Spermatocytic seminomas are solid tumors found solely in the testis of predominantly elderly individuals. We investigated these tumors using a genome-wide analysis for structural and numerical chromosomal changes through conventional karyotyping, spectral karyotyping, and array comparative genomic hybridization using a 32 K genomic tiling-path resolution BAC platform (confirmed by in situ hybridization). Our panel of five spermatocytic seminomas showed a specific pattern of chromosomal imbalances, mainly numerical in nature (range, 3-24 per tumor). Gain of chromosome 9 was the only consistent anomaly, which in one case also involved amplification of the 9p21.3-pter region. Parallel chromosome level expression profiling as well as microarray expression analyses (Affymetrix U133 plus 2.0) was also done. Unsupervised cluster analysis showed that a profile containing transcriptional data on 373 genes (difference of ≥3.0-fold) is suitable for distinguishing these tumors from seminomas/dysgerminomas. The diagnostic markers SSX2-4 and POU5F1 (OCT3/OCT4), previously identified by us, were among the top discriminatory genes, thereby validating the experimental set-up. In addition, novel discriminatory markers suitable for diagnostic purposes were identified, including Deleted in Azospermia (DAZ). Although the seminomas/dysgerminomas were characterized by expression of stem cell–specific genes (e.g., POU5F1, PROM1/CD133, and ZFP42), spermatocytic seminomas expressed multiple cancer testis antigens, including TSP50 and CTCFL (BORIS), as well as genes known to be expressed specifically during prophase meiosis I (TCFL5, CLGN, and LDHc). This is consistent with different cells of origin, the primordial germ cell and primary spermatocyte, respectively. Based on the region of amplification defined on 9p and the associated expression plus confirmatory immunohistochemistry, DMRT1 (a male-specific transcriptional regulator) was identified as a likely candidate gene for involvement in the development of spermatocytic seminomas. (Cancer Res 2006; 66(1): 290-302)
- Spermatocytic seminoma
- genomic profiling
- expression profiling
- cell of origin
- positional candidate gene(s)
- diagnostic markers
Spermatocytic seminomas are benign testicular tumors that exceptionally may progress to sarcoma ( 1– 4). It is becoming increasingly clear that these tumors are not a variant of (classic) seminoma (see ref. 5 for review). On the other hand, the histology of dysgerminomas of the ovary is virtually undistinguishable from seminomas of the testis (see ref. 6 for review). These types of germ cell tumors are thought to originate from a primordial germ cell or gonocyte, whereas the spermatocytic seminomas are suggested to originate from a more mature germ cell, likely a spermatogonium/spermatocyte. This is based on morphologic characteristics, as well as expression of markers, including placental-like alkaline phosphatase (PLAP; ref. 7), xeroderma pigmentosa protein A (XPA), synaptonemal complex protein 1 (SYCP1), and the synovial sarcoma protein on the X chromosome (SSX2-SSX4; ref. 8). Moreover, this has been supported by analysis of CHEK2, P53, p19INK4Ad (CDKN2D), the cancer testis antigen (CTA) MAGEA4 ( 9), and transcription factor POU5F1 (OCT3/OCT4), the latter related to pluripotency ( 10). The seminomas and spermatocytic seminomas differ also in their pattern of genomic imprinting ( 11); the seminomas show an erased pattern of genomic imprinting ( 12), whereas spermatocytic seminomas show a more paternal pattern ( 13, 14). These findings are in line with the model that the tumor types originate from germ cells at different stages of maturation, and that (most of) the characteristics found in the tumors are intrinsic to the cell of origin ( 6). Finally, the seminomas/dysgerminomas and spermatocytic seminomas show a different chromosomal constitution, as determined by DNA flow cytometry, karyotyping, and conventional comparative genomic hybridization (CGH; refs. 15, 16 and references therein). Although the seminomas are consistently aneuploid (around the hypertriploid range), the spermatocytic seminomas contain tumor cells with a diploid, tetraploid, and hypertetraploid DNA content ( 17). Gain of the short arm of chromosome 12 is the consistent finding in seminomas/dysgerminomas (see refs. 18, 19 for review), whereas additional copies of chromosome 9 has been found in all spermatocytic seminomas studied to date ( 15, 16).
Thus far, no systematic study has been done integrating both genome-wide DNA copy number profiling and gene expression analysis of spermatocytic seminomas. Therefore, we initiated a study on a unique set of frozen tumor samples. For the investigation of the genomic constitution, karyotyping, spectral karyotyping (SKY) as well as array CGH using a tiling-resolution 32 K BAC array platform, and in situ hybridization were applied. Expression analysis was done using the genome-wide Affymetrix U133 plus 2.0 array, as well as the comparative expressed sequence hybridization (CESH) technique (see ref. 20 for review). Based on integration of data obtained through this multiplatform approach, we show that the spermatocytic seminomas, in contrast to seminomas/dysgerminomas, originate from primary spermatocytes, and that have at least initiated prophase meiosis I. In addition, DMRT1 is identified as a likely candidate gene to explain the selective advantage of the observed consistent gain of chromosome 9 in these tumors.
Materials and Methods
Materials. Tissues use for the reported studies was approved by an institutional review board (MEC 02.981). Samples were used according to the “Code for Proper Secondary Use of Human Tissue in the Netherlands,” developed by the Dutch Federation of Medical Scientific Societies (version 2002; ref. 21).
In total, three dysgerminomas (of the ovary), four seminomas (of the testis), and five spermatocytic seminomas were included in this study. All were primary tumors, only treated with surgery, not with previous irradiation and/or chemotherapy. The spermatocytic seminomas (except case 5) have been reported previously ( 15). Tumors were collected in the Netherlands (Rotterdam) as well as abroad, as reported before ( 19). Representative parts of the tumor were snap frozen in liquid nitrogen, and/or were fixed in 10% buffered formalin for paraffin embedding. Diagnosis was done by a pathologist experienced in GCT pathology (J.W.O.) according to the classification of the WHO ( 22) and supported by immunohistochemistry using antibodies directed against germ cell-specific alkaline phosphatase (PLAP), α-fetoprotein, human chorionic gonadotropin, stem cell factor receptor (c-KIT), and the marker of pluripotency POU5F1 (OCT3/OCT4; ref. 10).
Tissue microarrays (TMA) were generated including 50 primary seminomas and 23 spermatocytic seminomas, according to standard procedures. Besides the invasive component, also the precursor lesions (i.e., carcinoma in situ/intratubular germ cell neoplasia unclassified and intratubular spermatocytic seminoma, respectively) were represented in the array. Representative parts of the paraffin-embedded tissues of all frozen tumors investigated in this study for their expression, and genomic imbalances were included in the TMA.
Immunohistochemistry. Immunohistochemical staining was done as described previously ( 8). Three-micrometer-thick paraffin-embedded tissue sections were incubated with the primary antibodies overnight at 4°C [PLAP (1:200, Cell Marque, Hot Springs, AR) and c-KIT (1:500, DakoCytomation, Glostrup, Denmark)], or for 2 hours at room temperature [POU5F1-OCT3/OCT4 (1:1000, Santa Cruz Biotechnology, Santa Cruz, CA)]. Antibodies against SSX2 to SSX4, XPA, and SYCP1 were used as described before ( 8). In addition, specific antibodies against DMRT1 (kindly provided by Drs. David Zarkower and Vivian Bardwell; dilution 1:1000) and Deleted in Azospermia (DAZ; kindly provided by Dr. Renee Reijo Pera, Howard Hughes Medical Institute, Whitehead Institute and Department of Biology, Massachusetts Institute of Technology, Cambridge, MA; ref. 23; dilution 1:400) were used. All slides for single staining experiments were counterstained with hematoxylin. New fuchine was used as chromogene for the detection of PLAP, c-KIT, and SSX, resulting in a red signal of cytoplasmic, membranous, and cytoplasmic/nuclear localization, respectively. 3,3′-Diaminobenzidine was used as chromogene for the detection of POU5F1-OCT3/OCT4, XPA, DMRT1, and DAZ, resulting in a brown nuclear signal.
DNA/RNA isolation and quality control. DNA was isolated from frozen tissue. For each sample, 10 pieces of 10-μm-thick histologic sections were cut from each representative block and used for standard DNA isolation (1 mg/mL proteinase K digestion/SDS, phenol extraction, and ethanol precipitation). Parallel 3-μm-thick sections were stained with H&E and alkaline phosphatase enzymatic detection for histologic control in case of seminoma/dysgerminoma [i.e., analysis of percentage of tumor cells (>80%)]. DNA was dissolved in 10 mmol/L Tris/1 mmol/L EDTA (pH 7.5).
High-quality total RNA was extracted with RNAqueous-4PCR kit (Ambion Europe, Huntingdon, United Kingdom) according to the manufacturer's instructions. RNA concentration, quality, and purity were examined using denaturing gel analysis. None of the samples showed RNA degradation [28S/18S RNA (rRNA) ratio ≥ 2] and/or contamination with DNA.
cDNA synthesis and real-time quantitative PCR. For reverse transcription, 1 μL oligo-d(T)12-18 primers (0.6 μg/μL; Invitrogen, Carlsbad, CA) and 0.5 μL oligo d(N)6 primers (0.6 μg/μL; Amersham, Piscataway, NJ) were added to 2 μg of total RNA, isolated as described above, in a final volume of 9 μL. Samples were subsequently incubated for 5 minutes at 70°C, 1 minute at room temperature, and chilled on ice. Samples were spun down using centrifugation and incubated with 6 μL 5× First Strand Buffer (Invitrogen), 2 μL DTT (0.1 mol/L), 1 μL deoxynucleotide triphosphates (10 mmol/L), 10 μL threhalose (1.7 mol/L; Sigma, St. Louis, MO), 0.5 μL RNaseOUT (40 units/μL; Invitrogen), and 1.5 μL Superscript II reverse transcriptase (200 units/μL; Invitrogen). The reverse transcription reaction was carried out for 50 minutes at 45°C and terminated by incubating the samples for 10 minutes at 70°C.
Quantitative PCR was done using the real-time PCR ABI PRISM 7700 sequence detector system (Applied Biosystems, Foster City, CA). The PCR reactions were done in a final volume of 25 μL containing cDNA synthesized from 2 μg total RNA (see above), 330 nmol/L primers, and 12.5 μL SYBR-green PCR master mix (Applied Biosystems). The input of cDNA for each tumor sample was determined based on expression analysis of the housekeeping gene hypoxanthine-guanine phosphoribosyltransferase (HPRT), being the concentration resulting in a threshold cycle (Ct) of 23. After 10 minutes of denaturation and activation of the Taq-DNA polymerase, PCR products were amplified in 35 cycles of 15 seconds of denaturation at 95°C, 30 seconds of annealing at 62°C, 10 seconds of ramping to 72°C, 20 seconds of extension at 72°C, 10 seconds of ramping to 79°C, and 20 seconds of extension at 79°C. A dissociation curve was run at the end of the reaction for product specificity. Quantitative values were obtained from the Ct. Each target mRNA was quantified relative to HPRT (target mRNA = 2(mean Ct HPRT − mean Ct target)). Intron spanning primers used are DMRT1 forward, 5′-GAGAACAATGGCAGTAACCCG-3′ and reverse, 5′-ACAGAGACGGCTGGTAGAAG-3′; DMRT3 forward, 5′-CTGCGTTGGACTGCCGAG-3′ and reverse, 5′-TCCACGGACACTATCTCAG-3′; HPRT forward, 5′-CGTGGGGTCCTTTTCACCAGCAAG-3′ and reverse, 5′-AATTATGGACAGGACTGAACGTC-3′.
Gene profiling and quality control. The five spermatocytic seminomas were investigated using the CESH technique ( 20), as described before ( 24). Normal testicular tissue RNA was used as reference and labeled in green, whereas tumor RNA was labeled in red. Self-to-self hybridizations were used to determine the cutoff ratios of 0.8 and 1.2 to define differential expression. Multiple metaphase spreads, with a minimum of six, were investigated per tumor.
All samples were analyzed using Affymetrix GeneChip Human Genome U133A plus 2.0 arrays. Each transcript on this chip is represented by a set of 11 probe pairs, called the probe set. The GeneChip contains >54,000 probe sets, representing 47,400 transcripts, including 38,500 genes. The intensity of hybridization of labeled mRNA to these sets reflects the level of expression of a particular transcript. Ten micrograms of total RNA were used to prepare anti-sense biotinylated RNA, according to the affymetrix GeneChip eukaryotic one-cycle target preparation protocol (Affymetrix, Santa Clara, CA). In short, single-stranded cDNA was synthesized using a T7-Oligo(dT) Promoter Primer followed by RNase H–facilitated second-strand cDNA synthesis, which was purified and served as a template in the subsequent in vitro transcription (IVT). The IVT reaction was carried out in the presence of T7 RNA polymerase and a biotinylated nucleotide analogue/ribonucleotide mix for cRNA. The biotinylated cRNA targets were then cleaned up and fragmented. The quantity of the fragmented-labeled cRNA was determined using standard spectrophotometric analysis, and the quality was checked on an Agilent 2100 Bioanalyzer (Agilent, Amstelveen, the Netherlands), using an RNA 6000 NANO assay. Fragmented biotinylated cRNA was hybridized to the GeneChip (45°C for 16 hours). Staining, washing, and scanning procedures were carried out as described in the GeneChip Expression Analysis Technical Manual (Affymetrix). All GeneChips were visually inspected for irregularities. The global method of scaling/normalization was applied. All additional measures of quality [percent genes present (50.6 ± 3.8), actin 3′ to 5′ ratio (1.24 ± 0.19), and glyceraldehyde-3-phosphate dehydrogenase 3′ to 5′ ratio (1.05 ± 0.14)] indicated high overall sample and assay quality.
Data normalization, analysis and visualization. All intensity values were scaled to an average value of 150 per GeneChip according to the GCOS method available in the Affymetrix Microarray Suite. For each probe set, the geometric mean of the hybridization intensities of all patient samples was calculated. The level of expression of each probe set in every sample was determined relative to this geometric mean and logarithmically transformed (on a base log 2 scale) to ascribe equal weight to gene expression levels with similar relative distances to the geometric mean. Deviation from the geometric mean reflects differential gene expression. The transformed expression data were subsequently imported into Omniviz software, version 3.6 (Omniviz, Inc., Mynard, MA), significance analysis of microarrays (SAM) software, version 1.21. Principle component analysis was done as part of the Spotfire software package (Spotfire, Inc., Sommerville, MA; Spotfire decision site V8.1).
Data analyses using Omniviz and SAM. The Omniviz package was used for performing and visualizing the results of unsupervised cluster analysis (an analysis that does not take into account external information such as histology). Genes whose level of expression differed from the geometric mean (reflecting up-regulation or down-regulation) in at least one patient were selected for further analysis. The clustering of molecularly recognizable specific groups of patients was investigated with each of the selected probe sets using the Pearson's Correlation and Visualization tool of Omniviz. The cutoff value of expression modulation was set to either 2.5- or 3.0-fold (resulting in 1,019 and 373 differentiating genes, respectively). This study will only focus on the data obtained from the analysis using the 3.0-fold change.
All supervised analyses were done using SAM software ( 25). Genes identified as being discriminating using Omniviz were used for the analysis. A supervised analysis correlates gene expression with an external variable being histology. SAM calculates a score for each gene based on the change in gene expression relative to the SD of all measurements. The criteria for identifying the top 50 genes for an assigned cluster were a minimum of a 2-, 10-, 20-, 50-, and 100-fold difference in gene expression between the assigned cluster and the other samples by a factor of at least 2.0 and a q < 2%. The q for each gene represents the probability that it is falsely called significantly deregulated.
SKY analysis. Cytogenetic preparations were pretreated with RNase for 60 minutes at 37°C and digested with pepsin for 10 minutes. After washing, cells were fixed in 1% formaldehyde/50 mmol/L MgCl2 in PBS for 10 minutes, washed, and dehydrated, as reported before ( 19). Hybridization for SKY analysis was done according to the manufacturer's protocol with minor adjustments (Applied Spectral Imaging, Migdal Ha'Emek, Israel). In short, after 2 days of hybridization, slides were washed for 10 minutes with 55% formamide/2× SSC (pH 7.0) followed by 1× SSC at 39°C and 4× SSC/0.05% Tween 20 at room temperature. Cells were counterstained with 4′,6-diamidino-2-phenylindole and mounted in antifade solution (Dabco-Vectashield 1:1). Using the Spectra Cube 300 system and Skyview analysis software (ASI), nine metaphase cells were examined ( 26).
Fluorescence in situ hybridization. For each case analyzed, 5-μm-thick tissue sections were cut and air-dried at 55°C overnight on starfrost microscope slides. Two adjacent sections were used for each histologic examination. Fluorescent in situ hybridization (ISH) was done as described previously ( 27). The following BACs were used: RP11-117J18, RP11-621O18, and RP11-344A7 (all three located on 9p21.3-9p23). Probes were labeled with biotin-16-dUTP and detected using streptavidin-CY3 (Jackson ImmunoResearch Laboratories, West Grove, PA). The centromere-specific probe for chromosomes 9 was labeled with digoxigenin-11-dUTP (Roche, Mannheim, Germany) and visualized with FITC-conjugated sheep-anti-digoxigenin (Roche). Results were confirmed with reverse labeling. Probes were verified to be specific for the regions of interest on metaphase spreads isolated from in vitro cultured human lymphocytes, both male and female.
Array CGH. We prepared a tiling-resolution microarray consisting of 32,447 overlapping BAC clones selected to cover the entire human genome ( 28, 29) and available through BACPAC resources 8 using methodology essentially as described before ( 30). In brief, genomic target DNAs were isolated from 1 mL bacterial cultures using an AutogenPrep 960 (Autogen, Holiston, MA), following the instructions of the manufacturer. Degenerate oligonucleotide primed (DOP) PCR was done on 50 ng DNA from all clones essentially as described before ( 31) with minor modifications ( 32). DOP-PCR products were dissolved at a concentration of 1 μg/μL in a 30% DMSO solution and spotted onto CMT-ULTRAGAPS-coated glass slides (Corning, Acton, MA) using an Omnigrid 100 arrayer (Genomic Solutions, Ann Arbor, MI).
Analysis of the microarray images obtained from the BAC hybridizations was done using the software package GenePix Pro 5.0 (Axon Instruments, Inc., Foster City, CA). For each spot, the median pixel intensity minus the median local background for both dyes was used to obtain a genomic copy number ratio. Data normalization was done in the software package SAS version 8.0 (SAS Institute, Cary, NC) for each array subgrid, by applying Lowess curve fitting with a smoothing factor of 0.3 to predict the log 2–transformed test-over-reference (T/R) value based on the average logarithmic fluorescent intensities. All mapping information regarding clone locations, cytogenetic bands, and gene content were retrieved from the University of California Santa Cruz genome browser (May 2004 freeze).
Five spermatocytic seminomas (including one pair of bilateral tumors from a single patient, cases 3 and 4), four testicular seminomas, and three ovarian dysgerminomas were investigated in this study. All spermatocytic seminomas were diagnosed as stage I and were found to stain positive by immunohistochemistry for the diagnostic markers XPA, SYCP1, VASA, and SSX2 to SSX4 and negative for the markers characteristic for seminoma/dysgerminoma, including PLAP, c-KIT, and POU5F1-OCT3/OCT4 ( 7, 8, 10). As expected, both the seminomas and dysgerminomas showed the opposite pattern (see below).
Expression profiling. All seminomas, dysgerminomas, and spermatocytic seminomas were expression profiled using Affymetrix U133 plus 2.0 arrays. Omniviz analysis using unsupervised clustering showed that the seminomas and dysgerminomas could not be distinguished based on mRNA expression (data not shown). Therefore, throughout this article, these two types of tumors will be considered together. Subsequently, the total sets of tumors (seminomas/dysgerminomas and spermatocytic seminomas) were compared. Principle component analysis showed that these tumors are subdivided into defined profile entities (see Fig. 1A ). This was confirmed using Omniviz, which separated these two histologic groups of tumors successfully based on 373 genes with an at least 3-fold difference in expression ( Fig. 1B). Correlation analyses showed that the spermatocytic seminomas were more similar to each other than the seminomas/dysgerminomas ( Fig. 1C, shown by the various color intensities). The bilateral cases (cases 3 and 4) were not more similar to each other than to the other tumors. This defined set of 373 genes was used for supervised analysis (SAM), in which the seminomas/dysgerminomas (group 1) were compared with the spermatocytic seminomas (group 2). The discriminating genes were classified based on their difference in level of expression (2-, 10-, 20-, 50- up to 100-fold), which revealed a log 2–association between the fold change and number of differentiating genes ( Fig. 1D). The top 50 differentiating genes [score(d) higher than 4.00 or lower than −4.00] were investigated in detail and are listed in Table 1 . The role of the protein encoded by the gene, or the process in which it is involved, is indicated.
Identification of significant biomarkers and cell of origin. Our expression profiling results show that some of the validated diagnostic markers previously identified by us and others for seminoma/dysgerminoma and spermatocytic seminomas are indeed in the most highly ranking distinguishing genes (see Table 1, shaded and Table 2 ). These include POU5F1-OCT3/OCT4 at position 1 for seminoma/dysgerminoma, being one of the six genes showing at least a 100-fold expression difference. The spermatocytic seminoma specific SSX2-SSX4 genes showed a 20-fold expression difference. Although the antibody used for immunohistochemistry only detects the SSX2 to SSX4 variants, expression data clearly show that also SSX1 (at position 10 of the top list of genes, P = 1.85e−8) and SSX5 (P = 0.004), as well as the related gene SSX2IP (P = 9.79e−4), are discriminators. Results of other suggested diagnostic markers are summarized in Table 2, of which the pattern of expression of these genes was fully in line with our prior expectations, thereby supporting the experimental set-up of this study. The genes CHEK2, c-KIT, PLAP, POU5F1-OCT3/OCT4, P53, SSX2 to SSX4, SYCP1, and XPA were indeed found to be significantly different in expression, whereas this was not the case for MAGEA4 and CDKN2D ( 33).
As it is reasonable to assume that genes with the highest expression difference might be the best discriminators for diagnostic purposes, we did immunohistochemistry to establish the diagnostic performance of DAZ. The gene encoding this protein is in the top 5 of discriminatory genes, specifically expressed in spermatocytic seminomas ( Table 1). As no specific antibodies are currently available for other interesting targets, like GAGE4/8/B1, GPC4, PEPP2, and SAGE, these markers could not be included in this survey. Fully in line with our expression profiling–based observation that DAZ (as well as DAZ2 and DAZ4) is specifically up-regulated in spermatocytic seminoma compared with seminoma/dysgerminoma, immunohistochemistry showed a specific staining for DAZ in all spermatocytic seminomas, including their intratubular precursor, with a nuclear localization (see Fig. 1E). This was found both on frozen as well as paraffin-embedded tissues, including those represented in the TMA (data not shown).
A considerable number of CTA genes, of which the SSX genes are representatives, are in the top 50 list of discriminatory genes (n = 11, 22%), being specifically expressed in spermatocytic seminoma (see Table 1). These genes are known to exhibit a maturation-stage specific expression pattern during spermatogenesis ( Fig. 2A ). Therefore, they can be informative to identify the cell of origin. TSP50 was predominantly found in spermatocytic seminomas (P = 0.018), in accordance to the previously reported down-regulation of this supposed protease in seminoma ( 34). CTCFL, which is also known as BORIS (position 26, Tables 1 and 2; ref. 35), is specifically up-regulated in spermatocytic seminomas compared with seminomas/dysgerminomas, whereas CTCF did not show a significant difference (P = 0.33). The CTA gene SYCP1, known to be involved in meiosis, is expressed (although at a rather low level) in spermatocytic seminomas (ref. 8; Fig. 2A; Table 2). Therefore, we checked the expression pattern of genes related to this specific type cell division in more detail ( Fig. 2A and B). In contrast to SYCP1, SYCP3 showed no difference (P = 0.38; Tables 1 and 2), whereas a number of genes related to meiotic prophase I (TCFL5, CLGN, and LDHc) are specifically expressed in spermatocytic seminomas compared with seminomas/dysgerminomas (P = 0.0055, 2.94e−5, and 0.0046, respectively; Fig. 2B). TCFL5 is a transcription factor–type basic helix-loop-helix ( 36); CLGN is a chaperone, interacting with TCFL5 ( 37); and LDHC is a germ cell lineage–specific variant of the lactate dehydrogenase enzyme ( 38).
Genomic imbalances in spermatocytic seminomas. Four of the spermatocytic seminomas have previously been characterized by conventional CGH analysis and one (case 1, Table 3 ) with karyotyping, of which the results were published elsewhere ( 15). The changes identified with conventional karyotyping are: 74-83<3n>:XY, add(X)(p11),+Y,+1,add(4)(p15.1),+5,+6,+add(6)(q21),−7,+9,+9,+10, inv(10)(q22q24) x2,del(12)(p11.2∼p12),+14,+17,+19,+20,−22,+mar. Here, we report additional SKY analyses on nine metaphase spreads of this previously karyotyped tumor. The results are illustrated in Fig. 3A and Table 3, of which the description is 57-83<3n> XXY,-X,+Y,+1,−2,-3,+5,+del(5q),+del(5q),+6,−7,−7,−,+9,+9, +10,+14,del(14q),+17,+19,+20,+20,−22 [cp9]. The previously reported ploidy heterogeneity of spermatocytic seminomas was confirmed by the identification of tumor cells with a different DNA content, based on metaphase spread analyses ( Table 3). The clonal relationship between this tumor cells was supported by a common del(5q) anomaly, which was confirmed with fluorescent ISH with a whole chromosome 5 paint (data not shown). The majority of changes identified using SKY were in line with the results of the other approaches (see Table 3, see also below).
Array CGH using a 32 K platform was done on all five spermatocytic seminomas ( Fig. 3B; Table 3). The seminomas and dysgerminomas were studied using a 3.7 K array, of which the results will be described elsewhere, as they were in agreement with previously described chromosomal imbalances, including gain of chromosome 7 and 8, as well as X, and the short arm of chromosome 12, and loss of chromosomes 4, 5 and 13 (see ref. 19 for review). Consistent with karyotyping, SKY, and conventional CGH, array CGH showed that the majority of imbalances in the spermatocytic seminomas affect entire chromosomes. Two spermatocytic seminomas (cases 1 and 5) showed a stepwise pattern of imbalances of multiple chromosomes. The imbalances resulting in the smallest numerical differences were not classified according to the criteria set as significant. Some of these nonsignificant changes affected chromosomes showing significant imbalances in the other tumors, like gain of chromosome 1 and gain of chromosome Y in case 1. Nineteen of the 21 significant changes detected by array CGH were also found using conventional CGH ( Table 3). Gain of chromosome 9 was the only consistent anomaly found, whereas gain of chromosome 20 was detected in four cases. Notably, array CGH revealed sub-chromosomal changes, being amplification of 9p21.3-pter in tumor 4 ( Fig. 3C), confirmed by fluorescent ISH ( Fig. 3D), estimated to be about 25 MB in size. In addition, tumor 2 showed loss of the 19q13.32-qter region.
Combined genome and expression analysis and identification of a positional functional candidate gene on chromosome 9. Using CESH, an overview of the expression profile on the chromosomal level was obtained. The results showed that the only genomic fragment with a consistent overexpression of genes compared with normal spermatogenesis is mapped to the short arm of chromosome 9, centered around band p22, but covering 9p13.3-pter. Representative examples are shown in Fig. 3E.
Because of the consistent gain of chromosome 9, as well as overexpression of genes from a restricted p-region in spermatocytic seminomas, and the identification of an overlapping, subchromosomal amplification of the 9p21.3-pter region (see above) this genomic fragment was investigated in more detail based on the Affymetrix array results. In total, 186 Affy-probe sets were represented in the region of interest, including known and hypothetical genes. To identify the gene(s) of interest, we compared the absolute expression level of the candidates in SS-4, containing the restricted 9p amplification, to the mean of expression of the other four spermatocytic seminomas. Only candidates showing a 2-fold overexpression were considered, and the candidate must show at least a 2-fold increase in level of expression in the tumor with the restricted amplification compared with the nonamplified tumor with the highest level of expression. These included 38 probe sets, representing 12 known and 11 hypothetical genes. Three genes were identified by nonspecific probe sets, or the different probe sets gave inconsistent results, which were therefore excluded from further analysis. Finally, eight genes from the 9p21.3-pter region showed overexpression in SS-4 compared with the others tumors: DMRT1, MPDZ, NIRF, PTPRD, RLN1, SH3GL2, SNAPC3, and MDS030. None of these were identified as a discriminatory gene in the top 50 list ( Table 1). Within the set of spermatocytic seminomas, DMRT1 shows the highest absolute level of expression (i.e., around 2,000) of the candidate genes. Based on the (supposed) function of the encoded protein, 9 DMRT1 was indeed considered as the most interesting candidate. DMRT3, closely related to DMRT1, is a discriminatory gene between spermatocytic seminoma and seminoma/dysgerminoma (position 31 in top 50; P = 4.25e−5, Tables 1 and 2). This in spite of the fact that this gene showed a low absolute level of expression, <100, like DMRT2 (see below). The expression pattern of DMRT1 and DMRT3 has been confirmed in an array technology-independent manner using quantitative PCR ( Fig. 4A ). Interestingly, DMRTC2, DMRTB1, and DMRT2 also showed a significant difference between spermatocytic seminomas and seminomas/dysgerminomas (P = 1.3e−4, 3.9e−4, and 0.044, respectively).
The presence of DMRT1 protein was investigated using immunohistochemistry of tissue sections (no antibody directed against DMRT3 is available), using an antibody directed against the mouse protein. As expected, a consistent nuclear staining of human spermatogonia and spermatocytes was observed ( Fig. 4B). Although all spermatocytic seminomas showed a positive staining, the seminomas/dysgerminomas were mainly negative ( Fig. 4C). This was found both on frozen tissue and formalin-fixed samples, including those present on the TMA (data not shown). The protein was specifically located in the nucleus. The tumor with the specific 9p amplification showed a particularly intense staining. A genetic change, reported in 46XY sex reversal individuals ( 39), Pro > Leu (codon 295, exon 4) was not found in the spermatocytic seminomas as determined by PCR amplification and sequencing (data not shown).
The inability of principle component, and unsupervised clustering analysis of genome-wide expression profiling data to distinguish seminomas from dysgerminomas, supports their common origin and pathogenesis. Their similar expression profile further supports the notion that these tumors indeed originate from the same embryonic germ cell (primordial germ cell/gonocyte), as was further supported by transcriptional interrogation of a number of stem cell markers, such as POU5F1-OCT3/OCT4 and NANOG, PROM1, ZNF42 (mRex1, also known as MZF-1), as well as markers related to the germ line, like PLAP, c-KIT, and VASA. Another interesting gene is NANOS3. In contrast to NANOS1 and NANOS2 (P = 0.267 and 0.18, respectively), NANOS3 is specifically up-regulated in all seminomas/dysgerminomas compared with spermatocytic seminomas (P = 4.4e−4), although it was not amongst the top 50 genes. In mouse, this gene is expressed in migrating germ cells, and its absence results in complete loss of germ cells ( 40). In this context, the apoptosis inhibiting role of POU5F1-OCT3/OCT4 in mouse primordial germ cells is also of interest ( 41). The previously reported expression of the “stem cell gene” HIWI (also called MIWIL) in seminoma was confirmed by this study (P = 0.06; ref. 42), adding to the previously mentioned list of stem cell–specific genes.
Here, we report on the first high resolution genome-wide analysis of chromosomal anomalies, as well as transcriptional profiling of a set of well-defined spermatocytic seminomas. The expression data were compared with observations made in seminomas/dysgerminomas. Principle component and unsupervised cluster analysis showed that spermatocytic seminomas are different from seminomas/dysgerminomas. Based on various attributes, including morphology, immunohistochemical characteristics, genomic imprinting, and chromosomal constitution, it has indeed been established that spermatocytic seminomas are different from seminomas ( 1– 3, 7– 10, 13, 15– 17). The results provide useful biomarkers that distinguish these two histologic groups, as shown by the finding of the known diagnostic markers identified previously by educated guesses or serendipitously. These included POU5F1-OCT3/OCT4 (positive in seminomas/dysgerminomas) and SSX2 to SSX4 (positive in spermatocytic seminomas; refs. 8, 10). Of the other earlier identified genes (encoding PLAP, c-KIT, XPA, SYCP1, MAGE4A2, p16INK4A, and CHEK2; ref. 33), none are found in our top 50 list, or even in the list of 373 distinguishing genes; however, some of them show a significant difference in expression between the two histologic subgroups ( Table 2). POU5F1-OCT3/OCT4 was one of the few genes showing a 100-fold positive difference between seminomas/dysgerminomas and spermatocytic seminomas. The second best gene, NANOG, also showed a high level of expression in seminoma, in agreement with previous studies ( 43– 45). Immunohistochemistry on primary tumors and a TMA containing 50 invasive seminomas and 23 invasive spermatocytic seminomas, including their precursor lesions, confirmed this finding. In conclusion, based on the availability of antibodies, POU5F1-OCT3/OCT4 is the best diagnostic marker for seminoma/dysgerminoma, in line with previous observations, and SSX2 to SSX4, XPA, and DAZ have high diagnostic value for spermatocytic seminoma.
In contrast to the observed pronounced expression of stem cell–specific related genes in seminoma/dysgerminoma, spermatocytic seminomas are characterized by expression of CTAs. These genes are of particular interest, because they can be used to identify the cell of origin and because the expression of CTAs is highly specific and tightly regulated during spermatogenesis (see ref. 46 for review). Based on the consistent pattern of expression of these genes, it prompted our conclusion that spermatocytic seminomas indeed show characteristics of germ cells at the differentiational stage of spermatocytes. The expression pattern of SYCP1, CTCFL, TCFL5, CLGN, and LDHc suggests that the tumor cells at least enter meiotic prophase I. These findings strongly suggest that human spermatocytic seminomas mimic germ cell tumors generated in Caenorhabditis elegans by knocking out of the PUF8 gene ( 47). These tumor cells undergo meiosis but do not finish meiotic division but, instead, return to the mitotic cell cycle. However, we did not observe absence of expression of PUM1 and PUM2 genes (the human homologues of PUF-8) in spermatocytic seminomas based on expression analysis as well as immunohistochemistry (data not shown). Interestingly, DAZ and DAZL have been reported to interact with PUM1 and PUM2 ( 48). This might be of relevance because we observed specific up-regulation of a number of DAZ genes in spermatocytic seminoma (DAZ, P = 3.29e−6; DAZ2, P = 6.07e−6; DAZ4, P = 7.34e−6). The up-regulation of DAZ was subsequently confirmed by immunohistochemistry and found to be an informative biomarker (see above). Although this provides an interesting novel diagnostic marker, the functional consequences of this observation remains to be elucidated. Besides the C. elegans germ cell tumors, also the spontaneously occurring canine seminomas deserve further investigation as an animal model for spermatocytic seminoma ( 17, 49).
Another striking difference between seminoma/dysgerminoma and spermatocytic seminoma is the specific expression of genes encoding proteins known to be chemotaxic for macrophages and (activated) T lymphocytes, including CRACC, CXCL10, and CXCL9 (refs. 50– 52; Table 1). Indeed, the seminomas/dysgerminomas are characterized by a consistent presence of lymphocytic infiltrates, which are absent in spermatocytic seminomas. The specific expression of APOC in seminomas/dysgerminomas might also be explanatory for this difference, which has been reported to act as an attractant for these cells of the immune system ( 53). The presence of IGHM and IGJ is most likely due to the presence of infiltrating lymphocytes in the seminomas/dysgerminomas.
Besides the unique data on expression profiling, this study also reports on the first comprehensive genomic screen for imbalances in spermatocytic seminomas, using a combination of complementary molecular cytogenetic approaches. All techniques applied (conventional and spectral karyotyping as well as array CGH) led to the same conclusion: spermatocytic seminomas are characterized by recurrent chromosomal imbalances that mainly affect complete chromosomes. Gain of chromosome 9 was the only consistent imbalance. The specific amplification of a subchromosomal region on the short arm of chromosome 9 (p21.3-pter), as well as the CESH results showing overexpression of genes from this region relative to normal testis, led us to hypothesize that a gene or genes of interest resides within this specific chromosomal fragment. Therefore, we systematically compared expression data with chromosomal data and found that a number of 9p genes were indeed up-regulated in the spermatocytic seminomas with the restricted 9p amplification compared with those with only gain of the whole chromosome 9. In particular, the DMRT genes are of interest. Various DMRT genes have been identified, of which DMRT1, DMRT2, and DMRT3 map to the region of interest. DMRT stands for Doublesex and mab-related transcription factor. These genes are related to sex determination, and at least DMRT1 and DMRT2 are candidates to explain 46XY sex reversal in individuals with monoallelic loss of one of the 9p24.3 regions during embryogenesis ( 54– 56). In addition, DMRTC2 (mapping to 19p13.3), DMRTB1 (mapping to 1p32), and DMRT2 (mapping to 9p24.3) all show a significant transcriptional up-regulation in spermatocytic seminoma (P = 1.3e−4, 3.0e−4, and 0.044). This might be of interest because of the evolutionary link between DMRT1 and DMRTC2 and DMRTB1 ( 57). DMRT1 is present during embryogenesis, in Sertoli and germ cells ( 58, 59). Thus far, the presence of DMRT1 during human development and adult life has not been elucidated yet. All three DMRT genes (DMRT1, DMRT2, and DMRT3) showed overexpression in spermatocytic seminomas, although the last two are at low level. Only DMRT1 was also overexpressed in the tumor with the subchromosomal amplification. Indeed, the protein was consistently detected by immunohistochemistry in spermatocytic seminomas and more heterogeneously in seminoma/dysgerminoma, possibly related to the underrepresentation of chromosome 9 in the majority of these tumors ( 6). The ability of the antibody directed against the mouse Dmrt1 protein to recognize the human protein, allows a careful investigation of DMRT1 presence during normal male and female development, and this is currently under investigation. In addition, the exact role of DMRT1 in the development of spermatocytic seminomas remains to be elucidated.
In summary, this study is the first comprehensive analysis of spermatocytic seminomas that includes genome wide analysis of genomic changes as well as expression profiling. The integrated results clearly illustrate the distinct pathogenesis of spermatocytic seminomas and profound differences compared with seminomas/dysgerminomas. Spermatocytic seminomas express genes related to specific stages of post-pubertal male germ cell maturation, which is in line with the testis-specific occurrence of this cancer. Our data identify the spermatocyte as the cell of origin for this type of germ cell tumor. Additional, supportive evidence is presented that these tumor cells undergo partial meiosis, possibly followed by a reentry into the mitotic cycle. Furthermore, spermatocytic seminomas show a specific pattern of chromosomal imbalances, in which gain of chromosome 9 is consistent, likely related to DMRT1, a gene expressed during embryogenesis and male germ cell maturation. Lastly, this study not only confirms known diagnostic markers but also discovers new markers that are available for immunohistochemical analysis and further biological studies.
In conclusion, spermatocytic seminomas are a separate entity of germ cell tumors, originating from primary spermatocytes, to be recognized as such in the upcoming WHO classification system. This study shows that a genome-wide investigation on well-characterized tumors is informative, even when done on a limited number of cases.
Grant support: Dutch Cancer Society, the Daniel den Hoed Cancer Center, Cancer Research UK, Lance Armstrong Foundation, and German Cancer Aid Max-Eder grant (D.T. Schneider).
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
We thank Prof. Dr. A. Grootegoed (Department of Reproduction and Development, Erasmus Medical Center Rotterdam) for the stimulating discussions regarding meiosis and Drs. David Zarkower and Vivian Bardwell (Department of Genetics, Cell Biology, and Development, and Cancer Center, University of Minnesota, Minneapolis, MN) for making the antibody directed against DMRT1 available for this study.
- Received August 17, 2005.
- Accepted October 26, 2005.
- ©2006 American Association for Cancer Research.