EXPRESSION OFSSXGENES IN HUMAN TUMORSOzlem TURECI1*, Yao-Tseng CHEN2,3, Ugur SAHIN1, Ali O. GURE3, Carsten ZWICK1, Carlos VILLENA 5,Solam TSANG3, Gerhard SEITZ4, Lloyd J. OLD3 and Michael PFREUNDSCHUH1

1Department of Internal Medicine, University of Saarland Medical School, Homburg-Saar, Germany2Cornell University Medical College, New York, NY, USA3Ludwig Institute for Cancer Research, New York Branch at Memorial Sloan-Kettering Cancer Center, New York, NY, USA4Department of Pathology, Klinikum Bamberg, Bamberg, Germany5Department of Gynecology, University of Saarland Medical School, Homburg-Saar, Germany

The HOM-MEL-40 antigen which is encoded by the SSX-2gene was originally detected as a tumor antigen recognizedby autologous IgG antibodies in a melanoma patient. Expres-sion analysis demonstrated that SSX-2 is a member of therecently described cancer/testis antigen (CTA) class as it isexpressed in a variety of different human neoplasms, but notin normal tissues with the exception of testis and a weakexpression in the thyroid. Further studies demonstrated thatSSX-2 belongs to a gene family consisting of at least 5homologous genes. We now report the analysis of the expres-sion of all 5 SSX genes in 325 specimens of human neoplasmsfrom various histological origins, using reverse transcriptionpolymerase chain reaction (RT-PCR). SSX-1, -2, and -4 werefound to be expressed in 8%, 15% and 15%, of the tumors,respectively, while the expression of the SSX-5 gene was rare(7/325), and SSX-3 expression was not detected. For definedtumor types, expression of at least one of the SSX familymembers was most frequently observed in head and neckcancer (75%), followed by ovarian cancer (50%), malignantmelanoma (43%), lymphoma (36%), colorectal cancer (27%)and breast cancer (23%), while leukemias and the few cases ofleiomyosarcomas, seminomas and thyroid cancers were foundnot to express any SSX gene. Int. J. Cancer 77:19–23, 1998.r 1998 Wiley-Liss, Inc.

Genes which are either exclusively or preferentially expressed inmalignant human tissues are a prerequisite for specific immuno-and genetherapeutic strategies. Some such genes have been shownto code for antigens which elicit a humoral and/or cellular immuneresponse in cancer-bearing patients (for review Tu¨reci et al.,1997;Boon and van der Bruggen, 1996). The cancer/testis antigen (CTA)class, which includes the MAGE- (van der Bruggenet al., 1991),BAGE- (Boel et al.,1995), GAGE-families (van den Eyndeet al.,1995), NY-ESO-1 (Chenet al., 1997) and HOM-MEL-40/SSX-2(Tureci et al., 1996) is a rapidly growing group of immunogenicproteins, the mRNA expression of which is characteristicallyrestricted to cancer and normal testis (Chenet al.,1996, Tureci etal., 1997, Sahinet al., 1997). For some cancer/testis antigens, aweak expression in additional normal tissues has been shown,which is detectable only by RT-PCR, but not Northern blotanalysis. The genes coding for most, but not all CTA, have beenassigned to the X chromosome. The HOM-MEL40 antigen wasoriginally identified in a malignant melanoma using SEREX, theserological analysis of antigens by recombinant expression cloning(Sahin et al., 1995). SEREX is based on the screening oftumor-derived cDNA expression libraries with sera from tumorpatients for the molecular characterization of human tumor anti-gens. HOM-MEL-40 has been shown to be encoded for by theSSX-2 gene which had previously been reported to be involved inthe t(X;18) translocation in synovial sarcoma (Clarket al., 1994;Crewet al.,1995). HOM-MEL-40/SSX-2 is expressed in tumors ofdifferent origins, but not in normal tissues except for testis and aweak expression in the thyroid gland. High-titered antibodies of theIgG class to HOM-MEL-40/SSX-2 are found in 10% of melanomapatients (Tu¨reci et al., 1996). SEREX analysis of an allogeneictesticular library with another melanoma serum identified theSSX-3 protein, which is encoded by a gene highly homologous toSSX-2 (Gure et al., 1997). This gene was also cloned outindependently by nucleotide hybridization with aSSX-2 probe from

a testicular library (de Leeuwet al.,1996). These findings as wellas the previous identification ofSSX-1 (Crewet al.,1995), anotherclosely related gene which is also involved in translocations insynovial sarcoma, indicated the presence of a multi-gene family,which was confirmed by Southern blot analysis (Gu¨reet al.,1997).By PCR cloning, we have identified 2 additional members of theSSXgene family, SSX-4 and SSX-5, increasing the number oftranscribedSSXgenes to 5 (Gu¨reet al.,1997). All 5 members of theSSXfamily share a strong sequence homology with nucleotidehomologies ranging from 88% to 95% and amino acid homologiesranging from 77% to 91%, respectively (Gu¨re et al., 1997). Thegenomic cloning ofSSX-2, a prototypeSSXgene, revealed that itconsists of 6 exons. AllSSX family genes were found to beexpressed in adult testis, but none of them significantly in othernormal tissues except for a faint expression in thyroid. All membersexcept forSSX-3 were found to be expressed at varying frequenciesin melanoma cell lines (Gu¨re et al., 1997). However, with theexception ofSSX-2, little is known about their distribution intumorsin vivo. We now describe an analysis of the expression ofthe expression pattern of 5SSXgenes in 325 human tumors usingreverse transcription and polymerase chain reaction amplification(RT-PCR). We show that with the exception ofSSX-3 the expres-sion of SSX genes is shared by a broad spectrum of humanneoplasms, suggesting them as promising candidates for immuno-and gene-therapeutic approaches.

MATERIAL AND METHODS

Tissues and cell lines

The study had been approved of by the local ethical review board(Ethikkommission der A¨ rztekammer des Saarlandes). Recombi-nant DNA work was done with the official permission andaccording to the rules of the State Government of Saarland. Tumortissues were obtained during routine diagnostic or therapeuticprocedures, from University of Saarland Medical School, Hom-burg, Germany, from Memorial Sloan-Kettering Cancer Center,New York, NY and from the Department of Pathology, Bamberg,Germany. Normal tissues were collected from autopsies of tumor-free patients. Tissues were stored at280°C until use.

Reverse transcription PCRTotal cellular RNA was prepared from frozen tissue specimens

using guanidinium-isothiocyanate for denaturation followed by anacidic phenol extraction and isopropanol precipitation (Chomczyn-ski and Sacchi, 1987). Total RNA (4 µg) was primed with a dT(18)oligonucleotide and reverse-transcribed with Superscript reverse

Grant sponsors: Sonderforschungbereich 399, Deutsche Forschungsge-meinschaft and Deutsche Krebshilfe

*Correspondence to: Innere Medizin I, Universita¨tskliniken des Saarlan-des, D-66421 Homburg/Saar, Germany. Fax: (49) 6841 -163092.E-mail: Tuereci@aol.com

Received 24 October 1997; Revised 9 February 1998

Int. J. Cancer:77,19–23 (1998)

r 1998 Wiley-Liss, Inc.

Publication of the International Union Against CancerPublication de l’Union Internationale Contre le Cancer

transcriptase (GIBCO BRL, Eggenstein, Germany) according tothe manufacturers instructions. Integrity of the obtained cDNA wastested by amplification ofb-actin transcripts in a 25-cycle standardPCR reaction as described elsewhere (Tu¨reciet al.,1997). For PCRanalysis of the expression of individualSSXgene transcripts 1 µlfirst-strand cDNA was amplified with transcript-specific oligo-nucleotides (10 pMol) using 2 U AmpliTaq Gold (Perkin Elmer,Weiterstadt, Germany), 10 nMol of each dNTP (dATP, dTTP,dCTP, dGTP) and 1.67 mM MgCl2 in a 30 µl reaction. The primers(MWG Biotech, Ebersberg, Germany) and lengths of PCR prod-ucts forSSXfamily members have been reported previously (Gu¨reet al.,1997) and are as follows:SSX1-58: 58-cta aag cat cag aga aga gaa gc;SSX1-38: 58-aga tct cttatt aat ctt ctc aga aa, annealing temperature 56°C, size 421 bp;SSX2-58: 58-gtg ctc aaa tac cag aga aga tc;SSX2-38: 58-ttt tgg gtccag atc tct cgt g, annealing temperature 67°C, size 435 bp;SSX3-58:58-gga aga gtg gga aaa gat gaa agt;SSX3-38: 58-ccc ctt ttg ggt ccagat atc a, annealing temperature 65°C, size 381 bp;SSX4-58: 58-aaatcg tct atg tgt ata tga agc t;SSX4-38: 58-ggg tcg ctg atc tct tca taa ac,annealing temperature 60°C, size 413 bp;SSX5-58: 58-gtt ctc aaatac cac aga aga tg;SSX5-38: 58-ctc tgc tgg ctt ctc ggg cg, annealingtemperature 66°C, size 324 bp; SYT-681: 58-aca gca tta cca agg acagca gcc acc, annealing temperature 60°C; SYT-911: 58-gcc aac agcaag atg cat acc agg gac, annealing temperature 60°C. SYT-681 andSYT-911 primers were used together with SSX1-38 and SSX2-38primers for the detection of the SYT/SSX fusion transcript reportedfor synovial sarcoma (Clarket al., 1994; Crewet al., 1995).Amplification was performed in a TRIO-Thermoblock (Biometra,Gottingen, Germany). After 12 min activation of AmpliTaq Goldpolymerase at 94°C for hot-start induction, 35 cycles of PCR wereperformed with 1 min at the respective annealing temperature asindicated above, 2 min at 72°C and 1 min at 94°C with a finalelongation step at 72°C for 8 min. A 15 µl aliquot of each reactionwas size-fractionated on a 2% agarose gel, visualized by ethidiumbromide staining and assessed for expected size.

RESULTS

We investigated a total of 325 tumor specimens which had beenassessed for cDNA integrity by amplification of an 800 bpb-actinproduct. Stringent amplification conditions and AmpliTaq Goldpolymerase were chosen to ensure the specific amplification of therespective individualSSX gene without concomitant cross-annealing of primers to other family members. This was achievedin all cases except for amplification withSSX-2 specific primerswhich also amplified a faint product from a control plasmid of thecloned SSX-3. However, sinceSSX-3was never found to beexpressed in tumors, this technical aspect was considered to beirrelevant. To exclude false-positive PCR products due to smallamounts of contaminating DNA in the RNA preparation, theindividual primer sets were chosen for sequences corresponding todifferent exons of the prototypeSSX-2genomic clone, anticipatingthat the genes of otherSSXfamily members might have a similarintron-exon structure. That these primers were trans-intronic wasconfirmed by the fact that purified genomic DNA from 2 testisspecimens generated no SSX PCR products under the sameexperimental conditions. That the SSX RT-PCR products wereindeed derived from cDNA and not from contaminating DNA wasalso assured by the observation that RNA preparations from 4 SSXtranscript-positive breast cancer cases were PCR-negative whenthe reverse transcription step was omitted. Each RT-PCR experi-ment was done in triplicate using the same poly-dT-primed cDNAsample together with the appropriate controls. Representativeexamples of PCR results are shown in Figure 1. Intensities of PCRproducts were found to be heterogeneous and some specimensyielded only faint amplicon bands. These were scored positive onlyif the result was reproduced by a repeated RNA extraction andspecific PCR from the same tumor specimen. Since these experi-ments were not designed to be quantitative PCR studies, andSSXtranscript turnover as well as correlation between transcript and

protein levels are not known at present, these cases were notdistinguished further in Table II. Cases with very low transcriptlevels, which were not reproducibly positive, were not regarded asbeing positive.

As summarized in Table I, 89 specimens (27%) were found toexpress at least one of theSSXfamily members. Expression ofSSXfamily members was most frequent in synovial sarcomas (3/4 or75%), head and neck cancer (8/14 or 57%), bladder cancer (5/9 or56%), and ovarian cancer (6/12 or 50%), followed by malignantmelanomas (16/37 or 43%), prostatic cancer (2/5 or 40%), andnon-Hodgkin’s lymphomas (4/11 or 36%). Expression was lessfrequent in colorectal cancers (16/58 or 27%), breast cancers(16/67 or 23%), lung cancers (5/24 or 21%), gliomas (5/31 or16%), endometrial cancers (1/8 or 13%), and renal cell cancer (1/22or 4%). No expression was found in specimens from leiomyosarco-mas, thyroid cancers, seminomas and leukemias. Of the differentmembers of theSSXfamily of genes,SSX-2 andSSX-4 were mostfrequently expressed (50 or 15%, and 48 or 15%, respectively). AllSSX-4 positive cases expressed only the longer variant of the 2alternatively spliced variants (Gu¨reet al.,1997). The expression ofSSX-1 was observed in 25 of the 325 cases (8%), while theexpression ofSSX-5 was rare (7 cases or 2%).SSX-3 expressionwas not detected at all.

Comparing tumors of different origins, it became evident thatspecific SSXgenes are preferentially expressed in some tumorentities. For example, expression in ovarian cancer was limited toSSX-4, and the only SSX gene expressed in non-Hodgkin’slymphomas wasSSX-2 (Table II). In contrast, no preference was

FIGURE 1 – Representative results of RT-PCR analysis for expres-sion ofSSXgenes in testis and human neoplasms (15 Br15; 25 Em5;3 5 Mm35; 45 Mm34; 55 Cr26; 65 Cr33; 75 Gl16; 85 Rc1; 95testis).

20 TURECIET AL.

observed in other tumor entities, with non-selective expression ofSSXgenes (except forSSX-3) being detected in head-and-neckcancer, colorectal cancer, glioma, breast cancer, bladder cancer,melanoma, lung cancer, endometrial cancer and synovial sarcomas.Most of theSSX-positive tumors expressed only one member of thegene family (Table II). However, co-expression of 2 or more geneswas found in several tumor types such as breast and colorectalcancer, and was most pronounced in melanomas, of which 4 out ofthe 16 positive tumors co-expressed 2, 6 expressed 3, and 1expressed 4 members of the gene family. Whether these genes areco-expressed simultaneously as a set in the same population oftumor cells, or whether different SSX genes are activated indifferent subpopulations as a consequence of tumor heterogeneity,remains to be shown.

The expression ofSSX genes in synovial sarcoma was ofparticular interest, since all synovial sarcoma cases have beenshown to carry either the SYT/SSX1 or the SYT/SSX2 transloca-tion at breakpoints flanked by the SSX primer sets designed in thisstudy (Clarket al.,1994; Crewet al.,1995). In order to determinethe translocational status of theSSXgenes in our cases and tocorrelate it with the SSX mRNA expression status, translocation-specific RT-PCRs were performed by coupling 58SYT primers and38SSX gene-specific primers (SYT-681/SSX1-38 and SYT-681/SSX2-38 confirmed by SYT-911/SSX1-38 and SYT-911/SSX2-38).Results showed SYT/SSX1 translocations in Sy1, Sy2 and Sy4,and a SYT/SSX2 translocation in Sy3. As Sy1, Sy2 and Sy3 werederived from male patients, one would have expected Sy1 and Sy2to be SSX1 full-length mRNA-negative and Sy3 to be SSX2-mRNA-negative. This was indeed observed (Table II). TheSSXgene expression, found in three of four cases, thus appears to beindependent of the SYT/SSX translocation event.

DISCUSSION

The analysis of the expression ofSSXfamily members in ourlarge series of malignant human tumors revealed thatSSX-1, -2, -4and-5 are all expressed in tumors of different origins with variousfrequencies. One notable exception, however, wasSSX-3, whichwas not detected in any of the 325 tumors tested. Our analysisconfirmed thatSSX-1, SSX-2, SSX-4andSSX-5 belong to a group ofgenes, the expression of which is restricted to malignant tumorsand testis as the only non-malignant tissue, with a weak expressionin the thyroid gland. The term ‘‘cancer/testis genes’’ has beencoined for these genes and the term cancer/testis antigens for theirproducts, if they induce an immune response. We detected theexpression ofSSXgenes in a wide variety of neoplasms such as

melanomas, head and neck cancers, breast cancers, ovarian can-cers, colorectal cancers and lymphomas. Leukemias and the fewcases of leiomyosarcoma, thyroid cancer and seminoma included inour study failed to expressSSXgenes. Immunogenicity ofSSXgeneproducts proving them as cancer/testis antigens has so far only beenshown forSSX-2 which codes for the HOM-MEL-40 antigen andwas originally detected by the analysis of the B cell repertoire of apatient with melanoma (Sahinet al.,1995; Tureci et al.,1996). Asonly a proportion of the patients with HOM-MEL-40/SSX-2positive tumors have detectable IgG antibodies in their sera (Tu¨reciet al., 1996), further studies will have to determine the role oftumor burden, human leucocyte antigen (HLA) haplotype and theintegrity of the antigen processing machinery of the tumor cells forthe induction of an antibody response against HOM-MEL-40/SSX-2. TheSSX-3 gene product was found by immunoscreening ofan allogeneic testis library with a melanoma patient’s serum.However, this might have been due to a crossreaction of theantibody induced by other SSX members in the serum of thatpatient, sinceSSX-3 derived transcripts were not detected in any ofthe 325 tumors tested. The finding that multiple members of theMAGE family are capable of eliciting humoral and/or cellularimmune responses (Sahinet al.,1995; Tureciet al.,1996; De Plaenet al.,1994) and the strong amino acid homology thatSSX-2 shareswith other members ofSSXfamily is supportive of the notion thatmost, if not all,SSXgenes code for immunogenic proteins, eitherspontaneously or upon specific vaccination. The frequent expres-sion of SSXgenes in the most common types of human tumors,such as colon and breast cancer, some of which rarely express othercancer testis genes, would enhance the numbers of patients whowould be potential candidates for a specific immunotherapy.

As SSX-1 andSSX-2 are involved in the t(X;18) translocation ofsynovial sarcomas (Clarket al., 1994; Crewet al., 1995) andnon-translocatedSSX-1, -2 and -4, and SSX-5 are aberrantlyexpressed in a variety of human tumors, one may speculate that theSSXgenes, which are constitutively silent in non-testicular normaladult tissue, may escape silencing by different mechanisms. In thet(X;18) translocation of synovial sarcomasSSX-1and -2 genesrecruit the active promotor of theSYTgene by fusion to its 5’ part.Other mechanisms must be operative in the activation of themajority of non-translocatedSSXgenes which we detected asfull-length genes using primers which flank the breakpoint, withthe 58 primer being as close to the transcription start as possible.One of the mechanisms discussed for such a reactivation of genes isdemethylation. The eukaryotic genome is methylated at the 5carbon of cytosines occurring in CpG islands (Razin and Cedar,

TABLE I – EXPRESSION OFSSXGENES BY HUMAN NEOPLASMS1

Tumor entity Tissuestested SSX1 SSX2 SSX3 SSX4 SSX5 At least

one positive %

Lymphoma 11 — 4 — — — 4 36Breast cancer 67 5 5 — 10 — 16 23Endometrial cancer 8 1 1 — 1 1 1 13Colorectal cancer 58 3 7 — 9 1 16 27Ovarian cancer 12 — — — 6 — 6 50Renal cell cancer 22 — 1 — — — 1 4Malignant melanoma 37 10 13 — 10 2 16 43Glioma 31 — 2 — 3 — 5 16Lung cancer 24 1 4 — 1 1 5 21Stomach cancer 3 — — — 1 — 1 33Prostatic cancer 5 — 2 — — — 2 40Bladder cancer 9 2 4 — 2 — 5 55Head-neck cancer 14 3 5 — 4 1 8 57Synovial sarcoma 4 — 2 — 1 1 3 75Leukemia 23 — — — — — 0 0Leiomyosarcoma 6 — — — — — 0 0Thyroid cancer 4 — — — — — 0 0Seminoma 2 — — — — — 0 0Total 325 25 50 0 48 7 89

1The number of tumor specimens of the respective entity expressingSSXgenes are shown.

21SSXGENE IN HUMAN TUMORS

1991; Bird 1992). A large body of evidence confirms that themethylation state within or near gene promotor or enhancersequences modulates the accessibility of the DNA to transcriptionfactors and thus the expression of the respective gene (Jost andBruhat, 1997). A constitutive genome-wide demethylation occursin spermatogonia (del Mazoet al., 1994; Martorellet al., 1997),and a global DNA demethylation is observed in tumor cell lines aswell as in tumor samples and has been shown to the reactivation ofsilenced genes and to genomic instability (Gartler and Goldman,

1994; Jones and Gonzalgo, 1997; Wachsman, 1997; Versteeg,1997). For theMAGE-1 gene reactivation CpG island demethyl-ation in the promotor elements have been shown to be crucial (DeSmetet al., 1996; Weberet al., 1994; Serranoet al., 1996), andpreliminary data suggest this mechanism to be operative also in theactivation of theBAGEandGAGEgenes (de Smetet al., 1996).While in some types of tumors,e.g., in melanomas, heterogeneousexpression of differentSSXgenes, with co-expression of 2 or moremembers of theSSXfamily occurs, other tumors express only one

TABLE II – EXPRESSION PATTERN OF INDIVIDUALSSXGENES IN SSX-POSITIVE TUMOR SAMPLES1

Breast cancer (67 specimens) SSX1 SSX2 SSX4 SSX5

51 specimens 2 2 2 27 specimens 2 2 1 24 specimens 2 1 2 22 specimens 1 2 2 22 specimens 1 2 1 21 specimens 1 1 1 2

Melanoma (37 specimens) SSX1 SSX2 SSX4 SSX5

21 specimens 2 2 2 25 specimens 1 1 1 24 specimens 2 1 2 22 specimens 2 1 1 21 specimen 1 2 2 21 specimen 1 1 2 21 specimen 1 2 1 21 specimen 1 2 1 11 specimen 1 1 1 1

Endomet. cancer (8 specimens) SSX1 SSX2 SSX4 SSX5

7 specimens 2 2 2 21 specimen 1 1 1 1

Glioma (31 specimens) SSX1 SSX2 SSX4 SSX5

25 specimens 2 2 2 23 specimens 2 1 2 22 specimens 2 2 1 2

Lung cancer (24 specimens) SSX1 SSX2 SSX4 SSX5

19 specimens 2 2 2 23 specimens 2 1 2 21 specimen 2 2 2 11 specimen 1 1 1 2

Colorectal cancer (58 specimens) SSX1 SSX2 SSX4 SSX5

42 specimens 2 2 2 27 specimens 2 1 2 25 specimens 2 2 1 23 specimens 1 2 1 21 specimen 2 2 1 1

Bladder cancer (9 specimens) SSX1 SSX2 SSX4 SSX5

4 specimens 2 2 2 22 specimens 2 1 2 21 specimen 2 2 1 21 specimen 1 1 2 21 specimen 1 1 1 2

Head-Neck cancer (14 specimens)SSX1 SSX2 SSX4 SSX5

6 specimens 2 2 2 22 specimens 1 2 2 22 specimens 2 1 1 21 specimen 2 1 2 21 specimen 2 2 1 21 specimen 1 1 2 21 specimen 2 1 1 1

Synovial sarcoma (4 specimens) SSX1 SSX2 SSX4 SSX5 SYT/SSX1 SYT/SSX2

Sy1 2 2 1 2 1 2Sy2 2 1 2 1 1 2Sy3 2 2 2 2 2 1Sy4 2 1 2 2 1 2

1Expression was determined by RT-PCR with family member specific primers.SYT/SSXfusions insynovial sarcomas were determined using SYT-681 and SYT-911 primers with SSX-1B or SSX-2B,respectively. Results for lymphoma, ovarian cancer, renal cell cancer, stomach and prostatic cancer are notshown, since positive tumors of these entities express only oneSSX-family member (Table I).

22 TURECIET AL.

given SSXgene (e.g., ovarian cancer onlySSX-4 and lymphomaonly SSX-2). For the latter cases we speculate that tumor type- andtissue type-specific activation of a transcription activator or theinactivation of an inhibitor/tumor suppressor may play a role in theearly development of the respective tumor. This suggests differenttranscriptional controls ofSSX gene family members which,together with the high homology of their transcripts and hence theirsimilar functional role, would allow for a fine-tuned differentialregulation of genes with a similar function in different temporospa-tial settings, not only in malignant tumors, but presumably also inembryonal development. Similar to Russoet al. (1995) andMulcahy et al. (1996) in the case of members of theMAGE genefamily, we also observed varying transcript levels ofSSXgenes asindicated by different intensities of PCR products which werereproducibly shown in identical tumor samples. Intensities of PCRproducts were found to be heterogenous and some specimensyielded only faint amplicon bands. Low expression levels byRT-PCR can be due to several reasons, such as a tumor-wide lowSSXgene expression level in the malignant cells, dilutional effectsof contaminating normal stroma and/or inflammatory cells, or alarge proportion of tumor cell variants in the tumor which do notexpress the respective gene, even though the expression levels in

positive cells might be high. While the restricted expression patternof theSSXgenes makes them promising candidates, much remainsto be done before they can be used as targets for immuno- and genetherapeutic interventions. As we described earlier, transcript levelsfor members of theSSXfamily as shown by Northern blots are high(Sahin et al., 1995; Tureci et al., 1996). However, to date nocorrelation between transcript levels and protein level has beenestablished forSSX. Only when specific antibodies for the respec-tive gene products are available will it be possible to determine theprotein expression level in tumors of different origins and atvarious stages, as well as in individual tumor cells of a given tumor.The analysis of the functional role of theSSXgene products willprovide insight into the therapeutic potential of specific genesilencing, and the analysis of the immune response against theindividual gene products in cancer patients should enable us todevelop and evaluate specific vaccine strategies for cancer treat-ment.

ACKNOWLEDGEMENTS

We thank Mrs. E. Vollmar, Mrs. K. Mack and Mr. M. Becker forexcellent technical assistance.

REFERENCES

BIRD, A., The essentials of DNA methylation.Cell, 70,5–12 (1992).BOEL, P., WILDMANN , C., SENSI, M.L., BRASSEUR, F., RENAULD, J.C., COULIE,P., BOON, T. andVAN DER BRUGGEN, P., BAGE: a new gene encoding anantigen recognized on human melanomas by cytolytic T lymphocytes.Immunity, 2, 167–175 (1995).BOON, T. andVAN DER BRUGGEN, P., Human tumor antigens recognized byT-lymphocytes. J. exp. Med.,183,725–729 (1996).CHEN, Y.T., SCANLAN, M., SAHIN, U., TURECI, O., GURE, A., TSANG, S.,WILLIAMSON , B., STOCKERT, E., PFREUNDSCHUH, M. and OLD, L.J., Cloningof a testicular antigen expressed in esophageal cancer by autologousantibody screening.Proc. nat. Acad. Sci (Wash.), 94,1914–1918 (1997).CHOMCZYNSKI, P. and SACCHI, N., Single step method of RNA isolation byacid guanidium thiocyanate-phenol-chloroform extraction.Anal. Biochem.,162,156–159 (1987).CLARK, J., ROQUES, P.J., CREW, A.J., GILL , S., SHIPLEY, J., CHAN, A.M.,GUSTERSON, B.A. and COOPER, C.S., Identification of novel genes,SYTandSSX, involved in the t(X;18)(p11.2;q11.2) translocation found in humansynovial sarcoma.Nature Genet., 7, 502–508 (1994).CREW, A.J., CLARK, J., FISHER, C., GILL , S., GRIMMER, R., CHand, A.,SHIPLEY, J., GUSTERSON, B.A. and COOPER, C.S., Fusion ofSYTto two genesSSX1 andSSX2, encoding proteins with homology to Kruppel-associatedbox in human synovial sarcoma.EMBO J., 14,2333–2340 (1995).DE LEEUW, B., BALEMANS, M. and GEURTS VAN KESSEL, A., A novelKruppel-associated box containing theSSX gene is not implicated int(X;18)-positive synovial sarcomas.Cytogenet. Cell Genet., 73, 179–183(1996).DE PLAEN, E. and 14OTHERS, Structure, chromosomal localization andexpression of 12 genes of theMAGEfamily. Immunogenetics, 40,360–369(1994).DE SMET, C., COURTOIS, S.J., FARAONI, I., LURQUIN, C., SZIKORA, J.P., DEBACKER, O. and BOON, T., The activation of human geneMAGE1 in tumorcells is correlated to genome wide demethylation.Proc. nat. Acad. Sci.(Wash.), 93,7149–7153 (1996).DEL MAZO, J., PRANTERA, G., MIGUEL, T. and FERRARO, M., DNAmethylation changes during mouse spermatogenesis.Chromosome Res., 2,147–152 (1994).GARTLER, S.M. and GOLDMAN, M.A., Reactivation of inactive X-linkedgenes.Develop. Genet., 15,504–514 (1994).GURE, A.O., TURECI, O., SAHIN, U., TSANG, S., SCANLAN, E., JAGER, M.,KNUTH, A., PFREUNDSCHUH, M., OLD, L.J. and CHEN, Y.T., SSX, a multigenefamily with several members transcribed in normal testis and human cancer.Int. J. Cancer, 72,965–971 (1997).JONES, P.A. and GONZALGO, M.L., Altered DNA methylation and genomeinstability: a new pathway to cancer.Proc. nat. Acad. Sci. (Wash.), 94,2103–2105 (1997).JOST, J.P. and BRUHAT, A., The formation of DNA methylation patterns and

the silencing of genes.Progr. Nucl. Acid Res. mol. Biol., 57, 217–248(1997).MARTORELL, M.R., NAVARRO, J., MARQUEZ, C., EGOZCUE, J. and BENET, J.,Hypomethylation of human sperm pronuclear chromosomes.Cytogenet.Cell Genet., 76,123–127 (1997).MULCAHY, K.A. and 12OTHERS, Infrequent expression of theMAGE genefamily in uveal melanomas.Int. J. Cancer, 66,738–742 (1996).RAZIN, A. and CEDAR, H., DNA methylation and gene expression.Micro-biol. Rev., 55,451–460 (1991).RUSSO, V., TRAVERSARI, C., VERRECCHIA, A., MOTTOLESE, M., NATALI , P.G.and BORDINGON, C., Expression of theMAGEgene family in primary andmetastatic breast cancer.Int. J. Cancer, 64,216–221 (1995).SAHIN, U., TURECI, O. and PFREUNDSCHUH, M., Serological identification ofhuman tumor antigens.Curr. Opin. Immunol., (1998) (in press).SAHIN, U., TURECI, O., SCHMITT, H., COCHLOVIUS, B., JOHANNES, T.,SCHMITS, R., STENNER, F., LUO, G., SCHOBERT, I. and PFREUNDSCHUH, M.,Human neoplasms elicit multiple immune responses in the autologous host.Proc. nat. Acad. Sci. (Wash.), 92,11810–11813 (1995).SERRANO, A., GARCIA, A., ABRIL, E., GARRIDO, F. and RUIZ-CABELLO, F.,Methylated CpG points identified withinMAGE-1 promotor are involved ingene repression.Int. J. Cancer, 68,464–470 (1996).TURECI, O., SAHIN, U. and PFREUNDSCHUH U., Serological analysis ofhuman tumor antigens: molecular definition and implications.Mol. Med.Today, 3, 342–349 (1997).TURECI, O., SAHIN, U., SCHOBERT, I., KOSLOWSKI, M., SCHMITT, H., SCHILD,H.J., STENNER, F., SEITZ, G., RAMMENSEE, H.G. and PFREUNDSCHUH, M., TheSSX2 gene, which is involved in the t(X;18) translocation of synovialsarcomas, codes for the human tumor antigen HOM-Mel-40.Cancer Res.,56,4766–4772 (1996).VAN DEN EYNDE, B., PEETERS, O., DE BACKER, O., GAUGLER, B., LUCAS, S.and BOON, T., A new family of genes coding for an antigen recognized byautologous cytolytic T lymphocytes on a human melanoma.J. exp. Med.,182,689–698 (1995).VAN DER BRUGGEN, P., TRAVERSARI, C., CHOMEZ, P., LURQUIN, C.,DE PLAEN,E., VAN DEN EYNDE, B., KNUTH, A. and BOON, T., A gene encoding anantigen recognized by cytolytic T lymphocytes on a human melanoma.Science, 254,1643–1647 (1991).VERSTEEG, R., Aberrant methylation in cancer.Amer. J. hum. Genet., 60,751–754 (1997).WACHSMAN, J.T., DNA methylation and the association between genetic andepigenetic changes: relation tp carcinogenesis.Mutation Res., 375, 1–8(1997).WEBER, J., SALGALLER, M., SAMID , D., JOHNSON, B., HERLYN, M., LASSAM,N., TREISMAN, J. and ROSENBERG, S.A., Expression of theMAGE1 tumorantigen is up-regulated by the demethylating agent 5-aza-2-deoxycytidine.Cancer Res., 54,1766–1771 (1994).

23SSXGENE IN HUMAN TUMORS