The C Terminus of the Synovial Sarcoma-Associated SSX Proteins Interacts with the LIM Homeobox Protein LHX4

Oncogene (2008) 27, 653–662 & 2008 Nature Publishing Group All rights reserved 0950-9232/08 $30.00 www.nature.com/onc ORIGINAL ARTICLE The C terminus of the synovial sarcoma-associated SSX proteins interacts with the LIM homeobox protein LHX4

DRH de Bruijn, AHA van Dijk, MP Willemse and A Geurts van Kessel

Department of Human Genetics, Radboud University Nijmegen Medical Centre, Nijmegen, The Netherlands

As a result of the synovial sarcoma-associated t(X;18) Introduction translocation, the SS18 gene on chromosome 18 is fused to either one of the three closely related SSX genes on the Synovial sarcoma is an aggressive soft tissue tumor X chromosome. The SS18 protein is thought to act as a predominantly affecting children and young adults. transcriptional co-activator, whereas the SSX proteins are Cytogenetically, it is characterized by a recurring thought to act as transcriptional corepressors. The main chromosomal translocation, t(X;18)(p11;q11), which is SSX-repression domain is located in its C terminus, a found in more than 95% of all cases (dos Santos et al., domain that is retained in the respective SS18–SSX fusion 2001; Ladanyi et al., 2002). As a result of this proteins. Both the SS18 and SSX proteins lack DNA- translocation, the SS18 gene (previously called SYT) binding domains. Previously, we found that the SS18 and on chromosome 18 is fused to either one of the three SS18–SSX fusion proteins may be tethered to DNA closely related SSX genes on the X chromosome, SSX1, targets via the SS18-interacting protein AF10. Here, we SSX2 or SSX4 (Clark et al., 1994; Crew et al., 1995; de set out to isolate proteins that interact with the SSX Leeuw et al., 1995; Skytting et al, 1999). The occurrence C-terminal repression domain using a yeast two-hybrid of either SS18–SSX1 or SS18–SSX2 fusions was found interaction trap. Of the positive clones isolated, two to correlate to histologic (de Leeuw et al., 1994) and/or corresponded to the LIM homeobox protein LHX4, a prognostic (Kawai et al., 1998; Ladanyi et al., 2002) DNA-binding protein that is involved in transcription parameters, although the latter observation could not be regulation. An endogenous interaction was subsequently confirmed (Guillou et al., 2004). established in mammalian cells via colocalization and The SS18 gene is well conserved over evolution and is coimmunoprecipitation of the respective proteins. Inter- expressed widely in embryonic and adult tissues (de estingly, the LHX4 gene was previously found to be Bruijn et al., 1996; 2001b). The human SS18 protein deregulated in various human leukemias. In addition, it resides in the nucleus where it displays a distinct was previously found that LIM homeobox proteins may punctate pattern (Brett et al., 1997; dos Santos et al., bind to and activate the glycoprotein-a (CGA) promoter. 1997). The SS18 protein is a modular protein, consisting Using LHX4 chromatin immunoprecipitation and CGA- of two functional domains, the SS18 N-terminal promoter assays, we found that endogenous LHX4 binds homology (SNH) domain and the glutamine, proline, to the CGA promoter and that LHX4-mediated CGA glycine and tyrosine-rich QPGY domain (Thaete et al., activation is enhanced by the SS18–SSX protein, but not 1999; de Bruijn et al., 2001a). Of these, the SNH domain by the SSX protein. Taken together, we conclude that this directly interacts with the acute leukemia-associated novel protein – protein interaction may have direct transcription factor AF10 (de Bruijn et al., 2001a), the consequences for the (de)regulation of SSX and/or SWI/SNF ATPases BRM and BRG1 (Thaete et al., SS18–SSX target genes and, thus, for the development 1999; Perani et al., 2003; de Bruijn et al., 2006), the of human synovial sarcomas. histone acetyltransferase p300 (Eid et al., 2000) and the Oncogene (2008) 27, 653–662; doi:10.1038/sj.onc.1210688; corepressor mSin3A (Ito et al., 2004). Indirectly, the published online 30 July 2007 SS18 protein associates with yet another SWI/SNF protein, INI1 (Debernardi et al., 2002; Kato et al., Keywords: synovial sarcoma; SSX; LHX4; SS18–SSX; 2002). The QPGY domain is a transcriptional co- transcription regulation activation domain and mediates multimerization of the SS18 protein (Brett et al., 1997; Thaete et al., 1999; Perani et al., 2003). Together, these findings indicate that the SS18 protein acts as transcriptional co-activator through chromatin modification mechanisms (de Bruijn et al., 2006). Correspondence: Dr A Geurts van Kessel, Department of Human The SSX genes constitute a family of at least nine Genetics 855, Radboud University Nijmegen Medical Centre, PO Box members which are located in two clusters on the X 9101, 6500 HB Nijmegen, The Netherlands. E-mail: [email protected] chromosome (de Leeuw et al., 1996; Gure et al., 1997; Received 3 January 2007; revised 13 June 2007; accepted 15June 2007; dos Santos et al., 2000b). All SSX genes encode proteins published online 30 July 2007 of 188 amino acids, which are rich in charged amino The SSX C terminus interacts with LHX4 DRH de Bruijn et al 654 acids and contain an acidic C-terminal tail. Normal SSX colocalization studies, we and others have found that expression has been found in testis and, at a low level, in the SSXRD is responsible for (i) SSX nuclear localiza- thyroid (Crew et al., 1995; de Leeuw et al., 1996; Gure tion, (ii) colocalization with PcG proteins, (iii) associa- et al., 1997, 2002). Ectopic SSX expression has been tion with mitotic chromosomes, and (iv) association observed in various tumor types, in particular in human with core histones (dos Santos et al., 2000a; Kato et al., melanomas (Tureci et al., 1998, 1996; dos Santos et al., 2002). Through yeast two-hybrid screening, we have 2000b). Homology between the SSX family members is previously identified the SSX2IP (also known as ADIP) high, ranging from 88 to 95% at the protein level. The and RAB3IP proteins as in vivo interactors of the SSX SSX proteins were found to repress transcription in KRAB-like domain (de Bruijn et al., 2002). Since no reporter assays (Brett et al., 1997; Thaete et al., 1999). direct interactions between SSX and any of the PcG Part of this activity was mediated by the N-terminal proteins have been demonstrated, we assume that other SSX moiety that exhibits homology to the family of proteins mediate the observed associations between SSX Kru¨ ppel-associated box (KRAB) repression domains, and PcG proteins. Taken together, these data suggest which were recently associated with heterochromatin that the SSX proteins function as transcriptional (Huntley et al., 2006; Vogel et al., 2006). However, corepressors, and exert these functions through interac- considerably stronger repression was achieved by the tions with, as yet, unknown DNA-binding proteins. SSX C-terminal 34 amino acids, which was therefore In the majority of SS18–SSX fusion proteins char- designated SSX-repression domain (SSXRD; Figure 1). acterized to date, the C-terminal eight amino acids of This acidic domain (pI 4.8) does not display obvious the SS18 protein are replaced by the C-terminal 78 homologies to known protein domains, but exhibits a amino acids of one of the SSX proteins (Clark et al., high degree of similarity between the various SSX family 1994; Crew et al., 1995; de Leeuw et al., 1995). As a members (Lim et al., 1998; dos Santos et al., 2000a). consequence, the SS18 QPGY domain is interrupted and Immediately upstream of the SSXRD, a divergent the complete KRAB domain of SSX is lost. The main domain (DD) has been defined that exhibits the highest body of the SS18 protein, including the SNH domain, is degree of divergence between the various SSX family fused to the highly polar C-terminal SSX tail, including members (dos Santos et al., 2000a). Although there is no the SSX-divergent (SXXDDs) and SSXRDs. This fusion evidence to suggest that the SSX proteins can bind DNA has been associated with neoplastic transformation directly, all SSX proteins are localized in the nucleus, (Nagai et al., 2001) and changes in the epigenetic being both diffusely distributed and concentrated in makeup of target genes (de Bruijn et al., 2006). Another nuclear dots (Brett et al., 1997; dos Santos et al., 1997, recent study indicated that the SS18–SSX1 and SS18– 2000a; Soulez et al., 1999). These dots have been found SSX2 fusion proteins interact preferentially with the to also harbor several members of the polycomb group transcription factors Snail and Slug, respectively (Saito (PcG) proteins, that is, HPC2, BMI1 and RING1 (Lim et al., 2006). These interactions were shown to be et al., 1998; dos Santos et al., 2000a). In line with the specific for the fusion proteins and might provide an transcriptional repressor function(s) of the SSX pro- explanation for the (disputed) correlation between teins, PcG proteins form multi-protein complexes that fusion type and histologic appearance (Guillou et al., maintain the repressed state of target genes by inducing 2004). Here, we have searched for interacting proteins changes in chromatin structure (Schuettengruber et al., through yeast two-hybrid screening using the SSX1 2007; Schwartz and Pirrotta, 2007). Through cellular C-terminal domains DD and SSXRD as a bait. Our

Figure 1 (a) Diagram depicting the SS18, SSX1 and LHX4 proteins with their annotated functional domains (see text). The arrowheads above the SS18 and SSX1 proteins mark the locations of the respective synovial sarcoma-associated break points. The solid lines below the SSX1 and LHX4 proteins represent the SSX1 bait fragment and the two independent LHX4 prey fragments that were isolated in the yeast two-hybrid screen. (b) Diagram depicting the LHX4 fragments which were used as baits in the yeast two- hybrid screen. The box (bottom) indicates the LHX4 domain conferring transcriptional activation in the yeast two-hybrid assay.

Oncogene The SSX C terminus interacts with LHX4 DRH de Bruijn et al 655 data indicate that the C terminus of the SSX1, SSX2 and this interaction is a general feature of all three SSX SSX4 proteins interact in vivo with the LIM homeobox proteins (not shown). protein LHX4. This transcription factor was previously To determine whether the uncharted LHX4 C found to be deregulated and/or translocated in chronic terminus harbors transcriptional repression and/or and acute leukemias (Wu et al., 1996; Dong et al., 1997; activation capacities, we set out to perform another Wu and Minden, 1997; Kawamata et al., 2002; yeast two-hybrid screen, this time using two LHX4 Yamaguchi et al., 2003). The putative implications of fragments as baits (Figure 1b; baits 1 and 2). Both these novel protein – protein interactions for the fragments were cloned into the pBD-GAL4 vector, development of human synovial sarcomas and other introduced into yeast strain pJ69-4A, and assayed for SSX-positive tumors are discussed. auto-activation of the His reporter gene mediated by the LHX4 baits alone. By doing so, we found that all clones in which the LHX4 C terminus was fused to the GAL4 DNA-binding domain sustained growth on histidine- Results deficient plates, indicating that the LHX4 C terminus indeed harbors a novel transcriptional activation do- The SSX1, 2 and 4 C termini interact with LHX4 main (Figure 1b). To search for novel SSX interactors, we used a testis cDNA interaction library for a yeast two-hybrid screen SSX1 and LHX4 colocalize in transfected cells as described before (de Bruijn et al., 2002, 2001a). For For a further validation of the SSX/LHX4 interaction, this screen, we cloned a SSX1 C-terminal fragment we determined the sub-cellular (co)localization of both (SSX1CT), encompassing amino acids 96–188 of the proteins. Therefore, full-length SSX1 and LHX4 SSX1 protein and including the SSXDD and SSXRD cDNAs were cloned in-frame with the cyan and yellow (Figure 1a), into a pBD-GAL4 vector (de Bruijn et al., fluorescent proteins ECFP and EYFP, respectively, and 2001a). Transformation of this SSX1CT bait alone in cotransfected into HeLa cells. After fixation, the yeast cells did not yield any growth on histidine-deficient transfected cells were analysed for fluorescent signals, plates, indicating that this SSX1 fragment does not using 40,6-diamidino-2-phenylindole as a nuclear coun- contain any activation domains. Next, this bait was ter stain. In the majority of EYFP–SSX1/ECFP–LHX4 cotransfected with the cDNA interaction library and double-transfected cells, we observed EYFP–SSX1 approximately 1 Â 106 Leu þ /Trp þ -independent yeast signals as nuclear dots combined with a diffuse nuclear transformants were generated. Replating of these staining (Figures 2b and f; green). In addition, we transformants on histidine-deficient plates yielded 30 observed that the size and number of dots varied per Leu þ /Trp þ /His þ clones, which were selected for further cell, ranging from 2 to B50 dots. These observations are analysis. Sequencing revealed that five of these clones completely in line with previously published data (dos contained two independent, but clonally expanded, Santos et al., 2000a). In all double-transfected cells, the cDNAs corresponding to the human LHX4 gene (preys ECFP–LHX4 signals displayed a similar nuclear pat- 3 and 4; Figure 1a). Retransformation of the SSX1CT tern, although with a somewhat lower intensity, with a bait together with these prey cDNAs confirmed the clear absence of nucleolar staining (Figures 2c and g; yeast two-hybrid interaction. Cotransfection of each red). In the overlay of the green and red images, almost prey with two control plasmids expressing lamin-B and all signals were yellow (Figures 2d and h), indicating a p53 turned out negative, thereby validating the yeast near complete colocalization of the SSX1 and LHX4 two-hybrid interaction observed. The LHX4 gene proteins, particularly in the nuclear dots. In a low encodes a 390 amino-acid protein with two tandem percentage of double-transfected cells, a configuration LIM domains (LIM1 and LIM2), followed by a single of many small nuclear SSX and LHX4 dots could be homeodomain (HOX) and a C terminus without observed (Figures 2j and k). The overlay of these images apparent functional domains (Figure 1). This member clearly indicated that colocalization in such nuclei was of the LIM homeobox protein family was found to be incomplete (Figures 2j–l; arrowheads). Similar experi- critical for pituitary and motor neuron development ments using ECFP–EYFP dye-swaps, transfection of (Sheng et al., 1997; Sharma et al., 1998). Further COS1 cells and/or expression of fluorescently tagged sequence analysis revealed that both prey cDNAs SSX1CT proteins yielded very similar results (not carried the LHX4 poly-A tail and started at nucleotide shown). We conclude that the exogenously expressed 685 (prey 3; amino acid 155) and 622 (prey 4; amino acid SSX1 and LHX4 proteins colocalize in both HeLa and 134) of the published LHX4 sequence, respectively. The COS1 cell nuclei. In a small percentage of, morpho- deduced proteins encompassed the HOX and the logically distinct, nuclei this colocalization was incom- uncharted LHX4 C terminus (Figure 1a). To test plete, indicating that the SSX/LHX4 interactions may whether the SSX/LHX4 interaction is specific for the be dynamic in nature. SSX1-derived C terminus, we cloned the corresponding C termini of the SSX2 and SSX4 genes into the pBD- GAL4 vector and tested them for interaction with SS18–SSX1 and LHX4 colocalize in nuclear domains LHX4 prey 3 in the yeast two-hybrid system. Using this Next, we tested whether the SS18–SSX1 fusion protein assay, we found that all three SSX C-terminal domains also colocalizes with LHX4. Therefore, a full-length were equally able to interact with LHX4, indicating that SS18–SSX1 cDNA was cloned in-frame with the cyan

Oncogene The SSX C terminus interacts with LHX4 DRH de Bruijn et al 656

Figure 2 Detection of EYFP–SSX1 (b, f and j; green) and ECFP–LHX4 full-length (c, g and k; red) in double-transfected HeLa cells. After overlay of these images an almost complete colocalization can be observed for EYFP–SSX1 and ECFP–LHX4 (d, h and l; yellow). 40,6-Diamidino-2-phenylindole (DAPI) counterstaining indicates a nuclear localization of all signals (a, e and i; blue). In morphologically distinct nuclei displaying numerous ﬁne dots (i–l), partial loss of SSX1/LHX4 colocalization can be observed (j–l; arrowheads).

fluorescent protein (ECFP) and combined with the same the LHX4 and AF10 proteins colocalized without EYFP–LHX4 construct that was used previously (see SS18–SSX1 staining, the presence of this fusion protein above). These constructs were either transfected into the appears to be crucial for the colocalization of LHX4 HeLa cell line or into the synovial sarcoma cell line and AF10 in the same nuclear domains. The cotrans- SYO1 (Kawai et al., 2004), and the fluorescent signals fections of HeLa cells yielded very similar results (not were analysed using confocal laser scanning microscopy. shown). Previously, we identified the AF10 protein as a bona fide interactor of the SS18 N terminus (de Bruijn et al., 2001a). Since the AF10 interacting domain is retained in Coimmunoprecipitation ofendogenous LHX4 the SS18–SSX fusion protein, we included an AF10 and SS18–SSX proteins expression construct encoding the full-length AF10 To establish a direct interaction between the SSX1 and protein fused to dsRED in our SYO1 cotransfection LHX4 proteins, we performed immunoprecipitations experiments. By doing so, we observed that all triple from HeLa cells, which were transfected with full-length labeled cells displayed nuclear EYFP, ECFP and SSX1 and LHX4 expression constructs. In this fashion, dsRED signals (Figure 3; green, blue and red, respec- we readily detected the LHX4 protein in anti-SSX tively). Most of the nuclei exhibited an almost full immunoprecipitates (not shown), indicating that the colocalization of all three proteins (Figure 3; top row), exogenous SSX1 and LHX4 proteins interact directly. while some showed less colocalization (Figure 3; bottom To confirm this interaction in vivo within the context of row). However, since we did not find any cells in which synovial sarcoma, we set out to coimmunoprecipitate

Oncogene The SSX C terminus interacts with LHX4 DRH de Bruijn et al 657

Figure 3 Confocal images of two triple-transfected SYO1 cells. From left to right: triple image overlay, EYFP–LHX4, ECFP–SS18– SSX1 and dsRED–AF10. In the upper nucleus, an almost complete colocalization of ECFP–SS18–SSX1, EYFP–LHX4 and dsRED– AF10 signals (nearly white in the overlay) can be observed in a nuclear dotted pattern. The lower nucleus displays regions with more prominent SS18–SSX1–AF10 colocalization (light purple in the overlay), alternating with domains displaying full colocalization (white in the overlay). the endogenous LHX4 and SS18–SSX (fusion) proteins. known target of other LHX proteins is the glycoprotein Since we did not observe any preference of SSX C hormone-a (CGA) gene. LHX-mediated CGA gene termini for interaction with the LHX4 protein (see expression depends on the presence of a LHX-binding above), we again employed the synovial sarcoma cell site (PGBE) within the CGA promoter (Roberson et al., line SYO1, which expresses the SS18–SSX2 and LHX4 1994). To establish direct binding of the LHX4 protein proteins endogenously, but does not express wild-type to this promoter, we performed ChIP with the anti- SSX, as judged by western blotting and reverse LHX4 antibody on SYO1 extracts. The recovered ChIP transcription (RT)–PCR (not shown). SYO1 nuclear DNAs were analysed by real-time quantitative PCR (Q- extracts were subjected to immunoprecipitation with a PCR) for enrichment of CGA-promoter sequences. By rabbit pre-immune serum (mock IP) and a rabbit doing so, we obtained a P/I value (see Materials and polyclonal anti-SSX antibody (B39; dos Santos et al., methods) of 6.18 (mean of three independent experi- 1997), which recognizes the SS18–SSX2 fusion protein. ments), indicating that in SYO1 cells the endogenous The input material and both immunoprecipitates were LHX4 protein binds to the CGA promoter. To analysed by western blotting using a polyclonal goat subsequently determine the effect of the SS18–SSX2 anti-LHX4 antibody. In the nuclear extract (NE) and in and/or SSX2 proteins on CGA expression, we cloned the anti-SSX immunoprecipitate (SS18–SSX2 IP), but two CGA-promoter fragments into a pGL3 luciferase not in the mock IP, the endogenous LHX4 protein was reporter construct. Of these fragments, one lacked the observed as a doublet of two bands (50 and 48 kDa, LHX-binding site PGBE (CGA1) whereas the other Figure 4a). Similarly, we performed immunoprecipita- contained this site (CGA2; Figure 4c). Together with a tions with the anti-LHX4 antibody. After western Renilla luciferase control plasmid, these CGA con- blotting using the anti-SSX antibody, the endogenous structs were transfected into SYO1 cells. Subsequently, 55 kDa SS18–SSX2 protein could readily be observed in the luciferase activities were measured, and normalized the nuclear extract (NE) and the anti-LHX4 immuno- to the Renilla values. To assess the effect of the LHX4- precipitate (LHX4 IP), but not in the mock IP binding site on CGA-promoter activation, the CGA2 (Figure 4b). From these results, we conclude that the values were normalized to the CGA1 values. In three endogenous LHX4 and SS18–SSX2 (fusion) proteins independent transfection experiments (each measured in interact in vivo. In synovial sarcoma cells, this interac- duplicate), we observed a threefold increase in luciferase tion may confer the transcriptional regulatory functions activity in the CGA2-transfected SYO1 cells as com- of the SS18–SSX fusion proteins to cognate LHX4 pared to the CGA1-transfected SYO1 cells (data not target genes. shown). In similar experiments, using an SS18–SSX- negative, LHX4-positive cell line (HEK293), we observed a twofold increase of CGA2-promoter activity SSX and SS18–SSX affect LHX4-mediated gene versus CGA1-promoter activity. These results indicate expression that the SS18–SSX2 protein, in conjunction with the To establish whether the SSX and/or SS18–SSX (fusion) LHX4 protein bound to the LHX4-binding site, proteins are able to affect the expression of known LHX signiﬁcantly enhances CGA transactivation. To assess target genes, we performed both chromatin immuno- further whether the wild-type SSX protein affects this precipitation (ChIP) and transactivation assays. A SS18–SSX2-mediated CGA transactivation, we included

Oncogene The SSX C terminus interacts with LHX4 DRH de Bruijn et al 658 that, in the context of synovial sarcoma, the SSX2 and SS18–SSX2 proteins exert opposite regulatory activities, probably by competing for binding to the LHX4 protein.

Discussion

Through yeast two-hybrid screening, using the synovial sarcoma-associated SSX1 C terminus as a bait, we isolated two cDNAs corresponding to the LHX4 protein, a transcription factor that is known to be involved in pituitary and motor neuron development (Sheng et al., 1997; Sharma et al., 1998; Machinis et al., 2001; Raetzman et al., 2002) and to be deregulated and/ or translocated in human chronic myeloid and acute lymphoblastic leukemias (Wu et al., 1996; Dong et al., 1997; Wu and Minden, 1997; Kawamata et al., 2002; Yamaguchi et al., 2003). In the yeast two-hybrid system, the LHX4 protein interacts equally well with the C terminus of all SSX proteins known to be involved in synovial sarcoma, that is SSX1, SSX2 and SSX4. We validated this interaction in synovial sarcoma cells via coimmunoprecipitation of the respective endogenous (fusion) proteins. Further support that LHX4 is a bona fide interactor of the SSX proteins came from cellular colocalization experiments, which were in conformity with our previously published data (dos Santos et al., 2000a). This colocalization, however, was not always complete, indicating that the SSX/LHX4 interaction may be dynamic in nature. Whereas most cells showed B Figure 4 Coimmunoprecipitation of the endogenous LHX4 and up to 50 nuclear dots in otherwise diffusely stained SS18–SSX2 (fusion) proteins from the SYO1 cell line, and CGA nuclei with an almost complete SSX/LHX4 colocaliza- transactivation assays in SYO1. (a) Western blot containing an tion, some cells showed a configuration of many SYO1 nuclear extract (NE), an anti-SSX immunoprecipitate relatively small nuclear dots with an incomplete SSX/ (SS18–SSX2 IP) and a rabbit pre-immune immunoprecipitate (mock IP), incubated with an anti-LHX4 antibody. The endogen- LHX4 colocalization. A similar, but less obvious, ous LHX4 protein is detected in the nuclear extract and the anti- pattern of incomplete colocalization was observed in SSX immunoprecipitate as a doublet of 50 and 48 kDa. (b) Western some triple (SS18–SSX1, LHX4 and AF10)-transfected blot containing a SYO1 nuclear extract (NE), an anti-LHX4 cells, indicating that the dynamic regulation may also immunoprecipitate (LHX4 IP) and a goat pre-immune immuno- relate to the SS18–SSX fusion proteins. The exact precipitate (mock IP), incubated with an anti-SSX antibody. The endogenous SS18–SSX2 protein is detected in the nuclear extract mechanisms underlying these dynamic protein localiza- and the LHX4 immunoprecipitate as a 55 kDa band. (c) Luciferase- tion patterns still remain to be identified. reporter constructs derived from the CGA promoter, without and The LHX4 protein exhibits features typical of with the LHX4-binding site PGBE (CGA1 and CGA2, respec- transcription factors, including a DNA-binding HOX tively). (d) CGA transactivation assays without (vector) and with exogenous SSX2 expression. Values are indicated as fold activation and two tandem LIM domains, each containing two of the CGA2 construct versus the CGA1 construct. In the presence zinc fingers. LIM HOX protein genes have frequently of SSX2, the CGA2-promoter activity is reduced significantly. been found to be involved in cancer. Ectopic expression of the LHX1 and LHX2 genes has, for example, been detected in various leukemias (Wu et al., 1996; Dong et al., 1997; Wu and Minden, 1997) and stable an SSX2 expression vector (and a control empty vector) expression of the LHX2 gene in hematopoietic stem in the same transfection experiments and determined the cells, engrafted in stem cell-deficient mice, has been luciferase activities as described above. In the control shown to cause myeloproliferative disorders and acute empty vector experiments, again an approximately leukemias (Richter et al., 2003). In cellular transforma- threefold increase in luciferase activity was observed in tion assays, using Rat1 and NIH3T3 cell lines, the CGA2-transfected versus CGA1-transfected SYO1 cells oncogenic activities of the LHX4 gene could not be (Figure 4d; vector). However, this effect was signifi- defined (Kawamata et al., 2002), but aberrant expres- cantly decreased upon exogenous SSX2 expression sion of this gene was observed in the case of chronic (Figure 4d; SSX2), indicating that the SSX2 protein myeloid leukemia in blast crisis and in the case of pre-B exerts an inhibitory effect on the SS18–SSX2/LHX4- acute lymphoblastic leukemia. Both cases carried mediated CGA transactivation. These data indicate a t(1;14)(q25;q32) chromosome translocation, which

Oncogene The SSX C terminus interacts with LHX4 DRH de Bruijn et al 659 resulted in juxtaposition of the respective LHX4 and dissociate them from their polycomb link. A preliminary IGH genes (Kawamata et al., 2002; Yamaguchi et al., indication for the existence of such a dynamic interac- 2003). The LHX4 gene was also found to be over- tion comes from our observations that in a certain expressed in the pre-B acute lymphoblastic leukemia cell percentage of the nuclei studied, the SSX1/LHX4 and line MD930 (Yamaguchi et al., 2003). During normal SS18–SSX1/LHX4 interactions are incomplete. development, the LHX4 gene is predominantly expres- Taken together, we have shown that the C-terminal sed in the central nervous system (Liu et al., 2002), and it domain of the SSX protein interacts in vivo with the was proposed that its overexpression may be associated LIM HOX protein LHX4 and that, through this with cellular proliferation and/or anti-apoptosis (Ka- interaction, the SSX protein may exert one or more wamata et al., 2002; Yamaguchi et al., 2003). Indeed, of its multiple functions, including that of a transcrip- mouse knockout studies revealed that LHX4 deficiency tional corepressor. Previously, we found that the N- leads to an increased apoptotic rate in Rathke’s pouch, terminal domain of the SS18 protein interacts with the resulting in an impaired pituitary development reminis- transcription factor AF10 and that, through this cent of human congenital pituitary hormone deficiency interaction, the SS18 protein may exert its function as (CPHD; Sheng et al., 1997; Machinis et al., 2001; a transcriptional co-activator (de Bruijn et al, 2001a). Raetzman et al., 2002). In addition, LHX4 was shown to In Figure 5, the consequences of these and other protein have a direct stimulatory effect on for example the – protein interactions are depicted for normal and prolactin gene promoter. This activation could be t(X;18)-positive synovial sarcoma cells, in which the enhanced by other HOX-containing transcription fac- SS18–SSX fusion protein is thought to act as a tors such as Pit-1, or be inhibited by interacting proteins transcriptional ‘activator – repressor’. This anomalous such as the selective LIM-binding protein SLB (Howard regulatory function is expected to lead to an aberrant and Maurer, 2000), indicating that the modulation of regulation of target genes and, ultimately, malignant LHX4 activity occurs through protein – protein inter- transformation. actions. In this study, we observed a similar, LHX4- mediated, regulation of the CGA gene. Through ChIP, we were able to show that endogenous LHX4 binds Materials and methods directly to the CGA promoter in synovial sarcoma cells. Furthermore, we found through promoter assays that Cloning and sequencing procedures the regulation of CGA gene expression is differentially All cloning procedures were essentially as described before (de affected by the LHX4-interacting proteins SSX and Bruijn et al., 1996; 2001a, b). Sequence analyses were SS18–SSX, that is, the CGA promoter was activated in performed at the DNA Sequence Facility of the Radboud the presence of the SS18–SSX fusion protein, but University Nijmegen Medical Centre. DNA and protein repressed upon exogenous expression of the wild-type databases were searched using the BLAST or BLAT search algorithms at the NCBI or UCSC, respectively. The SSX1, SSX protein. Therefore, we assume that this novel SSX/ SSX2 and SSX4 bait fragments, carrying amino acids 96–188 LHX4 interaction may act as an allosteric mechanism of either protein, were generated by PCR from full-length (Howard and Maurer, 2000) to regulate LHX4- and/or cDNAs and cloned in frame with the GAL4 DNA-binding SSX-associated functions. In synovial sarcoma cells, this domain into the pBD-GAL4-Cam vector (Stratagene, La mechanism may be affected by the SS18–SSX fusion Jolla, CA, USA). For efficient, in-frame cloning of selected protein, thereby, inhibiting apoptosis and/or promoting interactors (and deletion fragments thereof) with the GAL4 proliferation. Endogenous LHX4 (over)expression has transactivation or DNA-binding domains, we used the same been found to inhibit apoptosis in leukemias, while set of adapted pGAD10 (Clontech, Mountain View, CA, endogenous, ectopic (over)expression of wild-type SSX USA) and pBD-GAL4-Cam vectors as described previously genes has been observed in melanoma (dos Santos et al., (de Bruijn et al., 2001a). All SSX1 and LHX4-deletion fragments were generated by PCR with specific oligonucleotide 2001), osteosarcoma (Naka et al., 2002), breast cancer primers. PCR products were sub-cloned into pGEM-T (Mischo et al., 2006) and multiple myeloma (Taylor (Promega, Leiden, The Netherlands) and, after verification et al., 2005). Although the significance of SSX (over)- by sequencing, cloned into the correct yeast or mammalian expression in these tumor types has not yet been expression vector. addressed, one study has indicated that it may affect cell motility (Cronwright et al., 2005). Obviously, Yeast two-hybrid screening further studies are required to assess whether and to Yeast two-hybrid screening was performed using a human what extent the LHX4 and SSX genes are coexpressed in testis cDNA library (de Bruijn et al., 2002). For the these (and other) tumor types. construction of this library, poly-A( þ ) RNA was isolated Our current work, and that of others, indicates that using OLIGOTEX (Qiagen, Venlo, The Netherlands), reverse- the SSX C terminus is involved in multiple functions, transcribed and cloned into hybriZAP (Stratagene). The resulting primary phage library contained 4 Â 106 independent including histone association, transcriptional repression 6 (possibly mediated by the polycomb complex) and clones. Approximately 1 Â 10 clones were amplified once and 1 Â 109 of the resulting plaque forming units were mass-excised LHX4 protein interaction. The regulation of one or according to manufacturers’ instructions to generate the more of these functions may occur in a competitive and interaction cDNA library in pAD-GAL4. The average insert dynamic fashion. As such, interaction with the LHX4 size of this library was estimated to be around 1 kb. The yeast protein might be sufficient for SSX proteins to render strain used for all interaction traps was pJ69-4A (a kind gift their repression domains (SSXRD) inactive and/or from Philip James). Positive interactions in this yeast strain

Oncogene The SSX C terminus interacts with LHX4 DRH de Bruijn et al 660

Figure 5 Model depicting the synovial sarcoma-associated SS18, SSX and SS18–SSX (fusion) proteins and their respective (dynamic) interactions. In normal cells (left), the SSX proteins and their interactors SSX2IP and RAB3IP may associate with the polycomb complex and histones and in addition, through interaction with the LHX4 protein, bind to cognate DNA sites and affect target gene expression. The SS18 protein and its interacting proteins BRM, BRG1, Sin3A and p300 may associate with the SWI/SNF complex and in addition, through interaction with the AF10 protein, bind to cognate DNA sites and affect target gene expression. In synovial sarcoma cells (right), the SS18–SSX fusion proteins have lost the interaction domains for SSX2IP and RAB3IP, but have retained the interaction/association domains for both the SWI/SNF and the polycomb complexes. Through these interactions, the SS18–SSX fusion proteins may anomalously affect the regulation of these target genes and/or affect the regulation of other (novel) target genes, either through AF10, LHX4 or both.

can be selected for by adenine and histidine auxotrophy, next described previously (de Bruijn et al., 2001a, 2002). For the to b-galactosidase activity. This yeast strain was cotrans- endogenous SS18–SSX2/LHX4 coimmunoprecipitations, formed with the bait plasmid pBD-GAL4-SSX1CT and the nuclear extracts were prepared from SYO1 cells according to testis cDNA library (in pAD-GAL4), using the Yeastmaker the Lamond Lab protocol (http://www.lamondlab.com/f7pro- transformation kit (Clontech). Subsequently, all double tocols.htm). Approximately 500 mg of freshly prepared nuclear transfectants, containing a pBD and a pAD vector, were proteins were subjected to immunoprecipitations with 2 mgof selected on selective medium lacking leucine and tryptophan. the goat anti-LHX4 antibody (sc-22136, Santa Cruz, Heidel- A total of 1 Â 106 double transformants were harvested and berg, Germany), the affinity-purified B39 rabbit anti-SSX replated on selective medium lacking leucine, tryptophan and antibody, and the goat and rabbit pre-immune sera in the histidine, supplemented with 25m M of 3-aminotriazole (3-AT; presence of protein A/G agarose beads (Santa Cruz). After Sigma, Zwijndrecht, The Netherlands). LacZ activity was washing, the whole immunoprecipitates, the corresponding determined for all growing colonies using a filter lift assay mock-IP, and 50 mg of nuclear extract, were used to prepare according to the manufacturers’ instructions. The cDNA western blots according to standard procedures. These blots inserts of positive clones were isolated by direct PCR on yeast were incubated with either the anti-SSX (LHX4-IP) or the colonies with two pAD-GAL4 specific oligonucleotides, AD- anti-LHX4 (SSX-IP) antibodies. The signals were detected for (CTGTCACCTGGTTGGACGGACCAA) and pGAD- with the Odyssey Infrared Imaging System (LI-COR, Lincoln, rev (GTGAACTTGCGGGGTTTTTCAG). After sequencing NE, USA), using the Alexa Fluor 680 goat anti-rabbit IgG of the PCR products, the pAD plasmids from the yeast (Molecular probes, Leiden, The Netherlands) or IRDye-800 interaction clones were rescued in DH5a bacteria. The donkey anti-Goat IgG (Rockland, Gilbertsville, PA, USA) as interacting cDNA clones were retransformed together with secondary antibodies. the standard control plasmids expressing lamin-B and p53 and the SSX1 bait into pJ69-4A to confirm the previously found Chromatin immunoprecipitations and CGA transactivation interaction. assays Chromatin was extracted from SYO1 cells as described (Boyd Transfections, fluorescence and immunoprecipitation assays et al., 1998). Protein – DNA complexes were immunoprecipi- Human HeLa, green monkey kidney COS1 and human tated with a LHX4 antibody (see above) and a goat pre- synovial sarcoma SYO1 cells were grown in Dulbecco’s immune serum. The recovered genomic DNAs were analysed modification of Eagle’s medium containing 10% fetal calf by Q-PCR on an ABI 7700 sequence detection system, using a serum. Transient transfections of these cell lines were SYBR green PCR master mix (Applied Biosystems, Nieuwer- performed with the Amaxa Nucleofector in Amaxa buffer V kerk a/d ijssel, The Netherlands) with primer pairs spanning with program A28 (HeLa and COS1) or O17 (SYO1). the CGA promoter (F: GTACAAAGTGACAGCGTA Approximately 1 mg of each of the appropriate plasmid DNAs, CTCTCTTTTCATG; R: AGCTTCGTCTTATGAGTTCTC pECFP–SSX1/pEYFP–SSX1, pEYFP–LHX4/pECFP–LHX4 AGTAACTG), and an intronic region located 1.6 kb down- and/or pdsRED–AF10 were combined with 3 Â 106 cells in stream (F: GTTACAGCAGTGCCATTGTAATGTTGC; R: each transfection. The transfected cells were seeded on poly-L- CTGATGCTAAGATGGTGTCCTTCCAC). The Q-PCR lysine (Sigma)-coated glass slides and grown for another 16– data were calculated as promoter/intron ratios (P/I) to indicate 18 h, before fluorescent analysis. Fluorescent detection of the the enrichment of the CGA-promoter sequence in the LHX4 transfected ECFP, EYFP and dsRED-tagged proteins was ChIP. For the transactivation experiments, two CGA-promo- performed after a 30 min fixation in 3% formaldehyde as ter fragments (CGA1: À253 to þ 50 and CGA2: À752 to

Oncogene The SSX C terminus interacts with LHX4 DRH de Bruijn et al 661 þ 50) were generated by PCR and cloned into the pGL3 Acknowledgements luciferase-reporter vector (Clontech). These vectors were cotransfected in triplicate with a Renilla luciferase expression We thank Jack Fransen (Department of Cell Biology, vector (Promega) into the SYO1 cell line and seeded in Nijmegen Centre for Molecular Life Sciences) for assistance duplicate. One day after transfection, the cells were lysed and with the confocal laser scanning microscopy. This work was ﬁreﬂy and Renilla luciferase activities were measured using the supported by a grant from the Dutch Cancer Society (KWF Dual-Glo Luciferase System (Promega). kankerbestrijding).

References

Boyd KE, Wells J, Gutman J, Bartley SM, Farnham PJ. metaphase and interphase FISH: relationship to histologic (1998). c-Myc target gene specificity is determined by a post- subtypes. Cancer Genet Cytogenet 73: 89–94. DNA binding mechanism. Proc Natl Acad Sci USA 95: Debernardi S, Bassini A, Jones LK, Chaplin T, Linder B, de 13887–13892. Bruijn DR et al. (2002). The MLL fusion partner AF10 Brett D, Whitehouse S, Antonson P, Shipley J, Cooper C, binds GAS41, a protein that interacts with the human SWI/ Goodwin G. (1997). The SYT protein involved in the t(X;18) SNF complex. Blood 99: 275–281. synovial sarcoma translocation is a transcriptional activator Dong WF, Heng HH, Lowsky R, Xu Y, DeCoteau JF, localised in nuclear bodies. Hum Mol Genet 6: 1559–1564. Shi XM et al. (1997). Cloning, expression, and chromosomal Clark J, Rocques PJ, Crew AJ, Gill S, Shipley J, Chan AM localization to 11p12–13 of a human LIM/HOMEOBOX et al. (1994). Identification of novel genes, SYT and SSX, gene, hLim-1. DNA Cell Biol 16: 671–678. involved in the t(X;18)(p11.2;q11.2) translocation found in dos Santos NR, de Bruijn DR, Balemans M, Janssen B, human synovial sarcoma. Nat Genet 7: 502–508. Gartner F, Lopes JM et al. (1997). Nuclear localization of Crew AJ, Clark J, Fisher C, Gill S, Grimer R, Chand A et al. SYT, SSX and the synovial sarcoma-associated SYT-SSX (1995). Fusion of SYT to two genes, SSX1 and SSX2, fusion proteins. Hum Mol Genet 6: 1549–1558. encoding proteins with homology to the Kruppel-associated dos Santos NR, de Bruijn DR, Geurts van Kessel A. (2001). box in human synovial sarcoma. EMBO J 14: 2333–2340. Molecular mechanisms underlying human synovial sarcoma Cronwright G, Le Blanc K, Gotherstrom C, Darcy P, Ehnman development. Genes Chromosomes Cancer 30: 1–14. M, Brodin B. (2005). Cancer/testis antigen expression in dos Santos NR, de Bruijn DR, Kater-Baats E, Otte AP, human mesenchymal stem cells: down-regulation of SSX Geurts van Kessel A. (2000a). Delineation of the protein impairs cell migration and matrix metalloproteinase 2 domains responsible for SYT, SSX, and SYT-SSX nuclear expression. Cancer Res 65: 2207–2215. localization. Exp Cell Res 256: 192–202. de Bruijn DR, Allander SV, van Dijk AH, Willemse MP, dos Santos NR, Torensma R, de Vries TJ, Schreurs MW, Thijssen J, van Groningen JJ et al. (2006). The synovial de Bruijn DR, Kater-Baats E et al. (2000b). Heterogeneous sarcoma-associated SS18-SSX2 fusion protein induces epi- expression of the SSX cancer/testis antigens in human genetic gene (de)regulation. Cancer Res 66: 9474–9482. melanoma lesions and cell lines. Cancer Res 60: 1654–1662. de Bruijn DR, Baats E, Zechner U, de Leeuw B, Balemans M, Eid JE, Kung AL, Scully R, Livingston DM. (2000). p300 Olde Weghuis D et al. (1996). Isolation and characterization interacts with the nuclear proto-oncoprotein SYT as part of of the mouse homolog of SYT, a gene implicated in the the active control of cell adhesion. Cell 102: 839–848. development of human synovial sarcomas. Oncogene 13: Guillou L, Benhattar J, Bonichon F, Gallagher G, Terrier P, 643–648. Stauffer E et al. (2004). Histologic grade, but not SYT-SSX de Bruijn DR, dos Santos NR, Kater-Baats E, Thijssen J, van fusion type, is an important prognostic factor in patients den Berk L, Stap J et al. (2002). The cancer-related protein with synovial sarcoma: a multicenter, retrospective analysis. SSX2 interacts with the human homologue of a Ras-like J Clin Oncol 22: 4040–4050. GTPase interactor, RAB3IP, and a novel nuclear protein, Gure AO, Tu¨ reci O, Sahin U, Tsang S, Scanlan MJ, Jager E SSX2IP. Genes Chromosomes Cancer 34: 285–298. et al. (1997). SSX: a multigene family with several members de Bruijn DR, dos Santos NR, Thijssen J, Balemans M, transcribed in normal testis and human cancer. Int J Cancer Debernardi S, Linder B et al. (2001a). The synovial sarcoma 72: 965–971. associated protein SYT interacts with the acute leukemia Gure AO, Wei IJ, Old LJ, Chen YT. (2002). The SSX gene associated protein AF10. Oncogene 20: 3281–3289. family: characterization of 9 complete genes. Int J Cancer de Bruijn DR, Kater-Baats E, Eleveld M, Merkx G, 101: 448–453. Geurts van Kessel A. (2001b). Mapping and characterization Howard PW, Maurer RA. (2000). Identification of a conserved of the mouse and human SS18 genes, two human SS18-like protein that interacts with specific LIM homeodomain genes and a mouse Ss18 pseudogene. Cytogenet Cell Genet transcription factors. J Biol Chem 275: 13336–13342. 92: 310–319. Huntley S, Baggott DM, Hamilton AT, Tran-Gyamfi M, de Leeuw B, Balemans M, Geurts van Kessel A. (1996). A Yang S, Kim J et al. (2006). A comprehensive catalog of novel Kruppel-associated box containing the SSX gene human KRAB-associated zinc finger genes: insights into the (SSX3) on the human X chromosome is not implicated in evolutionary history of a large family of transcriptional t(X;18)-positive synovial sarcomas. Cytogenet Cell Genet 73: repressors. Genome Res 16: 669–677. 179–183. Ito T, Ouchida M, Ito S, Jitsumori Y, Morimoto Y, Ozaki T de Leeuw B, Balemans M, Olde Weghuis D, Geurts van Kessel et al. (2004). SYT, a partner of SYT-SSX oncoprotein in A. (1995). Identification of two alternative fusion genes, synovial sarcomas, interacts with mSin3A, a component of SYT-SSX1 and SYT-SSX2, in t(X;18)(p11.2;q11.2)-positive histone deacetylase complex. Lab Invest 84: 1484–1490. synovial sarcomas. Hum Mol Genet 4: 1097–1099. Kato H, Tjernberg A, Zhang W, Krutchinsky AN, An W, de Leeuw B, Suijkerbuijk RF, Olde Weghuis D, Meloni AM, Takeuchi T et al. (2002). SYT associates with human SNF/ Stenman G, Kindblom LG et al. (1994). Distinct Xp11.2 SWI complexes and the C-terminal region of its fusion breakpoint regions in synovial sarcoma revealed by partner SSX1 targets histones. J Biol Chem 277: 5498–5505.

Oncogene The SSX C terminus interacts with LHX4 DRH de Bruijn et al 662 Kawai A, Naito N, Yoshida A, Morimoto Y, Ouchida M, Saito T, Nagai M, Ladanyi M. (2006). SYT-SSX1 and SYT- Shimizu K et al. (2004). Establishment and characterization SSX2 interfere with repression of E-cadherin by snail and of a biphasic synovial sarcoma cell line, SYO-1. Cancer Lett slug: a potential mechanism for aberrant mesenchymal to 204: 105–113. epithelial transition in human synovial sarcoma. Cancer Res Kawai A, Woodruff J, Healey JH, Brennan MF, Antonescu CR, 66: 6919–6927. Ladanyi M. (1998). SYT-SSX gene fusion as a determinant Schuettengruber B, Chourrout D, Vervoort M, Leblanc B, of morphology and prognosis in synovial sarcoma. N Engl J Cavalli G. (2007). Genome regulation by polycomb and Med 338: 153–160. trithorax proteins. Cell 128: 735–745. Kawamata N, Sakajiri S, Sugimoto KJ, Isobe Y, Kobayashi Schwartz YB, Pirrotta V. (2007). Polycomb silencing mechan- H, Oshimi K. (2002). A novel chromosomal translocation isms and the management of genomic programmes. Nat Rev t(1;14)(q25;q32) in pre-B acute lymphoblastic leukemia Genet 8: 9–22. involves the LIM homeodomain protein gene, Lhx4. SharmaK,ShengHZ,LettieriK,LiH,KaravanovA,PotterS Oncogene 21: 4983–4991. et al. (1998). LIM homeodomain factors Lhx3 and Lhx4 Ladanyi M, Antonescu CR, Leung DH, Woodruff JM, Kawai assign subtype identities for motor neurons. Cell 95: 817–828. A, Healey JH et al. (2002). Impact of SYT-SSX fusion type Sheng HZ, Moriyama K, Yamashita T, Li H, Potter SS, on the clinical behavior of synovial sarcoma: a multi- Mahon KA et al. (1997). Multistep control of pituitary institutional retrospective study of 243 patients. Cancer Res organogenesis. Science 278: 1809–1812. 62: 135–140. Skytting B, Nilsson G, Brodin B, Xie Y, Lundeberg J, Uhlen Lim FL, Soulez M, Koczan D, Thiesen HJ, Knight JC. (1998). M et al. (1999). A novel fusion gene, SYT-SSX4, in synovial A KRAB-related domain and a novel transcription repres- sarcoma. J Natl Cancer Inst 91: 974–975. sion domain in proteins encoded by SSX genes that are Soulez M, Saurin AJ, Freemont PS, Knight JC. (1999). SSX disrupted in human sarcomas. Oncogene 17: 2013–2018. and the synovial-sarcoma-speciﬁc chimaeric protein SYT- Liu Y, Fan M, Yu S, Zhou Y, Wang J, Yuan J et al. (2002). SSX co-localize with the human Polycomb group complex. cDNA cloning, chromosomal localization and expression Oncogene 18: 2739–2746. pattern analysis of human LIM-homeobox gene LHX4. Taylor BJ, Reiman T, Pittman JA, Keats JJ, de Bruijn DR, Brain Res 928: 147–155. Mant MJ et al. (2005). SSX cancer testis antigens are Machinis K, Pantel J, Netchine I, Leger J, Camand OJ, expressed in most multiple myeloma patients: co-expression Sobrier ML et al. (2001). Syndromic short stature in patients of SSX1, 2, 4, and 5correlates with adverse prognosis with a germline mutation in the LIM homeobox LHX4. and high frequencies of SSX-positive PCs. J Immunother 28: Am J Hum Genet 69: 961–968. 564–575. Mischo A, Kubuschok B, Ertan K, Preuss KD, Romeike B, Thaete C, Brett D, Monaghan P, Whitehouse S, Rennie G, Regitz E et al. (2006). Prospective study on the expression of Rayner E et al. (1999). Functional domains of the SYT and cancer testis genes and antibody responses in 100 consecu- SYT-SSX synovial sarcoma translocation proteins and tive patients with primary breast cancer. Int J Cancer 118: co-localization with the SNF protein BRM in the nucleus. 696–703. Hum Mol Genet 8: 585–591. Nagai M, Tanaka S, Tsuda M, Endo S, Kato H, Sonobe H Tu¨ reci O, Chen YT, Sahin U, Gure AO, Zwick C, Villena C et al. (2001). Analysis of transforming activity of human et al. (1998). Expression of SSX genes in human tumors. synovial sarcoma-associated chimeric protein SYT-SSX1 Int J Cancer 77: 19–23. bound to chromatin remodeling factor hBRM/hSNF2 Tu¨ reci O, Sahin U, Schobert I, Koslowski M, Scmitt H, Schild alpha. Proc Natl Acad Sci USA 98: 3843–3848. HJ et al. (1996). The SSX-2 gene, which is involved in the Naka N, Araki N, Nakanishi H, Itoh K, Mano M, Ishiguro S t(X;18) translocation of synovial sarcomas, codes for et al. (2002). Expression of SSX genes in human osteosarco- the human tumor antigen HOM-MEL-40. Cancer Res 56: mas. Int J Cancer 98: 640–642. 4766–4772. Perani M, Ingram CJ, Cooper CS, Garrett MD, Goodwin GH. Vogel MJ, Guelen L, de Wit E, Peric-Hupkes D, Loden M, (2003). Conserved SNH domain of the proto-oncoprotein Talhout W et al. (2006). Human heterochromatin proteins SYT interacts with components of the human chromatin form large domains containing KRAB-ZNF genes. Genome remodelling complexes, while the QPGY repeat domain Res 16: 1493–1504. forms homo-oligomers. Oncogene 22: 8156–8167. Wu HK, Heng HH, Siderovski DP, Dong WF, Okuno Y, Shi Raetzman LT, Ward R, Camper SA. (2002). Lhx4 and Prop1 XM et al. (1996). Identiﬁcation of a human LIM-Hox gene, are required for cell survival and expansion of the pituitary hLH-2, aberrantly expressed in chronic myelogenous leu- primordia. Development 129: 4229–4239. kaemia and located on 9q33–34.1. Oncogene 12: 1205–1212. Richter K, Pinto do OP, Hagglund AC, Wahlin A, Carlsson Wu HK, Minden MD. (1997). Transcriptional activation of L. (2003). Lhx2 expression in hematopoietic progenitor/stem human LIM-HOX gene, hLH-2, in chronic myelogenous cells in vivo causes a chronic myeloproliferative disorder and leukemia is due to a cis-acting effect of Bcr-Abl. Biochem altered globin expression. Haematologica 88: 1336–1347. Biophys Res Commun 233: 806–812. Roberson MS, Schoderbek WE, Tremml G, Maurer RA. Yamaguchi M, Yamamoto K, Miura O. (2003). Aberrant (1994). Activation of the glycoprotein hormone a-subunit expression of the LHX4 LIM-homeobox gene caused by promoter by a LIM-homeodomain transcription factor. Mol t(1;14)(q25;q32) in chronic myelogenous leukemia in biphe- Cell Biol 14: 2985–2993. notypic blast crisis. Genes Chromosomes Cancer 38: 269–273.

Oncogene