Multiple markers for melanoma progression regulated by DNA

Multiple markers for melanoma progression regulated by DNA methylation:insights from transcriptomic studies

William M.Gallagher�,y, Orla E.Bergin1,y,Mairin Raffertyy, Zoe D.Kellyy, Ilse-Maria Nolan,Edward J.P.Fox, Aedin C.Culhane3, Linda McArdle5,Mario F.Fraga

6, Linda Hughes

2, Caroline A.Currid,

Fiona O’Mahony, Aileen Byrne, Alison A.Murphy4,Catherine Moss4, Susan McDonnell2,Raymond L.Stallings5, Jane A.Plumb7, Manel Esteller5,Robert Brown7, Peter A.Dervan1 and David J.Easty1

Department of Pharmacology, 1Department of Pathology and 2Department ofChemical Engineering, 3Bioinformatics Unit and 4Transcriptomics Core,Centre for Molecular Medicine, Conway Institute of Biomolecular andBiomedical Research, University College Dublin, Ireland, 5National Centrefor Medical Genetics, Our Lady’s Hospital for Sick Children, Dublin, Ireland,6Cancer Epigenetics Laboratory, Spanish National Cancer Centre (CNIO),Madrid, Spain and 7Centre for Oncology and Applied Pharmacology, CancerResearch UK Beatson Laboratories, University of Glasgow, UK

�To whom correspondence and reprint requests should be addressedTel: þ353 1 7166743; Fax: þ353 1 2692749;Email: [email protected]

The incidence of melanoma is increasing rapidly, withadvanced lesions generally failing to respond to conven-tional chemotherapy. Here, we utilized DNA microarray-based gene expression profiling techniques to identifymolecular determinants of melanoma progression within aunique panel of isogenic human melanoma cell lines. Whena poorly tumorigenic cell line, derived from an earlymelanoma, was compared with two increasingly aggressivederivative cell lines, the expression of 66 genes was signi-ficantly changed. A similar pattern of differential geneexpression was found with an independently derived meta-static cell line. We further examined these melanomaprogression-associated genes via use of a tailored TaqManLow Density Array (LDA), representing the majority ofgenes within our cohort of interest. Considerable concord-ance was seen between the transcriptomic profiles deter-mined by DNA microarray and TaqMan LDA approaches.A range of novel markers were identified that correlatedhere with melanoma progression. Most notable was TSPY,a Y chromosome-specific gene that displayed extensivedown-regulation in expression between the parental andderivative cell lines. Examination of a putative CpG islandwithin the TSPY gene demonstrated that this regionwas hypermethylated in the derivative cell lines, as wellas metastatic melanomas from male patients. Moreover,treatment of the derivative cell lines with the DNA methyl-transferase inhibitor, 20-deoxy-5-azacytidine (DAC),restored expression of the TSPY gene to levels comparablewith that found in the parental cells. Additional DNAmicroarray studies uncovered a subset of 13 genes from

the above-mentioned 66 gene cohort that displayedre-activation of expression following DAC treatment,including TSPY, CYBA andMT2A. DAC suppressed tumorcell growth in vitro. Moreover, systemic treatment of micewith DAC attenuated growth of melanoma xenografts, withconsequent re-expression ofTSPYmRNA.Overall, our datasupport the hypothesis that multiple genes are targeted,either directly or indirectly, by DNA hypermethylationduring melanoma progression.

Introduction

The incidence of cutaneous melanoma is at epidemic propor-tions, with rates steadily rising in Western countries over thepast few decades (1,2). However, effective treatmentfor patients with advanced melanoma is currently unavailable.Moreover, the prognosis of such patients is poor, with a 10%survival rate after 5 years. Less than a decade ago, cutaneousmelanoma was described as a black tumor and a black box (3).While considerable insights have recently been made withrespect to mapping out central events in melanoma develop-ment, the molecular basis of tumor progression in this diseaseremains ill defined. One approach towards understandingmelanoma is to compare gene expression patterns betweenmelanocytic cells from different stages of tumor progression.In this context, DNA microarray-based gene expression pro-filing has had a far-reaching impact on the study of numeroustumor types (4), including melanoma (5). Using this approach,new subgroups of melanoma have been identified (6), alongwith a wealth of marker genes that correlate with melanomaprogression and drug response (7–13).Cutaneous melanoma is a pigmented, readily accessible

lesion that has been well defined in histopathological terms(3). Early radial growth phase (RGP) melanomas can invadeinto the epidermis and papillary dermis, but have no capacityfor metastasis; resection at this stage is almost completelycurative. A subsequent vertical growth phase (VGP) denotesa transition to a more aggressive stage, which is capable ofmetastasis. Changes in gene expression occurring at the RGP/VGP transition are, thus, of great interest. However, compar-ative transcriptomic studies have so far been hindered in thisarena, as paired RGP/VGP biopsies are not normally available(since resection of RGP melanoma is often curative and noVGP develops). Here, we utilized a unique isogenic cell linemodel series that allows us to circumvent the lack of availab-ility of such paired samples from the clinic.The parental cell line in the series, WM793, was originally

isolated from a superficial spreading melanoma (14). Thepatient concerned has had no re-occurrence of the disease todate, suggesting that these cells had low metastatic potential.Accordingly, WM793 cells displayed poor tumorigenicity innude mice (15). Notably, the WM793 cell line was used as thebasis for in vivo selection of several aggressive, tumorigenic

Abbreviations: 5mC, 5-methylcytosine; CGH, comparative genomichybridization; COBRA, combined restriction bisulfite restriction analysis;DAC, 20-deoxy-5-azacytidine; FISH, fluorescence in situ hybridization; LDA,low density array; RGP, radial growth phase; VGP, vertical growth phase.

yThese four authors contributed equally to this work.

Carcinogenesis vol.26 no.11 # Oxford University Press 2005; all rights reserved. 1856

Carcinogenesis vol.26 no.11 pp.1856–1867, 2005doi:10.1093/carcin/bgi152Advance Access publication June 15, 2005

Dow

nloaded from https://academ

ic.oup.com/carcin/article/26/11/1856/2476034 by guest on 04 D

ecember 2021

sublines (15). In this respect, the derivative cell lines, WM793-P1 and WM793-P2, were established after inoculation in thepresence of Matrigel (a reconstituted basement membraneextract). These isogenic sublines exhibited properties ofcells isolated from advanced VGP melanoma: they werehighly tumorigenic in nude mice and displayed multi-cytokineresistance in vitro. A further cell line (1205-Lu) was derivedindependently at the Wistar Institute from a lung metastasisof WM793 after subcutaneous injection into the tail veins ofimmunodeficient mice; these cells exhibited spontaneousmetastasis (16).In this study, a high-density oligonucleotide array-based

approach was employed to identify genes that varied inexpression level between the parental and derivative celllines. We hoped, in particular, to elucidate key moleculardeterminants of the RGP–VGP transition, as well asobtain additional mechanistic insights into melanomaprogression.

Materials and methods

Cell lines

Conditions for culture of the melanoma cell lines used here and their originalsources have been described previously (17,18). In brief, WM793 and 1205-Lucells were a gift from Prof. Meenhard Herlyn (Wistar Institute, Philadelphia),whereas WM793-P1 and WM793-P2 cells [both N1 sublines in the WM793derivative series; (19) Kobayashi] were a gift from Prof. Robert Kerbel(University of Toronto, Canada). Cells were maintained in DMEM withGlutaMAX (Invitrogen Ltd., Paisley, UK), supplemented with 10% foetalcalf serum, 100 U/ml penicillin, 100 mg/ml streptomycin and 4 mg/ml insulin(Sigma–Aldrich, Dublin, Ireland).

Melanoma biopsies

Previously extracted DNA from 20 anonymized metastatic melanomas waskindly provided by the Department of Dermatology, University of Glasgow,UK, according to standardized ethical procedures set out by the University ofGlasgow.

Nucleic acid extraction

Genomic DNA and total RNA were extracted from monolayer cells in cultureand melanoma biopsies using the QIAamp DNA Mini (Qiagen, West Sussex,UK) kit. Total RNA was extracted from monolayer cells using the Tri Reagent(Sigma, Dublin, Ireland) kit.

Cell growth, migration and invasion assays

Cell growth rates were determined using an alamarBlue assay (BiosourceInternational, California, USA), which recorded cell proliferation each day.To assess cell migration, confluent monolayers were scratch wounded with apipette tip, with movement across the scratch documented by phase contrastmicroscopy. For invasion assays, cells were seeded into Transwell filters (poresize 8 mm; Corning-Costar Corporation, Cambridge, Massachusetts, USA)coated by Matrigel (1 mg/ml; 50 mg total from Becton, Dickinson andCompany, New Jersey, USA), with fibronectin (5 mg/ml; Sigma) used as achemoattractant. Cells were allowed to invade for 5 h, followed by carefulaspiration of the medium and removal of the filters. Cells on topside of thefilter were gently removed with a cotton bud. The filter, with cells remainingon the bottom side, was immersed in Toluidene Blue staining solution (5% w/vtoluidene blue, 5% w/v sodium borate; Sigma) and left overnight at roomtemperature. The filters were then washed several times with water to removebackground staining. Each filter was incubated with 300 ml of extraction buffer(20 mM Tris, 0.2% w/v SDS; Sigma) and shaken for 30 min at room temper-ature. Aliquots of the extracted material were measured at 540 nm using aVICTOR2 plate reader.

Flow cytometric analysis

A total of 1� 106 cells were used for flow cytometric analysis of DNA content.Exponentially growing cells were isolated by trypsinization, washed with500 ml phosphate-buffered saline (PBS), and fixed with 500 ml of ice-cold100% ethanol. Fixed cells were centrifuged and resuspended in 125 ml ofRNase solution (1 mg/ml in 1.12% w/v of sodium citrate) and incubated at37�C for 15 min. Following this, cells were incubated with 125 ml of prop-idium iodide solution (0.5 mg/ml in 1.12% of sodium citrate) for 30 min atroom temperature. Samples were then analysed on a Coulter Epics flowcytometer (Beckman Coulter (UK) Ltd. Buckinghamshire, UK).

CGH and FISH Analysis

Comparative genomic hybridization (CGH) experiments were carried outas previously described (19). Briefly, test and reference DNA sampleswere labeled by nick translation with spectrum green-dUTP and red-dUTP,respectively, under conditions recommended by the supplier (Vysis Inc.,Illinois, USA). Labeled test (melanoma cells) and reference (normal lympho-cyte) DNA (500 ng) were then denatured and hybridized to normal humanmetaphase chromosomes in a solution containing 50� Cot1 fractionated DNA,50% formamide, 1� SSC, and 10% dextran sulfate (Vysis). Images wereacquired and analyzed using hardware and software from Applied ImagingCorporation, California, USA. For fluorescence in situ hybridization (FISH)experiments, a dual-colored fluorescently labeled probe to specific regions ofchromosomes X (Spectrum Green; DXZ1) and Y (Spectrum Red; DYZ1) wasused according to the manufacturer’s instructions (Vysis). A probe to 11q(bacterial artificial chromosome clone RP11-163A13 from the Sanger Insti-tute, UK) was also used to determine ploidy.

DAC treatment in vitro

Seeded cells were treated with 2 mg/ml 20-deoxy-5-azacytidine (DAC) on days1, 3 and 5, with fresh drug-containing medium being added at each timepoint(20). On days 2 and 4, drug-containing medium was exchanged for drug-freemedium. On day 6, cells were harvested and total RNA extracted as above.

Global DNA methylation analysis

5-Methylcytosine (5mC) genomic content was determined by high-performance capillary electrophoresis, as previously described (21). Briefly,genomic DNA samples were boiled, treated with nuclease P1 (Sigma) for 16 hat 37�C, and with alkaline phosphatase (Sigma) for an additional 2 h at 37�C.After hydrolysis, total cytosine and 5mC content were measured by capillaryelectrophoresis using a P/ACE MDQ system (Beckman Coulter). Relative5mC content was expressed as a percentage of total cytosine content (methyl-ated and non-methylated).

DNA microarray analysis

Ten micrograms of total RNA from each cell line was reverse transcribed intosingle-stranded cDNA using the SuperScript Choice kit (Invitrogen). For thispurpose, an oligo-dT primer containing a T7 RNA polymerase promoter(Genset Corporation, California, USA) was utilized. Following double-stranded cDNA synthesis, biotin-labelled cRNA was generated by in vitrotranscription using the BioArray RNA labelling kit (Enzo Life Sciences Inc.,New York, USA). These complex cRNA targets, which are representative ofthe transcriptome of a particular sample, were hybridized against HuGeneFLarrays (7129 probe sets). Detection was accomplished via a streptavidin-labelled fluorochrome (phycoerythrin) and laser scanning. Normalisation ofdata and inter-array comparisons of gene expression profiles was carried outusing Microarray Analysis Suite (MAS) software v4.0 (Affymetrix, HighWycombe, UK), together with Microsoft Access. In more detail, DNAmicroarray experiments were analyzed using an approach based on theMann–Whitney pairwise comparison test (22). To identify differentiallyexpressed genes between any two samples, pairwise comparisons were per-formed using MAS. Lists of altered transcripts from different pairwise com-parisons were sorted via Microsoft Access. Additional bioinformatic analysiswas completed using publicly available annotation databases and softwaretools, notably TIGR Multiple Experiment Viewer v2.0. Available upstreamgenomic sequences of identified differentially expressed genes were automat-ically retrieved using either EZRetrieve or ENSEMBL. Putative promoter-associated CpG islands at or around presumed transcription start site wereidentified using CpGPlot. The raw DNA microarray data have been submittedfor public access to Gene Expression Omnibus (http://www.ncbi.nlm.nih.gov/geo/) and can be obtained using the following accession numbers: GSE1792and GSE1793.

RT–PCR analysis via cycle limitation

RNA extracts were pre-digested with DNase I prior to cDNA synthesis usingthe DNA-Free kit (Ambion (Europe) Ltd., Cambridgeshire, UK). Single-stranded cDNA was synthesized from 1 mg total RNA using the ImProm-IIReverse Transcription kit (Promega, Southampton, UK). Recombinant RNasinRibonuclease Inhibitor (20 U/20 ml reaction; Promega) was added to preventRNase-mediated degradation. Two negative controls were also utilised,namely minus reverse transcriptase (RT) enzyme control and minus templatecontrol. Following inactivation of RT at 70�C for 15 min, aliquots of generatedsingle-stranded cDNA were subjected to PCR amplification via a cycle lim-itation approach. The following primer pair combinations were used: 18 S—forward, 50-AGGGTTCGATTCCGGAG-30 and reverse, 50-ACCAGACTTG-CCCTCC-30 (195 bp amplicon);DCT—forward, 50-AGTGATTCGGCAGAA-CATCC-30 and reverse, 50-AGTTCCAGTAGGGCAAAGCA-30 (368 bpamplicon); GPNMB—forward, 50-TGCATAAAGCCAATGTAGTCCA-30

and reverse, 50-CAGGGACCTCATCTTTGGAA-30 (373 bp amplicon);

Epigenetic control of melanoma progression

1857

Dow



ecember 2021

http://www.ncbi.nlm.nih.gov/

TYR—forward, 50-CAGCTTTCAGGCAGAGGTTC-30 and reverse, 50-GCTTCATGGGCAAAATCAAT-30 (470 bp amplicon); TYRP1—forward,50-CCGAAACACAGTGGAAGGTT-30 and reverse, 50-ACCAGTAA--ACCAGTAACAAAGCGCCAAC-30 (406 bp amplicon); CYBA—forward,50-TTTGTGTGCCTGCTGGAGTA-30 and reverse, 50-CTCCTCGCTGGGC-TTCTT-30 (369 bp amplicon); MX1—forward, 50-AGCCACTGGACTGAC-GACTT-30 and reverse, 50-ACGGCACTCATGCTCCTAAA-30 (335 bpamplicon); HOXD4—forward, 50-TGACTCGCCAAGATTTTATGT-30 andreverse, 50-CACCTCGCTGGGCTCTAA-30 (190 bp amplicon); RARB—forward, 50-ACAAGGTCAAAGGAGGCAGA-30 and reverse, 50-TTC-ACAAGCCATTAGGGAAA-30 (188 bp amplicon); SIAT7B—forward,50-GGCACATCCTACCCCAGA-30 and reverse, 50-AAGCAACTAACCCC-CATCAA-30 (272 bp amplicon); PRK1—forward, 50-GGGCTGTTTCTTCA-CATCTTC-30 and reverse, 50-GTGGACTGGTGGGGACTG-30 (231 bpamplicon); TSPY—forward, 50-CACCACAACAGCAGCCTTA-30 andreverse, 50-TGCTCCATCATATTCAACTCA-30 (193 bp amplicon). PCRproducts were subcloned into the pCRII-TOPO vector via the Topo TACloning kit (Invitrogen), with insert-containing plasmids then subjected toautomated DNA sequencing via a commercial route (MWG-Biotech AG,Ebersberg, Germany).

Real-time RT–PCR analysis using TaqMan Low Density Arrays

Pre-designed TaqMan probe and primer sets for target genes were chosen froman on-line catalogue (Applied Biosystems). Once selected, the sets werefactory-loaded into the 384 wells of TaqMan Low Density Arrays (LDAs).Array format was customized on-line with two replicates per target gene.Expression levels of target genes were normalized to concentration of 18SrRNA. Samples were analyzed using the 7900HT system with a TaqMan LDAUpgrade (Applied Biosystems), according to the manufacturer’s instructions.In short, 2.5 ml of single-stranded cDNA (to final concentration of 100 ngstarting RNA; see above) was combined with 47.5 ml water and 50 ml TaqManUniversal PCR Master Mix, following by loading of 100 ml sample per port.Thermal cycling conditions were as follows: 50�C for 2 min, 94�C for 10 min,97�C for 30 s, and 59.7�C for 1 min. Gene expression values were calculatedbased on the DDCt method, where one sample was designated the calibrator,through which all other samples were analyzed. Briefly, DCt represents thethreshold cycle (Ct) of the target minus that of 18S rRNA and DDCt representsthe DCt of each target minus that of the calibrator. Relative quantities weredetermined using the equation; relative quantity ¼ 2�DDCt. For the calibratorsample (i.e. WM793 cells), the equation is relative quantity ¼ 2�0, which is 1;therefore, every other sample is expressed relative to this.

Northern analysis

Eight micrograms of total RNA from each cell line was subjected to electro-phoresis through an agarose–formaldehyde gel, followed by transfer to nitro-cellulose membranes (17). A TSPY-specific cDNA probe, derived from theabove-mentioned RT–PCR amplicon, was labeled by random hexanucleotidepriming, with hybridization conditions as described previously (17). Integrityand loading of RNA were determined by probing for GAPDH expression.

Southern blot analysis

EcoRI-digested DNA (15 mg) was subjected to agarose gel electrophoresis,followed by transfer to nylon membrane, as described previously (23). Use ofthe radiolabeled TSPY-specific cDNA probe was as detailed above. Loading ofDNA was determined by ethidium bromide staining.

COBRA

For combined restriction bisulfite restriction analysis (COBRA), genomicDNA was first subjected to bisulfite modification via the CpGenome DNAModification kit (Intergen, New York, USA); this process converts unmethyl-ated, but not methylated, cytosine residues to uracil. The genomic DNAsequence of the TSPY gene was retrieved from GenBank (accession number:M98524). A putative CpG island within the first exon of the TSPY gene wasidentified using CpGPlot (Figure 4C). The following primer pairs (designedagainst bisulfite-modified DNA) were used to amplify this region by PCR:forward, 50-GGTAGTATAGGTTTTGGTGTGTG-30 and reverse, 50-CCAACACCTCCTCCAATACAAAC-30. Amplified PCR products (269 bpin length) were incubated with BsiEI (20 U/20ml PCR product) for 2 h at 60�C,with restriction digests subsequently examined by agarose gel electrophoresis.In addition, COBRA was also performed on 100% methylated DNA, namelyCpGenome Universal Methylated DNA (Chemicon Europe Ltd., Hampshire,UK), which acted as a positive control.

Human tumor xenografts

Animal studies were carried out under an appropriate United Kingdom HomeOffice Project Licence. All work conformed to the UKCCR guidelines for thewelfare of animals in experimental neoplasia studies and was approved by theUniversity of Glasgow Ethics Committee. Approximately 1 � 106 cells in a

volume of 100 ml PBS were injected subcutaneously into the left or rightflank of athymic female CD-1 nude mice (Charles River Laboratories Inc.,Wilmington, Massachusetts, USA). After 19 days, when the mean tumorvolume was at least 0.1 cm3, the mice were weighed daily and tumor volumeswere estimated by two calliper measurements assuming spherical geometry(volume ¼ d3 � p/6). Seven days later, when the mean tumor volume wasat least 0.25 cm3, DAC (5 mg/kg/mouse) was administered by threeintraperitoneal injections at 3 h intervals over the course of a day (total dose,15 mg/kg/mouse). The control mice were injected with PBS where 10 ml ofPBS was given per gram of mouse bodyweight.

Results

Phenotype of cells in vitro

The behavior of the WM793-based isogenic cell line modelseries has been well documented in vivo (15,16). However,there are only limited data available with respect to character-istic features of these cell lines in vitro (15,24). The growthrate of the parental WM793 cells and three isogenic deriva-tives was examined over a period of 7 days (Figure 1A). Thederivative cell lines exhibited more rapid rates of cell growththan the parental cells. Interestingly, the 1205-Lu cells showedan intermediate rate of cell growth, which may be due to theapparent increased propensity of this cell type to detach fromthe surface in monolayer culture. The isogenic derivatives alsodisplayed increased invasive capacity over WM793 cells, withthe most striking difference seen between 1205-Lu andparental cells (Figure 1B).

A

B

0 50 100 150 2000

Hours

Gro

wth

Rat

e (A

U)

1 x 105

2 x 105

WM793WM793-P1WM793-P21205-Lu

*

P<0.05

Inva

sion

(Abs

540n

m)

WM793 WM793-P1 WM793-P2 1205-Lu

WM793

WM793-P1

WM793-P2

1205-Lu

0.0

0.2

0.4

0.6

0.8

1.0

1.2

Fig. 1. Phenotypic characteristics of melanoma cells in vitro. (A) Growth ofparental WM793 cells and isogenic derivatives (WM793-P1, WM793-P2and 1205-Lu). Cells (20 000/well) were seeded into 12-well tissue cultureplates and left to grow for 7 days. Growth rates were measured using analamarBlue assay for cell proliferation. AU, arbitrary units. (B) Invasivecapacity of melanoma cells. Invaded cells were stained with toluidene blue,which was extracted and evaluated at Abs540 nm. P-value obtained using aStudent t-test. In both A and B, error bars refer to the SEM of triplicatedeterminations.

W.M.Gallagher et al.

1858

Dow



ecember 2021

Genomic aberrations

A range of chromosomal abnormalities was observed in boththe parental and derivative cells (Table I). The four cell linesexhibited some identical or similar abnormalities, illustrativeof the underlying genetic relatedness of the series. Moreover,the particular generic abnormalities found match those com-monly present in melanoma biopsies (3,25,26). For example,there is a consistent gain of material on chromosomes 1, 6 and7, where gene amplification frequently occurs in melanoma. Inaddition, certain progression-related abnormalities were iden-tified, when the parental cells were compared with its variousderivative cell types. Crucially, the two main derivative types,WM793-P1/WM793-P2 and 1205-Lu, exhibit some commonalterations in this context (e.g. loss of 12q10q13 and 18p),supporting the concept of non-random changes. Overall,there is a marked increase in the number of abnormalitiesobserved in the derivative cells as compared with the parentalcell line. FISH and flow cytometric analysis showed evidenceof increased ploidy in association with progression in themodel system, which is further indicative of genomic instabil-ity (see Supplementary data, S1). As compared withWM793 cells, for example, FISH analysis showed increasesin X, Y and 11q chromosomal regions in the derivative cellsthat were consistent with an increase in the numbers of tetra-ploid, triploid or hyperdiploid cells found in these cell linepopulations.

Identification of differentially expressed genes

All four melanoma cell lines were subjected to gene expressionprofile analysis using high-density oligonucleotide arrays. Inthis case, total RNA from exponentially growing cultures wasused. When the parental cell line in the series, WM793, wascompared with the two increasingly aggressive derivative celllines, WM793-P1 and WM793-P2, 66 genes were commonlyaltered with respect to expression level (Figure 2A). Withinthis cohort, 44 genes were identified as being down-regulated.A similar pattern of differential gene expression was foundwith the independently derived metastatic cell line, 1205-Lu(Figure 2B). Altered expression of a subset of 10 genes wasconfirmed by conventional RT–PCR analysis (Figure 3A),with the results closely matching that obtained from the initialDNA microarray screen (Figure 2A). We also examined theexpression of 45 out of the 66 genes via a novel high-throughput quantitative RT–PCR assay (Figure 3B), whichemployed the use of a specifically designed TaqMan LDA. TheTaqMan LDA approach showed a high level of reproducibility(see Supplementary data, S2A). Moreover, considerable

concordance was seen between the transcriptomic profilesdetermined by DNAmicroarray and TaqMan LDA approaches(see Supplementary data, S2B).A considerable proportion of the identified differentially

expressed genes have been previously associated withmelanoma development and progression. Reduced expressionof the tumor suppressor genes, CDKN2A and IL-24, wasobserved in all three derivative cell lines as compared withthe parental cells. Moreover, several genes (TYR, TYRP1 andTYRP2) involved in melanin biosynthesis exhibited a similarmarked reduction in transcript levels in the derivative celltypes. A subset of immune-related genes (including BST2,C1S, C1R, HLA-DQA1, TNFAIP6 and LGALS3) andinterferon-related genes (including GIP1, GIP3, IFIT1,CASP1, ISGF3G, DAP and MX1) were also observed to bedown-regulated in the derivative cells. Several genes encodingfor tumor-associated antigens, such as MAGEA4 and GAGE1,displayed increased expression in the derivative cell lines.Intriguingly, the AIM1 gene, which has been previously sug-gested as a tumor suppressor based on correlation in expres-sion terms with experimental reversal of tumorigenicity viachromosome transfer, showed evidence of increased expres-sion at the RNA level in the derivative cells.In addition, a range of novel markers were identified that

correlated with melanoma progression. Most notable wasTSPY, a Y chromosome-specific gene that displayed markeddown-regulation in expression (between 137- and 317-fold, asdetermined from DNA microarray study) between the parentaland derivative cell lines (Figures 2, 3 and 4A). The TSPY genehas previously been shown to exhibit dysregulated expressionin a number of cancer types, including gonadoblastoma, aswell as testicular and prostate cancer (20,27–30). Although Ychromosome loss has been described for certain melanomas,this is not a common event. FISH analysis showed retention ofthis chromosome in all four cell lines under study (see Sup-plementary data, S1A). Moreover, Southern blot analysisshowed no evidence for deletions or gross rearrangement ofthe TSPY gene (data not shown). TSPY gene expressionis regulated by androgens and DNA methylation (20).This suggested that aberrant DNA methylation may have arole transcriptional silencing of TSPY gene expression betweenearly and advanced melanoma cell lines.

Regulation of gene expression by DNA methylation

Treatment of the derivative cell lines with DAC restoredexpression of the TSPY gene to varying degrees (Figure 4B).We then employed COBRA to determine whether the TSPYgene was directly methylated (Figure 4C–E). COBRAutilizes bisulfite treatment of the DNA (which converts non-methylated cytosines in CpG sites to uracil), together withPCR amplification and restriction enzyme analysis. In thiscase, the enzyme site lies on a predicted methylated cytosine.As a result, following initial bisulfite modification of DNA,restriction sites with unmethylated CpGs will be protectedfrom digestion, whereas those with methylated CpGs will beavailable for cleavage. Examination of a putative CpG islandwithin the TSPY gene demonstrated that this region washypermethylated in all three derivative cell lines (Figure 4Cand D). In addition, hypermethylation of the TSPY gene wasobserved in metastatic melanomas from five male patients(Figure 4E).Further DNA microarray studies uncovered a subset of 13

out of the 44 (29.5%) down-regulated genes that displayed

Table I. CGH analysis of melanoma cells

Cell line Genomic aberrationsa

WM793 enh 1q10qter, 2q10q35, 6q22q27, 7, 8q, 20q dim 10q24WM793-P1 enh 1q10q32, 2q10q35, 5p, 6q22q27, 7, 8q22, 13q32,

17q21q23, 20q dim 10q24, 12q10q13, 18pWM793-P2 enh 1q10q32, 2q10q35, 6q22q27, 6q10q16, 7, 13q32,

17q21q23, 20q dim 10q24, 12q10q13, 18, 18p1205-Lu enh 1q, 2, 5p, 7, 7pþ, 8q, 11p, 12p, 13, 17p13 dim 10q24,

12q10q13, 18, 18p

aenh, enhanced green to red fluorescent ratio of chromosomal region(gain); dim, diminished green to red fluorescent ratio of chromosomalregion (loss). Bold text refers to melanoma progression-relateddifferences, i.e. variation between parental and derivative cells.


1859

Dow



ecember 2021

Fig. 2. Comparative analysis of gene expression profiles. The 66 genes identified as consistently differentially expressed between WM793 cells and thederivative cell lines, WM793-P1 and WM793-P2, are listed above. (A) Duplicate data with respect to gene expression obtained from WM793, WM793-P1 andWM793-P2 cell lines (separate cultures; replicate data listed beside each other). Pairwise comparison of gene expression between different cell lines yielded thefollowing numbers of altered transcripts: WM793 versus WM793-P1 (129 transcripts); WM793 versus WM793-P2 (114 transcripts); WM793-P1 versusWM793-P2 (14 transcripts). Cross-comparison of the 129 and 114 transcript lists yielded 68 commonly altered transcripts, of which two (TAC1 and TYRP1) arerepresented by an additional probe set. On average, inter-array variability between biological replicates was observed to be 2.18% (155/7129 probe sets).(B) Independent dataset showing expression of 66 gene cohort in untreated (WM793 and 1205-Lu) and DAC treated (all four cell lines) cells. Gene expressionprofile information is represented using a color-coded scheme (key provided) in which light blue refers to genes expressed at a low level (below meanabsolute intensity) and bright red refers to genes expressed at a high level (above mean absolute intensity). From left to right, the tabular columns refer to theAffymetrix probe set identifier, the corresponding UniGene cluster (Build #166), associated chromosomal location and Human Genome Organisation-approvedgene symbol for each transcript. The additional three columns detail the fold change values between WM793 and respective derivative cells. With respect to thelast column: þ, CpG island found around presumed transcription start site or near upstream region; �, no CpG island found in these regions; NA, upstreamsequence is unknown. Black and red crosses signify that upstream genomic sequence obtained using EZRetrieve and ENSEMBL, respectively.


1860

Dow



ecember 2021

consistent re-activation of expression following DACtreatment across all three derivative cell lines, includingTSPY, CYBA and MT2A (Figure 2B and Table II). This is incontrast to only 3.24% of all transcripts represented on theDNA microarray that exhibited elevated expression in all threederivative cell lines following DAC treatment. However, 7 outof the 13 genes do not have 50 CpG islands (Figure 2), soare not likely to be directly regulated by DNA methylation.This phenomenon has been reported previously by Liang et al.(31) among others. In summary, multiple transcripts that arepotential markers for melanoma progression can be increasedfollowing DAC treatment suggesting that the relevant genesare suppressed in terms of expression by DNA methylation,either directly or indirectly, in the more aggressive derivativecell lines.

DAC-mediated inhibition of cell growth and migration in vitro

The growth rate of all four melanoma cell lines was examinedover a period of 7 days under DAC treated conditions(Figure 5A). As compared with untreated cells, the growth ofboth the parental and derivative cell lines was suppressed inresponse to DAC (Figure 5A), albeit with the latter cell typesshowing a slightly delayed response in this respect. The DAC-mediated inhibitory effect on cell growth was most marked inthe case of the 1205-Lu cells. The migration capacity of allfour melanoma cell lines was determined by a scratch woundhealing assay under both untreated and DAC treated condi-tions (Figure 5B). For this, the cells were grown to confluency,scratch wounded with a pipette tip and incubated for a further48 h to allow the cells to migrate into the scratch. Underuntreated conditions, the derivative cell lines exhibited morerapid rates of cell migration than the parental cells. Forexample, wounded areas were mostly fully replenished within48 h in the case of the derivative cells (Figure 5B), with thisprocess taking a further 48 h for the parental cells to complete(data not shown). Under DAC treated conditions, the migrationof both the parental and derivative cell lines was suppressed(Figure 5B). Overall, these data suggest that alteration of DNAmethylation has a significant effect on cellular phenotypein our model system. However, one cannot exclude the pos-sibility that the DAC-mediated inhibition of cell growth mayalso be, at least in part, due to a direct cytotoxic action, whichis independent of DNA methylation. Indeed, we readily

Fig. 3. Validation of DNA microarray results by two RT–PCR analysismethods. The vertical blue and red bars refer to genes that were identifiedfrom the prior DNA microarray study as either down-regulated or up-regulated, respectively, in derivative cells as compared with parentalWM793 cells. Genes listed according to same order shown in Figure 2. (A)RT–PCR analysis, via a cycle limitation method, of transcripts previouslyshown to be down-regulated (CYBA, GPNMB, DCT, TYRP1, TYR, MX1) orup-regulated (HOXD4, PRK1, SIAT7B, RARB) in the derivative cell lines.Level of 18S rRNA served as a loading control. (B) Real-time RT–PCRanalysis via TaqMan LDA. Forty-five of the sixty-six differentially expressedgenes were assessed for expression level in the WM793 series. Expressionmeasurements were normalized to 18S rRNA levels. The parental WM793cells were defined as the calibrator to which all three derivative cell lineswere compared. The three data columns to the right represent the meannormalized relative quantities of target gene expression between WM793cells and relevant derivative cells across duplicate measurements fromindependent stocks of exponentially growing cells. Gene expression profileinformation is represented to the right using a color-coded scheme (keyprovided below) in which light blue and bright red refer to down-regulatedand up-regulated genes (as compared withWM793 cells), respectively. Blackindicates no change in expression level.


1861

Dow



ecember 2021

envisage that the suppressive effects on cell growth in vitromay in part be due to induction of apoptosis.

Suppression of tumor growth in vivo following DAC treatment

To further examine the functional role of DNA methylation inregulating tumor growth of an aggressive derivative, weexamined the effect of systemic DAC treatment on 1205-Luxenografts in vivo (Figure 6A). Systemic DAC treatment(15 mg/kg) significantly attenuated tumor growth in compar-ison to untreated mice. Again, one cannot preclude the

possibility that a direct cytotoxic action by DAC on tumorgrowth also plays a part in this effect. The inhibitory effectof DAC on tumor growth was seen to cease 7 days post-treatment. This is consistent with the reversible nature ofdemethylation induced by DAC in xenograft models (32).Intriguingly, while the DAC treated tumors display renewedgrowth capacity following withdrawal of the demethylatingagent, re-expression of TSPY mRNA was seen to be main-tained in 3 out of 4 explanted tumors (Figure 6B), suggesting apotential use of this gene as a residual biomarker for DAC

269 bp191 bp

WM

793

WM

793-P1

WM

793-P2

1205-Lu

TSPY

GAPDHA

B

C

D

E

269 bp

269 bp

TSPY

1 2 3 4 5 6 7 8 9 10

11 12 13 14 15 16 17 18 19 20

191 bp

191 bp

TSPY

Exon 1TSPY

100% Control

- DAC

+ DACTSPY

H2OPlasmid Control

18S- DAC

+ DAC

P1

-

191 bp

191 bp

Exon 1TSPY

Fig. 4. Regulation of TSPY gene by DNA methylation. (A) Northern blot analysis of TSPY mRNA expression. Order of parental WM793 cells andderivatives indicated. (B) Effect of DAC treatment on TSPY gene expression in melanoma cell lines. Total RNAwas extracted from untreated (�DAC) and treated(þDAC) cells and subjected to RT–PCR analysis for TSPY and 18S rRNA via a cycle limitation method. TSPY cDNA-containing plasmid and H2O alone controlswere also included as positive and negative controls, respectively. The efficiency of DAC treatment was assessed by global DNA methylation analysis (seeSupplemental data, S3). Same order of samples as in A. (C) Schematic representation of 50 region of TSPY gene, including Exon 1 (black bar). Amplicon assessedby COBRA also indicated (light bar), along with relevant restriction enzyme (BsiEI) cleavage site. Vertical bars indicate CpG sites. (D) COBRA of TSPY geneusing genomic DNA extracted from melanoma cell lines. (E) COBRA of TSPY gene using genomic DNA extracted from metastatic melanomas. Presence of191 bp fragment signifies DNA methylation at CpG within BsiEI site. COBRA was also performed on commercially available 100% methylated DNA (as apositive control). For D, same order of samples as in A. For E, genomic DNA from 20 randomly assigned metastatic melanomas (5 male, 15 female). Samples inlanes 1, 2, 7, 8 and 13 were derived from male patients, with the remainder being from female patients. Only male-derived samples generated PCR products byCOBRA, which is as expected given the chromosomal localization of the TSPY gene.


1862

Dow



ecember 2021

activity. In summary, DAC treatment may facilitate tumori-genic reversion of advanced melanomas.

Discussion

The development of culture techniques permitting the estab-lishment and propagation of cells derived from histologicallydefined stages of melanoma progression has permitted majoradvances. However, comparative studies of RGP and VGPmelanoma have been limited since it has not generally provedpossible to establish cell lines from each of these stages fromone individual patient. Here, we have used a unique isogeniccell line-based model system that circumvents this problem.We compared the gene expression profile of an early melan-

oma cell type with a variety of isogenic derivative cell linesof increasing aggressiveness. Accordingly, the expressionpatterns of 66 genes were identified as correlating with melan-oma progression. Amongst these, we found a large numberof genes previously associated with melanoma, includingseveral tumor suppressor genes and antigenic markers, aswell as genes involved in melanin biosynthesis. Thesecommonly encountered alterations, together with the notedsimilarities in terms of chromosomal aberrations observedbetween the cell lines used here and melanoma biopsies,adds confidence to our current model of melanoma progres-sion. However, further work at the RNA level with melanomabiopsy samples would shed additional light on this collectionof putative melanoma progression-associated markers. Inthis respect, our specially designed high-throughput quantitat-ive RT-PCR assay, afforded by the use of TaqMan micro-fluidic cards, may facilitate fast-track validation of thesemarkers.Of the 66 genes examined, TSPY displayed the most striking

change in gene expression terms within the WM793 series.The TSPY gene is found in multiple copies (20–40 based oncurrent predictions) on both the long and short arm of chro-mosome Y (30,33). The TSPY gene is normally expressed inthe germ cells of the testis and distinct subsets of spermatogo-nia (34,35). Apart from an assumed activity in spermatogen-esis (34,36), the functional role of TSPY remains to beelucidated (33). Previous work had implicated TSPY as aputative oncogene based on its elevated expression in somegonadoblastomas, as well as testicular and prostate cancers(27–30,37). This contrasts with our observation of extensive

down-regulation of TSPY gene expression during melanomaprogression. In addition, Dasari et al. (20) found TSPY geneexpression to be suppressed in certain prostate cancer cells.While this study highlights the TSPY gene as a promisingmarker for melanoma progression and, potentially, DAC activ-ity, further work will be required to clarify its role in cancer. Inparticular, additional studies employing forward and reversegenetics approaches are necessary to determine if TSPY iscritical in modulating tumor progression in WM793 series,with these studies ongoing in our laboratory at present. Ourdata, however, also provide an intriguing hypothesis that theremay be sex-specific markers of melanoma, which may beuseful in discriminating differences in terms of disease pro-gression between males and females (1,2), with the TSPY genebeing a promising candidate in this respect.Detailed examination of the 66 gene cohort pointed towards

DNA methylation as having a potentially important role inmediating gene expression alterations between parental andderivative cell lines. Demonstration of hypermethylation at acandidate CpG island within the TSPY gene in the derivativecell lines provided initial support for this concept. Thiswas further backed by the noted de-repression of a significantproportion of down-regulated genes following DAC treatmentin vitro. Overall, our data support the hypothesis that multiplegenes are targeted, either directly or indirectly, by DNA hyper-methylation in this melanoma model system.The role of DNA methylation in carcinogenesis is complex:

global hypomethylation and region-specific hypermethylationco-exist (38). DNA hypermethylation at CpG islands is knownto be associated with epigenetic silencing of tumor suppressorgene expression and may increase genomic instability (39).DNA hypermethylation can occur at all stages of tumor devel-opment and progression (40). Previous studies have identifieda number of genes that are affected by alterations in DNAmethylation patterns in melanoma cells, including CDKN2A(41), PTEN (42), APAF-1 (43), MAGEA1 (44), TIMP3 (13),GAGED2 (45), various human leukocyte class I antigens (46),and CASP8 (47). While our data add a further collectionof putative methylation-sensitive genes in melanoma, addi-tional work will be required to ascertain which of the remain-ing 12 genes, apart from TSPY, are directly regulated by DNAmethylation.DNA microarray-based gene expression profiling technol-

ogy has been previously utilized in several cancer-related

Table II. Melanoma progression-related genes de-repressed by DAC treatmenta

P1b versus 793 P2 versus 793 Lu versus 793 793-DAC versus 793 P1-DAC versus 793 P2-DAC versus 793 Lu-DAC versus 793

TSPY �316.9 �316.8 �137.0 1.1 �7.0 �13.9 �13.0CYBA �101.0 �102.0 �73.0 1.9 �39.4 �14.9 1.0MT2A �48.2 �51.4 �6.0 2.5 �2.3 1.0 1.0BST2 �35.7 �40.9 �23.0 2.1 �2.8 �3.2 �4.0GIP3 �26.6 �15.9 �2.6 2.5 1.0 �1.6 1.4S100A1 �14.2 �16.2 �4.6 2.0 �1.4 1.0 2.5IFIT1 �11.7 �8.1 �13.9 9.2 �1.4 1.0 1.0MX1 �9.3 �7.4 �8.6 4.9 1.0 �1.6 4.0RGS3 �3.7 �4.1 �4.9 �1.9 �2.6 �3.2 �4.3ISGF3G �3.3 �3.0 �1.9 1.2 �1.4 1.0 1.0APOD �3.3 �3.7 �2.1 2.1 1.0 1.0 7.0RPL37A �2.3 �2.1 �1.7 �1.6 1.0 1.0 1.0HSPB1 �2.3 �2.4 �3.2 2.0 �1.5 1.3 1.0

aFold difference in gene expression indicated relative to untreated WM793 cells.bP1, WM793-P1; P2, WM793-P2; Lu, 1205-Lu; �DAC, DAC treated cells.


1863

Dow



ecember 2021

model systems to identify genes that are regulated by DNAmethylation (13,44,48,49). Notably, van der Velden et al. (13)identified 19 genes, including TIMP3 and TYRP1, that weredifferentially expressed between a demethylated derivativeclone of a primary uveal melanoma cell line and its untreatedcontrol. Our study provides further insight by linking changesin gene expression between early and advanced melanomawith DNA methylation.

It is clear that both epigenetic and genetic events contributeto determining tumor development and progression (50). Inkeeping with the proposed link between DNA hypermethyla-tion and genomic instability, CGH analysis demonstratedan increased number of chromosomal abnormalities in thederivative cell lines, as compared with the parental cells.Such large-scale cytogenetic changes would be expected toimpact significantly on the expression of many genes, which in

Fig. 5. Effect of DAC treatment on melanoma cells in vitro. (A) Cells (20 000/well) were seeded into 12-well tissue culture plates and left to grow for 7 days,while being simultaneously treated with and without DAC. Growth rates were measured using an alamarBlue assay for cell proliferation. Error bars refer toSEM of triplicate determinations. In (B), cells (grown to confluence in 12-well plates) were wounded by creating a scratch across the monolayer culture.Phase contrast images of this region were taken directly following injury and 48 h later.


1864

Dow



ecember 2021

turn may influence the aggressive nature of the derivatives.While our analyses revealed no major striking correlationsbetween gene expression and specific genomic aberrations inthe four cell lines examined here (Figure 2 and Table I), wholegenome CGH microarray studies might provide additionalclarification of this issue. It should be noted that the GeneOntology classes/categories exhibited by the �7000 genesrepresented on the DNA microarray used in this study showedstrong similarity with the whole genome (see Supplementaldata, S4).There is a rapidly increasing interest in the potential of

epigenetic modifiers in the treatment of cancer (51). In manycancer types, the use of DNA methyltransferase and histonedeacetylase inhibitors have shown to be useful in mediatingsuppression of tumor growth and increased activity of otheranti-cancer agents (32,52,53). However, the use of such agentsfor melanoma therapy has been inconclusive.In an early phase II clinical trial, Abele et al. (54) showed

only one response in a set of 20 patients with melanoma treatedwith DAC. Following this negative result, however, a numberof studies have provided more support for the use of DAC inthe treatment of melanoma (53,55,56). Anzai et al. (55) showeda synergistic effect between DAC and the topoisomeraseI inhibitor, topotecan, against melanoma cells in vitro, sug-gesting that combination therapies of DAC and other drugsmay have more beneficial effects than DAC alone. In agree-ment with this concept, Coral et al. (56) noted that DAC in

combination with the inhibitory cytokine, IFN-g, enhancedthe expression of human leucocyte class I antigens togetherwith certain co-stimulatory molecules, such as ICAM-1and LFA-3, in a panel of 12 metastatic melanoma cell lines.Moreover, DAC treatment yielded a persistent (460 days)expression of MAGE-1 in one of the melanoma cell lines.This DAC/IFN-g combination may enhance the immunogenicpotential of melanoma cells, thereby increasing the efficacy ofimmunotherapy. More recently, Kozar et al. (53) showed thatcombined treatment of DAC and IL-12 significantly attenuatedgrowth of B16F10 melanoma cell-derived tumors in vivo, incomparison with moderate anti-tumor effects when eitheragent was given alone. This further supports the use of DACas an immunomodulatory agent for complementation withinhibitory cytokines for the treatment of melanoma.Interestingly, Wang et al. (57) proposed that responsiveness

of melanomas to immunotherapy is predetermined and may bedeciphered from analysis of gene expression profiling data.Within the 66 gene cohort, we noted down-regulation of asubstantial cohort of immune-related and interferon-relatedgenes in the derivative cell lines, which may be illustrative ofthe previously documented resistance to various inhibitorycytokines (15,19).In our study, DAC suppressed tumor cell growth in vitro.

Moreover, systemic treatment of mice with DAC attenuatedgrowth of 1205-Lu-derived xenografts, with consequentre-expression of TSPY mRNA. While these data might, in

- DAC + DAC

TSPY

18 S

Time (days)

0

1

2

3

4

5

6

7

8

9

1 3 5 7 9 11 13 15 17 19 21

Control

Treated

P <

0.05

P < 0.05

Rel

ativ

e tu

mor

vol

ume

(cm

3 )

Treatment

Control

Treated

A

B

Fig. 6. Effect of DAC treatment on growth of xenografts and TSPY expression in vivo. (A) 1205-Lu cells were injected subcutaneously into athymic female CD-1nude mice. The relative tumor volume (volume of tumor/volume of tumor at day 1 of measurement) was recorded daily over 22 days, following an initialgrowth period (mean tumor volume�0.1 cm3). At day 7, animals were either treated with DAC (15 mg/kg/mouse) or PBS alone (n¼ 8 animals/group). Error barsrefer to the SEM of 8 determinations. There was no significant difference in body weight of treated and untreated groups over observation period (data notshown). Significant differences (P-values 5 0.05) in terms of tumor growth rates were observed between treated and control groups both on an overall basis(day 22) and over time from days 7 to 15. Statistical analysis was performed using an analysis of variance test. (B) RT–PCR analysis of TSPY mRNA expressionin tumours explanted from untreated and DAC treated mice using a cycle limitation method.


1865

Dow



ecember 2021

some part, be due to a direct cytotoxic action by DAC, ouroverall data point towards the concept that regional DNAhypermethylation at multiple loci is likely to be involved inthe epigenetic regulation of melanoma progression. DAC alsohas an inhibitory effect on growth and migration of WM793cells in vitro, as well as being able to mediate complex changesin gene expression in this cell type. Given that the WM793 cellline is representative of an early melanoma, this suggests thatthe phenotype of these cells is also controlled, to a certaindegree, by DNA methylation events. The relative enrichmentof methylation-responsive markers in the identified set ofdown-regulated melanoma progression-related genes acrossthe three derivative cells, however, suggests that further hyper-methylation has occurred as one progresses through theWM793 series. This hypothesis is in keeping with the currentmodel of acquisition of epigenetic marks during tumorprogression.Our data provides further support for the incorporation of

demethylation agents into clinical trials. Additional work isrequired to determine potential for synergy with other epigen-etic modifiers and conventional therapies in terms of alteringgene expression and therapeutic responses. In conclusion, abetter understanding of melanoma progression, as exemplifiedby this study, may translate into new therapeutic avenues forthis intractable disease.

Supplementary material

Supplementary material can be found at: http://carcin.oxfordjournals.org/

Acknowledgements

With respect to the drafting of this manuscript, the authors would like toacknowledge the constructive comments of Prof. Michael J.Duffy andDr Paul Moynagh. The authors would also like to thank the following fortechnical assistance and other support: Dr Rhona MacKie, Dr Jens Teodoridis,Dr Gordon Strathdee, and Mr Liam Faller.Grant support: Health Research Board, Enterprise Ireland and CancerResearch Ireland. The Conway Institute is funded by the Programme forResearch in Third Level Institutions (PRTLI), administered by the HigherEducation Authority (HEA) of Ireland.

Conflict of Interest Statement: Robert Brown has acted as a consultant forSupergen Inc. who produce Decitabine (20-deoxy-5-azacytidine) which wasused in the study. No other authors have declared any conflict of interest.

References

1.Tucker,M.A. and Goldstein,A.M. (2003) Melanoma etiology: where arewe? Oncogene, 22, 3042–3052.

2.Beddingfield,F.C (2003) The melanoma epidemic: res ipsa loquitur.Oncologist, 8, 459–465.

3.Chin,L., Merlino,G. and DePinho,R.A. (1998) Malignant melanoma:modern black plague and genetic black box. Genes Dev., 12, 3467–3481.

4.Chung,C.H., Bernard,P.S. and Perou,C.M. (2002) Molecular portraits andthe family tree of cancer. Nat. Genet., 32 (suppl.), 533–540.

5.Baldi,A., Santini,D., Uca Ade,L. and Paggi,M.G. (2003) cDNA arraytechnology in melanoma: an overview. J. Cell Physiol., 196, 219–223.

6.Bittner,M., Meltzer,P., Chen,Y. et al. (2000) Molecular classification ofcutaneous malignant melanoma by gene expression profiling. Nature, 406,536–540.

7.Clark,E.A., Golub,T.R., Lander,E.S. and Hynes,R.O. (2000) Genomicanalysis of metastasis reveals an essential role for RhoC. Nature, 406,532–535.

8.Su,Y.A., Bittner,M.L., Chen,Y. et al. (2000) Identification of tumor-suppressor genes using human melanoma cell lines UACC903,UACC903(þ6), and SRS3 by comparison of expression profiles. Mol.Carcinog., 28, 119–127.

9.Certa,U., Seiler,M., Padovan,E. and Spagnoli,G.C. (2001) Highdensity oligonucleotide array analysis of interferon-alpha2a sensitivity andtranscriptional response in melanoma cells. Br. J. Cancer, 85, 107–114.

10.Seftor,E.A., Meltzer,P.S., Kirschmann,D.A. et al. (2002) Moleculardeterminants of human uveal melanoma invasion and metastasis. Clin.Exp. Metastasis, 19, 233–246.

11.Wang,E., Miller,L.D., Ohnmacht,G.A. et al. (2002) Prospective molecularprofiling of melanoma metastases suggests classifiers of immuneresponsiveness. Cancer Res., 62, 3581–3586.

12.Baldi,A., Battista,T., De Luca,A. et al. (2003) Identification of genesdown-regulated during melanoma progression: a cDNA array study. Exp.Dermatol., 12, 213–218.

13.van der Velden,P.A., Zuidervaart,W., Hurks,M.H. et al. (2003) Expressionprofiling reveals that methylation of TIMP3 is involved in uvealmelanoma development. Int. J. Cancer, 106, 472–479.

14.Hsu,M.-Y., Elder,D.E. and Herlyn,M. (1999) Melanoma: the Wistar (WM)melanoma cell lines. In Masters,J.R.W. and Palsson,B. (eds) Human CellCulture. Kluwer Academic Publishers, London, pp. 259–274.

15.Kobayashi,H., Man,S., MacDougall,J.R., Graham,C.H., Lu,C. andKerbel,R.S. (1994) Variant sublines of early-stage human melanomasselected for tumorigenicity in nude mice express a multicytokine-resistantphenotype. Am. J. Pathol., 144, 776–786.

16. Juhasz,I., Albelda,S.M., Elder,D.E. et al. (1993) Growth and invasion ofhuman melanomas in human skin grafted to immunodeficient mice. Am. J.Pathol., 143, 528–537.

17.Easty,D.J., Mitchell,P.J., Patel,K., Florenes,V.A., Spritz,R.A. andBennett,D.C. (1997) Loss of expression of receptor tyrosine kinase familygenes PTK7 and SEK in metastatic melanoma. Int. J. Cancer, 71,1061–1065.

18.McArdle,L., Rafferty,M., Maelandsmo,G.M. et al. (2001) Protein tyrosinephosphatase genes downregulated in melanoma. J. Invest. Dermatol., 117,1255–1260.

19.Breen,C.J., Barton,L., Carey,A. et al. (1999) Applications of comparativegenomic hybridization in constitutional chromosome studies. J. Med.Genet., 36, 511–517.

20.Dasari,V.K., Deng,D., Perinchery,G., Yeh,C.C. and Dahiya,R.D. (2002)DNA methylation regulates the expression of Y chromosome specificgenes in prostate cancer. J. Urol., 167, 335–338.

21.Fragra,M.F., Uriol,E., Borja Diego,L. et al. (2002) High-performancecapillary electrophoretic method for the quantification of 5-methyl20-deoxycytidine in genomic DNA: application to plant, animal andhuman cancer tissues. Electrophoresis, 23, 1677–1681.

22. Jin,H., Yang,R., Awad,T.A. et al. (2001) Effects of early angiotensin-converting enzyme inhibition on cardiac gene expression after acutemyocardial infarction. Circulation, 103, 736–742.

23.Southern,E.M. (1975) Detection of specific sequences among DNAfragments separated by gel electrophoresis. J. Mol. Biol., 98, 503–517.

24.Kath,R., Jambrosic,J.A., Holland,L., Rodeck,U. and Herlyn,M. (1991)Development of invasive and growth factor-independent cell variants fromprimary human melanomas. Cancer Res., 51, 2205–2211.

25.Balaban,G., Herlyn,M., Guerry,D.IV. et al. (1984) Cytogenetics of humanmalignant melanoma and premalignant lesions. Cancer Genet. Cytogenet.,11, 429–439.

26.Balaban,G.B., Herlyn,M., Clark,W.H.Jr and Nowell,P.C. (1986)Karyotypic evolution in human malignant melanoma. Cancer Genet.Cytogenet., 19, 113–122.

27.Hildenbrand,R., Schroder,W., Brude,E., Schepler,A., Konig,R., Stutte,H.J.and Arnemann,H. (1999) Detection of TSPY protein in a unilateralmicroscopic gonadoblastoma of a Turner mosaic patient with a Y-derivedmarker chromosome. J. Pathol., 189, 623–636.

28.Lau,Y.F. (1999) Gonadoblastoma, testicular and prostate cancers, and theTSPY gene. Am. J. Hum. Genet., 64, 921–927.

29.Lau,Y. and Zhang,J. (2000) Expression analysis of thirty one Ychromosome genes in human prostate cancer. Mol. Carcinog., 27,308–321.

30.Lau,Y.F., Lau,H.W. and Komuves,L.G. (2003) Expression pattern of agonadoblastoma candidate gene suggests a role of the Y chromosome inprostate cancer. Cytogenet. Genome Res., 101, 250–260.

31.Liang,G., Gonzales,F.A., Jones,P.A., Orntoft,T.F. and Thykjaer,T. (2002)Analysis of gene induction in human fibroblasts and bladder cancer cellsexposed to the methylation inhibitor 5-aza-20-deoxycytidine. Cancer Res.,61, 961–966.


1866

Dow



ecember 2021

http://carcin

32.Plumb,J.A., Strathdee,G., Sludden,J., Kaye,S.B. and Brown,R. (2000)Reversal of drug resistance in human tumor xenografts by 20-deoxy-5-azacytidine-induced demethylation of the hMLH1 promoter. Cancer Res.,60, 6039–6044.

33.Krick,R., Jakubiczka,S. and Arnemann,J. (2003) Expression, alternativesplicing and haplotype analysis of transcribed testis specific protein(TSPY) genes. Gene, 302, 11–19.

34.Arnemann,J., Jakubiczka,J., Thuring,S. and Schmidtke,J. (1991) Cloningand sequence analysis of a human Y-chromosome-derived, testicularcDNA, TSPY. Genomics, 11, 108–114.

35.Schneiders,F., Dork,T., Arnemann,J., Vogel,T., Werner,M. andSchmidtke,J. (1996) Testis-specific protein, Y-encoded (TSPY) expressionin testicular tissues. Hum. Mol. Genet., 5, 1801–1807.

36.Zhang,J.S., Yang-Feng,T.L., Muller,U., Mohandas,T.K., de Jong,P.J. andLau,Y.F. (1992) Molecular isolation and characterization of an expressedgene from the human Y chromosome. Hum. Mol. Genet., 1, 717–726.

37.Dasari,V.K., Goharderakhshan,R.Z., Perinchery,G. et al. (2001)Expression analysis of Y chromosome genes in human prostate cancer.J. Urol., 165, 1335–1341.

38.Baylin,S.B., Herman,J.G., Graff,J.R., Vertino,P.M. and Issa,J.P. (1998)Alterations in DNA methylation: a fundamental aspect of neoplasia. Adv.Cancer Res., 72, 141–196.

39.Herman,J.G. and Baylin,S.B. (2003) Gene silencing in cancer inassociation with promoter hypermethylation. N. Engl. J. Med., 349,2042–2054.

40.Momparler,R.L. (2003) Cancer epigenetics. Oncogene, 22, 6479–6483.41.van der Velden,P.A., Metzelaar-Blok,J.A., Bergman,W. et al. (2001)

Promoter hypermethylation: a common cause of reduced p16(INK4a)expression in uveal melanoma. Cancer Res., 61, 5303–5306.

42.Zhou,X.P., Gimm,O., Hampel,H., Niemann,T., Walker,M.J. and Eng,C.(2000) Epigenetic PTEN silencing in malignant melanomas without PTENmutation. Am. J. Pathol., 157, 1123–1128.

43.Soengas,M.S., Capodieci,P., Polsky,D. et al. (2001) Inactivation of theapoptosis effector Apaf-1 in malignant melanoma. Nature, 409, 207–211.

44.Karpf,A.R., Lasek,A.W., Ririe,T.O., Hanks,A.N., Grossman,D. andJones,D.A. (2004) Limited gene activation in tumor and normal epithelialcells treated with the DNA methyltransferase inhibitor 5-aza-20-deoxycytidine. Mol. Pharmacol., 65, 18–27.

45.Zendman,A.J., van Kraats,A.A., den Hollander,A.I., Weidle,U.H.,Ruiter,D.J. and van Muijen,G.N. (2002) Characterization of XAGE-1b, ashort major transcript of cancer/testis-associated gene XAGE-1, induced inmelanoma metastasis. Int. J. Cancer, 97, 195–204.

46.Serrano,A., Tanzarella,S., Lionello,I. et al. (2001) Re-expression of HLAclass I antigens and restoration of antigen-specific CTL response in

melanoma cells following 5-aza-20-deoxycytidine treatment. Int. J.Cancer, 94, 243–251.

47.Fulda,S., Kufer,M.U., Meyer,E., van Valen,F., Dockhorn-Dworniczak,B.and Debatin,K.M. (2001) Sensitization for death receptor- or drug-inducedapoptosis by re-expression of caspase-8 through demethylation or genetransfer. Oncogene, 20, 5865–5877.

48.Suzuki,H., Gabrielson,E. and Chen,W. (2002) A genomic screen for genesupregulated by demethylation and histone deacetylase inhibition in humancolorectal cancer. Nat. Genet., 31, 141–149.

49.Shi,H., Wei,S.H., Leu,Y. et al. (2003) Triple analysis of the cancerepigenome: an integrated microarray system for assessing geneexpression, DNA methylation, and histone acetylation. Cancer Res., 63,2164–2171.

50. Jones,P.A. and Baylin,S.B. (2002) The fundamental role of epigeneticevents in cancer. Nat. Rev. Genet., 3, 415–428.

51.Brown,R. and Strathdee,G. (2002) Epigenomics and epigenetic therapy ofcancer. Trends Mol. Med., 8, S43–S48.

52.Smith,K.S., Folz,B.A., Adams,E.G. and Bhuyan,B.K. (1995) Synergisticand additive combinations of several antitumor drugs and other agentswith potent alkylating agent adozelsin. Cancer Chemother. Pharmacol.,35, 471–482.

53.Kozar,K., Kaminski,R., Switaj,T. et al. (2003) Interleukin 12-basedimmunotherapy improves the antitumour effectiveness of a low-dose 50-aza-20-deoxycitidine treatment in L1210 leukemia and B16F10 melanomamodels in mice. Clin. Cancer Res., 9, 3124–3133.

54.Abele,R., Clavel,M., Dodion,P. et al. (1987) The EORTC Early ClinicalTrials Cooperative Group experience with 5-aza-20-deoxycytidine(NSC 127716) in patients with colo-rectal, head and neck, renalcarcinomas and malignant melanomas. Eur. J. Cancer Clin. Oncol., 23,1921–1924.

55.Anzai,H., Frost,P. and Abbruzzese,J.L. (1992) Synergistic cytotoxicitywith 20-deoxy-5-azacytidine and topotecan in vitro and in vivo. CancerRes., 52, 2180–2185.

56.Coral,S., Sigalotti,L., Gasparollo,A., Cattarossi,I., Visintin,A. andCattelan,A. (1999) Prolonged upregulation of the expression of HLAclass I antigens and costimulatory molecules on melanoma cellstreated with 5-aza-20-deoxycytidine (5-AZA-CdR). J. Immunother., 22,16–24.

57.Wang,E., Miller,L.D., Ohnmacht,G.A. et al. (2002) Prospective molecularprofiling of melanoma metastases suggests classifiers of immuneresponsiveness. Cancer Res., 62, 3581–3586.

Received December 6, 2004; revised and accepted June 7, 2005


1867

Dow



ecember 2021

Multiple markers for melanoma progression regulated by DNA

Documents

Transcript of Multiple markers for melanoma progression regulated by DNA