0008-5472.CAN-09-2367.Full.Pdf

Published OnlineFirst October 20, 2009; DOI: 10.1158/0008-5472.CAN-09-2367 Published Online First on October 20, 2009 as 10.1158/0008-5472.CAN-09-2367

Molecular Biology, Pathobiology, and Genetics

Identification of Key Regions and Genes Important in the Pathogenesis of Sézary Syndrome by Combining Genomic and Expression Microarrays

Elisabetta Caprini,1 Cristina Cristofoletti,1 Diego Arcelli,1 Paolo Fadda,1 Mauro Helmer Citterich,1 Francesca Sampogna,1 Armando Magrelli,2 Federica Censi,2 Paola Torreri,2 Marina Frontani,1 Enrico Scala,1 Maria Cristina Picchio,1 Paola Temperani,3 Alessandro Monopoli,1 Giuseppe Alfonso Lombardo,1 Domenica Taruscio,2 Maria Grazia Narducci,1 and Giandomenico Russo1

1Istituto Dermopatico dell'Immacolata-Istituto di Ricovero e Cura a Carattere Scientifico and 2National Centre for Rare Diseases, Istituto Superiore di Sanità, Rome, Italy; and 3Università di Modena e Reggio Emilia, Unità di Ematologia, Dipartimento di Oncologia ed Ematologia, Modena, Italy

Abstract Conventional cytogenetics, allelotyping, and comparative genomic In this study, we used single nucleotide polymorphism and hybridization (CGH) analyses have shown that these tumor cells comparative genomic hybridization array to study DNA copy exhibit chromosomal instability, manifested as allelic gain, loss, number changes and loss of heterozygosity for 28 patients af- or rearrangement (2). Although numerous efforts have been made fected by Sézary syndrome (SS), a rare form of cutaneous T- to clarify how the genomic aberrations might predispose to the cell lymphoma (CTCL). Our data identified, further confirm- development of this disease, the identification of pathogenically ing previous studies, recurrent losses of 17p13.2-p11.2 and relevant genes still remains a challenge. Recently, the application 10p12.1-q26.3 occurring in 71% and 68% of cases, respectively; of high-throughput technologies for genome-wide surveys of genet- common gains were detected for 17p11.2-q25.3 (64%) and ic or expression profiles in CTCL are giving new insights into this – chromosome 8/8q (50%). Moreover, we identified novel ge- malignancy (3 6). Furthermore, there is an increasing tendency to nomic lesions recurring in >30% of tumors: loss of 9q13- integrate mapping and expression data to study the relationship q21.33 and gain of 10p15.3-10p12.2. Individual chromosomal between copy number changes and gene expression levels as a aberrations did not show a significant correlation with significant fraction of genes are altered in a manner consistent prognosis; however, when more than three recurrent chromo- with the underlying genomic alterations in a variety of tumors – somal alterations (gain or loss) were considered, a statistical (7 11) including CTCL (12, 13). The presence of recurrent chromo- association was observed using Kaplan-Meier survival analy- somal copy number alterations (CNA) that correlate with disease sis. Integrating mapping and transcriptional data, we were outcome suggests that changes in the expression of specific genes able to identify a total of 113 deregulated transcripts in aber- within these regions are critical to the disease process (14). rant chromosomal regions that included cancer-related genes Here, we describe a genome-wide analysis, at submegabase res- such as members of the NF-κB pathway (BAG4, BTRC, olution, of DNA copy number changes in 28 SS samples using the NKIRAS2, PSMD3, and TRAF2) that might explain its constitu- single nucleotide polymorphism (SNP) and array CGH (aCGH) tive activation in CTCL. Matching this list of genes with those technology. Our findings identified six regions of highly recurrent discriminating patients with different survival times, we copy number aberrations affecting chromosomes 8, 9, 10, and 17. identify several common candidates that might exert critical Additionally, the SNP technology allowed us to distinguish be- roles in SS, such as BUB3 and PIP5K1B. Altogether, our study tween loss of heterozygosity (LOH) associated with either copy confirms and maps more precisely the regions of gain and number changes, such as hemizygous deletions, or copy number loss and, combined to transcriptional profiles, suggests a neutral status, underlying the involvement of different genetic me- novel set of genes of potential interest in SS. [Cancer Res chanisms that lead to uniparental disomy (UPD) in SS (15). In the 2009;69(21):8438–46] attempt to correlate copy number data and clinical parameters, we find a relationship between complex pattern of chromosomal aberrations, involving at least three recurrent CNAs, and shorter surviv- Introduction al. We then combined copy number results with gene expression Sézary syndrome (SS) is an aggressive leukemic variant of data from the same SS patients to generate a signature of genes cutaneous T-cell lymphoma (CTCL) that typically presents with differentially expressed and located within the chromosomal erythroderma, peripheral lymphadenopathy, severe pruritus, and regions of interest that include novel potential candidates of SS + malignant circulating CD4 T lymphocytes: the Sézary cells (1). tumor development.

Note: Supplementary data for this article are available at Cancer Research Online Materials and Methods (http://cancerres.aacrjournals.org/). Requests for reprints: Giandomenico Russo and Maria Grazia Narducci, Istituto Patients. Peripheral blood samples from 28 SS-affected individuals were Dermopatico dell'Immacolata-Istituto Di Ricovero e Cura a Carattere Scientifico, Via analyzed. All the patients were enrolled in clinical protocols approved by dei Monti di Creta 104, 00167 Rome, Italy. Phone: 39-6-66464798; Fax: 39-6-66462430; the Ethical Committee of the Istituto Dermopatico dell'Immacolata, and E-mail: [email protected] and [email protected]. ©2009 American Association for Cancer Research. informed consent was obtained in accordance with the Declaration of doi:10.1158/0008-5472.CAN-09-2367 Helsinki. Diagnosis of SS was based on described criteria (16). The major

Cancer Res 2009; 69: (21). November 1, 2009 8438 www.aacrjournals.org

Table 1. Highly recurrent CNAs in SS

Chr Copy number loss Mb* Genes in SS patients Copy Mb* Genes in SS patiens region† (%) number gain region† (%) Start Stop Size Start Stop Size

8 8p23.3-q24.3 1,13 142,1 140,94 1029 14/28 (50%) 9 9q13-q21.33 69,11 86,66 17,56 115 9/28 (32%) 10 10p12.1-q26.3 25,1 133,8 108,70 1016 19/28 (68%) 10p15.3-10p12.2 0,14 24,02 23,88 187 10/28 (36%) 17 17p13.2-p11.2 3,57 20,15 16,58 388 20/28 (71%) 17p11.2-q25.3 17,35 76,40 59,05 1171 18/28 (64%)

*Numerical position based on human genome assembly hg 18 (National Center for Biotechnology Information Build 36.1). † Number of coding transcripts based on genomic sequence information available on Mar 24, 2008 (National Center for Biotechnology Information Build 36.3).

clinical and immunologic characteristics, including T-cell receptor clonal hybridization were performed following the Agilent manufacturer's proto- analysis (17) of the samples, are listed in Supplementary Table S1. cols (version 5.0). Images were analyzed using Agilent Feature Extraction Tumor, normal cell isolation, and genomic DNA extraction. Periph- Software version 9.5.1 and data were imported into Agilent CGH analytics eral blood mononuclear cells were separated by Ficoll-Histopaque density software version 3.5 for a graphical overview and analysis. Analysis was per- gradient centrifugation (Sigma-Aldrich). Lymphomonocytes and granulo- formed using the ADM-2 algorithm (Agilent Technologies) with a threshold cytes were separately collected from the gradient fractions. Lymphomono- of 5.3. Experiments showing a derivative log ratio spread value of >0.3 were cytes were subsequently purified by positive selection using anti–human excluded from the analysis. CD3-conjugated dynabeads (Oxoid Ltd.). Granulocytes (normal matched RNA isolation, labeling, and gene expression analysis. Five micro- cells) were obtained collecting the upper phase overlaying the Ficoll density grams of total RNA extracted from sorted T lymphocytes, according to Af- gradient sediment; residual erythrocytes were then lysed by repetitive fymetrix procedure, were reverse transcribed, synthesized in cRNA, washes with a solution of 10 mmol/L Tris-HCl (pH 7.6), 5 mmol/L MgCl2, fragmented, labeled, and hybridized to Human Genome U133A arrays. and 10 mmol/L NaCl. Tumor genomic DNA was isolated from sorted CD3+/ The scanned data were analyzed with a customized script that uses Bio- CD4+ T lymphocytes of SS patients, whereas normal matched DNA was ex- conductor packages6 based on the R language,7 for quality control assess- tracted from the granulocytes of the same individuals according to pub- ment, data normalization, unsupervised and supervised clustering analysis, lished protocols (17). The calculated percentage of tumor cells after and identification of differentially expressed transcripts (19). This R-script purification is indicated for each sample in Supplementary Table S1 and provides utilization of different Bioconductor packages; gcrma package was ranged from 35% to 91%. used for normalization and background correction, and Genefilter package SNP genotyping, LOH, and DNA copy number change. The samples was used to filter genes with highly variance by interquartile range method; were analyzed on the GeneChip Human Mapping 10K Array Xba 142 samr-package, significance analysis of microarrays (20) was used for detect- (Affymetrix) containing ∼10,000 tiled SNPs (median intermarker distance, ing significantly expressed genes and controlling the false discovery rate 105 Kb). DNA was prepared for hybridization using the GeneChip Human between groups of tumor samples. Mapping 10K Xba Assay kit (Affymetrix). Genotype calls and signal infor- Survival and statistical analyses. A time-to-event analysis was per- mation were obtained using GCOS 1.4 and GTYPE 4.0 software (Affyme- formed using nonparametric Kaplan-Meier product-limit survival trix). SNP arrays were then analyzed using dChipSNP software,4 allowing estimates, and differences between Kaplan-Meier survival curves were an- the simultaneous measurement of LOH and DNA copy number changes alyzed using the Mantel-Haenszel log-rank test. Statistical analyses were (18). To compute LOH, in a pair of normal and tumor DNA, dChipSNP performed using Statistical Package SPSS v. 13.0. assigned the following calls: L, Loss (AB in normal A or B in tumor); R, Quantitative real-time PCR. Quantitative RT-PCR was performed us- retention (AB both in normal and tumor); and N, noninformative or no call ing a SensiMix DNA kit (Quantace Ltd) on a ABI PRISM 7000 Sequence (homozygous A or B in normal and tumor; AB, A, B, and no call in normal Detection System (Applied Biosystem). Primers, designed by using Primer3 or tumor); the LOH status, inferred from the observed L calls, used Hidden Input 0.4.0, were synthesized by Eurofins MWG Operon. Quantification of Markov Model. For copy number determination (15) after normalization to tumor DNA alterations was performed by comparing the target locus to the a baseline array using the invariant set normalization method, a model- reference Line-1 (21), and copy number changes of target genes were de- based method (PM/MM) was used to compute the signal values for each termined as described (15). Primers for each target are shown in Supple- SNP in each array. To infer the DNA copy number from the raw signal data, mentary Table S2. We have used reverse quantitative RT-PCR to validate the Hidden Markov Model was used applying subinteger inferred step of gene expression data as previously described (16) using, for each gene, a 0.2. Genomic gains and losses were defined as inferred copy number of reference RNA pool chosen from those tumors with no CNA; primers' >2.4 and <1.8, respectively; these threshold values of dChipSNP-copy sequences are indicated in Supplementary Fig. S1. number determination were chosen as they correspond to copy number measures made by aCGH. Results Mapping information of SNP loci and cytogenetic band are derived from Affymetrix and University of California Santa Cruz5 (hg 17 human genome CNAs: detection of copy number loss and gain. Tumor and assembly). paired normal samples from 28 SS patients were genotyped using aCGH. aCGH analysis was performed using the Human Genome CGH 10K SNP arrays. The abnormalities found by SNP chip analyses in Microarray kit 44B (Agilent Technologies) comprising 60-mer oligonucleo- 24 of these patients were verified by comparison to those detected ∼ tides spaced 75 kb density over the full genome. Digestions, labeling, and by aCGH studies. The copy number regions were determined

4 http://www.dchip.org 6 http://www.bioconductor.org 5 http://genome.ucsc.edu 7 http://www.r-project.org www.aacrjournals.org 8439 Cancer Res 2009; 69: (21). November 1, 2009

Downloaded from cancerres.aacrjournals.org on September 23, 2021. © 2009 American Association for Cancer Research. Published OnlineFirst October 20, 2009; DOI: 10.1158/0008-5472.CAN-09-2367 Cancer Research according to the hybridization intensity data generated from each Lastly, we identified chromosome 10p15.3-p12.2 commonly affect- genotype probe on the mapping array comparing normal and ed by copy number gain as 10 of 28 samples (36%) display this tumor samples. The result of the genome-wide analysis of CNAs, chromosomal abnormality. It may be considered a novel recurrent showing the most frequent regions involved in gain or loss of genetic lesion in CTCL because it has not been described yet. The genetic material is represented, as summary plot, in Supplementary extension of copy number gain of chromosome 10p in our sample Fig. S2. The computed copy number summary score identified three group ranged from 2.76 Mb (P23), within 10p15.1-p14, to a 26-Mb regions of copy number loss and three regions of gain as highly re- region (P42) spanning 10p15.3-p12.1 (Table 2; Fig. 1). In our data current (>30%). The relative sizes and boundaries were then mapped set, chromosome 10 was affected by complex allelic imbalances so referring to the SNP analysis (Table 1). that tumors with gain of 10p showed concomitant loss of short Copy number loss. The most common allelic loss, occurring in arm regions in two cases (P23, P45); similarly, 10q24.32-q26.3 dele- 71% of tumors, was on the short arm of chromosome 17. Nineteen tion cooccurred with 10p gain in patients P28 and P35. In six cases cases showed broad regions of deletion largely overlapping within (P2, P25, P36, P39, P42, and P48), all the above-mentioned allelic 17p13.2-p11.2 and one sample (P49) exhibited whole chromosome imbalances were simultaneously present. 17 loss (Table 2). Loss of genetic material spanning 10p12.1-q26.3 Analysis of LOH and DNA copy number change. The SNP was found in 68% of cases. As outlined in Fig. 1 and detailed in mapping array enabled us to assess the concurrent status of Table 2, seven tumors (P2, P22, P32, P36, P42, P48, and P49) dis- LOH and CNAs. We included in the LOH analysis tumor samples played one single deletion involving the long arm and the pericen- displaying consistent regions of deletion defined as more than tromeric region that, in one case (P49), was extended to the short three contiguous deleted SNPs (22). We found a good correlation arm. Six additional cases (P5, P8, P15, P25, P34, and P39) exhibit the between LOH and copy number data for all the above-described cooccurrence of two discrete areas of loss, one within 10p12.1-q21.3 recurrent deletions affecting chromosomes 9, 10, and 17 (Fig. 2). and the second in 10q23.1-q26.3. Three samples showed regions of However, we observed that some individuals, displaying LOH, do deletion encompassing only 10p12.1-q11.21 (P23, P27, and P45) and not exhibit copy number changes, a condition that might be ex- three further cases solely 10q23.2-q26.3 (P7, P28, and P35). We also plained with copy-neutral LOH events also known as UPD (23). found a highly recurrent region of loss within 9q13-q21.33, present The highest incidence, with six samples displaying UPD events, in 9 of 28 samples (32%), which was not previously reported among was detected for chromosome 10, whereas not more than two the common deletions in CTCL (Table 1; Fig. 1). cases have been detected for other chromosomes such as 17 and Genome-wide analysis of copy number losses enable also the 9 (boxed areas in Fig. 2). These regions generally span interstitial detection of homozygous deletions (HD) that are of particular areas of LOH or telomeric ends and they usually interest relatively interest as they may unveil the presence of tumor suppressor small areas (∼10–60 SNP markers), with the exception of UPDs genes. Two HDs were unequivocally identified in patient P28: the exhibited by patient P28 for 10p12.1-q24.3 (303 SNPs over 78 Mb; first one, encompassing 34 consecutive SNP loci with inferred copy Fig. 2) and 13q12.11-q14.2 (143 SNPs in ∼29 Mb; data not shown). number of 0.2 over 6.4 Mb, occurred on 9p21.3-p21.2 locus and Besides UPD, we found one case (P23) showing a chromosomal included tumor suppressor genes CDKN2A and CDKN2B (Fig. 2); area of ∼7.3 Mb (24 SNPs) on 10p12.31-p12.1 with inferred copy the second HD (∼2 Mb segment with three consecutive SNPs of number value of >5. This patient, however, exhibits concomitant inferred copy number values of 0.2) was identified on chromosome LOH of this region (Fig. 2), suggesting that the LOH might be due 13q14.2 encompassing the RB1 locus (data not shown). Both these to the amplification, rather than loss, of one of the allele. The use HDs have been confirmed by quantitative RT-PCR experiments of SNP technique allowed for a more complete understanding performed on tumor DNA from patient P28 (Supplementary of the complex genetic rearrangements underlying SS pathology. Table S2). Furthermore, a potential HD, defined as inferred copy Identification of genes differentially expressed in regions of number of 0.6 in 21 consecutive markers, has been identified in genomic gain and loss. Because we had expression data for patient P48; the flanking SNPs describe a 3.97-Mb area at 9p21.3 the majority of our samples (20 of 28 cases, indicated in bold in that, similar to P28 case, includes the CDKN2A and CDKN2B. Final- Table 2), we sought to determine the impact of CNAs on gene ex- ly, all the samples exhibited the loss of the T-cell receptor α and δ pression. To this end, we have dichotomized our data set for each chains locus due to the physiologic recombination of V and J gene of the highly recurrent chromosomal imbalances, grouping the segments, confirming, therefore, the homogenous selection of T samples on the basis of the presence or absence of each specific lymphocytes in our tumor samples (data not shown). CNA (see Supplementary Table S3 and Info for details). As a result, Copy number gain. The most frequent region of copy number we were able to identify a total of 113 genes (136 probe sets) dif- gain, present in 64% of samples, was on the long arm of chromo- ferentially expressed in regions of genomic imbalances (two- some 17 (17p11.2-q25.3). We observed that 16 of 18 patients exhibit sample t test P < 0.01): (a) 53 genes within chromosome 17, 94% the gain of 17q region associated with the concurrent deletion of of which matched the CNAs of this chromosome; (b) 21 encom- 17p (Table 2). This suggests the presence of an isochromosome 17q passing CNAs of chromosome 10, all of which correspondingly up- (iso17q), a chromosomal abnormality that shows a relatively high regulated or downregulated; (c) 24 genes spanning chromosome 9, frequency in our group of SS patients (57%). High-resolution map- the 85% were concordantly underexpressed; (d) 15 overexpressed ping of the breakpoints associated with loss of 17p and gain of 17q transcripts (100%) matched chromosome 8 gain. A complete list of revealed scattering of the breakpoints over 7.2 Mb in 17p12-q11.1 these genes is reported in Supplementary Table S4. The means of (15.2–22.4 Mb). Another recurrent gain involved chromosome 8 in the observed fold change values of gene expression were 1.33 (log2 14 of 28 (50%) cases; the whole chromosome 8 (8p23.3-q24.3) ratio = 0.41) for copy number gains and 0.75 (log2 ratio = −0.43) for resulted affected by copy number gain in eight tumor samples copy number loss consistent with a dosage effect [2-fold change in (P2, P7, P22, P27, P35, P42, P50, and P51), whereas the remaining DNA copy number corresponds to 1.4- to 1.5-fold change in mRNA six cases showed a gain of chromosomal regions varying between level (24)]. Real-time PCRs further confirmed some of these data 53.12 and 107.84 Mb mainly involving the long arm (Table 2; Fig. 1). (Supplementary Fig. S1). The search for cancer-related genes,

Cancer Res 2009; 69: (21). November 1, 2009 8440 www.aacrjournals.org

Table 2. Recurrent chromosomal changes as obtained from SNP and aCGH based copy number

SNP copy number loss aCGH loss SNP copy number gain aCGH gain

Tumor Cytoband Number Value Size Cytoband Size Cytoband Number Value Size Cytoband Size sample consecutive (Mb) (Mb) consecutive (Mb) (Mb) markers markers

P02 9p13.2-q31.2 141 1,8 69,26 N.d. N.d. 8p23.3-q24.3 472 2,6 140,94 N.d. N.d. 10p12.2-q26.3 429 1.4-1.6 110,41 N.d. N.d. 10p15.3-p12.2 99 2,8 23,25 N.d. N.d. 17p13.2-p11.2 51 1,4 16,58 N.d. N.d. 17q11.1-q25.3 115 2,6 53,96 N.d. N.d. P05 10p11.23-p11.21 16 1,4 4,62 N.d. N.d. 10q23.32-q25.2 69 1,2 19,49 N.d. N.d. 17p11.2-q25.3 116 2,6 56,25 N.d. N.d. 17p13.2-p11.2 50 1,4 13,79 N.d. N.d. P07* 10q23.2-q25.2 111 1.4-1.6 26,09 N.d. N.d. 8p23.3-q24.3 547 2,6 138,44 N.d. N.d. 17p13.3-p11.2 59 1,4 19,21 N.d. N.d. 17q11.1-q25.3 126 2,6 54,08 N.d. N.d. P08 9p24.3-q33.1 449 1,8 119,04 9q21.11-q33.2 51,98 17p11.2-q25.3 117 2,6 59,05 17q11.1-q25.3 55,62 10p14-q11.21 114 1,4 31,90 10p12.1-q11.21 17,49 10q25.1-q26.3 108 1,4 25,34 10q24.32-q26.3 29,36 17p13.2-p12 49 1,8 12,05 17p13.2-p11.2 17,73 P11 17p13.2-p11.2 51 1,6 16,58 17p13.3-p11.2 21,92 17q11.1-q25.3 115 2,6 53,96 17q11.2-q25.3 55,77 P15 9q21.13-q21.33 48 1,4 11,77 9q21.13-21.33 9,95 17p11.2-q25.3 116 2,6 56,25 17p11.2-q25.3 59,48 9q22.31-q31.1 31 1,2 11,51 9q22.3-q31.1 9,62 10p12.31-q11.21 66 1,4 21,56 10p12.31-q11.21 21,92 10q23.2-q25.1 79 1,6 21,89 10q23.21-q25.1 21,42 17p13.3-p11.2 18,76 P22 9p24.3-q22.32 320 1,4 94,13 9p24.3-q22.32 22,27 8p23.3-q24.3 472 2,6 140,94 8 146,1 10p13-q26.3 467 1,4 120,01 10p12.2-qter 109,912 17p13.2-p12 49 1,4 12,05 17p13.3-p11.2 18,54 P23 9p23-p21.1 110 1,2 19,33 9p23-p21.1 19,04 8p11.21-q24.3 314 2.4-2.8 101,47 8p11.21-q24.3 105 10p12.1-q11.1 41 1,2 13,17 10p12.1-p11.21 10,35 10p15.1-p14 8 3,4 2,76 10p15.1-p14 2,53 17p13.2-p12 46 1,2 11,49 17p13.3-p12 14,7 17p12-q24.3 89 2,6 48,51 17p11.2-q24.1 39,12 P25 9p13.2-q21.13 31 1,6 37,66 9p21.11-p21.13 6,67 10p12.1-q21.3 173 1,8 44,05 10p12.1-p11.23 2,553 10p15.3-12.1 110 2,4 24,96 10p15.3-p12.33 19,08 10q23.1-q26.3 200 1,8 50,67 10q23.1-q26.3 2,392 17p11.2-q21.33 43 2,8 27,25 17p11.2-q21.33 25,93 P27 10p12.1-q11.21 52 1,6 18,04 10p12.1-q11.21 18,17 8p23.3-q24.3 472 2,6 140,94 8 146,1 17p13.2-p11.2 51 1,8 16,58 17p13.3-p11.2 21,05 17q11.1-q25.3 53,79 P28 9p21.3-p21.1 8,512 9p13.2-q31.3 154 1,2 70,46 9p13.1-q31.3 72,17 10q24.32-q26.3 121 1,4 30,71 10q24.32-q26.3 31,71 10p15.3-p12.31 98 2,6 22,43 10p15.3-p12.31 21,12 17p13.2-p12 47 1,2 11,67 17p13.3-p12 15,44 P30 9p21.3 2,184 9q13-q34.3 255 1,2 67,34 9p13.2-q34.3 102,8 P32 10p12.31-q26.3 439 1.4-1.6 113,81 10p12.32-q26.3 114,9 17p11.2-q12 27 2,4 12,52 17p11.2-q12 12,3 17p13.2-p11.2 50 1,4 13,79 17p13.3-p11.2 18,76 P33 17p13.3-p12 18,03 17q12-q25.3 93 2,4 46,69 17p11.2-q25.3 59,48 P34 10p12.1-p11.21 27 1,4 7,82 10p12.1-p11.22 6,362 10q23.33-q26.3 138 1,4 36,99 10q23.33-q26.3 38,3 P35 10q24.32-q26.3 121 1 8 30, 71 10q23.31-q26.3 42,06 8p23.3-q24.3 460 2,4 138,32 8 146,1 17p13.3-p12 18,03 10p15.3-p12.2 101 2,4 23,88 P36 8q12.1-q24.3 268 2,6 82,31 8q12.1-q24.3 86,36 10p12.2-q26.3 427 1,8 109,77 10p12.1-q26.13 108,9 10p15.3-p12.2 101 2,4 23,88 10p15.3-p14 11,41 17p13.2-p11.2 50 1,8 13,79 17p13.3-p11.2 18,66 17p11.2-q23.3 79 2,4 38,99 17p11,2-q24.2 43,26 P37 17p13.2-p12 49 1,8 12,05 17p13.3-p11.2 16,22 17p11.2-q25.3 117 2,4 59,05 P38 P39 10p12.1-p11.21 32 1,2 7,12 10p12.12-q26.3 106,1 10p13-p12.1 52 2,4 11,83 - - 10p11.21-q26.3 365 1,2 95,71 17p13.2-p11.2 50 1,2 13,79 17p13.3-p11.2 18,8 17p11.2-q12 20 2,8 9,04 17p11.2-q12 14,59 P41 P42 9p24.2-q33.2 125 8p23.3-q24.3 472 2,4 140,94 8 146,1 10p12.1-q26.3 407 1,8 107,35 10q22.21-q26.3 109 10p15.3-p12.1 120 2,4 26,00 10p15.3-p12, 1 23,77 17p13.2-p11.2 50 1,8 13,79 17p13.3-p11.2 19,57 17p11.2-q25.3 116 2,4 56,25 17p11.2-q25.3 57,61 P43 8p12-q24.3 333 2,6 107,84 8p12-q24.3 111,8

(Continued on the following page)

www.aacrjournals.org 8441 Cancer Res 2009; 69: (21). November 1, 2009

Table 2. Recurrent chromosomal changes as obtained from SNP and aCGH based copy number (Cont'd)

SNP copy number loss aCGH loss SNP copy number gain aCGH gain

Tumor Cytoband Number Value Size Cytoband Size Cytoband Number Value Size Cytoband Size sample consecutive (Mb) (Mb) consecutive (Mb) (Mb) markers markers

17p13.2-p12 49 1,4 12,05 17p13.3-p12 21,96 17p11.2-q25.3 117 2,6 59,05 17q11.1-q25.3 55,18 P45 8p12-q24.3 333 2,6 107,84 8p12-q24.3 112,2 10p12.1-p11.23 11 1,4 1,42 10p12.1-p11.23 4,914 10p13-p12.1 52 2,4 11,83 - - 17p13.2-p11.2 51 1,4 16,58 17p13.3-p11.1 22,08 17q11.1-q25.3 115 2,8 53,96 17q11.1-q25.3 56,2 P48 9p23-p22.3 27 1,4 3,13 9p23-p22.3 3,77 9p22.1-q22.32 187 0.4-1.4 75,51 9p22.1-q22.32 79,45 9q32-q33.1 31 1,6 4,78 9q32-q33.1 3,08 8p12-q24.3 320 2,4 105,23 8p12-q24.3 95,32 10p11.23-q25.1 282 1,6 78,50 10p11.23-q25.1 91,44 10p15.3-p13 61 2,6 14,03 10p15.3-p13 13,25 17p13.2-p11.2 50 1,4 13,79 17p13.3-p11.2 18,8 17p11.2-q25.1 110 2,6 50,59 17q11.1-q25.1 50,53 P49 9p24.3-q34.3 483 1,8 134,82 9p24.2-q34.2 137,1 10p15.3-q23.33 376 1,8 93,66 10p15.3-q23.1 84,97 8p23.3-q11.2 184 2,4 53,12 8p21.3-q12.1 37,43 17p13.2-q25.3 166 1,8 72,83 17p13.3-q25.3 78,01 P50 17p13.2-p11.2 50 1,8 13,79 N.d. N.d. 8p23.3-q24.3 472 2,4 140,94 N.d. N.d. 17p11.2-q25.3 116 2,4 56,25 N.d. N.d. P51 9q21.3 0,889 8p23.3-q24.3 472 2,4 140,94 8 146, 1

NOTE: Bold indicate patient sample profiled for gene expression analysis. Abbreviation: N.d. not determined. *Patient P07 has been analyzed with 10K SNP array version 1. through the CancerGenes database resource (25), led to the iden- prised three deregulated genes (PIP5K1B, FLJ10232,andSET), of tification of 26 candidates (in bold in Supplementary Table S4). which PIP5K1B mapped to the recurrent region of deletion on Analysis of CNAs, expression profile, and survival. We then 9q13-q21.33; four probe sets (three genes) spanned the 10q22.3- asked if, as in other leukemias, a correlation existed between CNAs q26.3 genomic interval of copy number loss (MGEA5, PDCD11, and clinical outcome. Taken singularly, or in several combinations, and BUB3); two genes, TIMM22 and ABR, map head-to-head within any CNAs showed a significant association with survival time in the common deleted 17p13 locus, whereas 17q gained region en- our data set (data not shown). However, when we divided the compasses PSMD3 and PSMC5; finally, two genes were found on patients on the basis of a complex aberration pattern, defined as chromosome 8 (ADRA1A, DPYS). Although further investigations three or more recurrent CNAs, a decreased overall survival was are required to correlate genomic alterations and the clinical observed for individuals with ≥3alterations(P = 0.05; Fig. 3A). course of SS, these genes might represent novel candidates in- No significant difference, using the Mann-Whitney test, was seen volved in the pathogenesis of this neoplasm. between the observation times (i.e., copy number analysis), relative to the disease onset, for patients with <3 (median, 93.2 months) and ≥3 CNAs (median, 66.5 months). Therefore, we used the avail- Discussion able expression data to identify genes differentially expressed be- Allelic imbalance is a common feature of many malignancies. tween tumors harboring more (14 patients) or less (6 patients) We have measured allelic imbalance in a series of SS patients using than three CNAs. A total of 352 gene probes were found statistical- SNP oligonucleotide microarrays and aCGH. Both these techniques ly significant (two-sample t test P < 0.05); this list has been then enable genotyping with a comparable high-level resolution com- used in a supervised hierarchical clustering of a larger data set that prised between 75 and 105 Kb. The SNP analysis for copy number included five normal samples, two Mycosis fungoides (MF) pa- determination identified three common regions of loss within tients (P40, P44) displaying a normal chromosomal profile (data 9q13-q21.33, 10p12.1-q26.3 and 17p13.2-p11.2, whereas copy num- not shown), and eight additional SS tumors (P01, P04, P16, P18, ber gain were defined for 8p23.3-q24.3, 10p15.3-p12.2, and 17p11.2- P19, P21, P26, and P46) for whom mapping data were not available. q25.3. These chromosomes have long been established to be As illustrated in Fig. 3B, this set of 352 genes was able to clearly among the mostly affected by cytogenetic aberrations in CTCL us- discriminate two sample groups where healthy individuals (R1- ing conventional techniques (2, 26–31). More recently, aCGH has R5) and MF patients clustered with the tumor samples harboring been used by Vermeer and colleagues (3) in a genome-wide survey less than three CNAs, whereas only one sample with known com- of 20 SS cases; their findings identified 17p13.1-p11.2 loss in 75% of plex genomic aberrations (P36) was included in the same samples cases, a complex pattern of deletions of chromosome 10 with eight set. This suggests that genes with predictive value of a different SS discrete regions of highly recurrent loss (40–60%). Gains were clinical course might be comprised in the list. We then compared observed for 17q21.31-q23 (80%) and 8q24.1-q24.3 (75%); several these differentially expressed genes to those located in aberrant scattered minimal common gained areas across the long arm of regions of CNAs identifying 12 genes. They were distributed across chromosome 8 were also reported. Our results show a substantial the identified regions of recurrent aberration: chromosome 9 com- agreement with these data; however, main differences were

Cancer Res 2009; 69: (21). November 1, 2009 8442 www.aacrjournals.org

Downloaded from cancerres.aacrjournals.org on September 23, 2021. © 2009 American Association for Cancer Research. Published OnlineFirst October 20, 2009; DOI: 10.1158/0008-5472.CAN-09-2367 SNP and Expression Profiling of Sézary Syndrome observed for chromosome 10 losses and for regions of gain. change in copy number. In cancer biology, it might unveil genes Although our data confirmed the complex profile of alterations that are downregulated through epigenetic mechanisms or muta- affecting chromosome 10, we identified two prevalent lost regions tions, but still retain a diploid copy number (32). Events of UPD of variable length within 10p12.1-q22.1 and 10q22.3-q26.3 that might be due to: (a) mitotic nondisjunction, leading to UPD of a very often cooccur in tumor samples. Concerning 17q, we detected whole chromosome, (b) recombination between chromatids, a wider region of recurrent gain extending from the pericentro- resulting in UPD of a chromosomal arm or telomeric ends, (c) meric (17p11.2-q11.1) to the subtelomeric area of the long arm multiple recombination events, determining interstitial regions of (17q21.33-q24.3); furthermore, in our data set, we observed a prev- UPD (9, 23). In the SS cases analyzed, we could find several small alence of 17p loss (71%) with respect to 17q gain (64%), and a large and few large UPDs involving interstitial or telomeric regions of fraction of our cases (57%) had imbalances compatible with the LOH occurring mainly for chromosome 10 and to a lesser extent presence of iso17q. Regarding chromosome 8, we were not able for chromosome 9 and 17. This suggests that in SS cells, UPD might to identify minimal common regions of gain; we observed a rather arise through mechanisms of mitotic recombination rather than homogeneous alteration of the whole chromosome 8 or 8q involv- nondisjunction events and that DNA double-strand breaks repair ing about the 50% of our sample group. Finally, we were able to might be important in the etiology of this disease. This obviously detect two additional regions of allelic imbalance: 10p15.3-p12.2 requires the investigation of a larger number of individuals and gain, further increasing the pattern of complexity observed for this the identification of mutations within key SS-specific genes chromosome, and 9q13-q21.33 loss; to our knowledge, these were contained within the runs of UPD. In addition to UPD, we were also not previously reported in SS and may, therefore, be considered able to detect LOH and simultaneous copy number gain of novel recurrent genetic lesions. The slight discrepancies observed 10p12.31-p12.1 in one patient (P23) that is explained by the overrep- between the studies may reflect methodologic approaches using resentation of one of the allele rather than true LOH. Noticeably, different resolutions; however, altogether, these studies identified this is the only focal high-amplitude alteration that occurred in the most common allelic imbalances in SS. our data set of tumor samples, suggesting that high-level gene Paired tumor/control samples allow the identification of LOH by amplification events are not a feature of SS. Our results indicate true allelic imbalance (i.e., LOH associated with copy number loss) that large regions of low-level copy number gains (inferred copy or acquired UPD, a mechanism responsible for LOH without a number between 2.4 and 3.4), likely corresponding to duplication

Figure 1. Recurrent regions ofchromosomal gain and loss in SS samples as determined by SNP and aCGH. In the aCGH view, gains are indicated to the left ofthe ideogram in red and deletions in green, foreach sample. In the 10K SNP view, copy number abnormalities are shown in the gray panel to the right ofth e ideogram, where the threshold value (red line) of 2.4 is for copy number gain and 1.8 for copy number loss; blue line, the proportion ofsamples with gain or loss, respectively; cytobands in blue font, the recurrent regions exceeding the threshold.

www.aacrjournals.org 8443 Cancer Res 2009; 69: (21). November 1, 2009

Downloaded from cancerres.aacrjournals.org on September 23, 2021. © 2009 American Association for Cancer Research. Published OnlineFirst October 20, 2009; DOI: 10.1158/0008-5472.CAN-09-2367 Cancer Research events, are prevalent in this malignancy. Similarly, HDs were not fre- suggests that they might represent sporadic events further worsen- quently observed in our series of SS patients, and although all those ing the pattern of chromosomal aberrations of the tumor cell. identified matched well-known tumor suppressor genes (CDKN2A, We have then used the fine-mapped regions of recurrent CNAs CDKN2B, RB1) already described in CTCL (30, 33, 34), they do not to perform a comparison of expression level between classes of occur in the highly recurrent regions of copy number loss. This tumors with a given CNA and tumors lacking that CNA. This approach of combining copy number and gene expression data has been successful in other indications (7, 9, 11, 35); the impact of copy number on gene expression pattern shows generally more dramatic effects in the case of high-level copy number increase, whereas low-level copy number gains and losses, although have significant influence on expression of genes in the regions affected, display more subtle effects on a gene-by-gene basis (24, 36). There- fore, we applied stringent parameters of statistical significance to identify genes whose differential expression is more likely to be at- tributable to underlying low-level gene copy number changes; this led to the identification of 113 genes differentially expressed in regions of genomic gain and loss. Importantly, several of these genes have been already implicated in the molecular pathogenesis of CTCLsuchasMXI1 (3), HSF1, NFE2L1, SUPT4H1, DDX42,and ICAM2 (5, 13), providing, therefore, a measure of validation for the described study. Other genes have been previously reported in this lymphoma as important candidates for cancer development such as NME1 and RPA1 (3, 37). For this latter gene, in vivo studies showed that, if mutated, it results in defective DNA double-strand break repair, chromosomal instability, and development of lym- phoid tumors (38). A key molecular feature of CTCL is the constitutive activation of NF-κB that is involved in survival and apoptosis of SS cells (39, 40); interestingly, in our list of deregulated genes, we found a total of 5 genes related to the TNFα/NF-κB pathway (BAG4 8p12, BTRC 10q24.32, NKIRAS2 17q21.2, PSMD3 17q12-q21.1, and TRAF2 9q34), thereby suggesting that the recurrent low-level copy number changes, and the concomitant imbalance in gene expression, might disrupt critical stochiometric relationships in SS cell, possibly promoting tumor development or progression. It should be also noted that these analyses may represent an underestimate of the mRNA variation levels directly attributed to variation in gene copy number as this is affected not only by true variation in the expression programs of the tumor cells themselves, but also by the variable presence of no-tumor cell types within clinical samples (24). Therefore, a larger number of genes in SS could be deregulated as a consequence of copy number changes. Previous studies in CTCL have shown the agreement between aberration rate and change of clinical conditions and outcome (28, 41). Similar to Espinet and colleagues (27), we have found that as few as three recurrent CNAs are sufficient to reduce the overall survival; however, aberrations in a number of individual chromosome did not result in a significant correlation with prognosis. This might be due to the sample size (28 patients) that prevents to run robust statistical analysis, but also to the requirement of alteration or biallelic inactivation (e.g., point mutation, epigenetic inactivation) of one or more specific critical genes. Comparing the expression profiles of SS cases harboring three or more CNAs with those exhibiting less than three genetic lesions, we have found a set of 352 genes that, when further tested in a larger data set of CTCL Figure 2. LOH and copy number changes. SNP analysis ofSS tumor samples and matched constitutional DNA showing LOH on the left and copy number and healthy samples, seem to have a prognostic value. Interesting- on the right ofeach chromosome ideogram representing chromosomes 9 ( A), ly, this list of transcripts comprised 12 deregulated genes located in 10 (B), and 17 (C). Each sample is depicted as a series ofvertical bars in both the LOH and copy number panels. Blue areas, regions ofLOH; yellow areas, the recurrent regions of imbalance. Among them, particularly im- retention ofheterozygosity. The color intensities ofinferredmarkers decline to portant is BUB3 (10q26), as this gene is an essential component of the white as the distance from the nearest informative markers increases (18). the mitotic checkpoint pathway whose function ensures genomic Copy number is marked by shades of red ranging from light red for copy of ≤1to dark red for copies of ≥3 (see scale at the bottom ofthe panel); white color is stability (42). Mutations in essential checkpoint proteins lead to for 0 copies corresponding to HD. Boxed regions, UPD events. chromosome instability and promote carcinogenesis (43). Since

Cancer Res 2009; 69: (21). November 1, 2009 8444 www.aacrjournals.org

Figure 3. A, calculated survival curves for SS patients using the Kaplan-Meier method showing a better survival of individuals with less than three recurrent CNAs compared with individuals with greater than or equal to three CNAs. Twenty-six patients were used for this analysis because one died of unrelated disease and the other was lost at follow-up. B, supervised hierarchical clustering showing two groups oftumor samples: green, tumor samples with greater than or equal to three CNAs; blue, tumors with less than three CNAs. R1-R5 normal CD4+ T cells from healthy donors, P40, P44, are MF patients with normal genomic copy number values. P4b and P26b are replicates. No genomic data are available for the other SS samples. we and others have shown that SS genome harbors a high rate of Taken together, our data support the hypothesis that SS is a chromosomal changes, BUB3 might play a critical role in the etiol- complex disease and occurs as a consequence of multiple genetic ogy of this neoplasm. Also important are cancer genes such as SET lesions: we found a correlation between complex profiles of (9q34), ADRA1A (8p21-p11.2), and the members of the 26S protea- aberration and overall survival in our series, as well as a novel some enzyme complex PSMD3 (17q12-q21.1) and PSMC5 (17q23- signature of genes of potential pathogenetic interest within regions q25). The first one is a well-known oncoprotein involved in of loss/gain. chromatin modeling, transcription, cell cycle control, and protein – phosphatase 2A inhibition (44 47). The second has been implicat- Disclosure of Potential Conflicts of Interest ed in growth-promoting pathways (48), whereas the ubiquitin- proteasome pathway is required for the activation of NF-κB (40). No potential conflicts of interest were disclosed. One transcript, PIP5K1B, maps to the novel recurrent region of deletion 9q13-q21.33; it has been found to regulate directional Acknowledgments leukocyte motility (49). This activity, if deregulated, might result Received 6/26/09; revised 8/7/09; accepted 9/3/09; published OnlineFirst 10/20/09. in the dysfunction of leukocyte responses, such as extravasation Grant support: Ministero della Salute, Ministero dell' Università e della Ricerca Scientifica, Progetto Ialia-USA: Malatie Rare (G. Russo) and from Associazione Italiana and antigen recognition, but also induction and/or inhibition of Ricerca sul Cancro (M.G. Narducci). cellular programs that lead to activation, proliferation, and survival The costs of publication of this article were defrayed in part by the payment of page (50). A detailed characterization of these genes may provide new charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. biological insights into SS and might lead to the development of We thank Dr. Giannandrea Baliva for continuous support and valuable discussion novel clinical tools. and the nurse personnel of the III Dermatological Division.

References Young BD, Whittaker S. Molecular cytogenetic charac- 4. Tracey L, Villuendas R, Dotor AM, et al. Mycosis fun- terization of Sezary syndrome. Genes Chromosomes goides shows concurrent deregulation of multiple genes 1. Willemze R, Jaffe ES, Burg G, et al. WHO-EORTC clas- Cancer 2003;36:250–60. involved in the TNF signaling pathway: an expression sification for cutaneous lymphomas. Blood 2005;105: profile study. Blood 2003;102:1042–50. – 3. Vermeer MH, van Doorn R, Dijkman R, et al. Novel 3768 85. and highly recurrent chromosomal alterations in Sezary 5. Kari L, Loboda A, Nebozhyn M, et al. Classification 2. Mao X, Lillington DM, Czepulkowski B, Russell-Jones R, syndrome. Cancer Res 2008;68:2689–98. and prediction of survival in patients with the leukemic www.aacrjournals.org 8445 Cancer Res 2009; 69: (21). November 1, 2009

phase of cutaneous T cell lymphoma. J Exp Med 2003; of microarrays applied to the ionizing radiation re- genes important in the pathogenesis of T-cell prolym- 197:1477–88. sponse. Proc Natl Acad Sci US A 2001;98:5116 –21. phocytic leukemia with inv(14)(q11q32). Leukemia 6. van Doorn R, Dijkman R, Vermeer MH, et al. Aberrant 21. Wang TL, Maierhofer C, Speicher MR, et al. Digital 2007;21:2153–63. expression of the tyrosine kinase receptor EphA4 and karyotyping. Proc Natl Acad Sci US A 2002;99:16156 –61. 36. Hyman E, Kauraniemi P, Hautaniemi S, et al. Impact the transcription factor twist in Sezary syndrome iden- 22. Janne PA, Li C, Zhao X, et al. High-resolution single- of DNA amplification on gene expression patterns in tified by gene expression analysis. Cancer Res 2004;64: nucleotide polymorphism array and clustering analysis breast cancer. Cancer Res 2002;62:6240–5. – 5578 86. of loss of heterozygosity in human lung cancer cell lines. 37. Mao X, Lillington D, Scarisbrick JJ, et al. Molecular 7. Nigro JM, Misra A, Zhang L, et al. Integrated array- Oncogene 2004;23:2716–26. cytogenetic analysis of cutaneous T-cell lymphomas: comparative genomic hybridization and expression 23. Kotzot D. Complex and segmental uniparental dis- identification of common genetic alterations in Sezary array profiles identify clinically relevant molecular omy (UPD): review and lessons from rare chromosomal syndrome and mycosis fungoides. Br J Dermatol 2002; subtypes of glioblastoma. Cancer Res 2005;65: complements. J Med Genet 2001;38:497–507. 147:464–75. – 1678 86. 24. Pollack JR, Sorlie T, Perou CM, et al. Microarray anal- 38. Wang Y, Putnam CD, Kane MF, et al. Mutation in 8. Tsafrir D, Bacolod M, Selvanayagam Z, et al. Relation- ysis reveals a major direct role of DNA copy number Rpa1 results in defective DNA double-strand break re- ship of gene expression and chromosomal abnormali- alteration in the transcriptional program of human pair, chromosomal instability and cancer in mice. Nat – ties in colorectal cancer. Cancer Res 2006;66:2129 37. breast tumors. Proc Natl Acad Sci US A 2002;99: Genet 2005;37:750–5. 9. Walker BA, Leone PE, Jenner MW, et al. Integration of 12963–8. 39. Izban KF, Ergin M, Qin JZ, et al. Constitutive expres- κ global SNP-based mapping and expression arrays re- 25. Higgins ME, Claremont M, Major JE, Sander C, Lash sion of NF- B is a characteristic feature of mycosis fun- veals key regions, mechanisms, and genes important AE. CancerGenes: a gene selection resource for cancer goides: implications for apoptosis resistance and – in the pathogenesis of multiple myeloma. Blood 2006; genome projects. Nucleic Acids Res 2007;35:D721–6. pathogenesis. Hum Pathol 2000;31:1482 90. – 108:1733 43. 26. Batista DA, Vonderheid EC, Hawkins A, et al. Multi- 40. Sors A, Jean-Louis F, Pellet C, et al. Down-regulating κ 10. Wolf M, Mousses S, Hautaniemi S, et al. High-resolution color fluorescence in situ hybridization (SKY) in mycosis constitutive activation of the NF- B canonical pathway analysis of gene copy number alterations in human pros- fungoides and Sezary syndrome: search for recurrent overcomes the resistance of cutaneous T-cell lymphoma – tate cancer using CGH on cDNA microarrays: impact chromosome abnormalities. Genes Chromosomes to apoptosis. Blood 2006;107:2354 63. of copy number on gene expression. Neoplasia 2004;6: Cancer 2006;45:383–91. 41. Karenko L, Sarna S, Kahkonen M, Ranki A. Chromo- – 240 7. 27. Espinet B, Salido M, Pujol RM, et al. Genetic charac- somal abnormalities in relation to clinical disease in 11. Lastowska M, Viprey V, Santibanez-Koref M, et al. terization of Sezary's syndrome by conventional cytoge- patients with cutaneous T-cell lymphoma: a 5-year – Identification of candidate genes involved in neuro- netics and cross-species color banding fluorescent follow-up study. Br J Dermatol 2003;148:55 64. blastoma progression by combining genomic and ex- in situ hybridization. Haematologica 2004;89:165–73. 42. Logarinho E, Resende T, Torres C, Bousbaa H. The pression microarrays with survival data. Oncogene human spindle assembly checkpoint protein Bub3 is re- – 28. Fischer TC, Gellrich S, Muche JM, et al. Genomic 2007;26:7432 44. aberrations and survival in cutaneous T cell lympho- quired for the establishment of efficient kinetochore-mi- – 12. Mao X, McElwaine S. Functional copy number mas. J Invest Dermatol 2004;122:579–86. crotubule attachments. Mol Biol Cell 2008;19:1798 813. changes in Sezary syndrome: toward an integrated mo- 29. Karenko L, Kahkonen M, Hyytinen ER, Lindlof M, 43. Chan GK, Yen TJ. The mitotic checkpoint: a signaling lecular cytogenetic map III. Cancer Genet Cytogenet Ranki A. Notable losses at specific regions of chromo- pathway that allows a single unattached kinetochore to – – 2008;185:86 94. somes 10q and 13q in the Sezary syndrome detected inhibit mitotic exit. Prog Cell Cycle Res 2003;5:431 9. 13. van Doorn R, van Kester MS, Dijkman R, et al. by comparative genomic hybridization. J Invest Derma- 44. Canela N, Rodriguez-Vilarrupla A, Estanyol JM, et al. Oncogenomic analysis of mycosis fungoides reveals tol 1999;112:392–5. The SET protein regulates G2/M transition by modulat- major differences with Sezary syndrome. Blood 2009; 30. Scarisbrick JJ, Woolford AJ, Russell-Jones R, Whittaker ing cyclin B-cyclin-dependent kinase 1 activity. J Biol – – 113:127 36. SJ. Allelotyping in mycosis fungoides and Sezary syn- Chem 2003;278:1158 64. 14. Popescu NC, Zimonjic DB. Molecular cytogenetic drome: common regions of allelic loss identified on 9p, 45. Cervoni N, Detich N, Seo SB, Chakravarti D, Szyf M. characterization of cancer cell alterations. Cancer Genet 10q, and 17p. J Invest Dermatol 2001;117:663–70. The oncoprotein Set/TAF-1β, an inhibitor of histone – Cytogenet 1997;93:10 21. 31. Thangavelu M, Finn WG, Yelavarthi KK, et al. Recur- acetyltransferase, inhibits active demethylation of 15. Zhao X, Li C, Paez JG, et al. An integrated view of ring structural chromosome abnormalities in peripheral DNA, integrating DNA methylation and transcriptional – copy number and allelic alterations in the cancer ge- blood lymphocytes of patients with mycosis fungoides/ silencing. J Biol Chem 2002;277:25026 31. nome using single nucleotide polymorphism arrays. Sezary syndrome. Blood 1997;89:3371–7. 46. Li M, Makkinje A, Damuni Z. The myeloid leukemia- – Cancer Res 2004;64:3060 71. 32. Fitzgibbon J, Smith LL, Raghavan M, et al. Associa- associated protein SET is a potent inhibitor of protein – 16. Narducci MG, Scala E, Bresin A, et al. Skin homing of tion between acquired uniparental disomy and homozy- phosphatase 2A. J Biol Chem 1996;271:11059 62. Sezary cells involves SDF-1-CXCR4 signaling and down- gous gene mutation in acute myeloid leukemias. Cancer 47. Okuwaki M, Nagata K. Template activating factor-I regulation of CD26/dipeptidylpeptidase IV. Blood 2006; Res 2005;65:9152–4. remodels the chromatin structure and stimulates tran- – 107:1108 15. 33. Mao X, Orchard G, Vonderheid EC, et al. Heteroge- scription from the chromatin template. J Biol Chem – 17. Scala E, Narducci MG, Amerio P, et al. T cell recep- neous abnormalities of CCND1 and RB1 in primary cu- 1998;273:34511 8. β tor-V analysis identifies a dominant CD60+ CD26- taneous T-Cell lymphomas suggesting impaired cell 48. Thebault S, Roudbaraki M, Sydorenko V, et al. α1- CD49d- T cell clone in the peripheral blood of Sezary cycle control in disease pathogenesis. J Invest Dermatol adrenergic receptors activate Ca(2+)-permeable cat- – syndrome patients. J Invest Dermatol 2002;119:193 6. 2006;126:1388–95. ionic channels in prostate cancer epithelial cells. – 18. Lin M, Wei LJ, Sellers WR, Lieberfarb M, Wong WH, 34. Scarisbrick JJ, Woolford AJ, Calonje E, et al. Frequent J Clin Invest 2003;111:1691 701. Li C. dChipSNP: significance curve and clustering of abnormalities of the p15 and p16 genes in mycosis fun- 49. Lacalle RA, Peregil RM, Albar JP, et al. Type I phos- SNP-array-based loss-of-heterozygosity data. Bioinfor- goides and sezary syndrome. J Invest Dermatol 2002; phatidylinositol 4-phosphate 5-kinase controls neutro- – matics 2004;20:1233 40. 118:493–9. phil polarity and directional movement. J Cell Biol – 19. Arcelli D, Palmieri A, Pezzetti F, Brunelli G, Zollino I, 35. Durig J, Bug S, Klein-Hitpass L, et al. Combined sin- 2007;179:1539 53. Carinci F. Genetic effects of a titanium surface on osteo- gle nucleotide polymorphism-based genomic mapping 50. Vicente-Manzanares M, Sanchez-Madrid F. Role of – blasts: a meta-analysis. J Oral Sci 2007;49:299 309. and global gene expression profiling identifies novel the cytoskeleton during leukocyte responses. Nat Rev 20. Tusher VG, Tibshirani R, Chu G. Significance analysis chromosomal imbalances, mechanisms and candidate Immunol 2004;4:110–22.

Cancer Res 2009; 69: (21). November 1, 2009 8446 www.aacrjournals.org

Identification of Key Regions and Genes Important in the Pathogenesis of Sézary Syndrome by Combining Genomic and Expression Microarrays

Elisabetta Caprini, Cristina Cristofoletti, Diego Arcelli, et al.

Cancer Res Published OnlineFirst October 20, 2009.

Updated version Access the most recent version of this article at: doi:10.1158/0008-5472.CAN-09-2367

Supplementary Access the most recent supplemental material at: Material http://cancerres.aacrjournals.org/content/suppl/2009/10/16/0008-5472.CAN-09-2367.DC1

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

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

Permissions To request permission to re-use all or part of this article, use this link http://cancerres.aacrjournals.org/content/early/2009/10/20/0008-5472.CAN-09-2367. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.