Impaired Expression of the Cell Cycle Regulator BTG2 Is Common in Clear Cell Renal Cell Carcinoma

[CANCER RESEARCH 64, 1632–1638, March 1, 2004] Impaired Expression of the Cell Cycle Regulator BTG2 Is Common in Clear Cell Renal Cell Carcinoma

Kirsten Struckmann, Peter Schraml, Ronald Simon, Katja Elmenhorst, Martina Mirlacher, Juha Kononen, and Holger Moch Institute for Pathology, University of Basel, Basel, Switzerland

ABSTRACT carcinoma-1 locus at 3p12, which all have been shown to play a role in the biology of cRCC (12–15). In contrast to loss of 3p, which is The prognosis of patients with renal cell carcinoma (RCC) is poor. A associated with initiation of cRCC (10, 11), loss of 9p and 14q have full understanding of the molecular genetics and signaling pathways been shown to be linked to progression of cRCC (9, 16, 17). Addi- involved in renal cancer development and in the metastatic process is of central importance for developing innovative and novel treatment options. tional cytogenetic alterations in cRCC are losses of 4q, 6q, 13q, and In this study, BD Atlas Human Cancer 1.2 cDNA microarrays were used Xq and gains of 5q, 17p, and 17q (9, 11), suggesting many still to identify genes involved in renal tumorigenesis. By analyzing gene unknown genes involved in the initiation and progression of cRCC. expression patterns of four clear cell RCC (cRCC) cell lines and normal The development of microarray technology platforms allows rapid renal tissue, 25 genes were found differentially expressed. To determine screening and evaluation of molecular markers and signaling path- the relevance of these genes, RNA in situ hybridization was performed on ways important in human cancer. Using cDNA microarrays, the a tissue microarray generated from 61 snap-frozen primary renal cell expression levels of thousands of genes could be assessed in a limited carcinomas and 12 normal renal cortex biopsies. B-cell translocation gene number of samples enabling molecular classification of cancer and the 2 (BTG2), a negative cell cycle regulator, which was expressed in normal identification of molecular signatures that might facilitate prediction renal tissue but down-regulated in cRCC cell lines and primary cRCCs, of disease outcome and response to treatment (18–20). Tissue mi- was selected for additional experiments. Quantitative BTG2 mRNA expression analysis in 42 primary cRCCs and 18 normal renal cortex croarrays (TMAs) have been designed to analyze simultaneously new biopsies revealed up to 44-fold reduced expression in the tumor tissues. cancer-related genes in hundreds of tumors on the DNA, RNA, and Decrease of BTG2 expression was not associated with tumor stage, grade, protein level (21, 22). Consequently, a combination of cDNA mi- and survival. Cell culture experiments demonstrated that BTG2 expres- croarray and TMA technology is particularly suitable for rapid iden- sion was weakly inducible by the phorbolester 12-O-tetradecanoylphor- tification and subsequent validation of potential novel cancer markers bol-13-acetate in one of four cRCC cell lines. In contrast, increasing cell and prognostic parameters. density led to elevated BTG2 mRNA expression in three of four cRCC cell In this study, we used a combination of cDNA microarray analysis lines. In both experiments, BTG2 mRNA levels did not reach values and RNA in situ hybridization (RISH) on TMAs made from snap- observed in normal renal tissue. These data suggest that down-regulation frozen tissue specimens to identify tumor-relevant genes for cRCC. of BTG2 is an important step in renal cancer development. The B-cell translocation gene 2 (BTG2), coding a negative cell cycle regulator, was further analyzed in fresh frozen primary cRCCs and INTRODUCTION normal renal cortex biopsies by quantitative reverse transcription- Renal cell cancer (RCC) accounts for ϳ2% of all human cancers PCR (RT-PCR). Additionally, the regulation of the BTG2 gene was worldwide with an incidence of 189.000 and a mortality of 91,000 in studied in cRCC cell lines. the year 2000 (1). RCC is characterized by absence of early warning signs leading to a rather high percentage of advanced, already metastatic tumors at first presentation. Additionally, 40% of nonmetastatic MATERIALS AND METHODS tumors will become metastatic during the course of disease (2). To Cell Cultures. Human cRCC cell lines Caki-1, Caki-2, 786-O, and 769-P date, there is no known cure for metastatic RCC because those tumors and the human cervix carcinoma cell line HeLa were obtained from American are unresponsive to conventional systemic therapies (3, 4). The 5-year Type Culture Collection. All cell lines were grown in Optimem (Invitrogen, survival rate of patients suffering from metastatic RCC is Ͻ10% (5). San Diego, CA), which was supplemented with 10% FCS (Amimed, Basel, Histopathological tumor stage and grade are well-established prog- Switzerland) and 1% penicillin/streptomycin (Amimed). nostic markers for RCC (6). However, the clinical behavior of RCC is Primary Tumors and Normal Renal Tissues. All primary renal tumors variable and often unpredictable. Identification of new molecular and normal renal tissues used for the TMA construction and the quantitative markers would facilitate outcome predictions and improve therapeutic RT-PCR experiments were taken from our frozen tissue archives. Normal options. For this reason, many efforts have been made to understand tissue samples stemmed exclusively from the renal cortex parenchyma, which predominantly consists of proximal tubules (23). Tumor stage and histological the genetic background of initiation and progression of RCC. Com- subtype were defined according to the recommendations of the Union Inter- plete or partial loss of 3p is linked to clear cell RCC (cRCC)—the national Contre Cancer (24) and the recent RCC classification (7). Histological ϳ most common subtype of RCC accounting for 75% of all RCCs grading was done according to a three-tiered grading system (25). H&E- (7)—and is the most frequent alteration in this renal tumor subtype stained sections were prepared from all tissue samples and were reviewed by (8–11). The short arm of chromosome 3 harbors several potential one pathologist (H. M) to ensure the integrity of the tissue. Representative tumor suppressor genes, including the von Hippel-Lindau gene at tissue areas were marked and used for TMA construction or RNA extraction, 3p25, the RASSF1A gene at 3p21.3, and the nonpapillary renal respectively. RNA Extraction. Total RNA from cRCC cell lines, primary cRCCs, and normal renal cortex biopsies was extracted with TRIzol reagent (Invitrogen) Received 6/10/03; revised 12/22/03; accepted 12/31/03. Grant support: Swiss National Science Foundation Grant 31-63923.00. according to the instructions of the manufacturer. DNase treatment of total The costs of publication of this article were defrayed in part by the payment of page RNA was done using DNase I system in combination with the RNeasy kit charges. This article must therefore be hereby marked advertisement in accordance with (Qiagen, Hilden, Germany). RNA concentrations were determined with a 18 U.S.C. Section 1734 solely to indicate this fact. spectrophotometer. Requests for reprints: Dr. Holger Moch, Institute for Pathology, Schoenbeinstrasse 40, CH-4031 Basel, Switzerland. Phone: 41-61-265-2980; Fax: 41-61-265-3194; E-mail: cDNA Microarray Analysis. Gene expression patterns of cRCC cell lines [email protected]. Caki-1, Caki-2, 786-O, and 769-P and normal renal tissue (Invitrogen) were 1632

Fig. 1. The frozen renal tissue microarray. A, overview of the renal tissue microarray (TMA). B, composition of the renal TMA; cRCC, clear cell renal cell carcinoma; pRCC, papillary RCC; chRCC, chromophobe RCC. C, H&E-stained frozen section of the renal TMA. analyzed using BD Atlas Human Cancer 1.2 cDNA microarrays (BD Bio- Air-dried TMA sections were hybridized with 2–4 ng of radiolabeled probe sciences Clontech, Palo Alto, CA). For each experiment, 5 ␮g of total RNA in hybridization mix (50% formamide, 10% dextransulfate, 2ϫ SSC) in a were used for single-stranded cDNA synthesis using the Atlas Pure Total RNA moist chamber at 42°C overnight. After hybridization, slides were washed in ␣ 32 ϫ ϫ Labeling System (BD Biosciences Clontech) and ( - P) dATP (Amersham 1 SSC at 55°C(4 15 min), incubated in distilled H2O, 60 and 90% ethanol Biosciences, Buckinghamshire, United Kingdom) as a label. Unincorporated (30 s each), and air-dried. Slides were exposed to high-resolution screens for nucleotides were removed using the QIAquick Nucleotide Removal kit (Qia- 48 h (Packard Bioscience Company) prior scanning. gen). Prehybridization, hybridization, and washing of the cDNA microarrays ArrayVision software package (Imaging Research Inc., St. Catharines, were done according to standard protocols. Arrays were exposed to a high- Ontario, Canada) was used for image analysis. In brief, for each experiment resolution screen (Packard Bioscience Company, Toronto, Ontario, Canada) signal density (sDens) values were obtained for all tissue spots on the TMA as for 24 h and scanned (Cyclone; Packard Bioscience Company). AtlasImage well as for the background (measured between tissue spots of a given slide and 1.01a software (BD Biosciences Clontech) was used for digital image analysis. averaged). The mean background sDens values were nearly identical for all Background-corrected signal density (sDens) values were calculated for each slides. To safely distinguish between true positivity and background signals, array spot and were normalized according to the AtlasImage sum method. only those tissue spots were considered positive for gene expression that Genes with aberrant expression were identified by calculating ratios between showed sDens values Ն 2-fold mean background value. signal densities of each spot on the cell line (test) arrays and the corresponding After image analysis, slides were incubated in Hypercoat LM-1 emulsion array spot on the normal renal tissue (reference) array. Only genes showing (Amersham Biosciences) for 2–4 weeks to evaluate the hybridization speci- sDens ratios Ն ϩ4orՅ Ϫ4 in at least two renal cancer cell lines were ficity by direct autoradiography. After development of the emulsion, slides considered as significantly differentially expressed. were counterstained with hematoxylin to facilitate microscopical reevaluation TMA. A TMA was constructed from frozen tissue samples of 51 cRCCs, of the TMAs. Spots that lost Ͼ40% cells during the experimental process were 4 papillary RCCs, 4 chromophobe RCCs, 2 oncocytomas, and 12 tissues from considered to be not analyzable and were therefore excluded from additional normal renal cortex. There were 32 pT1, 4 pT2, and 15 pT3 cRCCs. Twelve statistical analysis. cRCCs were grade 1, 32 grade 2, and 7 grade 3. The TMA was constructed in Northern Analysis. PolyAϩ RNA was extracted from total RNA of frozen Tissue-Tek OCT compound (Miles Laboratories, Naperville, IL) as Caki-1, Caki-2, 786-O, and 769-P and normal renal tissue (Invitrogen) using described previously (22). We optimized a commercialized microarray device the polyAϩ RNA extraction kit from Qiagen. One ␮g polyAϩ RNA of each (Beecher Instruments, Sun Prairie, WI) by using a 0.6-mm drill for recipient sample was separated by gelelectrophoresis and blotted to a Hybond- whole making instead of the conventional hollow needle. Fig. 1A–C shows the Nϩ membrane (Amersham Biosciences) following standard protocols (26). renal TMA generated for this study. Single-stranded DNA probes were generated from 25 ng of BTG2 and RISH. For RISH experiments, two to four oligonucleotides/gene were G3PDH PCR products using 25 ng sequence-specific reverse primers for designed using the Vector NTI software package (InforMax, Inc., Frederick, BTG2 or G3PDH, respectively, (␣-32P) dATP (Amersham Bioscience), 0.5 mM 1 MD). Specificity of the probes was confirmed using the BLAST program. dCTP, dGTP, and dTTP, and Klenow enzyme (Promega). QIAquick Nucleo- ␣ 33 Each of the oligonucleotides were labeled separately with ( - P) dATP tide Removal kit was applied to remove unincorporated nucleotides (Qiagen). (Amersham Biosciences) using terminal deoxynucleotidyltransferase (Pro- Prehybridization, hybridization, and washing of the blot were done according mega, Madison, WI) following the instructions of the manufacturer. Oligonu- to conventional protocols (26). The Northern was exposed to a high-resolution cleotide probes were pooled and unincorporated nucleotides were removed screen (Packard Bioscience Company) before scanning. using the QIAquick Nucleotide Removal kit (Qiagen). Quantitative RT-PCR. PCR standard curves for BTG2 and the reference G3PDH were generated to allow quantification of the copy numbers of both 1 Internet address: http://www.ncbi.nlm.nih.gov. genes. Serial dilutions of BTG2 and G3PDH in vitro generated transcripts 1633

(1012–101 copies) were mixed with rRNA from mice (Roche, Basel, Switzer- determine BTG2 and G3PDH mRNA copy numbers by quantitative RT-PCR land) to a final concentration of 0.5 ␮g/␮l. as described above. Two ␮l of each dilution were reverse transcribed using 50 ng of random Statistics. In RISH experiments, analyzable (Ͼ60% representative cells) hexamers (Invitrogen), 1 mM of each deoxynucleotide triphosphate, 5 mM tissue spots showing sDens values Ն 2-fold mean background sDens value

MgCl2, 1.5 mM Tris-HCl (pH 8.0), 500 mM KCl, 100 units of Moloney murine were considered positive for gene expression, whereas the remaining evaluable leukemia virus reverse transcriptase (Invitrogen), and 20 units of RNAsin tissue spots were considered negative for gene expression. Results obtained by (Roche) in a total volume of 19.5 ␮l for 10 min at 25°C and 60 min at 37°C. RISH were than evaluated in two ways. In a first approach, contingency table After heat inactivation of the enzymes, the remaining RNA was treated with (␹2) analysis was used to search for differences in the expression frequency of 0.5 ␮l (1 unit) RNase H (Invitrogen) at 37°C for 20 min. a given gene. For this purpose, the percentage of spots positive for gene Five ␮l of a 1:2.5 dilution of each first strand product was amplified using expression was compared between normal renal tissue and cRCC and also 2 ␮l of LightCycler-FastStart DNA Master SYBR Green I (Roche), BTG2 within the subset of cRCCs. In a second approach, average sDens values were (forward: 5Ј-ctcacctgcaagaaccaagtg-3Ј; reverse: 5Ј-agttccccaggttgaggtatgt-3Ј) calculated from gene expression positive spots to obtain the mean expression and G3PDH (forward: 5Ј-gaaatcccatcaccatcttcc-3Ј; reverse: 5Ј-cagagatgatgac- level of a given gene. Those expression levels were than compared between Ј ␮ ccttttgg-3 ) primer pairs at a final concentration of 1 M each and MgCl2 at a normal renal tissue and cRCC and also within the subset of cRCCs using final concentration of 2 mM for BTG2 and3mM for G3PDH. PCR was ANOVA analysis. performed in a total volume of 20 ␮l for 10 min at 95°C and 40 cycles with Data obtained by quantitative RT-PCR were evaluated by (a) ANOVA 15sat95°C,10sat58°Cand7sat72°C. PCR efficiencies were calculated analysis to search for differences in the BTG2 mRNA expression between from the standard curves (Fig. 3, A and B). Melting curve analysis (LightCycler normal renal tissue and primary cRCC and within the subset of cRCCs and (b) software package) was applied to ensure the specificity of the PCR reaction Kaplan Meyer analysis to search for associations between the BTG2 mRNA (Fig. 3C). expression and patient survival. BTG2 and G3PDH copy numbers were evaluated in 42 primary cRCCs and 18 randomly selected not matching normal renal cortex biopsies. There were 20 pT1, 7 pT2, and 15 pT3 cRCCs. Eight cRCCs were grade 1, 28 grade 2, and RESULTS 6 grade 3. One ␮g of total RNA from each sample was reverse transcribed cDNA Microarray Analysis. Of 1176 genes localized on the prior quantitative PCR on the LightCycler exactly following the protocols described for the preparation of the standard curves. BTG2 copy numbers were cDNA microarray used in this study, the expression patterns of 231 normalized to G3PDH copy numbers to correct for differences in the RNA (19.6%) were analyzable. Thirteen genes showed reduced expression quality and quantity. levels, whereas 12 were stronger expressed in at least two renal cancer Induction of BTG2 Expression by 12-O-Tetradecanoylphorbolester-13- cell lines compared with normal renal tissue. Those 25 differentially Acetate (TPA). Caki-1, Caki-2, 786-O, 769-P, and HeLa (positive control) expressed genes are listed in Table 1. As an example, BTG2 showed were plated at 500,000 cells/25 ml culture vessel. Medium renewal was done reduced expression in all four cRCC cell lines (Fig. 2, A and B). 24 h after plating. RNA was extracted after 48 h from each cell line to RISH. Twenty-five genes, displaying altered mRNA expression determine the BTG2 and G3PDH mRNA expression level prior to treatment. levels on the cDNA microarray, were further examined by RISH on a The remaining cells were treated with various concentrations of TPA (25, 50, TMA generated from frozen renal tumors and normal renal cortex 75, or 100 ng/ml culture medium) or DMSO alone (0.1% final concentration). biopsies. Nineteen genes had detectable mRNA expression levels on RNA was extracted 70 min after addition of TPA/DMSO. BTG2 and G3PDH mRNA copy numbers were determined by quantitative RT-PCR following the the TMA (Table 2). protocol described above. In a first step, expression frequencies of each gene were compared BTG2 mRNA Expression and Cell Density. Caki-1, Caki-2, 786-O, and between normal renal tissues and cRCCs. BTG2, CHES1, GAS6, and 769-P were plated at 100,000, 200,000, and 500,000 cells/well in 6-well plates. LITAF were significantly more frequently expressed in normal renal Medium renewal was done 24 h after plating. RNA was extracted after 48 h to tissue than in cRCC (Table 2, Fig. 2C).

Table 1 Differentially expressed genes identified by cDNA array analysis Abbreviation Chromosomal Accession Expression status Gene name in texta localization numberb Increased in cRCC cell lines Centromeric protein F CENPF 1q32–q41 NM_016343 CDC28 protein kinase regulatory subunit 1B CKS1B 8q21 NM_001826 Chondroitin sulfate proteoglycan 2 CSPG2 5q12–q14 NM_004385 Fibronectin 1 FN1 2q34 NM_002026 FOS-like antigene 1 FOSL1 11q13 NM_005438 High mobility group AT-hook 1 HMGA1 6p21 NM_002131 Integrin ␣ 3 ITGA3 17 NM_002204 Integrin ␤ 8 ITGB8 7p15.3 NM_002214 Tubulin ␣ ubiquitous K-␣-1 12q13.11 NM_006082 Plasminogen activator inhibitor type 1 PAI1 7q21.3–q22 NM_000602 Transforming growth factor ␤ induced TGFB1 19q13.1 NM_000358 Vimentin VIM 10p13 NM_003380 Reduced in cRCC cell lines Basigin BSG 19p13.3 NM_001728 B-cell translocation gene 2 BTG2 1q32 NM_006763 CD74 antigen CD74 5q32 NM_004355 CD9 antigen CD9 12p13 NM_001769 Checkpoint suppressor 1 CHES1 14q24.3–q31 NM_005197 Cathepsin D CTSD 11p15.5 NM_001909 Early growth response 1 EGR1 5q31.1 NM_001964 FC fragment of IgG, receptor transporter, ␣ FCGRT 19q13.3 NM_004107 Growth arrest-specific 6 GAS6 13q34 NM_000820 Integrin ␣ 6 ITGA6 2q31.1 NM_000210 LPS-induced TNF ␣ factor LITAF 16p13.3–p12 NM_004862 Nuclear hormone receptor NR0B2 1p36.1 NM_021969 Tissue inhibitor of matrixmetalloproteinases 3 TIMP3 22q22.1–q13.2 NM_000362 a Abbreviation for the gene name which is used in the text. b National Center for Biotechnology Information gene bank accession number. 1634

Fig. 2. BTG2 expression in renal cancer cell lines, normal renal tissues, and primary renal cell carcinoma (RCC). A, cDNA microarray hybridization of normal renal tissue and the renal cancer cell line 786-O. BTG2 and G3PDH cDNA spots are indicated. B, magnification of BTG2 and G3PDH cDNA array spots after hybridization. C, RNA in situ hybridization of BTG2 and ␤-actin (positive control) on a tissue microarray generated from frozen tissue specimens. Positions of analyzable normal renal tissue spots and BTG2-positive cRCC spots are indicated by rectan- gles and circles, respectively. BTG2 is more frequently expressed in normal renal tissue than in the subset of clear cell RCC (cRCC). In contrast, strong ␤-actin expression is shown by all evaluable tissue spots. D, Northern blot analysis of BTG2 and G3PDH. BTG2 mRNA expression in normal renal tissue is indicated by an arrowhead.

In a second step, we searched for associations between the expres- tissue but not in renal cancer cell lines, confirming the results from our sion frequencies of the analyzable 19 genes and tumor stage and grade cDNA microarray experiment (Fig. 2D). in the subset of cRCCs. There was an association between expression Quantitative RT-PCR. mRNA expression levels of BTG2 in fro- frequency and tumor stage for BTG2 and CD9. Both genes were zen primary cRCCs and normal renal cortex biopsies were assessed by detectable in pT1/pT2 (BTG2: 4 of 24, 17%; CD9: 6 of 20, 30%) but quantitative RT-PCR using the LightCycler system (Fig. 3A–C). not in pT3 tumors (BTG2:0of11;CD9: 0 of 9). There was an Compared with the mean normalized BTG2 mRNA copy number of association between expression frequency and tumor grade for normal renal tissue, cRCC showed an average of a 6.7-fold (1.6–44- TIMP3, which was significantly less frequent expressed in grade 3 (1 fold) reduced expression (P Ͻ 0.0001; Fig. 3D) However, there was of 5; 20%) than in grade 2 (17 of 23; 74%) and grade 1 cRCC (7 of no significant correlation between BTG2 expression and tumor stage, 8; 88%; P ϭ 0.03). grade, or other clinical parameters in primary cRCC. We also searched for differences in the expression levels between Induction of BTG2 mRNA Expression by TPA. Because TPA is normal renal tissue and cRCC and also within the subset of cRCCs for a potent inductor for BTG2 expression in different cell lines, including all 19 analyzable genes. VIM was significantly stronger expressed in HeLa (27, 28), we studied the inducibility of BTG2 expression in cRCC than in normal renal tissue (P ϭ 0.04). CD74 was significantly cRCC cell lines by TPA. All four cRCC cell lines and the cervix higher expressed in pT3 cRCC than in pT1/2 cRCC (P ϭ 0.03). carcinoma cell line HeLa (positive control) were treated with various Northern Blot. Northern analysis was performed to verify the concentrations of TPA. In contrast to the BTG2 copy numbers, which results of the cDNA microarray experiment in an independent ap- were comparable with those obtained for primary cRCCs, G3PDH proach. The 2.7-kb mRNA of BTG2 was expressed in normal renal copy numbers were strongly increased in the cRCC cell lines but

Table 2 Gene expression frequency evaluated by RNA in situ hybridization Clear cell renal cell Expression status Gene Normal renal tissuea % carcinomaa % Pb Increased in cRCC cell lines CENPF 2/5 40 4/33 12 n.s. FN1 2/5 40 16/36 44 n.s. ITGA3 0/5 0 3/35 9 – ITGB8 5/5 100 29/36 81 n.s. K-␣-1 2/5 40 15/33 45 n.s. PAI1 1/5 20 9/35 26 n.s. VIM 5/5 100 36/36 100 n.s. Reduced in cRCC cell lines BSG 4/5 80 15/32 47 n.s. BTG2 4/5 80 4/35 11 0.0003 CD74 5/5 100 33/33 100 n.s. CD9 3/5 60 6/29 21 0.07 CHES1 3/5 60 5/35 14 0.02 CTSD 5/5 100 32/36 89 n.s. EGR 4/4 100 19/35 54 0.08 FCGRT 0/5 0 6/36 17 – GAS6 5/5 100 14/36 39 0.01 ITGA6 0/5 0 5/35 14 – LITAF 5/5 100 10/35 29 0.002 TIMP3 5/5 100 25/36 69 n.s. a Number of tissue spots with gene expression/total number of evaluable tissue spots. b P was calculated using contingency table analysis; n.s., not significant. 1635

Fig. 3. Quantitative determination of BTG2 expression in fresh frozen clear cell RCC (cRCC) and normal renal tissues. A,108–104 copies of reverse- transcribed G3PDH in vitro transcript after amplification on the LightCycler system. B, standard curve for G3PDH PCR calculated by the Light- Cycler software. PCR efficiency ϭ 10-1/-slope ϭ 1.84. Similar results were obtained for the BTG2 standard curve. C, melting curve analysis of G3PDH (a) and BTG2 (b). D, significantly reduced BTG2 expression (mean ratio of BTG2 to G3PDH copy numbers) in cRCC compared with normal renal tissue.

remained stable throughout the experiment (data not shown). There ization (35–38). However, these TMAs are not optimally suited for was a 15-fold increase of the BTG2 expression in HeLa cells. Among RISH studies. Processing of tissue and tissue fixation using formalin the renal cancer cell lines, only Caki-1 showed a slightly elevated causes degradation and chemically modifies RNA, respectively (39, (2.3-fold) increase of BTG2 expression after TPA treatment (Fig. 4). 40). It was shown that TMAs generated from frozen tissue specimens BTG2 Expression and Cell Density. BTG2 mRNA has been (22) are a better tool for studying mRNA expression patterns of tumor shown to be preferentially expressed in quiescent cells, whereas relevant genes. exponentially growing cells showed reduced BTG2 expression levels As the focus of biomedical research is shifting from genomics to (29, 30). This finding tempted us to study the association between cell proteomics, TMAs from frozen tissue are also likely to prove useful density and BTG2 mRNA expression in renal carcinoma cell lines. for mass-scale testing of antibodies that are not working on formalin- Here again, BTG2 mRNA copy numbers were comparable with those fixed, paraffin embedded tissue. detected in primary cRCCs, and G3PDH copy numbers were as The frozen TMA generated in our study, served to demonstrate the obtained in the TPA induction experiment described above. Caki-2, utility and feasibility of RISH on TMAs. This method is convenient 786-O, and 769-P cells showed a 4.4-, 1.8-, and 1.7-fold increase in for further validating RNA expression of most candidate genes iden- BTG2 mRNA expression, respectively, in cultures with highest densities (initial cell number 500,000 cells/well) compared with cultures with lowest cell densities (initial cell number 100,000 cells/well). In contrast, Caki-1 cells showed a 1.7-fold reduced BTG2 expression in cultures with highest densities compared with cultures with lowest densities. Fig. 5 shows the correlation between cell density and BTG2 expression for all analyzed cell lines.

DISCUSSION cDNA microarray-based expression profiling of human tumors produces large amounts of data from a limited number of tissue samples. One of the major bottlenecks in cancer research is evaluating the clinical relevance of candidate genes uncovered by cDNA microarray analysis. Our recently developed TMA technology has been designed to facilitate such studies. The use of TMAs allows analysis of genes of interest in hundreds of clinical sample (21). It was Fig. 4. Inducibility of BTG2 expression by the phorbolester TPA. BTG2 expres- ϭ ϭ demonstrated that TMAs are optimally suited to evaluate the clinical sion ratio of BTG2 to G3PDH mRNA copies. T0 BTG2 expression before TPA addition; TPA ϭ BTG2 expression 70 min after addition of 12-O-tetradecanoylphorbo- significance of genes identified by cDNA array experiments (31–34). lester-13-acetate (TPA); control ϭ BTG2 expression 70 min after addition of DMSO. TMAs, usually made from formalin-fixed, paraffin-embedded tis- Because there was no association between the concentration of TPA and BTG2 expression, mean BTG2 expression values after TPA treatment are given in this figure. Addition sue, are an excellent tool for studying protein expression by immu- of TPA reveals a 15-fold increase of BTG2 expression in HeLa and a 2.3-fold increase in nohistochemistry and gene aberrations by fluorescence in situ hybrid- Caki-1. 1636

very weak (15-fold versus 2.3-fold) suggesting that BTG2 induction is regulated by different mechanism in cervical and renal cancers. The putative function of BTG2 as a cell cycle regulator was demonstrated recently by the observation that BTG2 expression is low in exponentially growing hepatocarcinoma, breast tumor, and normal prostate cell lines, whereas normal quiescent cells showed high BTG2 expression levels (29, 30). Our experiments showed that BTG2 expression was positively associated with increasing cell densities in Caki-2, 786-O, and 769-P but not in Caki-1 cultures. Because Caki-1 derives from a skin metastasis of a cRCC it is tempting to speculate that the ability of BTG2 induction changes during the metastatic process. BTG2 expression levels in the four cRCC cell lines were comparable with those of the 42 primary tumors. Interestingly, neither TPA Fig. 5. BTG2 expression and cell density. BTG2 expression ϭ ratio of BTG2 to G3PDH nor high cell densities were able to raise BTG2 expression to the level mRNA copies; low ϭ initial cell number of 100,000/well; medium ϭ initial cell number observed in normal renal tissues. This finding corroborates our theory ϭ of 200,000/well; high initial cell number of 500,000/well. BTG2 expression is posi- that down-regulation of BTG2 is an important mechanism involved in tively associated with cell density in Caki-2, 786-O, and 769-P. renal cancer development. The mechanisms leading to reduced BTG2 expression are yet tified by cDNA array analysis. A minor fraction of genes was ana- unknown. Further studies will be necessary to elucidate whether lyzable on the cDNA microarray but showed no detectable mRNA aberrations in the up-stream signaling pathways of BTG2 or methyl- expression on the TMA. There are two explanations for the observed ation of the BTG2 promoter are responsible for impaired expression in discrepancies: (i) Because the cDNA microarray experiments were cRCC. performed with RNA extracted from cRCC cell lines, expression of Beside BTG2, other genes identified in our study might play an some genes might be cell line specific. (ii) Because the absolute important role in renal tumor biology. For example, reduced expres- number of mRNA target sequences in a 0.6 mm diameter TMA spot sion was also seen for two other genes, CHES1 and LITAF, in cRCC is below the number of cDNA sequences present in a cDNA microar- cell lines and primary cRCC. Human CHES1, a member of the family ray spot some genes showing weak expression on the cDNA microar- of forkhead/winged transcription factors, is suggested to be involved ray remain undetected by RISH. in DNA damage response because this protein is able to suppress On the basis of the data obtained from the cDNA microarray and multiple yeast checkpoint mutations (45). Reduced expression levels RISH experiments BTG2 was selected for further analysis. Strong of CHES1 might result in genomic instability, which is one of the BTG2 expression in normal cells but reduced or absent expression in hallmarks of cancer. LITAF is involved in the activation of the human cRCC cell lines and primary tumors suggest that BTG2 inactivation is TNF-␣ gene (46), which encodes for a multifunctional cytokine ca- important for cRCC development. Several lines of evidence in the pable of inducing apoptosis, by binding to TNF receptor (47). literature indicate that this gene has tumor suppressive properties: (i) CD74, which was down-regulated in cRCC cell lines, was ex- It was shown that BTG2 is involved in negative cell cycle regulation pressed in primary renal tumors with expression levels similar to and differentiation (29, 41, 42). (ii) BTG2 is also involved in DNA normal renal tissue. High expression levels were significantly associ- damage response and is one of the primary targets of p53 (29, 43). (iii) ated with advanced tumor stage in primary cRCC suggesting a po- In the normal kidney, BTG2 protein is strongly expressed in the tential role of this gene in tumor progression. The possible oncogenic epithelial cells of the proximal tubules from which cRCC arise (25, properties of CD74 in RCC would be consistent with findings from 44). Unfortunately, we were not able to proof the latter finding by other groups because (i) CD74 is localized on chromosome 5q which RISH because of the insufficient resolution of our detection system. is frequently gained in cRCC (9), (ii) strong and frequent CD74 However, since each normal kidney biopsy represented on our TMA protein expression in primary cRCC was recently described (48, 49), exclusively derived from the renal cortex parenchyma we believe that and (iii) increased CD74 protein expression is associated with tumor the strong BTG2 mRNA expression stems from proximal tubules, progression in colon and gastric cancers (50, 51). which is the major component of this cortical substructure (23). In summary, the combination of cDNA microarray and TMA Using quantitative RT-PCR, we confirmed the data obtained from technologies enabled us to identify three novel renal cancer gene the cDNA microarray, RISH, and Northern experiments. Compared candidates, BTG2, CHES1, and LITAF. Inactivation of BTG2, a cell with primary cRCC there was an average 6.7-fold stronger expression cycle regulator, suggests a potential mechanism required for renal of BTG2 mRNA in normal renal tissue. BTG2 mRNA expression level tumorigenesis. Further studies on the identified genes will contribute was independent of tumor stage and grade suggesting that the reduc- to a better understanding of renal cancer biology. tion of BTG2 mRNA expression is not associated with tumor progression and might be an early event in renal tumorigenesis. To address the possible tumor suppressive function of BTG2 in ACKNOWLEDGMENTS renal tumors, we studied the regulation of BTG2 mRNA expression in We gratefully acknowledge the technical assistance of the staff from the renal cancer. Recently, it has been shown that BTG2 is an immediate Institute for Pathology (University of Basel) and would like to thank Christoph early response gene, whose transcription can be induced by the Moroni (Institute of Medical Microbiology, Basel) and Nancy Hynes phorbolester TPA in human HeLa cells and also in a variety of rodent (Friedrich Mischer Institute, Basel) for helpful discussions and support. cell lines of different origin (27, 28). These findings prompted us to investigate the inducibility of BTG2 expression by TPA in the cRCC cell lines Caki-1, Caki-2, 786-O, and 769-P. Among all tested cRCC REFERENCES cell lines, merely Caki-1 cells showed a response to TPA. However, 1. Parkin, D. M., Bray, F., Ferlay, J., and Pisani, P. Estimating the world cancer burden: compared with HeLa, the increase in BTG2 mRNA expression was Globocan 2000, Int. J. Cancer, 94: 153–156, 2001. 1637

2. Linehan, W. M., Shipley, W. U., and Parkinson, D. R. Cancer of the kidney and 27. Fletcher, B. S., Lim, R. W., Varnum, B. C., Kujubu, D. A., Koski, R. A., and ureter. In: V. T. DeVita, S. Hellmann, and S. A. Rosenberg (eds.), Cancer: Principle Herschman, H. R. Structure and expression of TIS21, a primary response gene and Practice of Oncology, pp. 1023–1051. Philadelphia: J. B. Lippincott, 1993. induced by growth factors and tumor promoters, J. Biol. Chem., 266: 14511–8, 1991. 3. Yagoda, A., Abi-Rached, B., and Petrylak, D. Chemotherapy for advanced renal cell 28. Bradbury, A., Possenti, R., Shooter, E. M., and Tirone, F. Molecular cloning of PC3, carcinoma: 1983–1993. Semin. Oncol., 22: 42–60, 1995. a putatively secreted protein whose mRNA is induced by nerve growth factor and 4. Motzer, R. J., and Russo, P. Systemic therapy for renal cell carcinoma, J. Urol., 163: depolarization. Proc. Natl. Acad. Sci. USA, 88: 3353–3357, 1991. 408–417, 2000. 29. Rouault, J. P., Falette, N., Guehenneux, F., Guillot, C., Rimokh, R., Wang, Q., 5. Kosary, C., and McLaughlin, J. Kidney and Renal Pelvis, NIH Publication No. Berthet, C., Moyret-Lalle, C., Savatier, P., Pain, B., Shaw, P., Berger, R., Samarut, J., 93-2789, Vol. XI, pp. 1–22. Bethesda, MD: National Cancer Institute, 1993. Magaud, J. P., Ozturk, M., Samarut, C., and Puisieux, A. Identification of BTG2, an 6. Pantuck, A. J., Zisman, A., and Belldegrun, A. S. The changing natural history of antiproliferative p53-dependent component of the DNA damage cellular response renal cell carcinoma, J. Urol., 166: 1611–1623, 2001. pathway. Nat. Genet., 14: 482–486, 1996. 7. Kovacs, G., Akhtar, M., Beckwith, B. J., Bugert, P., Cooper, C. S., Delahunt, B., Eble, 30. Walden, P. D., Lefkowitz, G. K., Ficazzola, M., Gitlin, J., and Lepor, H. Identification J. N., Fleming, S., Ljungberg, B., Medeiros, L. J., Moch, H., Reuter, V. E., Ritz, E., of genes associated with stromal hyperplasia and glandular atrophy of the prostate by Roos, G., Schmidt, D., Srigley, J. R., Storkel, S., van den Berg, E., and Zbar, B. The mRNA differential display. Exp. Cell Res., 245: 19–26, 1998. Heidelberg classification of renal cell tumours. J. Pathol., 183: 131–133, 1997. 31. Moch, H., Schraml, P., Bubendorf, L., Mirlacher, M., Kononen, J., Gasser, T., 8. Zbar, B., Brauch, H., Talmadge, C., and Linehan, M. Loss of alleles of loci on the short Mihatsch, M. J., Kallioniemi, O. P., and Sauter, G. High-throughput tissue microarray arm of chromosome 3 in renal cell carcinoma. Nature (Lond.), 327: 721–727, 1987. analysis to evaluate genes uncovered by cDNA microarray screening in renal cell 9. Moch, H., Presti, J. C., Jr., Sauter, G., Buchholz, N., Jordan, P., Mihatsch, M. J., and carcinoma. Am. J. Pathol., 154: 981–986, 1999. Waldman, F. M. Genetic aberrations detected by comparative genomic hybridization are 32. Sallinen, S. L., Sallinen, P. K., Haapasalo, H. K., Helin, H. J., Helen, P. T., Schraml, associated with clinical outcome in renal cell carcinoma. Cancer Res., 56: 27–30, 1996. P., Kallioniemi, O. P., and Kononen, J. Identification of differentially expressed genes 10. Presti, J. C., Jr., Moch, H., Gelb, A. B., Huynh, D., and Waldman, F. M. Initiating in human gliomas by DNA microarray and tissue chip techniques. Cancer Res., 60: genetic events in small renal neoplasms detected by comparative genomic hybridiza- 6617–6622, 2000. tion, J. Urol., 160: 1557–1561, 1998. 33. Sanchez-Carbayo, M., Socci, N. D., Charytonowicz, E., Lu, M., Prystowsky, M., 11. Jiang, F., Desper, R., Papadimitriou, C. H., Schaffer, A. A., Kallioniemi, O. P., Childs, G., and Cordon-Cardo, C. Molecular profiling of bladder cancer using cDNA Richter, J., Schraml, P., Sauter, G., Mihatsch, M. J., and Moch, H. Construction of microarrays: defining histogenesis and biological phenotypes. Cancer Res., 62: 6973– evolutionary tree models for renal cell carcinoma from comparative genomic hybrid- 6980, 2002. ization data. Cancer Res., 60: 6503–6509, 2000. 34. Giordano, T. J., Thomas, D. G., Kuick, R., Lizyness, M., Misek, D. E., Smith, A. L., 12. Latif, F., Tory, K., Gnarra, J., Yao, M., Duh, F. M., Orcutt, M. L., Stackhouse, T., Sanders, D., Aljundi, R. T., Gauger, P. G., Thompson, N. W., Taylor, J. M., and Kuzmin, I., Modi, W., Geil, L., et al. Identification of the von Hippel-Lindau disease Hanash, S. M. Distinct transcriptional profiles of adrenocortical tumors uncovered by tumor suppressor gene. Science (Wash. DC), 260: 1317–1320, 1993. DNA microarray analysis. Am. J. Pathol., 162: 521–531, 2003. 13. Herman, J. G., Latif, F., Weng, Y., Lerman, M. I., Zbar, B., Liu, S., Samid, D., Duan, 35. Bubendorf, L., Kononen, J., Koivisto, P., Schraml, P., Moch, H., Gasser, T. C., Willi, D. S., Gnarra, J. R., Linehan, W. M., et al. Silencing of the VHL tumor-suppressor N., Mihatsch, M. J., Sauter, G., and Kallioniemi, O. P. Survey of gene amplifications gene by DNA methylation in renal carcinoma. Proc. Natl. Acad. Sci. USA, 91: during prostate cancer progression by high-throughout fluorescence in situ hybrid- 9700–9704, 1994. ization on tissue microarrays. Cancer Res., 59: 803–806, 1999. 14. Morrissey, C., Martinez, A., Zatyka, M., Agathanggelou, A., Honorio, S., Astuti, D., 36. Simon, R., Nocito, A., Hubscher, T., Bucher, C., Torhorst, J., Schraml, P., Bubendorf, L., Mihatsch, M. M., Moch, H., Wilber, K., Schotzau, A., Kononen, J., and Sauter, G. Morgan, N. V., Moch, H., Richards, F. M., Kishida, T., Yao, M., Schraml, P., Latif, Patterns of her-2/neu amplification and overexpression in primary and metastatic F., and Maher, E. R. Epigenetic inactivation of the RASSF1A 3p21.3 tumor suppres- breast cancer. J. Natl. Cancer Inst. (Bethesda), 93: 1141–1146, 2001. sor gene in both clear cell and papillary renal cell carcinoma. Cancer Res., 61: 37. Halvorsen, O. J., Haukaas, S. A., and Akslen, L. A. Combined Loss of PTEN and p27 7277–7281, 2001. expression is associated with tumor cell proliferation by Ki-67 and increased risk of 15. Lovell, M., Lott, S. T., Wong, P., El-Naggar, A., Tucker, S., and Killary, A. M. The recurrent disease in localized prostate cancer. Clin. Cancer Res., 9: 1474–1479, 2003. genetic locus NRC-1 within chromosome 3p12 mediates tumor suppression in renal 38. Hendriks, Y., Franken, P., Dierssen, J. W., De Leeuw, W., Wijnen, J., Dreef, E., Tops, cell carcinoma independently of histological type, tumor microenvironment, and VHL C., Breuning, M., Brocker-Vriends, A., Vasen, H., Fodde, R., and Morreau, H. mutation. Cancer Res., 59: 2182–2189, 1999. Conventional and tissue microarray immunohistochemical expression analysis of 16. Schraml, P., Muller, D., Bednar, R., Gasser, T., Sauter, G., Mihatsch, M. J., and mismatch repair in hereditary colorectal tumors. Am. J. Pathol., 162: 469–477, 2003. Moch, H. Allelic loss at the D9S171 locus on chromosome 9p13 is associated with 39. Masuda, N., Ohnishi, T., Kawamoto, S., Monden, M., and Okubo, K. Analysis of progression of papillary renal cell carcinoma. J. Pathol., 190: 457–461, 2000. chemical modification of RNA from formalin-fixed samples and optimization of 17. Wu, S., Hafez, G., Xing, W., Newton, M., Chen, X., and Messing, E. The correlation molecular biology applications for such samples. Nucleic Acids Res., 27: 4436– between the loss of chromosome 14q with histologic tumor grade, pathologic stage, 4443, 1999. and outcome of patients with nonpapillary renal cell carcinoma. Cancer (Phila.), 77: 40. Werner, M., Chott, A., Fabiano, A., and Battifora, H. Effect of formalin tissue fixation 1154–1160, 1996. and processing on immunohistochemistry. Am. J. Surg. Pathol., 24: 1016–1019, 18. Bittner, M., Meltzer, P., Chen, Y., Jiang, Y., Seftor, E., Hendrix, M., Radmacher, M., 2000. Simon, R., Yakhini, Z., Ben-Dor, A., Sampas, N., Dougherty, E., Wang, E., Marin- 41. Guardavaccaro, D., Corrente, G., Covone, F., Micheli, L., D’Agnano, I., Starace, G., cola, F., Gooden, C., Lueders, J., Glatfelter, A., Pollock, P., Carpten, J., Gillanders, Caruso, M., and Tirone, F. Arrest of G1-S progression by the p53-inducible gene PC3 E., Leja, D., Dietrich, K., Beaudry, C., Berens, M., Alberts, D., and Sondak, V. is Rb dependent and relies on the inhibition of cyclin D1 transcription. Mol. Cell. Molecular classification of cutaneous malignant melanoma by gene expression pro- Biol., 20: 1797–1815, 2000. filing. Nature (Lond.), 406: 536–540, 2000. 42. Iacopetti, P., Michelini, M., Stuckmann, I., Oback, B., Aaku-Saraste, E., and Huttner, 19. Alizadeh, A. A., Eisen, M. B., Davis, R. E., Ma, C., Lossos, I. S., Rosenwald, A., W. B. Expression of the antiproliferative gene TIS21 at the onset of neurogenesis Boldrick, J. C., Sabet, H., Tran, T., Yu, X., Powell, J. I., Yang, L., Marti, G. E., identifies single neuroepithelial cells that switch from proliferative to neuron-gener- Moore, T., Hudson, J., Jr., Lu, L., Lewis, D. B., Tibshirani, R., Sherlock, G., Chan, ating division. Proc. Natl. Acad. Sci. USA, 96: 4639–4644, 1999. W. C., Greiner, T. C., Weisenburger, D. D., Armitage, J. O., Warnke, R., Staudt, 43. Kannan, K., Amariglio, N., Rechavi, G., Jakob-Hirsch, J., Kela, I., Kaminski, N., L. M., et al. Distinct types of diffuse large B-cell lymphoma identified by gene Getz, G., Domany, E., and Givol, D. DNA microarrays identification of primary and expression profiling, Nature (Lond.), 403: 503–511, 2000. secondary target genes regulated by p53. Oncogene, 20: 2225–2234, 2001. 20. Sorlie, T., Perou, C. M., Tibshirani, R., Aas, T., Geisler, S., Johnsen, H., Hastie, T., 44. Melamed, J., Kernizan, S., and Walden, P. D. Expression of B-cell translocation gene Eisen, M. B., van de Rijn, M., Jeffrey, S. S., Thorsen, T., Quist, H., Matese, J. C., 2 protein in normal human tissues. Tissue Cell, 34: 28–32, 2002. Brown, P. O., Botstein, D., Eystein Lonning, P., and Borresen-Dale, A. L. Gene 45. Pati, D., Keller, C., Groudine, M., and Plon, S. E. Reconstitution of a MEC1- expression patterns of breast carcinomas distinguish tumor subclasses with clinical independent checkpoint in yeast by expression of a novel human fork head cDNA, implications. Proc. Natl. Acad. Sci. USA, 98: 10869–10874, 2001. Mol. Cell. Biol., 17: 3037–3046, 1997. 21. Kononen, J., Bubendorf, L., Kallioniemi, A., Barlund, M., Schraml, P., Leighton, S., 46. Myokai, F., Takashiba, S., Lebo, R., and Amar, S. A novel lipopolysaccharide- Torhorst, J., Mihatsch, M. J., Sauter, G., and Kallioniemi, O. P. Tissue microarrays induced transcription factor regulating tumor necrosis factor ␣ gene expression: for high-throughput molecular profiling of tumor specimens. Nat. Med., 4: 844–847, molecular cloning, sequencing, characterization, and chromosomal assignment. Proc. 1998. Natl. Acad. Sci. USA, 96: 4518–4523, 1999. 22. Schoenberg Fejzo, M., and Slamon, D. J. Frozen tumor tissue microarray technology 47. Idriss, H. T., and Naismith, J. H. TNF ␣ and the TNF receptor superfamily: for analysis of tumor RNA, DNA, and proteins. Am. J. Pathol., 159: 1645–1650, structure-function relationship(s). Microsc. Res. Tech., 50: 184–195, 2000. 2001. 48. Saito, T., Kimura, M., Kawasaki, T., Sato, S., and Tomita, Y. MHC class II 23. Clapp, W. L. Adult kidney. In: S. S. Sternberg (ed.), Histology for Pathologists, pp. antigen-associated invariant chain on renal cell cancer may contribute to the anti- 677–707. New York: Raven Press, 1992. tumor immune response of the host. Cancer Lett., 115: 121–127, 1997. 24. Union International Contre Cancer. TNM Classification of Malignant Tumors, Ed. 5. 49. Young, A. N., Amin, M. B., Moreno, C. S., Lim, S. D., Cohen, C., Petros, J. A., New York: Wiley-Liss, 1997. Marshall, F. F., and Neish, A. S. Expression profiling of renal epithelial neoplasms: 25. Thoenes, W., Stoerkel, S., and Rumpelt, H. Histopathology and classification of renal a method for tumor classification and discovery of diagnostic molecular markers. cell tumors (adenomas, oncocytomas, and carcinomas): the basic cytological and Am. J. Pathol., 158: 1639–1651, 2001. histopathological elements and their use for diagnostics. Pathol. Res. Pract., 181: 50. Jiang, Z., Xu, M., Savas, L., LeClair, P., and Banner, B. F. Invariant chain expression 125–143, 1986. in colon neoplasms. Virchows Arch., 435: 32–36, 1999. 26. Ausubel, F., M., Brent, R., Kingston, R. E., Moore, D. D., Seidman, J. G., Smith, 51. Ishigami, S., Natsugoe, S., Tokuda, K., Nakajo, A., Iwashige, H., Aridome, K., J. A., and Struhl, K. Current Protocols in Molecular Biology, Vol. 1. Etobicoke, Hokita, S., and Aikou, T. Invariant chain expression in gastric cancer. Cancer Lett., Ontario: John Wiley & Sons, Inc., 1994. 168: 87–91, 2001. 1638

Downloaded from cancerres.aacrjournals.org on September 29, 2021. © 2004 American Association for Cancer Research. Impaired Expression of the Cell Cycle Regulator BTG2 Is Common in Clear Cell Renal Cell Carcinoma

Kirsten Struckmann, Peter Schraml, Ronald Simon, et al.

Cancer Res 2004;64:1632-1638.

Updated version Access the most recent version of this article at: http://cancerres.aacrjournals.org/content/64/5/1632

Cited articles This article cites 44 articles, 17 of which you can access for free at: http://cancerres.aacrjournals.org/content/64/5/1632.full#ref-list-1

Citing articles This article has been cited by 14 HighWire-hosted articles. Access the articles at: http://cancerres.aacrjournals.org/content/64/5/1632.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/64/5/1632. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.