Integrated Genomic Analysis of Sézary Syndrome

Sézary syndrome (SS) is a rare variant of primary cutaneous T-cell lymphoma. Little is known about the underlying pathogenesis of S. To address this issue, we used Affymetrix 10K SNP microarray to analyse 13 DNA samples isolated from 8 SS patients and qPCR with ABI TaqMan SNP genotyping assays for the validation of the SNP microarray results. In addition, we tested the impact of SNP loss of heterozygosity (LOH) identified in SS cases on the gene expression profiles of SS cases detected with Affymetrix GeneChip U133A. The results showed: (1) frequent SNP copy number change and LOH involving 1, 2p, 3, 4q, 5q, 6, 7p, 8, 9, 10, 11, 12q, 13, 14, 16q, 17, and 20, (2) reduced SNP copy number at FAT gene (4q35) in 75% of SS cases, and (3) the separation of all SS cases from normal control samples by SNP LOH gene clusters at chromosome regions of 9q31q34, 10p11q26, and 13q11q12. These findings provide some intriguing information for our current understanding of the molecular pathogenesis of this tumour and suggest the possibility of presence of functional SNP LOH in SS tumour cells.

Single nucleotide polymorphism (SNP) denotes a single base in the DNA sequence that differs from the usual base at that position. Millions of SNPs have been catalogued in the human genome, which may be responsible for disease such as cancer (http://www.ncbi.nlm.nih.gov/sites/entrez?db=snp). SNP microarray is a powerful genome research technique designed for the identification of SNP and dose change within the whole genome [29][30][31]. This method has been widely used to investigate SNP copy number change and LOH in cancer genome in a variety of medical and biological subjects [32][33][34][35][36][37][38][39][40][41][42]. However, there has been no report describing the use of SNP microarray to assess genomewide SNP copy number change and LOH in SS in the literature (http:// www.ncbi.nlm.nih.gov/sites/entrez).
To address this issue we initially performed SNP microarray on DNA samples isolated from biopsies of skin lesions and peripheral blood mononuclear cells (PBMCs) of patients with SS by using Affymetrix GeneChip Mapping 10K Xba assay. Then we assessed gene-specific SNP copy number changes in same DNA samples with ABI TaqMan SNP genotyping assays to compare and verify the findings from Affymetrix 10K SNP microarray analysis. Finally, we conducted a literature review on LOH findings of patients with SS. Our findings are presented here in this paper. GeneChip Human Mapping 10K Array (http://www.affymetrix.com/products/arrays/specific/10k.affx) and analysed by using GOLF software (https://bioinformatics.cancerresearchuk.org/cazier01). Here each red-coloured line represents copy number gain of individual SNP, greencoloured line and block stand for SNP copy number loss, black-coloured line and block denote balanced/neutral SNP copy number or uniparental disomy. Similar SNP homozygosity profile is present in paired DNA samples from the biopsied skin lesion (S) and peripheral blood (B) isolated from individual patient with SS on the right end of the diagram. Frequent SNP homozygous changes (>3) are seen on almost all autosomes consistent with SNP copy number alterations shown in Figure 2. The remarkable changes on X chromosome were used as the internal quality control for the 10K SNP array experiment as green represents X chromosome copy number loss or monosomy indicating male sex, and black signifies balanced or disomy X chromosome denoting female.  Figure 2). Additional 2 SNP genotyping assays amplifying beta-2-microglobulin (B2M, rs935885, C 12080829 10) and glyceraldehyde-3-phosphate dehydrogenase (GAPDH, rs3741918, C 27510362 10) were used as the internal quality controls. qPCR reactions contained 900 nmol primers, 200 nmol probes, 400 mM each dATP, dCTP and dGTP, 800 mM dUTP, 1 U Amplitaq Gold DNA polymerase, 0.2 U AmpErase uracil N-glycosylase (UNG), and 13 TaqMan buffer in a total volume of 25 μL in 96-well microtiter plate (Applied Biosystems). After a decontamination step at 50 • C, a two-step protocol was followed for 50 cycles: 95 • C for 15 s and 61 • C for 1 min. SNP qPCR data analysis was conducted as previously described [45] with ABI SNPbrowser Software (http://marketing.appliedbiosystems.com/mk/get/snpb landing).

SNP Loss of Heterozygosity Expression Profiling.
We have previously put forward a concept of functional copy number changes in cancer cells [46]. To further test our hypothesis in this study, we assume a SNP LOH gene cluster in SS is bioinformatically significant or functional if it is capable of separating all SS cases from normal control (NC) samples. This test was carried out through a data mining experiment by using a bioinformative method developed by us [46]. In this experiment, the effect of SNP LOH detected in this study on gene expression profiling in SS cases conducted in previous studies [23,24,47] was tested as before [46]. The procedures were briefly described as follows. We initially created gene lists from 9 chromosomal regions showing frequent LOH (>3 per region) at SNP level seen in SS (Table 1) by using IdeogramBrowser Software and UCSC Human Gene Sorter (http://www.genome.ucsc.edu/). Then we imported these 9 gene-lists into GeneSpring software version 7 (http://www.sigenetics.com/), which contained a genome/raw data of gene expression profile from 6 SS patients who were treated with photopheresis therapy and 2 healthy individuals generated by using Affymetrix GeneChip U133A (http://www.affymetrix.com/products/arrays/specific/hgu133av2.affx). Of 6 SS cases analysed with Affymetrix gene expression microarray the 3 cases were also tested by above Affymetrix SNP microarray. Finally, we analysed the selected 9 gene lists by using the clustering tool of Gene-Spring to generate gene tree and condition tree, a supervised clustering, which enables direct visualisation of sample congregation and separation [46].

SNP Microarray.
In this study, paired DNA samples from PBMCs and skin lesions from 5 of 8 SS cases were analysed with Affymetrix 10 K SNP microarray, which showed similar SNP homozygous changes between the paired samples indicating they are coming from the same individual. There was a difference in SNP fingerprint among 8 different SS cases signifying the absence of sample contamination. As Figure 1 shows, frequent homozygous changes (>3 per chromosome) including gain, loss, and uniparental disomy (UPD) were seen to distribute throughout almost an entire genome except of chromosomes 14, 19 and 22. As Figure 2 shows there were frequent SNP copy number changes (>3 per chromosome) at chromosomes 1, 2p, 3, 4q, 5q, 6, 7p, 8, 9, 10, 11, 12q, 13, 14, 16q, 17, and 20, which are not only consistent with above homozygous changes but also in line with previous CGH findings in SS [10].
In this study, a weak and/or smear band of TCR gene was detected in skin DNA samples, while a strong monoclonal TCR gene band was seen in PBMC DNA samples (data not shown), suggesting that only a few tumour cells were present in skin samples and sample contamination was negligible. Overall, a frequent LOH was noted in 5 paired SS cases at 9 chromosomal regions including 1p36p33, 2q21q24, 8q22q24, 9q31q34, 10p11q26, 11q21q25, 13q11q12, 16q21q23, and 17p13q11 (Table 1), which are to some extent similar to previous reports on LOH at certain chromosome regions [9,[15][16][17][18][19].

SNP LOH Expression
Profiling. The impact of SNP LOH gene cluster on gene expression pattern was further tested in this study. As shown in Figure 4, the clusters of 2 NCs appeared at the right end of the heatmap and the clusters of 6 SS cases were present in the middle and at the left end of the heatmap. This clearly showed that two groups were separated and different. Over the 9 SNP LOH gene lists analysed in this study, 3 (33%) showed gene expression patterns of the separation of 6 SS cases from 2 NCs (Table 1). This included SNP LOH gene clusters at 9q31q34 (40 genes), 10p11q26 (111 genes), and 13q11q12 (15 genes) (Figure 4). Each individual gene in these 3 gene clusters is listed in Table 2. However, the remaining 6 SNP LOH gene lists revealed a mixed gene expression pattern without the separation of SS from NCs (Table 1).

Refining Previous SS LOH Data and Comparing with
This Study. Previous studies have shown LOH on several chromosomes including chromosomes 1 and 10q [9,[15][16][17][18][19]48]. The 10K SNP microarray gene chip used in this study is equivalent to the genetic distance covered by 400 MS. This allows us to compare SNP LOH identified in this study with MS LOH described in those previous studies. Frequent LOH at MS loci D1S247 (rs11372930) at 1p36 and D10S562 (rs4128597) at 10q25.3 ( Figure 5) reported in previous studies [9,18,19] were seen in this study.

Discussion
This integrated genomic study has for the first time revealed genomewide SNP copy number change and LOH as well as SNP LOH gene expression profiling in SS cases. Although the number of SS cases tested in this study was small, the patterns of SNP copy number change and LOH identified in SS cases are generally in line with previous metaphase-/array-CGH [6][7][8][9][10][11][12][13][14] and MS LOH studies [9,15,[17][18][19]48], suggesting the high efficiency and reliability of the methods used. In addition some novel findings emerged. This includes the presence of SNP copy number loss at the FAT gene (4q35) in most SS cases and the separation of all SS cases from NC samples by SNP LOH gene clusters at chromosome regions of 9q31q34, 10p11q26, and 13q11q12. Although all of these results will need to be validated in larger cohorts of samples from SS patients in the future, they do provide some intriguing information for our current understanding of the molecular pathogenesis of this type of lymphoma and suggest the possibility of presence of functional SNP LOH in SS tumour cells at least from the point of view of bioinformatics.
Although there are different definitions for loss of heterozygosity in literature, LOH is generally defined as a locus or loci at which a deletion or other process has converted the locus from heterozygosity to homozygosity or hemizygosity. In terms of cancer, LOH signifies that in tumour cells carrying a mutated allele of a tumour suppressor gene, the gene becomes fully inactivated when the cell loses a large part of the chromosome carrying the wild-type allele (http://www.ncbi.nlm.nih.gov/cancerchromosomes). LOH is traditionally used as a molecular tool to identify a tumoursuppressor gene but it also represents one of the most common molecular features of cancer cells [47,[50][51][52]. Previous studies have shown LOH at MS loci on chromosomes 1p, 9p, 10q, 12q, 13q, 17p, and 19 in CTCL including SS [9,[15][16][17][18][19]48]. This was supported by previous conventional cytogenetics and metaphase and array-based CGH studies, which revealed DNA copy number losses at these chromosomal regions [6][7][8][9][10][11][12][13][14]48]. In this study LOH at SNP level on 9 chromosomal regions (1p, 2q, 8q, 9q, 10, 11q, 13q, 16q, and 17)    Figure 5: A summary of reanalysis of previously published LOH at 18 microsatellite loci on 10q21q26 in SS and MF [19] with this study, in which each microsatellite locus is finely mapped and linked to individual gene and SNP via NCBI ENTREZ GENE and ENTREZ SNP database (http://www.ncbi.nlm.nih.gov/). Here black bar represents LOH, gray bar stands for noninformative, and white bar indicates microsatellite marker that is not available. One SNP within microsatellite locus D10S562 (bold) is also present in the SNP LOH cluster at 10p11q26 in this study (Table 1).     [15-17, 19, 48]. In addition, copy number losses were described in these 9 chromosomal regions by using molecular cytogenetic techniques as discussed above. Furthermore, these findings were subsequently validated by extensive SNP qPCR test in this study. Taking together all of these results suggest that some of these SNP LOH and copy number changes are likely to be associated with the pathogenesis of SS. The FAT gene, which is expressed at high levels in a number of foetal epithelia, is an ortholog of the Drosophila fat gene. FAT encodes a tumour suppressor essential for controlling cell proliferation during Drosophila development. Its product is a member of the cadherin superfamily, a group of integral membrane proteins characterised by the presence of 34 tandem cadherin-type repeats as well as five epidermal growth factor-like repeats and one laminin A-G domain with possible functions as an adhesion molecule and/or signalling receptor, which are likely to be associated with developmental processes and cell communication (http://www.ncbi.nlm.nih.gov/gene/2195). The FAT gene is located on human chromosome 4q35. Previous molecular cytogenetic studies have revealed copy number losses at this chromosomal region in a variety of cancers (http://www.progenetix.org/cgi-bin/pgCasesMatrix-Plotter.cgi). Furthermore deletion/LOH and absent expression of FAT have also been found in oral cancer [53] and human astrocytic tumours [54]. Intriguingly, in this study SNP copy number loss of FAT was observed in three-quarter of SS cases and reduced gene expression of FAT was also noted in most cutaneous squamous cell carcinoma cases (data not shown) no spare SS RNA samples are available for qPCR testing. These results further indicate the tumour suppressor role of FAT and also suggest that loss of FAT is relevant to SS pathology.
Interindividual copy number variation/polymorphism (CNV/P) is thought to be the manifestation of a considerable and unanticipated plasticity of the human genome. CNV/P constitutes a major source of interindividual genetic variation that could explain variable penetrance of inherited diseases and variation in the phenotypic expression of aneuploidies and sporadic traits [55,56]. There is increasing evidence showing de novo CNV/P as a major cause of mental retardation and several other complex disorders [38-40, 43, 57]. Cytogenetically CNV/P or copy number changes (CNCs) have long been described in cancer cells (http:// www.cgap.nci.nih.gov/Chromosomes/Mitelman, http://www .progenetix.net/). As discussed above SS has been found with frequent CNCs at several chromosomal regions [6][7][8][9][10][11][12][13]. Some of these CNCs may have a net effect on gene expression as previous expression profiling studies on SS have revealed up-and downregulations of several genes, which are mapped within the chromosomal regions harbouring these CNCs [20][21][22][23][24][25][26]. From, bioinformatic point of view these CNCs may be functional. Whilst the remaining CNCs (CNV/P) may not have any net effect at all and are therefore nonfunctional. To test this hypothesis we developed a simple bioinformatics strategy through the analysis of the expression profiling of CNCs gene clusters commonly presented in SS to see if SS cases can be separated from NCs. We found separated expression patterns in 4 of 17 CNCs gene clusters [46]. In this study, we expanded the scope of our bioinformatics experiment by examining the impact of common SNP LOH gene clusters identified in SS cases with Affymetrix 10 K SNP microarray on gene expression patterns in SS cases detected using gene expression microarray with Affymetrix GeneChip U133A and dedicated software. There was a separation of all 6 SS cases from 2 NCs by 3 of 9 SNP LOH gene clusters. This includes 9q31q34, 10p11q26, and 13q11q12, which lie within the chromosomal regions showing frequent LOH and/or DNA copy number loss in SS. In addition, decreased expression of MXI1 in SNP LOH gene cluster at 10p11q26 has been described in SS previously [22]. All of these findings suggest that the 3 SNP LOH gene clusters may be functional in terms of bioinformatics and relevant to SS pathology. There are also a number of genes in these 3 SNP LOH gene lists which have not been described in SS before. Further functional studies, including proteomic analysis, are therefore necessary to elucidate the oncogenic role of these genes in the initiation and progression of SS and other types of CTCL.