The in Vitro and in Vivo Effects of Re-Expressing Methylated Von Hippel-Lindau Tumor Suppressor Gene in Clear Cell Renal Carcinoma with 5-Aza-2؅-Deoxycytidine

Vol. 10, 7011–7021, October 15, 2004 Clinical Cancer Research 7011

The In vitro and In vivo Effects of Re-Expressing Methylated von Hippel-Lindau Tumor Suppressor Gene in Clear Cell Renal Carcinoma with 5-Aza-2؅-deoxycytidine

Wade G. Alleman,1,2 Ray L. Tabios,2 Well described phenotypic changes of VHL expression in- Gadisetti V. R. Chandramouli,3 cluding decreased invasiveness into Matrigel, and decreased Olga N. Aprelikova,3 Carlos Torres-Cabala,2 vascular endothelial growth factor and glucose transport- 4 5 er-1 expression were observed in the treated lines. VHL Arnulfo Mendoza, Craig Rodgers, methylated ccRCC xenografted tumors were significantly 2 2 Nikolai A. Sopko, W. Marston Linehan, and reduced in size in mice treated with 5-aza-dCyd. Mice bear- James R. Vasselli2 ing nonmethylated but VHL-mutated tumors showed no 1Howard Hughes Medical Institute, Chevy Chase, MD; 2Urology tumor shrinkage with 5-aza-dCyd treatment. Branch, 3Laboratory of Biosystems and Cancer, and 4Pediatric Conclusion: Hypo-methylating agents may be useful in Oncology Branch, Center for Cancer Research, National Cancer 5 the treatment of patients having ccRCC tumors consisting of Institute, Bethesda, Maryland; The Brady Urologic Institute, The cells with methylated VHL. Johns Hopkins Hospital, Baltimore, Maryland INTRODUCTION ABSTRACT An estimated 31,900 people are diagnosed annually with Purpose: Clear cell renal carcinoma (ccRCC) is cancer of the kidney in the United States, of which the majority strongly associated with loss of the von Hippel-Lindau (VHL) are clear cell type. In sporadic clear cell renal carcinoma tumor suppressor gene. The VHL gene is functionally lost (ccRCC), 50–85% of patients are found to have biallelic loss of through hypermethylation in up to 19% of sporadic ccRCC the von Hippel-Lindau (VHL) tumor suppressor gene (1–3), cases. We theorized that re-expressing VHL silenced by located on chromosome 3p25 (4, 5). It has been reported that in methylation in ccRCC cells, using a hypo-methylating agent, up to 19% of sporadic ccRCC, VHL function is lost because of may be an approach to treatment in patients with this type hyper-methylation of a CpG island in the promoter region of the of cancer. We test the ability of two hypo-methylating agents VHL gene (3, 6). DNA methylation is the result of DNA meth- to re-express VHL in cell culture and in mice bearing human yltransferase enzymatic activity transferring a methyl group ccRCC and evaluate the effects of re-expressed VHL in these from S-adenosyl-methionine to the C-5 position of cytosine (7). models. The majority of available hypo-methylating agents are Experimental Design: Real-time reverse transcription- structural variations of cytidine with a modified position 5 of the PCR was used to evaluate the ability of zebularine and pyrimidine ring. After phosphorylation, the compound is incor- 5-aza-2؅-deoxycytidine (5-aza-dCyd) to re-express VHL in porated into DNA where it forms an irreversible covalent bond four ccRCC cell lines with documented VHL gene silencing with DNA methyltransferase. Total enzyme levels decrease after through hypermethylation. We evaluated if the VHL re- DNA replication, leading to global hypo-methylation (8, 9). The expressed after hypo-methylating agent treatment could re- hypo-methylating agent with the greatest selectivity for DNA is create similar phenotypic changes in ccRCC cells observed 5-aza-2Ј-deoxycytidine (5-aza-dCyd), first synthesized in 1964 when the VHL gene is re-expressed via transfection in cell (10). It has been shown to induce expression of silenced genes culture and in a xenograft mouse model. Finally we evaluate (11) and suppress growth of tumor cells in vitro (12). Despite global gene expression changes occurring in our cells, using some efficacy as a treatment for leukemia, minimal success has microarray analysis. been seen in the treatment of solid tumors (13–15). Zebularine, Results: 5-Aza-dCyd was able to re-express VHL in our a cytidine deaminase inhibitor, has been described recently as an cell lines both in culture and in xenografted murine tumors. effective hypo-methylating agent (16). Functioning in a similar manner to 5-aza-dCyd, zebularine has the advantage of being stable in aqueous solution up to a pH of 12 (17, 18) and having oral bioavailibility (16). Received 3/15/04; revised 6/24/04; accepted 7/7/04. In this report we evaluate the ability of 5-aza-dCyd and The costs of publication of this article were defrayed in part by the zebularine to re-express VHL in several human hyper-methyl- payment of page charges. This article must therefore be hereby marked ated VHL ccRCC cell lines. Furthermore, we assess the ability advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. of the re-expressed VHL to cause biochemical and phenotypic Note: Supplementary data for this article can be found at Clinical alterations in the cells observed when the VHL gene is trans- Cancer Research Online (http://clincancerres.aacrjournals.org). fected back into the cells. We also document the ability of Requests for reprints: W. Marston Linehan, Chief, Urologic Oncology 5-aza-dCyd to re-express VHL in xenografted, VHL-methyl- Branch, CCR National Cancer Institute, Building 10, Room 2B47, Bethesda, Maryland 20892-1501. Phone: 301-496-6353; Fax: 301-402- ated, ccRCC cells that have formed small tumors in mice. 0922; E-mail: [email protected]. We then evaluate the effects of re-expressing VHL in ©2004 American Association for Cancer Research. established tumors consisting of VHL-negative cells. Finally,

using cDNA microarrays, we evaluate the changes in global The technique used for bisulfite modification and subsequent gene expression in our cell lines after they are exposed to RT-PCR analysis has been described previously (23). Briefly, 5-aza-dCyd. we subjected 1 ␮g of sample DNA to sodium bisulfite modification, using the CpGenome DNA Modification Kit (Chemicon, Temecula, CA). We used universally methylated DNA to gen- MATERIALS AND METHODS erate a standard curve, based on which we calculated the amount Cells and Cell Culture. Six cell lines from sporadic of DNA in the samples. White blood cell DNA was used as a clear cell renal carcinomas were analyzed: UOK 108, UOK 121, negative control (unmethylated). Amplification of MYOD1 was UOK 127, UOK 143, UOK 171, and 786–0. The methylation used to determine total amount of amplifiable DNA in the status of UOK 108, UOK 121, UOK 127, and UOK 143 has samples. Our PCR reaction mix contained 1 ␮L of DNA in 25 been described previously (3). UOK 121 and UOK 143 have one ␮L of total volume, which also included 2.5 ␮Lof10ϫ PCR ␮ hyper-methylated copy and one silent copy of VHL. UOK 127 buffer II, 5 mmol/L MgCl2, 250 mol/L deoxynucleoside has two hyper-methylated copies of VHL. UOK 108 has one triphosphate, 1 ␮mol/L of each primer, 0.2 ␮mol/L probe, 1.25 ␮ silent and one partially methylated copy of VHL, indicating units Taq gold, and 10.25 LH2O. The PCR buffer, MgCl2, and cellular heterogeneity within the culture. UOK 171 is a non- Taq gold are parts of the “AmpliTaq Gold with GeneAmp” kit methylated VHL ϩ/ϩ cell line, whereas 786–0 is a nonmethy- from Applied Biosystems (Roche, Basel, Switzerland). The lated VHL Ϫ/Ϫ line, exhibiting loss of one VHL allele and PCR was done at 95°C for 8 minutes, followed by 45 cycles of inactivation of the other by truncation after amino acid 104. 95°C for 15 seconds, 60°C for 1 minute, and 72°C for 1 minute. Additionally, UOK 121 wild-type, is the UOK 121 cell line with Every sample was run in triplicate. A “methylation index” was a stably transfected VHL (19). The 786–0 ccRCC cell line is calculated that is the ratio between amount of DNA for the VHL VHL negative through deletion of one allele and nonsense gene and amount of MYOD1 that allowed us to calculate the mutation causing a shortened protein (amino acid 1–104) in the relative levels of VHL gene methylation between 5-aza-dCyd- second allele (20). Cells were grown in DMEM containing 10% treated and untreated cells.

fetal bovine serum (FBS) and incubated at 37°C in 5% CO2. Western Blot Analysis. To evaluate if we could detect UOK 121 wild-type was grown in the same media with 1 mg/ml VHL protein re-expression in the UOK 121 cell line after neomycin antibiotic. 5-aza-dCyd exposure, we did a Western blot analysis. The UOK Real-Time Reverse Transcription-PCR Analysis. For 121 cell line stably transfected with VHL and UOK 121 trans- the studies on the renal cancer lines, two million viable cells fected with empty plasmid were each plated into several 15-cm counted by trypan blue exclusion method were plated in 15-cm plates. One plate of each cell line was exposed to 3 ␮mol/L tissue culture dishes. 5-Aza-dCyd (Sigma, St. Louis, MO) and 5-Aza-dCyd as described above for the RT-PCR experiments, zebularine (a generous gift of Dr. Victor E. Marquez) were and the remaining plates were allowed to grow normally without dissolved in water. Cells were incubated in zebularine at 50 treatment. At the appropriate time points the cells were washed ␮mol/L, 500 ␮mol/L, and 1,000 ␮mol/L for 24 or 48 hours and with ice cold PBS and trypsinized into a pellet. Whole cell were given 24, 48, and 72 hours for cell division. Incubation extracts were generated with lysis buffer containing 1% Igepal, with 5-aza-dCyd at 0.5, 3.0, and 5.0 ␮mol/L concentrations 0.25% sodium deoxycholate, 1% SDS, 50 mmol/L Tris (pH occurred for 24 hours; cells were then allowed 72 hours for cell 7.4), 150 mmol/L NaCl, 10 mmol/L EDTA, 1 mmol/L sodium division. We also exposed a bladder cancer cell line, T24, that orthovanadate, 1 mmol/L sodium fluoride, 1 mmol/L phenyl- has a known methylated p16 gene, to 500 ␮mol/L zebularine for methylsulfonyl fluoride, 10 ␮mol/L leupeptin, and 0.10 ␮g/ml 24 hours and then gave 72 hours for cell division (21). At the aprotinin. The lysate was incubated at 4°C on a rotator for 30 appropriate times, total RNA was isolated with Trizol reagent minutes and centrifuged to recover the total protein. Bio-Rad (Invitrogen, Carlsbad, CA) and purified with three phenolchlo- protein assay kit (Bio-Rad, Hercules, CA) was used to quantify roform extractions. Using TaqMan Reverse Transcription kit the protein, and equal amounts of protein were fractionated on (ABI, Foster City, CA), 2 ␮g of total RNA were used to form 4 to 20% Tris-Glycine Bio-Rad Gels and transferred to polyvi- cDNA in a 25 ␮L solution, following the manufacturer’s sug- nylidene difluoride membranes. The VHL antibody Ig32 (BD gested protocol. Real-time quantitative PCR was carried out Biosciences-Phramingen, San Diego, CA) at a 1:250 v/v con- with ABI Prism 7000 machine. Assays-on-demand probe and centration was used to probe for the VHL protein. Protein bands primer master mixes from ABI were used for all reverse tran- were detected with horseradish peroxidase conjugated goat an- scription (RT)-PCR reactions except for the probe and primers tirabbit or goat antimouse secondary antibodies (Amersham for VHL that were described previously (22). The comparative Pharmacia, Piscataway, NJ).

CT (threshold cycle) method described in the manufacturer’s In vitro Proliferation Analysis. In vitro proliferation protocol was used to do relative quantitation of expression, was measured for UOK 121, UOK 121 stably transfected with comparing treated samples to concordant untreated controls and VHL, and 786–0 cell lines. Cells were treated with 3 ␮mol/L the internal control, ␤-actin. Each sample was run in triplicate 5-aza-dCyd for 24 hours at 37°C, and then collected with and statistical analysis done using Microsoft Excel and one- CellStripper (CellGro, Herndon, VA). The trypan blue exclusion sided student t tests. method was used to count cells, and 10,000 cells suspended in DNA Methylation Analysis. We isolated DNA with the 100 ␮L of 10% FBS media added to each well of a 96-well Puregene DNA Isolation kit (Puregene, Ashby Park, United plate. Eight identical wells were measured for each cell line, and Kingdom) from untreated UOK 121 cells or cells treated with a different 96-well plate was used for each day of measurement. 500 ␮mol/L 5-aza-dCyd for 24 hours than grown for 72 hours. The CellTiter 96 nonradioactive cell proliferation assay (Pro-

mega, Madison, WI) was used according to the manufacturer’s step Tukey’s Biweight estimate) and detection calls based on protocol. This assay is based on the conversion of a tetrazolium Wilcoxon’s signed rank test. The signal values were normalized salt, and represents a measure of the in vitro metabolism of the to a trimmed (excluding 2% lowest and 2% highest signals) cells. Promega Aqueous One, the indicator dye, was added to mean level of 500 within each array. Expression ratios of post- each well (20 ␮L/well) and allotted three hours to develop. The to pre-5-aza-dCyd exposure and statistical significance of ex- absorbance was then measured at 490 nm by a SpectraMax Plus pression change as “increase,” “decrease,” and “no change” 96-well plate reader (Molecular Devices, Sunnyvale, CA), every were determined by comparison analysis of MAS5. Because two 24 hours for three total measurements. separate arrays were done for all data points, we had two gene Branching Morphogenesis Assays. A description of the profiles pre- and two gene profiles post-5-aza-dCyd treatment in branching morphogenesis assay has been described previously each cell line tested. We thus were able to make four compar- (24). Briefly, UOK 108, UOK 127, and 786–0 cells growing in isons for each cell line; for example, we compared pretreatment log phase were either treated with 3 ␮mol/L 5-aza-dCyd for 24 gene profile A to post-treatment profile A, then pretreatment hours or kept in standard 10% FBS media (controls), then profile A to post-treatment profile B and so on. To be included collected with CellStripper and resuspended in 10% FBS media. in the analysis, a gene had to increase or decrease calls. We Cells were counted with the trypan blue exclusion method, and considered a gene to be affected by 5-aza-dCyd treatment if its two million cells were plated into wells of a 96-well plate in a mean expression had changed by 2-fold using the criteria listed 50:50 mixture of Matrigel and 10% FBS media in a volume of above in at least three of the six cell lines tested. We categorized 130 ␮L. The Matrigel cellular mixture was allowed to harden at the above significant genes by Gene Ontologies using Expres- 37°C for 30 minutes. Once the Matrigel hardened, 130 ␮Lof sion Analysis Systematic Explorer software (27). The number of media alone or media with 100 ng/mL recombinant hepatocyte genes in the significant list was compared with the population of growth factor (HGF, R&D systems, Minneapolis, MN) was all genes on the microarray. Over-representation analysis was added to the appropriate wells. All experiments were carried out done with a modified Fisher’s test to estimate the probability of in triplicate. Cells were incubated at 37°C for 72 hours before finding significant genes in a Gene Ontology term from the gene being photographed. population on array. Expression Analysis Systematic Explorer Microarray Assays. Total RNA from each of the six cell score, which is upper boundary of the distribution of Jackknife lines exposed to 3 ␮mol/L 5-aza-dCyd, analyzed in the real-time Fisher exact probabilities, was calculated for each Gene Ontol- quantitative PCR experiments described previously, was also ogy term in significant list, and the Gene Ontology terms were used for the microarray analysis. The gene expression from the 5- ranked by significance. aza-dCyd-exposed cells was compared with total RNA isolated Murine Tumor Xenograft Experiments. Immunocom- from the matched cell lines grown under similar conditions but not promised male SCID/beige mice were used to develop a sub- exposed to 5-aza-dCyd. Reverse transcription was carried out with cutaneous tumor xenograft model with UOK 121 cells. In the 5 ␮g of total RNA with the Superscript II Reverse Transcription kit first murine study, ten million cells were suspended in 250 ␮L

(Invitrogen); the T7-d(T)24 primer was used in place of the oli- of Hank’s balanced salt solution and injected subcutaneously godeoxythymidylic acid primer supplied in the kit. Purification was into the left inguinal region of 6-week-old mice. The tumors done with the Genechip Sample Clean-up Module (Qiagen, Valen- were allowed to grow for 15 days at which time the mean tumor cia, CA). In vitro transcription was carried out to form biotin- diameter was 15 mm. Three mice received one intraperitoneal labeled cRNA with BioArray High Yield RNA Transcription La- dose (2.5 mg/kg) of 5-aza-dCyd whereas two mice received beling kit (Enzo, New York, NY) and purified with the Qiagen saline injections and after five days were sacrificed. Tumors Cleanup Module kit. Fragmentation was done with 5ϫ fragmen- were frozen to Ϫ70°C and crushed to a fine powder with mortar tation buffer at 95°C for 35 minutes, and the reaction was chilled on and pestle while on dry ice; the powder from the treated mice ice. The fragmented cRNA was then applied to HG-U133A array and the powder from the untreated mice were combined to form chips (Affymetrix, Santa Clara, CA) and hybridized to the array two samples. Total RNA was isolated with TRIzol and real-time probes for 16 hours. After washing and staining with streptavidin- quantitative PCR carried out in quadruplicate to determine VHL phycoerythrin, the chip was scanned with the GeneArray Scanner message levels. (Hewlett-Packard, Palo Alto, CA). Duplicate microarray experi- In the second murine study, 20 mice received injection with ments were done for each cell line, both with and without 5- five million UOK 121 cells, and 10 mice received injection with aza-dCyd exposure; thus a total of 24 microarray experiments were five million 786–0 cells in the same manner described above. done for this analysis. Subcutaneous tumors were allowed to develop for 15 days Microarray Data and Statistical Analysis. The proto- (UOK 121) and 20 days (786–0) before starting the 5-aza-dCyd cols for statistical analysis of Affymetrix microarray data have injections. For the 20 mice bearing UOK 121 tumors, 10 re- been described previously (25, 26). Microarray Suite version 5.0 ceived an intraperitoneal injection of 2.5 mg/kg 5-aza-dCyd (MAS5) by Affymetrix [Statistical Algorithms Reference dissolved in PBS, and 10 received only PBS, once a week for 4 Guide, Part No. 701110 Rev 1, Affymetrix Inc. (2001) Santa consecutive weeks. Five mice with 786–0 tumors received 2.5 Clara, CA 95051]5 was used to determine signal values (one- mg/kg 5-aza-dCyd injection, and the remaining 5 received PBS placebo injections at the same dosing schedule as the mice with UOK 121 tumors. Mice were weighed and tumors measured in the two largest dimensions every 3 to 4 days by a technician 5 http://www.affymetrix.com/support/technical/technotes/statistical_ref- blinded to the experiment. Tumor volume (TV) was calculated erence_guide.pdf according to the following equation: TV ϭ (D)(d2)(␲/6), where

D is the larger diameter and d is the smaller diameter (28). After 28 days of treatment (1 week after the last 5-aza-dCyd dose), all mice were sacrificed, and tumors were dissected away from normal tissue and weighed. The tumors were formalin fixed and paraffin embedded. Histologic and Terminal Deoxynucleotidyl Transferase Assay Evaluation. Sequential thin (6 ␮m) tissue sections from the UOK 121 cell line tumors of the 5-aza-dCyd-treated and -untreated mice were mounted on noncharged slides for terminal deoxynucleotidyl transferase (TUNEL) assay and H&E staining. The TUNEL assay was performed using the Dead End TUNEL assay kit (Promega) as described by the manufacturer, without modification.

RESULTS Effects of 5-Aza-dCyd and Zebularine on VHL Expres- sion. Treatment with 0.5 ␮mol/L, 3.0 ␮mol/L, and 5.0 ␮mol/L of 5-aza-dCyd significantly increased VHL expression in the four hyper-methylated VHL lines (UOK 108, UOK 121, UOK 127, and UOK 143) as determined by real-time RT-PCR (Fig. 1A). The bisulfite modification analysis showed that VHL methylation became half as prevalent in the UOK 121 line after 3 ␮mol/L 5-aza-dCyd treatment as described above (supplemental Fig. A). UOK 121 showed a 160-fold increase in VHL transcript expression that was the largest increase of all cell lines tested. In each of the four cell lines, VHL expression significantly increased at all three doses, increasing by as much as 7.8-fold in 127, 11.6-fold in 143, and 1.9-fold in 108 (P Ͻ 0.05 in all four lines). VHL protein re-expression was clearly evident on West- ern blot in the UOK 121 cell line after exposure to 5-aza-dCyd (Fig. 2). Exposure to 5-aza-dCyd did not significantly change VHL expression in the non-VHL–methylated UOK 171 or 786–0 cell lines. Zebularine was similarly tested for its ability to increase VHL expression in UOK 121 and UOK 127 at 50 ␮mol/L, 500 ␮mol/L, and 1,000 ␮mol/L concentrations and at six different time points after exposure. However, zebularine Fig. 1 Effects of 5-aza-dCyd on gene expression. A, real-time quanti- failed to consistently increase VHL expression at any dose or tative PCR analysis of VHL expression for six cell lines. The 1st column time point in the renal lines. To establish that our zebularine for each cell line represents the control sample with no 5-aza-dCyd; the remaining three represent increasing doses of the drug at 0.5, 3, and 5 treatment regimen could re-express a methylated gene in a cell ␮mol/L. B, expression analysis of VEGF. C, expression analysis of line, we showed an up-regulation of the expression of p16 in the Glut-1. B and C represent downstream products of HIF, and decreased T24 cell line after the cells were exposed to 500 ␮mol/L of expression is hypothetically a result of increased VHL expression. All zebularine for 24 hours (supplemental Fig. B). experiments were done in triplicate and error bars represent SEM in all 5-Aza-dCyd is an S-phase-specific agent; thus, the duration three graphs. of VHL re-expression is dependent upon the speed at which each cell line divides in vitro. UOK 127 was treated with 4 ␮mol/L of 5-Aza-dCyd for 24 hours; then, every two days, RNA was isolated from the treated cells and VHL expression was vascular endothelial growth factor (VEGF; Fig. 1B and C) were measured. VHL expression increased rapidly, reaching a plateau evaluated. Glut-1 expression significantly decreased across all at 3 days from removal of the agent. Expression remained four methylated cell lines: UOK 108 expression decreased to relatively constant until 7 days, followed by a steady decline 83% of the original, UOK 121 to 33%, UOK 127 to 67%, and until 15 days, when the expression of VHL returned to control UOK 143 to 63% (P Ͻ 0.05 in all cases). Glut-1 expression also levels (data not shown). decreased in 786–0, with expression after treatment down to Re-expressed VHL Effects on Vascular Endothelial 15% of control levels (P Ͻ 0.005). VEGF expression decreased Growth Factor and Glut-1 Expression. To establish in all six cell lines as well. In the hyper-methylated VHL lines, whether VHL re-expression resulting from 5-aza-dCyd exposure UOK 108 expression decreased to 72% of control, UOK 121 to is sufficient to decrease expression of the downstream hypoxia- 34%, UOK 127 to 68%, and UOK 143 to 53% (P Ͻ 0.05). In inducible factor (HIF) targets, transcript levels of Glut-1 and UOK 171 and 786–0, the VEGF expression significantly de-

1 gene in the 108, and 12 (0.1%) genes in the UOK 143 line. In the non-VHL–methylated lines, UOK 171 and 786–0, 132 (0.9%) and 394 (2.8%) of 13,900 genes are down-regulated. The UOK 108 cell line has only 26 genes that are Ն2-fold up- regulated and one gene down-regulated after 5-aza-dCyd exposure. Only two genes, ␣1 collagen and matrix metalloproteinase 1, were Ն2-fold up-regulated and no genes were Յ2-fold down- Fig. 2 Western blot of UOK 121 cell lysates stably transfected with empty plasmid or VHL into the cells. The cells were exposed to regulated across all six cell lines after 5-aza-dCyd treatment, as 5-aza-dCyd as described in the text. defined by the array analysis. The VHL signal is too low across all of the arrays for us to make any determination of the VHL expression status by array analysis, but the RT-PCR had already verified the VHL re-expression. In a few cell lines (e.g., UOK creased to 52% and 48%, respectively, of control levels (P Ͻ 108 and UOK 143), many genes have low expression change 0.005). after 5-aza-dCyd exposure. Therefore to establish a larger cohort In vitro Proliferation and Branching Morphogenesis. of genes with altered expression after exposure to 5-aza-dCyd in We evaluated the effects of VHL re-expression on UOK 121 our ccRCC lines by array analysis, we reduced our criteria for proliferation, using a colorimetric assay based on the metabo- inclusion and identified genes with Ͼ2-fold expression change lism of a tetrazolium compound by viable cells. The non-VHL– post-5-aza-dCyd treatment in three or more of the six cell lines. methylated 786–0 cells were used as a control. The in vitro Using the criteria outlined in the methods, we identified 151 proliferation of the hyper-methylated VHL line UOK 121 was (unique of 173) genes Ն2-fold up-regulated and four genes Ն2 not significantly inhibited post-5-aza-dCyd exposure (compared down-regulated in at least three of the six cell lines (supplemen- with non-5-aza-dCyd exposed cells). For comparison we also tal information, Table A). We chose a set of 15 genes from the show that the growth of UOK 121 stably expressing VHL had 157 identified above that affect a broad spectrum of cellular no effect on proliferation rate compared with the parental UOK 121 line. Treatment with 5-aza-dCyd did not significantly alter the proliferation of 786–0 cells in culture (Fig. 3). When VHL Ϫ/Ϫ RCC cell lines, such as 786–0, are transfected with the VHL gene, the ability of the HGF-Met signaling cascade to cause in vitro branching morphogenesis into Matrigel is prevented (24). We evaluated whether re- expression of VHL by 5-aza-dCyd is adequate to prevent branching morphogenesis in the VHL-methylated UOK 108 and UOK 127 cell lines. Substantial branching morphogenesis was observed 72 hours after HGF exposure in UOK 108 and UOK 127 cells whereas these cells pretreated with 5-aza-dCyd remained spherical and showed no branching, similar to HGF- negative controls (Fig. 4). The 786–0 cells continued to exhibit significant branching even after exposure to 5-aza-dCyd, strongly suggesting that the inhibition of branching in the UOK 108 and UOK 127 cell lines was a result of increased VHL expression and not a product of toxicity from 5-aza-dCyd. Gene Expression Analysis of 5-aza-dCyd–Exposed RCC Cell Lines. The microarray assays were done to identify other genes, besides VHL, re-expressed by 5-aza-dCyd in the clear cell RCC lines. We began by looking at the overall (global) expression changes associated with exposure to 5-aza-dCyd in each of our 6 cell lines. The Affymetrix HG-U133A oligonucleotide microarray has 22,283 probe-sets with 13,903 unique genes on each chip. Treatment with 3 ␮mol/L 5-aza-dCyd for 24 hours resulted in a Ͼ2-fold up-regulation in 1,043 (7.5%) of 13,900 genes evaluated in the UOK 121 cell line, 242 (1.7%) in the 127 line, 26 (0.2%) in the UOK 108 and 140 (1.0%) in the Fig. 3 Effects of 5-aza-dCyd on cell proliferation. Cell proliferation at UOK 143 line. In the cell lines where VHL is not methylated, 24 and 72 hours after plating into 96-well plates measured via absorp- UOK 171 and 786–0, 258 (1.9%) and 774 (5.6%) of 13,900 tion at 490 nm after addition of an indicator dye with and without genes are up-regulated, respectively. Because the re-expression 5-aza-dCyd treatment. A, the proliferation rates of UOK 121 hyper- of some genes can down-regulate others, we next identified the methylated VHL lines alone, exposed to 3 ␮mol/L 5-aza-dCyd for 24 hours before plating and UOK 121 stably transfected with VHL. B, genes 2-fold down-regulated in our cell lines. Three hundred proliferation of 786–0 cells, a nonmethylated VHL Ϫ/Ϫ control un- and thirty-seven (2.4%) genes are Ͼ2-fold down-regulated in treated and exposed to 5-aza-dCyd for 24 hours before plating in the the UOK 121 cell line, 501 (3.6%) genes in the UOK 127 line, 96-well plates. Error bars represent SEM.

Fig. 4 Effects of 5-aza-dCyd treatment on branching morphogenesis. Compari- son of branching morphogenesis in two ccRCC VHL-methylated cell lines, UOK 108 and UOK 127, and the ccRCC VHL- negative but non-VHL–methylated 786–0 cell line. The cell lines are organized into three rows across, and each column shows a different treatment condition. Column A, cells growing in Matrigel for 72 hours with no treatment and no HGF. Column B, cells exposed to 100 ␮gof HGF for 72 hours and no prior treatment with 5-aza-dCyd. Column C, cells exposed to 100 ␮g of HGF for 72 hours and exposed to 3 ␮mol/L 5-aza-dCyd for 24 hours before plating into Matrigel. All photos taken on inverted microscope at 200ϫ magnification.

processes and confirmed the increased expression found by have already established tumors in the mouse at the time of VHL array analysis using real-time quantitative RT-PCR analysis gene re-expression in the tumor cells. (Table 1). Eight of the 15 genes listed in Table 1 were found to Our first experiment verified the in vivo ability of 5-aza- have CpG islands in the promotor region of the gene by per- dCyd to re-express VHL in established tumors. Tumor bearing forming a combination of literature search and internet gene mice were given a single intraperitoneal injection of 5-aza-dCyd analysis tools (Genomatix and CpG Island Searcher6; ref. 29– and sacrificed 5 days later as described in “Materials and Meth- 37). The RNA used for the RT-PCR was isolated from two of ods.” Tumors were dissected and surrounding normal tissue the three cell lines where the gene was found up-regulated by cleared from the tumor. Total RNA was extracted from the the array analysis. All samples tested show an increased expres- tumor specimen and real-time quantitative PCR performed. sion Ͼ2-fold, equal to or exceeding the results of the microarray Treated mice demonstrated approximately 1.5-fold higher VHL data. The Expression Analysis Systematic Explorer analysis expression than PBS controls (P Ͻ 0.05; Fig. 5A). described in the methods was done on the array data to find The second murine experiment evaluated the in vivo effects “classes” of genes disproportionately up- or down-regulated of 5-aza-dCyd and re-expression of VHL on the established compared with other classes of gene represented on the array, VHL-methylated cell line tumors. UOK 121 and 786–0 cells post-5-aza-dCyd exposure. Unexpectedly, we found that the were injected as described, and tumors were allowed to develop major categories of genes up-regulated after 5-aza-dCyd expo- for 15 days for UOK 121 and 20 days for 786–0 at which time sure are the nucleosome and nucleosome assembly class of the approximate mean tumor volumes were 200 mm3 (UOK genes, which includes many histone genes (Table 2). This 121) and 150 mm3 (786–0). Mice were then given intraperito- suggests that 5-aza-dCyd may disproportionately increase the neal injections of 5-aza-dCyd (2.5 mg/kg) or PBS placebo every expression of several classes of histone genes as compared with 7 days, for a total of four injections. Tumor growth was signif- other classes of genes in our cell lines. A list of all Gene icantly reduced in the treatment group of UOK 121 compared Ontologies identified by the Expression Analysis Systematic with the control group by 20 days after the 5-aza-dCyd injec- Explorer analysis is included in the supplementary information tions started (P Ͻ 0.05), and tumor weight was significantly (Table B). decreased at 28 days (P Ͻ 0.05; Fig. 5B and C). In vivo, Effects of 5-aza-dCyd in Murine Xenograft Models. It 5-aza-dCyd treatment did not have a significant effect on either has been shown previously that VHL-negative cell lines lose tumor growth or tumor weight in the 786–0 tumors (Fig. 5B and their ability to form tumors in vivo with reintroduction of D; P Ͼ 0.05). wild-type VHL gene (38, 39) before injection into mice; this Histology and TUNEL Assay. There were clear differ- work was done with 786–0 cells. Our series of experiments ences in the histology from the UOK 121 cell line tumors of the evaluates the biological behavior of VHL-negative cells that 5-aza-dCyd-treated mice versus the untreated mice. The tumors from the untreated mice showed solid sheets of poorly differ- entiated clear cells with very few areas of necrosis or fibrosis. The tumors from the treated mice had more extensive areas of 6 Genomatix, www.genomatix.de/; CpG Island Searcher, www.uscnorris. fibrosis and necrosis (Fig. 6). The TUNEL assay showed no com/cpgislands/cpg.cgi. difference in the amounts of apoptosis in the tumors of the

Table 1 List of genes up-regulated by 2-fold or higher after 5-aza-dCyd exposure in at least three of the six cell lines chosen and confirmed by RT-PCR Cell lines GenBank CpG islands accession Gene RT-PCR present in number symbol Gene name Location Array analysis confirmation promoter NM_005504.1 BCAT1 Branched chain aminotransferase 12pter-q12 121, 171, 786–0 121, 171 Yes 1, cytosolic NM_004360.1 CDH1 Cadherin 1, type 1, E-cadherin 16q22.1 108, 121, 786–0 108, 786–0 Yes34 (epithelial) NM_001423.1 EMP1 Epithelial membrane 12p12.3 121, 127, 171 121, 127 No protein 1 NM_000640.1 IL13RA2 Interleukin 13 receptor, ␣2 Xq13.1-q28 143, 171, 786–0 171, 786–0 Yes36 AF003934.1 PLAB Prostate differentiation factor 19p13.1–13.2 121, 127, 786–0 121, 786–0 No NM_002773.1 PRSS8 Protease, serine, 8 (prostasin) 16p11.2 121, 127, 786–0 121, 786–0 Yes37 NM_002615.1 SERPINF1 Pigment epithelium derived 17p13.1 121, 143, 786–0 121, 143 No factor (PEDF) X57348 SFN Stratifin 1p35.3 121, 143, 786–0 121, 786–0 Yes32,33 AL574096 TFPI2 Tissue factor pathway inhibitor 2 7q22 121, 171, 786–0 121, 786–0 Yes30,31 AF001294.1 TSSC3 Tumor suppressing 11p15.5 121, 143, 786–0 121, 786–0 Yes29,35 subtransferable candidate 3 NM_000376.1 VDR Vitamin D (1,25- 12q12-q14 108, 127, 171, 786–0 108, 786–0 Yes dihydroxyvitamin D3) receptor NM_001785.1 CDA Cytidine deaminase 1p36.2-p35 121, 127, 143, 171, 786–0 143, 786–0 No NM_000088.1 COL1A1 Collagen, type I, ␣ 1 17q21.3-q22.1 All six cell lines 121, 786–0 No NM_000930.1 PLAT Plasminogen activator, tissue 8p12 108, 121, 127, 171, 786–0 121, 786–0 No NM_002961.2 S100A4 S100 calcium binding protein A4 1q21 121, 127, 171, 786–0 127, 171 No

treated and untreated mice. Both groups showed only very small RT-PCR analysis. Four cell lines from the tumors of patients amounts of apoptosis compared with positive controls. with clear cell RCC in which the VHL gene is hyper-methylated were used in this study. Exposure to 5-aza-dCyd re-expresses DISCUSSION VHL in all four hyper-methylated cell lines in vitro; the largest Re-expression of the VHL gene in VHL-negative ccRCC increase occurred in the UOK 121 cell line with over 100-fold cell lines is associated with a constellation of biochemical and increase in VHL expression at the 3 and 5 ␮mol/L 5-aza-dCyd phenotypic changes in vitro and in vivo. The effects of re- concentrations. The smallest increase of VHL expression oc- expressing VHL in VHL-negative ccRCC cells have been de- curred in the 108 cell line with approximately 2-fold increase at fined previously by stably transfecting the VHL gene into cells the highest 5-aza-dCyd concentration. The incomplete nature of and then comparing phenotypic characteristics of the transfected the methylation of the VHL gene in the UOK 108 line may cells with the nontransfected cells. In this report, we used the account for the relatively small increase in VHL expression after hypo-methylating agent 5-aza-dCyd to re-express VHL in 5-aza-dCyd exposure as compared with the other three cell lines. ccRCC cell lines having methylated VHL, both in vitro cell We were also able to re-express VHL in the UOK 121 line culture and in vivo murine models. after these cells had formed small tumors in mice. The VHL We began by showing that hyper-methylated VHL can be expression levels attained were 1.5- to 1.8-fold higher than in re-expressed in cell lines exposed to 5-aza-dCyd with real-time untreated mice with UOK 121 cell line tumors of similar size.

Table 2 The nucleosome assembly cohort of genes as identified by EASE analysis 5-aza-dCyd exposed cell line vs. the untreated control cell line log2 signal ratios for each gene Gene symbol GenBank Accession Gene description Location 108 121 143 171 127 786–0 TSPY NM_003308.2 Testis specific protein, Y-linked Yp11.2 6.55 5.6 5.375 H2AFB AI218431 H2A histone family, member B Xq28 3.2 4.125 4.55 4.825 HIST1H2BC NM_003526.1 Histone 1, H2bc 6p21.3 1.3 0.75 1.575 4.075 HIST1H2AE NM_021052.1 Histone 1, H2ae 6p22.2-p21.1 4.075 1.5 2.875 HIST1H1C BC002649.1 Histone 1, H1c 6p21.3 Ϫ0.025 2.3 0.225 1.225 2.175 3.1 HIST1H3D NM_003530.1 Histone 1, H3d 6p21.3 2.55 1.2 2.8 HIST1H2BG BE271470 Histone 1, H2bg 6p21.3 2.55 1.475 2.225 AL353759 Ϫ0.1 2.3 Ϫ0.05 1.125 0.175 1 HIST1H2BH NM_003524.1 Histone 1, H2bh 6p21.3 1.15 0.75 1.75 1.9 NOTE. Genes are at least 2ϫ up-regulated in three of the six lines evaluated.

Fig. 5 In vivo tumor effects with IP 5-aza-dCyd administration. A, real-time quantitative PCR comparing in vivo expression of VHL from mRNA isolated out of UOK 121 cell line tumor xenografts in mice receiving a dose of 5-aza-dCyd to PBS (control). B, comparison of the xenografted tumor masses in mice treated with 5-aza-dCyd to mice given PBS for tumors arising from UOK 121 and 786–0 cell lines. C, comparison of xenografted tumor volumes in mice treated with 5-aza-dCyd to mice given PBS for tumors arising from UOK 121 cells. The difference in tumor volume becomes significant (P Ͻ 0.05) by day 20 after 5-aza-dCyd was started (day 35 after the cells were injected) P values obtained with one-sided t tests. All error bars represent SEM. D, comparison of 786–0 xenografted tumor volumes exactly as done for UOK 121 in C above. There is no significant difference in 786–0 tumor volume.

The reason for the lower levels of VHL expression in the mice expression was decreased to one third the pre-5-aza-dCyd mes- compared with what we saw in vitro may be attributable to a sage levels; this coincides with the UOK 121 cell line showing lower concentration of 5-aza-dCyd delivered to the UOK 121 the largest increase in VHL expression after 5-aza-dCyd expo- cells in the mice. Also, the mouse tumors were grossly dissected sure. We cannot directly attribute the decreases in these HIF- from the mice, and other cells, contributed by the mouse such as regulated genes to VHL re-expression alone because VEGF and vasculature and interstitial stromal cells, were included in our GLUT-1 expression also decrease in the non-VHL–methylated samples. lines. Other genes beyond HIF may regulate the expression of We next showed that 5-aza-dCyd re-expressed VHL can VEGF and GLUT-1, and 5-aza-dCyd may be causing the hypo- modulate the expression of the HIF-regulated genes, VEGF and methylation of other regulatory genes causing the decreased glucose transporter (GLUT)-1. Under normoxic conditions, the expression noted in the UOK 171 and 786–0 lines. VHL gene product forms a multimeric complex with elongin B, The next series of studies evaluates the phenotypic changes elongin C, Cul2, and Rbx1 (40–44), which in turn acts as a in our cell lines after 5-aza-dCyd exposure. We chose the ubiquitin ligase targeting HIF for proteosomal degradation (45, phenotypic characteristics evaluated in this study because they 46). Loss of normal VHL function leads to a failure to degrade represent well-described characteristics observed in VHL-nega- HIF, resulting in constitutive expression of HIF target genes (20, tive lines when VHL function is re-established by gene trans- 38, 47–50). The message levels of VEGF and GLUT-1 signif- fection. It is well documented that 5-aza-dCyd has a general- icantly decreased in the four methylated cell lines after 5-aza- ized, nonspecific cytotoxic effect on cells attributable, in part, to dCyd exposure correlating to the levels of VHL re-expression the formation of enzyme-DNA adducts (51). To document that noted in these lines. The largest decreases in expression oc- there are no cytotoxic effects of 5-aza-dCyd at the concentra- curred in the UOK 121 cell line where the VEGF and GLUT-1 tions used, we control each experiment by exposing the non-

Fig. 6 Histology of tumors in mice. Histology of UOK 121 cell line tumors grown in SCID/beige mice. Representative H&E-stained 6 ␮m-thick sections from UOK 121 cell tumors excised 43 days after 5 million UOK 121 cells were injected subcutaneously (original magnification, ϫ200). A, section from a mouse that received PBS. B, section from a mouse that received 5-aza-dCyd at 2.5 mg/kg once a week for 4 weeks.

VHL–methylated 786–0 ccRCC cell line to 5-aza-dCyd. The mice treated with 5-aza-dCyd as they did in mice receiving no 786–0 line is an excellent control because it is VHL-negative treatment, the UOK 121 VHL-methylated line tumors were through allele loss and mutation, and it shows similar pheno- significantly reduced in size and weight in the treated mice. The typic changes observed in the VHL-methylated lines after VHL mice did not seem to have any morbidity from receiving 5-aza- plasmid transfection. When 786–0 and UOK 121 cells are dCyd at the dosages given because there were no remarkable grown in Matrigel and exposed to HGF they become highly differences between the treated and untreated mice. At the time invasive into the Matrigel 3-dimensional matrix, and their mor- the 5-aza-dCyd was begun in the mice, the UOK 121 VHL- phology changes extensively to that of a highly branched form. methylated tumors were approximately 200 mm3 in size. Over When VHL is transfected into these lines, HGF exposure has the course of the 25-day 5-aza-dCyd-treatment period, the tu- little to no effect on either cell line. The UOK 121 line did not mors in the treated mice remained 200 mm3 whereas the un- invade into Matrigel after 5-aza-dCyd exposure whereas the treated tumors more than doubled in volume. Although VHL 786–0 continued to branch after 5-aza-dCyd. We next evaluated inhibits xenografted tumor growth in mice it does not slow the the proliferation rates of cells with 5-aza-dCyd exposure. The growth of cells in vitro. However, the re-expression of VHL is re-expression of VHL in the 786–0 and UOK 121 cell lines did associated with a decrease in expression of growth factors such not change their proliferation rates in vitro. Similarly, the pro- as VEGF, transforming growth factor-␣, and platelet-derived liferation rate of the 786–0 and UOK 121 line was unaffected growth factor in the tumor cells, which may enable the tumor to by 5-aza-dCyd treatment. The above experiments show that the grow to sizes that require neo-vascularization allowing adequate ability of 5-aza-dCyd to re-express VHL in our methylated lines oxygenation. Thus we would expect there to be areas of necrosis can recreate the phenotypic changes shown in these cells that and fibrosis in the 5-aza-dCyd-treated tumors (where the VHL occur when the VHL gene is transfected into them. Furthermore, gene is re-expressed), which was observed in these tumors the cytotoxic effects of 5-aza-dCyd played little role in causing without significant amounts of apoptosis. The observation that the observed phenotypic changes because the 3 ␮mol/L concen- the VHL-methylated UOK 121 lines were significantly smaller tration had no effect on the non-VHL–methylated 786–0 line. with 5-aza-dCyd treatment whereas the 786–0 lines were vir- Perhaps a most notable aspect of the role of the VHL gene tually unaffected suggests that the cytotoxic effects of 5-aza- as a tumor suppressor gene is its ability to inhibit VHL-negative dCyd had little role in the inhibition of tumor growth. cells from forming tumors in mice when these cells are trans- The array analysis was done to identify global changes in fected with the VHL gene before injection into the mouse. In this gene expression and categorize the major classes of genes with study we now show that re-expressing VHL in cells that have expression changes after 5-aza-dCyd exposure in our cell lines. established a tumor in a mouse can stop the growth of the tumor. Five of the genes listed in Table 1 that were confirmed by Mice were injected with the UOK 121 VHL-methylated line or RT-PCR, E-Cadherin, TFPI2, Stratifin, S100A4 and prostasin the 786–0 non-VHL–methylated controls. After small tumors are silenced by methylation in other solid tumors; one of these had grown, the mice were treated with 5-aza-dCyd as described genes, E-cadherin was previously shown to be methylated in the and the tumor size and weight at sacrifice were measured. We 786–0 line (34). In RCC a larger percentage of higher staged based our 5-aza-dCyd dosing in the mice on what had been used cancers had methylated E-cadherin compared with lower staged in previous studies and chose the lowest doses that had been tumors and the same is true for TFPI-2 in glioblastoma (31). The shown to be effective (52–54). We picked a low dose to mini- entire list of genes up- and down-regulated in our cell lines after mize the cytotoxic effects of 5-aza-dCyd on tumor growth. 5-aza-dCyd exposure is available as supplemental information. Whereas the 786–0 cells produced tumors of similar weight in The most notable finding from the microarray analysis may be

that many histone genes were found to be disproportionately 11. Jones PA. Altering gene expression with 5-azacytidine. Cell 1985; up-regulated by5-aza-dCyd. It has been described that 5-aza- 40:485–6. dCyd can modify histone protein methylation but not that the 12. Bender CM, Pao MM, Jones PA. Inhibition of DNA methylation by Ј levels of histone RNA message increase after 5-aza-dCyd ex- 5-aza-2 -deoxycytidine suppresses the growth of human tumor cell lines. Cancer Res 1998;58:95–101. posure (55). 5-aza-dCyd has been shown to decrease lysine-9 13. Braakhuis BJ, van Dongen GA, van Walsum M, Leyva A, Snow methylation whereas increasing lysine-4 methylation on histone GB. Preclinical antitumor activity of 5-aza-2Ј-deoxycytidine against 3. It is not known how 5-aza-dCyd modifies histone methylation human and neck cancer xenografts [corrected]. Investig New Drugs at present. Whether histone hypo-methylation by 5-aza-dCyd is 1988;6:299–304. related to the observed increase in histone message level is 14. Abele R, Clavel M, Dodion P, et al. The EORTC Early Clinical unknown. The significance of having a disproportionate up- Trials Cooperative Group experience with 5-aza-2Ј-deoxycytidine (NSC regulation of histone genes occurring across our cell lines after 127716) in patients with colo-rectal, head and neck, renal carcinomas and malignant melanomas. Eur J Cancer Clin Oncol 1987;23:1921–4. 5-aza-dCyd exposure will require additional investigation. 15. Goffin J, Eisenhauer E. DNA methyltransferase inhibitors-state of Clearly the interaction of promoter hypo-methylation and alter- the art. Ann Oncol 2002;13:1699–1716. ations in histone methylation is complex, and genotypic and 16. Cheng JC, Matsen CB, Gonzales FA, et al. Inhibition of DNA phenotypic changes occurring from this interaction is difficult to methylation and reactivation of silenced genes by zebularine. J Natl correlate. Large effects on gene expression might be attributable Cancer Inst (Bethesda) 2003;95:399–409. to the presence of multiple gene copies. However, of the cohort 17. Barchi J, Musser S, Marquez VE. The decomposition of 1-(beta-D- of histone genes listed in Table 2, only the TSPY gene is known ribofuranosyl)-1,2-dihydropyrimidin-2-one (zebularine) in alkali - mechanism and products. J Org Chem 1992;57:536–41. to exist in multiple copies with multiple CpG islands (56). 18. Kelley JA, Driscoll JS, McCormack JJ, Roth JS, Marquez VE. Therapy based on the re-expression of the VHL tumor Furanose-pyranose isomerization of reduced pyrimidine and cyclic urea suppressor gene in the tumor cells of patients that have lost VHL ribosides. J Med Chem 1986;29:2351–8. function is a logical hypothesis, if it could be accomplished in 19. Kuzmin I, Geil L, Ge H, et al. Analysis of aberrant methylation of vivo in an effective manner. The use of 5-aza-dCyd in the mice the VHL gene by transgenes, monochromosome transfer, and cell fu- is, to our knowledge, the first experimental test of such a model sion. Oncogene 1999;18:5672–9. in which the VHL gene is re-expressed in an established ccRCC 20. de Paulsen N, Brychzy A, Fournier MC, et al. Role of transforming Ϫ Ϫ VHL-negative tumor. VHL hyper-methylation occurs in up to growth factor-alpha in von Hippel–Lindau (VHL)( / ) clear cell renal carcinoma cell proliferation: a possible mechanism coupling VHL tumor 19% of patients with sporadic clear cell RCC, and the use of a suppressor inactivation and tumorigenesis. Proc Natl Acad Sci USA hypo-methylating agent in these patients affords us an unusual 2001;98:1387–92. opportunity to achieve VHL re-expression in the tumor cells. 21. Cheng JC, Weisenberger DJ, Gonzales FA, et al. Continuous zebularine treatment effectively sustains demethylation in human bladder cancer cells. Mol Cell Biol 2004;24:1270–8. REFERENCES 22. Maranchie JK, Vasselli JR, Riss J, Bonifacino JS, Linehan WM, 1. Gnarra JR, Tory K, Weng Y, et al. Mutations of the VHL tumor Klausner RD. The contribution of VHL substrate binding and HIF1- suppressor gene in renal carcinoma. Nat Genet 1994;7:85–90. alpha to the phenotype of VHL loss in renal cell carcinoma. Cancer Cell 2002;1:247–55. 2. Foster K, Prowse A, van den BA, et al. Somatic mutations of the von Hippel-Lindau disease tumour suppressor gene in non-familial clear cell 23. Yegnasubramanian S, Kowalski J, Gonzalgo ML, et al. Hyperm- renal carcinoma. Hum Mol Genet 1994;3:2169–73. ethylation of CpG islands in primary and metastatic human prostate cancer. Cancer Res 2004;64:1975–86. 3. Herman JG, Latif F, Weng Y, et al. Silencing of the VHL tumor- suppressor gene by DNA methylation in renal carcinoma. Proc Natl 24. Koochekpour S, Jeffers M, Wang PH, et al. The von Hippel-Lindau Acad Sci USA 1994;91:9700–4. tumor suppressor gene inhibits hepatocyte growth factor/scatter factor- 4. Latif F, Tory K, Gnarra J, et al. Identification of the von Hippel- induced invasion and branching morphogenesis in renal carcinoma cells. Lindau disease tumor suppressor gene. Science (Wash D C) 1993;260: Mol Cell Biol 1999;19:5902–12. 1317–20. 25. Lockhart DJ, Dong H, Byrne MC, et al. Expression monitoring by 5. Seizinger BR, Rouleau GA, Ozelius LJ, et al. Von Hippel-Lindau hybridization to high-density oligonucleotide arrays. Nat Biotechnol disease maps to the region of chromosome 3 associated with renal cell 1996;14:1675–80. carcinoma. Nature (Lond) 1988;332:268–9. 26. Golub TR, Slonim DK, Tamayo P, et al. Molecular classification of 6. Clifford SC, Prowse AH, Affara NA, Buys CH, Maher ER. Inacti- cancer: class discovery and class prediction by gene expression moni- vation of the von Hippel-Lindau (VHL) tumour suppressor gene and toring. Science (Wash D C) 1999;286:531–7. allelic losses at chromosome arm 3p in primary renal cell carcinoma: 27. Hosack DA, Dennis G, Jr., Sherman BT, Lane HC, Lempicki RA. evidence for a VHL-independent pathway in clear cell renal tumouri- Identifying biological themes within lists of genes with EASE. Genome genesis. Genes Chromosomes Cancer 1998;22:200–9. Biol 2003;4:R70. 7. Robertson KD. DNA methylation, methyltransferases, and cancer. 28. Tomayko MM, Reynolds CP. Determination of subcutaneous tumor Oncogene 2001;20:3139–55. size in athymic (nude) mice. Cancer Chemother Pharmacol 1989;24: 8. Santini V, Kantarjian HM, Issa JP. Changes in DNA methylation in 148–54. neoplasia: pathophysiology and therapeutic implications. Ann Intern 29. Dao D, Walsh CP, Yuan L, et al. Multipoint analysis of human Med 2001;134:573–86. chromosome 11p15/mouse distal chromosome 7: inclusion of H19/IGF2 9. Lubbert M. DNA methylation inhibitors in the treatment of leuke- in the minimal WT2 region, gene specificity of H19 silencing in Wilms’ mias, myelodysplastic syndromes and hemoglobinopathies: clinical re- tumorigenesis and methylation hyper-dependence of H19 imprinting. sults and possible mechanisms of action. Curr Top Microbiol Immunol Hum Mol Genet 1999;8:1337–52. 2000;249:135–64. 30. Hube F, Reverdiau P, Iochmann S, Rollin J, Cherpi-Antar C, Gruel 10. Pliml I, Sorm F. Synthesis of 2Ј-deoxy-D-ribofuranosyl-5-aza-cyto- Y. Transcriptional silencing of the TFPI-2 gene by promoter hyperm- sine. Coll Czech Chem Commun 1964;29:2576–8. ethylation in choriocarcinoma cells. Biol Chem 2003;384:1029–34.

31. Konduri SD, Srivenugopal KS, Yanamandra N, et al. Promoter tural resemblance of the pVHL/elongin BC/hCUL-2 complex with the methylation and silencing of the tissue factor pathway inhibitor-2 (TFPI- ubiquitin ligase complex SKP1/cullin/F-box protein. Proc Natl Acad Sci 2), a gene encoding an inhibitor of matrix metalloproteinases in human USA 1999;96:9533–8. glioma cells. Oncogene 2003;22:4509–16. 44. Kamura T, Koepp DM, Conrad MN, et al. Rbx1, a component of the 32. Lehmann U, Langer F, Feist H, Glockner S, Hasemeier B, Kreipe H. VHL tumor suppressor complex and SCF ubiquitin ligase. Science Quantitative assessment of promoter hypermethylation during breast (Wash D C) 1999;284:657–61. cancer development. Am J Pathol 2002;160:605–12. 45. Cockman ME, Masson N, Mole DR, et al. Hypoxia inducible 33. Lodygin D, Yazdi AS, Sander CA, Herzinger T, Hermeking H. factor-alpha binding and ubiquitylation by the von Hippel-Lindau tumor Analysis of 14-3-3sigma expression in hyperproliferative skin diseases suppressor protein. J Biol Chem 2000;275:25733–41. reveals selective loss associated with CpG-methylation in basal cell 46. Ohh M, Park CW, Ivan M, et al. Ubiquitination of hypoxia-induc- carcinoma. Oncogene 2003;22:5519–24. ible factor requires direct binding to the beta-domain of the von Hippel- 34. Nojima D, Nakajima K, Li LC, et al. CpG methylation of promoter Lindau protein. Nat Cell Biol 2000;2:423–7. region inactivates E-cadherin gene in renal cell carcinoma. Mol Car- 47. Kourembanas S, Hannan RL, Faller DV. Oxygen tension regulates cinog 2001;32:19–27. the expression of the platelet-derived growth factor-B chain gene in 35. Schwienbacher C, Angioni A, Scelfo R, et al. Abnormal RNA human endothelial cells. J Clin Investig 1990;86:670–4. expression of 11p15 imprinted genes and kidney developmental genes in 48. Siemeister G, Weindel K, Mohrs K, Barleon B, Martiny-Baron G, Wilms’ tumor. Cancer Res 2000;60:1521–5. Marme D. Reversion of deregulated expression of vascular endothelial 36. Wu AH, Low WC. Molecular cloning and identification of the growth factor in human renal carcinoma cells by von Hippel-Lindau human interleukin 13 alpha 2 receptor (IL-13Ra2) promoter. Neuro- tumor suppressor protein. Cancer Res 1996;56:2299–301. oncol 2003;5:179–87. 49. Iliopoulos O, Levy AP, Jiang C, Kaelin WG Jr, Goldberg MA. 37. Yu JX, Chao L, Ward DC, Chao J. Structure and chromosomal local- Negative regulation of hypoxia-inducible genes by the von Hippel- ization of the human prostasin (PRSS8) gene. Genomics 1996;32:334–40. Lindau protein. Proc Natl Acad Sci USA 1996;93:10595–9. 38. Gnarra JR, Zhou S, Merrill MJ, et al. Post-transcriptional regulation 50. Semenza G. Signal transduction to hypoxia-inducible factor 1. of vascular endothelial growth factor mRNA by the product of the VHL Biochem Pharmacol 2002;64:993–8. tumor suppressor gene. Proc Natl Acad Sci USA 1996;93:10589–94. 51. Karpf AR, Jones DA. Reactivating the expression of methylation 39. Iliopoulos O, Kibel A, Gray S, Kaelin WG, Jr. Tumour suppression silenced genes in human cancer. Oncogene 2002;21:5496–503. by the human von Hippel-Lindau gene product. Nat Med 1995;1:822–6. 52. Covey JM, Zaharko DS. Comparison of the in vitro cytotoxicity 40. Duan DR, Pause A, Burgess WH, et al. Inhibition of transcription (L1210) of 5-aza-2Ј-deoxycytidine with its therapeutic and toxic effects elongation by the VHL tumor suppressor protein. Science (Wash D in mice. Eur J Cancer Clin Oncol 1985;21:109–17. C)1995;269:1402–6. 53. Momparler RL. Molecular, cellular and animal pharmacology of 41. Kibel A, Iliopoulos O, DeCaprio JA, Kaelin WG, Jr. Binding of the 5-aza-2Ј-deoxycytidine. Pharmacol Ther 1985;30:287–99. von Hippel-Lindau tumor suppressor protein to Elongin B and C. 54. Vesely J, Cihak A. Incorporation of a potent antileukemic agent, Science (Wash D C) 1995;269:1444–6. 5-aza-2Ј-deoxycytidine, into DNA of cells from leukemic mice. Cancer 42. Pause A, Lee S, Worrell RA, et al. The von Hippel-Lindau tumor- Res 1977;37:3684–9. suppressor gene product forms a stable complex with human CUL-2, a 55. Kondo Y, Shen L, Issa JP. Critical role of histone methylation in member of the Cdc53 family of proteins. Proc Natl Acad Sci USA tumor suppressor gene silencing in colorectal cancer. Mol Cell Biol 1997;94:2156–61. 2003;23:206–15. 43. Pause A, Peterson B, Schaffar G, Stearman R, Klausner RD. Study- 56. Dechend F, Williams G, Skawran B, et al. TSPY variants in six loci ing interactions of four proteins in the yeast two-hybrid system: struc- on the human Y chromosome. Cytogenet Cell Genet 2000;91:67–71.

Downloaded from clincancerres.aacrjournals.org on September 25, 2021. © 2004 American Association for Cancer Research. The In vitro and In vivo Effects of Re-Expressing Methylated von Hippel-Lindau Tumor Suppressor Gene in Clear Cell Renal Carcinoma with 5-Aza-2 ′-deoxycytidine

Wade G. Alleman, Ray L. Tabios, Gadisetti V. R. Chandramouli, et al.

Clin Cancer Res 2004;10:7011-7021.

Updated version Access the most recent version of this article at: http://clincancerres.aacrjournals.org/content/10/20/7011

Cited articles This article cites 56 articles, 20 of which you can access for free at: http://clincancerres.aacrjournals.org/content/10/20/7011.full#ref-list-1

Citing articles This article has been cited by 2 HighWire-hosted articles. Access the articles at: http://clincancerres.aacrjournals.org/content/10/20/7011.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://clincancerres.aacrjournals.org/content/10/20/7011. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.