK2 copy number but no detectable MDM2 amplification, shows only slight nuclear staining of MDM2. E, gene expression patterns sorted by DYRK2 expression levels for 86 lung adenocarcinomas and 10 normal lung samples using our array database (30) demonstrate substantially elevated expression (red) of DYRK2 in at least 16 individual tumors. The expression patterns of the other genes located within the 12q14 amplicon are also included, and the comparison of the expression patterns of these genes with the patterns of DYRK2 indicates that overexpression of DYRK2 in tumors is independent from other closely linked genes within the 12q14 amplicon. 4141

Downloaded from cancerres.aacrjournals.org on September 27, 2021. © 2003 American Association for Cancer Research. DYRK2 GENE AMPLIFICATION IN ESOPHAGEAL ADENOCARCINOMAS carcinomas and can mediate the survival of colon carcinoma cell lines 20. Forus, A., Florenes, V. A., Maelandsmo, G. M., Meltzer, P. S., Fodstad, O., and under stress induced conditions (53). A recent study has reported that Myklebost, O. Mapping of amplification units in the q13–14 region of chromosome 12 in human sarcomas: some amplica do not include MDM2. Cell Growth Differ., 4: DYRK2 mRNA was down-regulated in neuroblastoma cells by treat- 1065–1070, 1993. ing the cells with an anticancer agent, 8-Cl-cAMP, a site-selective 21. Nilbert, M., Rydholm, A., Mitelman, F., Meltzer, P. S., and Mandahl, N. Character- ization of the 12q13–15 amplicon in soft tissue tumors. Cancer Genet. Cytogenet., 83: cyclic AMP analogue exhibiting growth inhibition in a broad spec- 32–36, 1995. trum of human cancer lines (54). These observations, together with 22. Reifenberger, G., Ichimura, K., Reifenberger, J., Elkahloun, A. G., Meltzer, P. S., and our findings, indicate a potential role of DYRK2 in lung and esopha- Collins, V. P. Refined mapping of 12q13–q15 amplicons in human malignant gliomas suggests CDK4/SAS and MDM2 as independent amplification targets. Cancer Res., geal tumor development and/or progression. 56: 5141–5145, 1996. 23. Hatada, I., Hayashizaki, Y., Hirotsune, S., Komatsubara, H., and Mukai, T. A genomic scanning method for higher organisms using restriction sites as landmarks. ACKNOWLEDGMENTS Proc. Natl. Acad. Sci. USA, 88: 9523–9527, 1991. 24. Hirotsune, S., Hatada, I., Komatsubara, H., Nagai, H., Kuma, K., Kobayakawa, K., We thank Dr. Jiayuh Lin for the generous gift of the MDM2 overexpressing Kawara, T., Nakagawara, A., Fujii, K., Mukai, T., et al. New approach for detection of amplification in cancer DNA using restriction landmark genomic scanning. Cancer cell line (0SAS). We also thank Rork Kuick and Jean-Marie Rouillard for their Res., 52: 3642–3647, 1992. assistance using the VGS tool and Dr. Guoan Chen for his help with the array 25. Rouillard, J. M., Erson, A. E., Kuick, R., Asakawa, J., Wimmer, K., Muleris, M., study. Petty, E. M., and Hanash, S. Virtual genome scan: a tool for restriction landmark- based scanning of the human genome. Genome Res., 11: 1453–1459, 2001. 26. Blin, N., and Stafford, D. W. A general method for isolation of high molecular weight REFERENCES DNA from eukaryotes. Nucleic Acids Res., 3: 2303–2308, 1976. 27. Kuick, R., Asakawa, J., Neel, J. V., Satoh, C., and Hanash, S. M. High yield of 1. Devesa, S. S., Blot, W. J., and Fraumeni, J. F. Changing patterns in the incidence of restriction fragment length polymorphisms in two-dimensional separations of human esophageal and gastric carcinoma in the United States. Cancer (Phila.), 83: 2049– genomic DNA. Genomics, 25: 345–353, 1995. 2053, 1998. 28. Kallioniemi, A., Kallioniemi, O-P., Sudar, D., Rutovitz, D., Gray, J. W., Waldman, 2. Bollschweiler, E., Wolfgarten, E., Gutschow, C., and Ho¨lscher, A. H. Demographic F., and Pinkel, D. Comparative genomic hybridization for molecular cytogenetic variations in the rising incidence of esophageal adenocarcinoma in white males. analysis of solid tumors. Science (Wash. DC), 258: 818–821, 1992. Cancer (Phila.), 92: 549–555, 2001. 29. Giordano, T. J., Shedden, K. A., Schwartz, D. R., Kuick, R., Taylor, J. M., Lee, N., 3. Farrow, D. C., and Vaughan, T. L. Determinants of survival following the diagnosis Misek, D. E., Greenson, J. K., Kardia, S. L., Beer, D. G., Rennert, G., Cho, K. R., of esophageal adenocarcinoma (United States). Cancer Causes Control, 7: 322–327, Gruber, S. B., Fearon, E. R., and Hanash, S. Organ-specific molecular classification 1996. of primary lung, colon, and ovarian adenocarcinomas using gene expression profiles. 4. Sarr, M. G., Hamilton, S. R., Marrone, G. C., and Cameron, J. L. Barrett’s esophagus: Am. J. Pathol., 159: 1231–1238, 2001. its prevalence and association with adenocarcinoma in patients with symptoms of 30. Beer, D. G., Kardia, S. L., Huang, C. C., Giordano, T. J, Levin, A. M., Misek, D. E., gastroesophageal reflux. Am. J. Surg., 149: 187–193, 1985. Lin, L., Chen, G., Gharib, T. G., Thomas, D. G., Lizyness, M. L., Kuick, R., 5. Jankowski, J. A., Wright, N. A., Meltzer, S. J., Triafilopoulos, G., Geboes, K., Hayasaka, S., Taylor, J. M., Iannettoni, M. D., Orringer, M. B., and Hanash, S. Casson, A. G., Kerr, D., and Young, L. S. Molecular evolution of the metaplasia- Gene-expression profiles predict survival of patients with lung adenocarcinoma. Nat. dysplasia-adenocarcinoma sequence in the esophagus. Am. J. Pathol., 152: 965–973, Med., 8: 816–824, 2002. 1999. 31. Thoraval, D., Asakawa, J., Wimmer, K., Kuick, R., Lamb, B., Richardson, B., 6. Lagergren, J., Bregstrom, R., Lindgren, A., and Nyren, O. N. Symptomatic gastroe- Ambros, P., Glover, T., and Hanash, S. Demethylation of repetitive DNA sequences sophageal reflux as a risk factor for esophageal adenocarcinoma. N. Engl. J. Med., in neuroblastoma. Genes Chromosomes Cancer, 17: 234–244, 1996. 340: 825–831, 1999. 32. Costello, J. F., Fruhwald, M. C., Smiraglia, D. J., Rush, L. J., Robertson, G. P., Gao, 7. Altorki, N. K., Oliveria, S., and Schrump, D. S. Epidemiology and molecular biology X., Wright, F. A., Feramisco, J. D., Peltomaki, P., Lang, J. C., Schuller, D. E., Yu, L., of Barrett’s adenocarcinoma. Sem. Surg. Onc., 13: 270–280, 1997. Bloomfield, C. D., Caligiuri, M. A., Yates, A., Nishikawa, R., Su Huang, H., Petrelli, 8. Hughes, S. J., Glover, T. W., Zhu, X. X., Kuick, R., Thoraval, D., Orringer, M. B., N. J., Zhang, X., O’Dorisio, M. S., Held, W. A., Cavenee, W. K., and Plass, C. Beer, D. G., and Hanash, S. A novel amplicon at 8p22–23 results in overexpression Aberrant CpG-island methylation has non-random and tumour-type-specific patterns. of cathepsin B in esophageal adenocarcinoma. Proc. Natl. Acad. Sci. USA, 95: Nat. Genet., 24: 132–138, 2000. 12410–12415, 1998. 33. Reifenberger, G., Liu, L., Ichimura, K., Schmidt, E. E., and Collins, V. P. Amplifi- 9. Lin, L., Aggarwal, S., Glover, T. W., Orringer, M. B., Hanash, S., and Beer, D. G. cation and overexpression of the MDM2 gene in a subset of human malignant gliomas A minimal critical region of the 8p22–23 amplicon in esophageal adenocarcinomas without p53 mutations. Cancer Res., 53: 2736–2739, 1993. defined using sequence tagged site-amplification mapping and quantitative polymer- 34. Esteve, A., Lehman, T., Jiang, W., Weinstein, I. B., Harris, C. C., Ruol, A., Peracchia, ase chain reaction includes the GATA-4 gene. Cancer Res., 60: 1341–1347, 2000. A., Montesano, R., and Hollstein, M. Correlation of p53 mutations with epidermal 10. Lin, L., Prescott, M. S., Zhu, Z., Singh, P., Chun, S. Y., Kuick, R. D., Hanash, S. M., growth factor receptor overexpression and absence of mdm2 amplification in human Orringer, M. B., Glover, T. W., and Beer, D. G. Identification and characterization of esophageal carcinomas. Mol. Carcinog., 8: 306–311, 1993. a 19q12 amplicon in esophageal adenocarcinomas reveals cyclin E as the best 35. Kovar, H., Auinger, A., Jug, G., Aryee, D., Zoubek, A., Salzer-Kuntschik, M., and candidate gene for this amplicon. Cancer Res., 60: 7021–7027, 2000. Gadner, H. Narrow spectrum of infrequent p53 mutations and absence of MDM2 11. Younes, M., Lebovitz, R. M., Lechago, L. V., and Lechago, J. p53 protein accumu- amplification in Ewing tumours. Oncogene, 8: 2683–2690, 1993. lation in Barrett’s metaplasia, dysplasia, and carcinoma: a follow-up study. Gastro- 36. Kessis, T. D., Slebos, R. J., Han, S. M., Shah, K., Bosch, X. F., Munoz, N., Hedrick, enterology, 105: 1637–1642, 1993. L., and Cho, K. R. p53 gene mutations and MDM2 amplification are uncommon in 12. Hamelin, R., Flejou, J. F., Muzeau, F., Potet, F., Laurent-Puig, P., Fekete, F., and primary carcinomas of the uterine cervix. Am. J. Pathol., 143: 1398–1405, 1993. Thomas, G. TP53 gene mutations and p53 protein immunoreactivity in malignant and 37. Quesnel, B., Preudhomme, C., Fournier, J., Fenaux, P., and Peyrat, J. P. MDM2 gene premalignant Barrett’s esophagus. Gastroenterology, 107: 1012–1018, 1994. amplification in human breast cancer. Eur. J. Cancer, 30A: 982–984, 1994. 13. Oliner, J. D., Kinzler, K. W., Meltzer, P. S., George, D. L., and Vogelstein, B. 38. Ohnishi, H., Kawamura, M., Hanada, R., Kaneko, Y., Tsunoda, Y., Hongo, T., Amplification of a gene encoding a p53-associated protein in human sarcomas. Bessho, F., Yokomori, K., and Hayashi, Y. Infrequent mutations of the TP53 gene and Nature (Lond.), 358: 80–83, 1992. no amplification of the MDM2 gene in hepatoblastomas. Genes Chromosomes 14. Leach, F. S., Tokino, T., Meltzer, P., Burrell, M., Oliner, J. D., Smith, S., Hill, D. E., Cancer, 15: 187–190, 1996. Sidransky, D., Kinzler, K. W., and Vogelstein, B. p53 Mutation and MDM2 ampli- 39. Tejedor, F., Zhu, X. R., Kaltenbach, E., Ackermann, A., Baumann, A., Canal, I., fication in human soft tissue sarcomas. Cancer Res., 53: 2231–2234, 1993. Heisenberg, M., Fischbach, K. F., and Pongs, O. minibrain: a new protein kinase 15. Soslow, R. A., Altorki, N. K., Yang, G., Xie, D., and Yang, C. S. mdm-2 expression family involved in postembryonic neurogenesis in Drosophila. Neuron, 14: 287–301, correlates with wild-type p53 status in esophageal adenocarcinoma. Mod. Pathol., 12: 1995. 580–586, 1999. 40. van Es, S., Weening, K. E., and Devreotes, P. N. The protein kinase YakA regulates 16. Taniere, P., Martel-Planche, G., Maurici, D., Lombard-Bohas, C., Scoazec, J., g-protein-linked signaling responses during growth and development of Dictyoste- Montesano, R., Berger, F., and Hainaut, P. Molecular and clinical differences between lium. J. Biol. Chem., 276: 30761–30765, 2001. adenocarcinomas of the esophagus and of the gastric cardia. Am. J. Pathol., 158: 41. Becker, W., Weber, Y., Wetzel, K., Eirmbter, K., Tejedor, F. J., and Joost, H. G. 33–40, 2001. Sequence characteristics, subcellular localization, and substrate specificity of DYRK- 17. Moskaluk, C. A., Hu, J., and Perlman, E. J. Comparative genomic hybridization of related kinases, a novel family of dual specificity protein kinases. J. Biol. Chem., 273: esophageal and gastroesophageal adenocarcinomas shows consensus areas of DNA 25893–25902, 1998. gain and loss. Genes Chromosomes Cancer, 22: 305–311, 1998. 42. Song, W. J., Sternberg, L. R., Kasten-Sportes, C., Keuren, M. L., Chung, S. H., Slack, 18. van Dekken, H., Geelen, E., Dinjens, W. N. M., Wijnhoven, B. P. L., Tilanus, H. W., A. C., Miller, D. E., Glover, T. W., Chiang, P. W., Lou, L., and Kurnit, D. M. Tanke, H. J., and Rosenberg, C. Comparative genomic hybridization of cancer of the Isolation of human and murine homologues of the Drosophila minibrain gene: human gastroesophageal junction: deletion of 14Q31–32.1 discriminates between esophageal homologue maps to 21q22.2 in the Down syndrome “critical region.” Genomics, 38: (Barrett’s) and gastric cardia adenocarcinomas. Cancer Res., 59: 748–752, 1999. 331–339, 1996. 19. Mitchell, E. L., White, G. R., Santibanez-Koref, M. F., Varley, J. M., and Heighway, 43. Chen, H., and Antonarakis, S. E. Localisation of a human homologue of the Dro- J. Mapping of gene loci in the Q13–Q15 region of chromosome 12. Chromosome sophila mnb and rat Dyrk genes to chromosome 21q22.2. Hum. Genet., 99: 262–265, Res., 3: 261–262, 1995. 1997. 4142

44. Miyata, Y., and Nishida, E. Distantly related cousins of MAP kinase: biochemical 49. Matsuo, R., Ochiai, W., Nakashima, K., and Taga, T. A new expression cloning properties and possible physiological functions. Biochem. Biophys. Res. Commun., strategy for isolation of substrate-specific kinases by using phosphorylation site- 266: 291–295, 1999. specific antibody. J. Immunol. Methods., 247: 141–151, 2001. 45. Smith, D. J., Stevens, M. E., Sudanagunta, S. P., Bronson, R. T., Makhinson, M., 50. Mao, J., Maye, P., Kogerman, P., Tejedor, F. J., Toftgard, R., Xie, W., Wu, G., and Watabe, A. M., O’Dell, T. J., Fung, J., Weier, H. U., Cheng, J. F., and Rubin, E. M. Wu, D. Regulation of Gli1 transcriptional activity in the nucleus by Dyrk1. J. Biol. Functional screening of 2 Mb of human chromosome 21q22.2 in transgenic mice Chem., 277: 35156–35161, 2002. implicates minibrain in learning defects associated with Down syndrome. Nat. Genet., 51. Lee, J., Platt, K. A., Censullo, P., and Ruiz i Altaba, A. Gli1 is a target of Sonic hedgehog 16: 28–36, 1997. that induces ventral neural tube development. Development, 124: 2537–2552, 1997. 46. Okui, M., Ide, T., Morita, K., Funakoshi, E., Ito, F., Ogita, K., Yoneda, Y., Kudoh, 52. Dahmane, N., Lee, J., Robins, P., Heller, P., and Ruiz i Altaba. A. Activation of the J., and Shimizu, N. High-level expression of the Mnb/Dyrk1A gene in brain and heart transcription factor Gli1 and the Sonic hedgehog signalling pathway in skin tumours. during rat early development. Genomics, 62: 165–171, 1999. Nature (Lond.), 389: 876–881, 1997. 47. Rahmani, Z., Lopes, C., Rachidi, M., and Delabar, J. M. Expression of the mnb (dyrk) 53. Lee, K., Deng, X., and Friedman, E. Mirk protein kinase is a mitogen-activated protein in adult and embryonic mouse tissues. Biochem. Biophys. Res. Commun., protein kinase substrate that mediates survival of colon cancer cells. Cancer Res., 60: 253: 514–518, 1998. 3631–3637, 2000. 48. Woods, Y. L., Rena, G., Morrice, N., Barthel, A., Becker, W., Guo, S., Unterman, 54. Park, G. H., Choe, J., Choo, H. J., Park, Y. G., Sohn, J., and Kim, M. K. Genome- T. G., and Cohen, P. The kinase DYRK1A phosphorylates the transcription factor wide expression profiling of 8-chloroadenosine- and 8-chloro-cAMP-treated human FKHR at Ser329 in vitro, a novel in vivo phosphorylation site. Biochem. J., 355: neuroblastoma cells using radioactive human cDNA microarray. Exp. Mol. Med., 34: 597–607, 2001. 184–193, 2002.

4143

Downloaded from cancerres.aacrjournals.org on September 27, 2021. © 2003 American Association for Cancer Research. Amplification and Overexpression of the Dual-Specificity Tyrosine-(Y)-Phosphorylation Regulated Kinase 2 ( DYRK2) Gene in Esophageal and Lung Adenocarcinomas

Charles T. Miller, Sanjeev Aggarwal, Theodore K. Lin, et al.

Cancer Res 2003;63:4136-4143.

Updated version Access the most recent version of this article at: http://cancerres.aacrjournals.org/content/63/14/4136

Cited articles This article cites 54 articles, 17 of which you can access for free at: http://cancerres.aacrjournals.org/content/63/14/4136.full#ref-list-1

Citing articles This article has been cited by 8 HighWire-hosted articles. Access the articles at: http://cancerres.aacrjournals.org/content/63/14/4136.full#related-urls

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

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

Permissions To request permission to re-use all or part of this article, use this link http://cancerres.aacrjournals.org/content/63/14/4136. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.