1999 Oxford University Press 585–591Human Molecular Genetics, 1999, Vol. 8, No. 4

Functional domains of the SYT and SYT–SSXsynovial sarcoma translocation proteins andco-localization with the SNF protein BRM in thenucleusC. Thaete+, D. Brett +, P. Monaghan , S. Whitehouse , G. Rennie , E. Rayner, C. S. Cooperand G. Goodwin*

Institute of Cancer Research, Molecular Carcinogenesis Section, The Haddow Laboratories, 15 Cotswold Road,Sutton, Surrey SM2 5NG, UK

Received September 8, 1998; Revised and Accepted January 21, 1999

The t(X;18)(p11.2;q11.2) chromosomal translocationcommonly found in synovial sarcomas fuses the SYTgene on chromosome 18 to either of two similar genes,SSX1 or SSX2, on the X chromosome. The SYT proteinappears to act as a transcriptional co-activator and theSSX proteins as co-repressors. Here we have investi-gated the functional domains of the proteins. The SYTprotein has a novel conserved 54 amino acid domainat the N-terminus of the protein (the SNH domain)which is found in proteins from a wide variety ofspecies, and a C-terminal domain, rich in glutamine,proline, glycine and tyrosine (the QPGY domain),which contains the transcriptional activator se-quences. Deletion of the SNH domain results in a moreactive transcriptional activator, suggesting that thisdomain acts as an inhibitor of the activation domain.The C-terminal SSX domain present in SYT–SSX trans-location protein contributes a transcriptional repres-sor domain to the protein. Thus, the fusion protein hastranscriptional activating and repressing domains. Wedemonstrate that the human homologue of theSNF2/Brahama protein BRM co-localizes with SYT andSYT–SSX in nuclear speckles, and also interacts withSYT and SYT–SSX proteins in vitro . This interactionmay provide an explanation of how the SYT proteinactivates gene transcription.

INTRODUCTION

Chromosomal translocations that result in oncogenic transform-ation frequently involve the fusion of two genes to create a hybridgene that expresses a novel chimeric protein. In many cases, thisprotein is an aberrant transcription factor. In the case of synovialsarcomas, we have described the characterization of the genes

involved in the commonly observed t(X;18)(p11.2;q11.2) trans-location. This results in the fusion of the SYT gene onchromosome 18 to either of two closely related genes SSX1 andSSX2 on chromosome X. The resulting chimaeric genes expressSYT–SSX1 or SYT–SSX2 fusion proteins in which the C-terminal amino acids of SYT are replaced by amino acids fromthe C-terminus of the SSX proteins (1–3).

The normal SYT gene is expressed in a wide variety of cell typesduring embryogenesis (4) and in the adult, but the function of theprotein is not known. The protein is localized in the cell nucleusbut it has no recognizable nucleic acid-binding motifs. Althoughthe functional domains within SYT currently have not beencharacterized, the C-terminal two-thirds of the protein is rich inglutamine, proline, glycine and tyrosine residues, resembling thecomposition of a number of transcription activators. In agreementwith this, we have shown that the SYT protein acts as a potenttranscriptional activator (5).

The N-terminal half of both the normal SSX proteins (which isnot retained in the fusion proteins) has a domain related to theKRAB domain which has been described previously in a numberof transcription repressors containing zinc finger DNA-bindingdomains (6). We have shown in previous experiments (5) that theentire SSX proteins have transcriptional repressing activity whenassayed as GAL4 fusions, but the precise region responsible forthis activity has not been established. Since both the SYT andSSX proteins do not have recognizable DNA-binding domains,they presumably function to modulate gene transcription throughassociation with other nuclear proteins that recruit them to theirtarget promoters. In this study, we have carried out deletionanalyses of the SYT and SSX proteins to investigate the functionof the different domains in more detail.

Using green fluorescent protein (GFP) fusions, we havedemonstrated that the SYT, SSX and SYT–SSX proteins arenuclear proteins (5). Similar results have been obtained byimmunofluorescent studies (7). Whilst the SSX1 protein has auniform nuclear distribution, the SYT protein has a speckleddistribution in the cell nucleus, and this distribution is retainedwith the SYT–SSX2 fusion protein. The SYT speckles did not

*To whom correspondence should be addressed. Tel: +44 181 643 8901; Fax: +44 181 770 7290; Email: goodwin@icr.ac.uk+These authors contributed equally to this work

Dow

nloaded from https://academ

ic.oup.com/hm

g/article-abstract/8/4/585/2896768 by guest on 03 April 2019

Human Molecular Genetics, 1999, Vol. 8, No. 4586

co-localize with the promyelocitic leukaemia bodies or spliceo-somes. In this study, we have extended our analysis to examinewhether SYT co-localizes with other nuclear proteins known toexhibit heterogeneous or speckled distribution, including theprotein BRM (8), one of the human homologues of theDrosophila Brahma and the yeast SNF2 proteins (9–12). BRM isone of components of the multiprotein complexes called the SNFcomplexes (13,14). The SNF complexes function as transcrip-tional activators of genes that are repressed by chromatinstructural proteins such as histones and polycomb-group proteins(11). One model for their mode of action is that sequence-specificDNA-binding proteins [e.g. nuclear hormone receptors(9,10,15)] can recruit the SNF complex to target promoters and,once bound, the complex disrupts the nucleosomes in its vicinity,thereby allowing the binding of other transcription factorsrequired to initiate transcription. The mammalian SNF complexis heterogeneous and the complexes are composed of 9–12proteins, termed BAFs, many of which have now been cloned(13,14). It is likely that the SNF/BAF proteins interact with othernuclear proteins, either to target the SNF complexes to specificpromoters or, once bound to these promoters, to interact withother factors that regulate gene expression. As we show here, theSYT protein may be one of these proteins or may be a corecomponent of the SNF complex (i.e. a BAF).

RESULTS

Domain structure of the SYT protein

Examination of the SYT protein sequence reveals three regionswithin the protein that may have functional significance. Adatabase search revealed 11 protein sequences and translations ofexpressed sequence tag (EST) sequences with significant homo-logy with a domain near the N-terminus of SYT comprisingamino acids 20–73 (Fig. 1, the SNH domain). One of these iscontained in what appears to be a new human SYT familymember (SYT homologue 1), the homology with SYT spanningthe whole length of the two proteins (∼60% homology). A secondhuman homologue (SYT homologue 2) was also identified inthese searches. The SNH domain was identified in predictedprotein sequences from plants, nematodes and fish, as well asmammals, demonstrating that this region is conserved across abroad range of species. The function of this novel domain in theseproteins has not been described.

In the C-terminal half of SYT, between amino acids 187 and387, lies a transcriptional activation domain. The sequence iscomposed predominantly of glutamine, proline and glycine, withtyrosine residues occurring at variable intervals (Fig. 1b, theQPGY domain). The abundance of glutamine, glycine andtyrosine in this region is similar to that observed in the N-terminalactivating region of the EWS/FUS/TLS family of proteins, whichcontain (S/G)YQQ(S/Q) repeats (16). However, we could notidentify a specific repeat peptide around the tyrosine residues asfound in the EWS/FUS/TLS proteins. Nevertheless, the C-terminal tyrosines in the two human and the mouse SYThomologues (4) are totally conserved, as are most of theglutamines. The SNH and QPGY domains are separated by asequence of 114 amino acids which is highly conserved (96%) inthe mouse SYT and characterized by an unusually high contentof methionine (16%). Although this region is not closely relatedto sequences in other proteins, it appears to be functionallysignificant, as we describe below.

We have shown previously that when the N- and C-terminalhalves of SYT are fused to GAL4 and assayed for their ability toactivate a luciferase reporter containing a 5× upstream activatingsequence (UAS) upstream of the thymidine kinase (TK) promoter(consisting of TATA and CAAT boxes plus two Sp1-bindingsites), the C-terminal half has strong transactivating activitywhilst the N-terminal has little activating or repressing activity. Adeletion analysis of the C-terminal half of SYT revealed that theactivating activity was spread throughout this region of theprotein (data not shown). This was confirmed with a differentreporter promoter: when two halves of the SYT C-terminaldomain were assayed for their ability to activate a promoterconsisting of the adenovirus TATA box region plus 5×UASupstream of luciferase, both constructs exhibited transactivatingactivity (Fig. 2, constructs C and D). The results of Figure 2 alsoreveal that removal of the N-terminal 60 amino acids from thefull-length SYT protein, which removes most of the SNHdomain, results in a 30-fold increase in the ability of the proteinto activate the adenovirus promoter (construct A), stronglysuggesting that the SNH domain may have an inhibitory role. Afurther deletion, removing most of the methionine-containingregion, results in little further change in transactivating activity(construct B).

It is possible that the differences in activities of some of theconstructs are due to differences in protein stability of the GAL4fusions, but attempts to assess this using a number of commercialGAL4 antibodies were unsuccessful. Nevertheless, it is stillapparent that the two non-overlapping C-terminal constructs Cand D both have substantial activities, demonstrating that theactivation domain is spread throughout the C-terminal half of theprotein. It is also apparent that the N-terminal SNH domain maycontrol the activity of the protein either as an inhibitory sequenceor as an element that controls protein stability.

The SSX C-terminal domain in SYT–SSX hasrepressing activity

We have demonstrated previously that the SSX1 protein, whenfused to GAL4, acts as transcriptional repressor (5). We havefound that the C-terminal region of SSX that is involved in thetranslocation reproducibly represses the TK promoter 3-fold but,surprisingly, the N-terminal part of the protein containing theKRAB-like domain does not repress (Fig. 3). Thus, the SSXC-terminal addition to the SYT protein in the translocation maybe contributing a repression domain. This may explain why theSYT–SSX fusion always has 3- to 5-fold lower activity whencompared with full-length SYT (5). It would appear, therefore,that the SYT–SSX protein has both transcriptional activating andrepressing domains. The histone deacetylase inhibitor trichostatinA (TSA) has no effect on the repressing activity of either thefull-length SSX or the C-terminal SSX domain (data not shown),indicating that the repression is not mediated by histone ortranscription factor deacetylation.

SYT and SYT–SSX co-localize with BRM in the cellnucleus

We (5) and dos Santos et al. (7) have shown previously that SYTand SYT–SSX, when expressed as GFP fusion proteins ordetected by immunofluorescence, are localized in discretespeckles in the cell nucleus. In order to see whether any othertranscription factors that are known to localize to discrete regions

Dow

nloaded from https://academ

ic.oup.com/hm

g/article-abstract/8/4/585/2896768 by guest on 03 April 2019

587

Nucleic Acids Research, 1994, Vol. 22, No. 1Human Molecular Genetics, 1999, Vol. 8, No. 4587

Figure 1. Sequence of the SYT protein and homologues. (a) The SNH domain identified by homology to sequences in the databases: 1, Brassica rapa (EST 887272);2, Arabidopsis thaliana (gi22552866); 3, rice (EST 702013); 4, human SYT; 5, mouse SYT (4); 6, zebra fish (EST 2225596); 7, human SYT homologue 1 (gi 3327200);8, human SYT homologue 2 (AA86859); 9, rat (EST 2862851); 10, Brugia malayi (EST 1172331); 11, Caenorhabditis elegans (EST 2388279). EST homologies werefrom the EST database at NCBI and the full-length sequences were from GenBank (the accession numbers are GenBank IDs). (b) The human SYT sequence showingthe SNH domain and the QPGY domain that encompasses the region of the protein that has the high concentration of the four amino acids glutamine (27%), proline(19%), glycine (19%) and tyrosine (12%). The arrows represent the positions of translocations in synovial sarcoma.

in the nucleus might co-localize with SYT in these bodies, weinvestigated a number of possible candidates in co-transfectionexperiments, including the protein BRM which recently has beenshown to give a speckled distribution in the cell nucleus (8). Asshown in Figure 4, we found that the human homologue of theSNF2 protein, BRM, co-localizes with GFP–SYT and withGFP–SYT–SSX. When a construct expressing epitope-taggedBRM protein is co-transfected into cells and detected byimmunofluorescence, the BRM protein (red fluorescence) co-lo-calizes with the green fluorescence of the SYT and SYT–SSXproteins in these speckles (Fig. 4A and B). These results wereobtained using both the anti-haemagglutinin (HA) monoclonalantibody (Fig. 5) and anti-BRM polyclonal antibody (data notshown) in a number of cell types (SW480, kidney 293, Cos7 andHT1080). In control experiments, no red speckles are seen whencells transfected with the GFP–SYT–SSX alone are stained withthe anti-HA monoclonal antibodies (Fig. 4C). In addition, whenBRM is transfected into the cells with the GFP–SIP fusion

protein, a protein that we have found also to give nuclear speckles,the red speckles of BRM do not co-localize with the green SIPspeckles (Fig. 4D). The SIP protein is a nuclear protein that werecently have identified and shown to be similar to members ofthe EWS/TLS/FUS family (P. Antonson and G. Goodwin,unpublished data). Similarly, no co-localization was seen with theSSX protein (data not shown). Deletion of the N-terminal 60amino acids from the SYT protein, which removes most of theSNH domain, had little effect on the ability of the protein toco-localize with BRM in nuclear speckles (Fig. 5). Deletion of theN-terminal half of SYT resulted in diminished association suchthat <30% of the SYT speckles had BRM associated, but we wereable to detect some residual binding to some of the SYT speckles.These results suggest that BRM interacts principally with themethionine-containing central region of SYT, but there may alsobe a secondary binding site in the C-terminal half of the protein.

These results therefore show that the SYT co-localizes withBRM in the cell nucleus, and the addition of the SSX portion or

Dow

nloaded from https://academ

ic.oup.com/hm

g/article-abstract/8/4/585/2896768 by guest on 03 April 2019

Human Molecular Genetics, 1999, Vol. 8, No. 4588

Figure 2. Transcriptional activation of SYT deletions. Expression constructs of GAL4–SYT deletions of the full-length SYT (387 amino acids) and GAL4DNA-binding domain (GAL4 DBD) were transfected into SW480 cells together with the 5×UAS–adenovirus TATA box–firefly luciferase reporter and the pRL-SV40Renilla luciferase reporter. Firefly luciferase activities of the SYT constructs were corrected for transfection efficiency, and these values have been normalized betweendifferent experiments such that all constructs are compared with the activity of the full-length SYT which is given the value 100. (The absolute firefly luciferase activityof the full-length GAL4–SYT was 16–22-fold above the background obtained with GAL4 DBD.) All six constructs were assayed in duplicate in two experiments,and the standard deviations are given in parentheses.

Figure 3. Repression domain in the C-terminus of the SSX1 protein. TheC-terminal 78 amino acids (which are translocated to SYT in sarcomas), theremaining N-terminal domain and full-length SSX1 were expressed as GAL4fusions in SW480 cells co-transfected with the 5×UAS TK-luciferase or withthe TK-luciferase. Luciferase activities were divided by that obtained withGAL4 DBD alone to give the fold repression of the two reporters.

the removal of the inhibitory SNH domain, although altering itstranscriptional activating properties, does not alter the associationof SYT with BRM. No difference was detected between the SYTand the SYT–SSX fusion in their ability to bind BRM in thenuclear co-localization assay. However, it is important to pointout that in experiments that were carried out with HT1080 andNIH 3T3 cells which have a larger cytoplasm than SW480 cells,it was apparent that whilst the SYT–SSX and BRM proteins werelocalized solely in the nucleus, a variable fraction of the SYT(10–50%) was observed as speckles in the cytoplasm. Thus, anadditional role for the SSX C-terminus in the fusion protein maybe to alter the cellular distribution of the SYT by conferringanother nuclear localization signal. This is supported by theobservation that a GFP fusion containing just the C-terminaltranslocated domain of SSX localizes to the cell nucleus (data notshown).

SYT and BRM bind directly and specifically with oneanother in vitro

In vitro protein-binding experiments provided evidence that theco-localization of SYT and BRM may in fact be explained by thedirect interaction between the proteins. GST and GST–cBRMimmobilized on glutathione beads were purified from Escheri-chia coli and incubated with radioactive SYT, SYT–SSX, andluciferase as a control. The beads were washed and the boundprotein analysed by PAGE and detected with a phosphoimager(Fig. 6). The results show that both SYT and SYT–SSX bind toGST–cBRM and not to GST, and that luciferase binds to neither.Both proteins appeared to bind with similar affinity to the cBRM.Further evidence that the interaction is specific was provided bydemonstrating that SYT and SYT–SSX do not bind to beadscontaining the N-terminal activation domain of TFE3 fused toGST (data not shown), and the TFIID protein TAF55 does notbind to GST–cBRM beads (see below).

To map the interacting domains, three portions of the cBRMprotein, BRM-A, -B and -C, were expressed as GST fusions.BRM-A consists of amino acids 1–872, BRM-B amino acids873–1433 (most of the central ATPase domain and the Rb-bind-ing domain), and BRM-C, the C-terminal 135 amino acidscontaining part of the bromodomain. As can be seen (Fig. 7), thelabelled SYT–SSX protein binds strongly only to the N-terminalBRM-A, as compared with the labelled BRM-B and -C and theluciferase and TAF55 controls.

DISCUSSION

When considered together, the data presented here demonstratethat the SYT protein co-localizes with BRM, probably as aconsequence of direct interactions between the central region ofSYT and the N-terminal half of the BRM protein. BRM is one ofthe human homologues of the Drosophila Brahma and yeastSNF2 proteins (9,10), proteins that are components of themultiprotein complexes called the SNF complexes. The SNFcomplexes function as transcriptional activators of genes that arerepressed by chromatin structural proteins such as histones andpolycomb-group proteins (11). One model for their mode ofaction is that certain sequence-specific DNA-binding proteins(such as nuclear hormone receptors) that are able to bind to their

Dow

nloaded from https://academ

ic.oup.com/hm

g/article-abstract/8/4/585/2896768 by guest on 03 April 2019

589

Nucleic Acids Research, 1994, Vol. 22, No. 1Human Molecular Genetics, 1999, Vol. 8, No. 4589

Figure 4. (A) Co-localization of GFP–SYT–SSX (green) and HA-tagged BRM(red). (B) Co-localization of GFP–SYT (green) and HA-tagged BRM (red). (C)Experiment carried out as in (A), except that the cells were transfected withGFP–SYT–SSX constructs only. (D) Localization of GFP–SIP (green) andHA-tagged BRM (red). Transfections were into SW480 cells. For all studies,the left hand panels shows GFP fluorescence, the central panel shows the resultsabove with anti-HA monoclonal antibody and the right hand panel shows thecombined image. The white scale bar is 10 µm.

specific sequences even when complexed in nucleosomes canrecruit the SNF complex and, once bound, the complex disruptsthe nucleosomes in its vicinity, thereby allowing the binding of

other transcription factors required to initiate transcription. Analternative model has been put forward from studies in yeastwhich suggest that the SNF complex may be recruited topromoters through its association with the RNA polymeraseholoenzyme, and whether or not a particular promoter respondsto the SNF complex is dependent on the nucleoprotein structureof the promoter (17). Evidence that the first model is operative inhigher eukaryotes comes from studies that show that componentsof the SNF complex co-immunoprecipitate with the glucocorti-coid receptors (15) and the recent demonstration that the humantrithorax-related protein MLL, a DNA-binding protein, interactswith the hSNF5 protein, one of the components of the SNFcomplex (18). This is consistent with the findings in Drosophilathat the trithorax-group proteins, which include trithorax andBrahma, are required to maintain the pattern of active homeoboxgenes of the ANT-C and BX-C loci during embryonic develop-ment by counteracting the chromatin repressor polycomb (12).

The mammalian SNF complex is composed of 9–12 proteins,termed BAFs, many of which have now been cloned (13,14).Biochemical analysis has revealed that the complex is in factheterogeneous; the SNF complexes are present in multiple formswhich differ in their protein composition. For example, com-plexes differ in whether they contain BRG1 or BRM (the twohuman Brahma homologues) and which of the different BAF60proteins they contain. The functions of the proteins in the SNFcomplexes are not known other than that BRM and BRG1 haveDNA-dependent ATPase activities required for nucleosomedisruption, (11) that they interact with the transcriptionalrepressor Rb, co-operating to induce cell cycle arrest (19), andthat the hSNF5 protein binds MLL (18). It seems likely that theSNF proteins interact with other nuclear proteins, either to targetthe complex to specific promoters, or, once bound to thesepromoters, to interact with other factors that regulate geneexpression. The SYT protein may be one of these proteins or maybe a core component of the SNF complex (i.e. a BAF). Since mostof the mammalian BAFs have now been cloned and none has anysequence homology with SYT, the latter seems unlikely.

It is of interest to find that the SYT–SSX protein has atranscriptional activation domain in the C-terminal half and arepression domain contributed by the SSX C-terminus. After thismanuscript was submitted, Lim et al. (21) also reported that theC-terminus of SSX has repressing activity. The mechanismwhereby the SSX portion of the protein represses is not known.It does not appear to do so by binding histone deacetylases sincethe inhibitor TSA has no effect on the repressing activity offull-length SSX or the C-terminal SSX domain, and it is temptingto speculate that the SSX C-terminus counteracts the effect ofBRM by recruiting the polycomb complex to the SNF complex.The effect of the translocation in synovial sarcomas may be thealteration of an SNF complex containing the SYT protein from anactivating complex to an inactive complex or even a repressingcomplex, depending on the relative activating and repressingactivities of the SYT–SSX fusion in synovial sarcoma cells. Thus,genes that normally are activated by SNF complexes containingthe SYT protein may be down-regulated in synovial sarcoma cellsby SNF complexes containing the SYT–SSX protein. Alterna-tively, genes that normally are repressed by SSX proteincomplexes could be activated by the recruitment of the SNFcomplexes containing the SYT–SSX proteins.

An alternative model for the function of the SYT andSYT–SSX proteins comes from studies that show that BRM and

Dow

nloaded from https://academ

ic.oup.com/hm

g/article-abstract/8/4/585/2896768 by guest on 03 April 2019

Human Molecular Genetics, 1999, Vol. 8, No. 4590

Figure 5. Co-localization of BRM with N-terminal deletions of SYT. GFP fusions of SYT deletions were transfected with BRM into HT1080 cells and the extent oflocalization of the BRM with the SYT speckles observed. +++ indicates that >80% of SYT nuclear speckles are also stained with BRM, + indicates <30%co-localization.

Figure 6. Binding of SYT and SYT–SSX to chicken BRM in vitro. LabelledSYT, SSX and luciferase were incubated with GST or GST–cBRM beads andthe bound proteins analysed by SDS–PAGE and phosphoimaging. Lanes 1–3show the input proteins and lanes 4–9 the proteins bound to the beads.

BRG1, as well as interacting with components of the SNFcomplex, are known to interact with the tumour repressor Rb toinduce cell growth arrest (19) and with the nuclear hormonereceptors to augment their transcriptional activities (9,10,15). Itis possible, therefore, that the SYT and SYT–SSX proteinsfunction by modulating the activities of these complexes or theassociation of these proteins with one another.

Analysis of the repression activities of the SSX1 constructssuprisingly showed that the N-terminal construct containing theKRAB domain had no repressing activity. Nevertheless thefull-length SSX protein had higher repressing activity than theC-terminal translocated domain, and this suggests that the KRABdomain may augment the activity of the C-terminal domain eventhough it has no intrinsic activity of its own.

The function of the SNH domain within SYT may be to controlthe activity of the C-terminal activation domain. Deleting most ofthe SNH did not affect nuclear localization nor its associationwith BRM, but did result in a more potent transactivator. TheN-terminal half of SYT, when assayed as a separate fragmentfused to GAL4, does not exhibit repressing activity. Thusenhanced transcriptional activity of the SYT N-terminal deletionsmay be explained by postulating that the SNH domain interactswith and masks the C-terminal activation domain. Alternatively,the SNH domain could control turnover of the protein, so thatremoval of the domain results in a more stable protein; this wasnot obvious in the GFP fusions but remains to be analysed morecarefully.

MATERIALS AND METHODS

Transient transfections of GAL4–SYT fusion constructs togetherwith transcription reporter constructs into SW480 cells were carriedout using SuperFect transfection reagent (Qiagen, Crawley, UK)using the manufacturer’s methods. All constructs were assayed induplicate in each transfection experiment, normalized for transfec-tion efficiency, and all experiments were repeated at least twice.Typically, Gal4–SYT construct (0.4 µg), pG5luc firefly luciferasereporter (1.0 µg) and pRL-SV40 Renilla luciferase (250 ng) weretransfected into the cells grown in 6-well plates. After 48 h, the cellswere harvested and the extracts were analysed for firefly luciferaseand Renilla luciferase using the dual reporter assay system.GAL4–SYT constructs are plasmids expressing fragments of SYTfused to the DNA-binding domain of GAL4 (5). pG5luc (Promega,Southampton, UK) contains the adenovirus major late promoterTATA box and 5×UAS upstream of the firefly luciferase gene.pRL-SV40 contains the Renilla luciferase gene downstream of theSV40 enhancer–promoter and is used as a measure of transfectionefficiency to normalize the firefly luciferase values.

GAL4–SSX constructs were assayed for repression as de-scribed previously (5) using luciferase reporters driven by the TKpromoter with and without 5×UAS.

Confocal microscopy was carried out as follows. SW480 cellswere transfected with GFP fusion or HA-tagged BRM expressionconstructs using lipofectamine as described previously (5). Thecells were fixed with 4% paraformaldehyde in phosphate-buffered saline (PBS) for 30 min. After a PBS wash, the cells werepermeabilized for 10 min in 0.5% Triton X-100 in PBS followedby a PBS wash. HA-BRM was detected with anti-HA monoclonalantibody diluted 1:40 or 1:200 in Leibovitz L15 mediumcontaining 5% fetal calf serum (L15/FCS). After PBS washes (3× 5 min), bound antibody was detected using anti-mouse IgGconjugated to Alexa568 (Molecular Probes, Cambridge Bio-science, Cambridge, UK) diluted 1:200 in L15/FCS. Coverslipswere mounted after washing in Vectashield (Vector Laboratories,Peterborough, UK) and sealed with varnish. Slides were imagedwith a Leica TCS SP confocal microscope with Ar and Kr lasers.GFP-labelled proteins and HA-BRM (Alexa 568) were imagedconcurrently. No image processing was used. Controls werecarried out to ensure there was no bleed-through between green(GFP) and red (Alexa568) signals. In addition, cells transfectedwith GFP-labelled constructs alone were labelled using the HAantibody and Alexa 568 to check that the HA antibody did notdetect transfected GFP–SYT.

Protein binding experiments were carried out by incubatingGST, GST–cBRM (20) or GST–TFE3–N-terminal activationdomain immobilized on glutathione beads, with in vitro [35S]me-thionine-labelled SYT, SYT–SSX, TAF55 or luciferase in 100mM NaCl, 10 mM Tris–HCl pH 7.5, 5 mM MgCl2, 0.1% Triton

Dow

nloaded from https://academ

ic.oup.com/hm

g/article-abstract/8/4/585/2896768 by guest on 03 April 2019

591

Nucleic Acids Research, 1994, Vol. 22, No. 1Human Molecular Genetics, 1999, Vol. 8, No. 4591

Figure 7. SYT–SSX binds to the N-terminal half of cBRM. Labelled SYT–SSX and controls luciferase and TAF55 were incubated with GST or GST fused to cBRM orto three fragments of cBRM (A, B and C). Bound proteins were analysed by SDS–PAGE and phosphoimaging. Lanes 1–3, labelled input proteins. Proteins bound to GST(lanes 4–6), GST–BRM-A fragment (lanes 7–9), GST–BRM-B fragment (lanes 10–12), GST–BRM-C fragment (lanes 13–15) and full-length GST–BRM (lanes 16–18).

X-100, 0.1% bovine serum albumin (BSA) for 1 h, washed fourtimes with the same buffer without BSA and bound proteinseluted with SDS sample solvent. The eluted proteins wereanalysed by electrophoresis and detected using a phosphoimager.

ACKNOWLEDGEMENTS

We would like to thank M. Yaniv and C. Muchardt for supplyingus with the vector expressing HA-tagged BRM and the anti-BRMantibodies. This work was supported by the Cancer ResearchCampaign

REFERENCES

1. Clark, J., Rocques, P.J., Crew, A.J., Gill, S., Shipley, J., Chan, A.M.,Gusterson, B.A. and Cooper, C.S. (1994) Identification of novel genes, SYTand SSX, involved in the t(X;18)(p11.2;q11.2) translocation found in humansynovial sarcoma. Nature Genet., 7, 502–508.

2. Crew, A.J., Clark, J., Fisher, C., Gill, S., Grimer, R., Chand, A., Shipley, J.,Gusterson, B.A. and Cooper, C.S. (1995) Fusion of SYT to two genes, SSX1and SSX2, encoding proteins with homology to the Kruppel-associated boxin human synovial sarcoma. EMBO J., 14, 2333–2340.

3. De Leeuw, B., Balemans, M., Weghuis, D.O. and van Kessel, A.G. (1995)Identification of 2 alternative fusion genes, SYT–SSX1 and SYT–SSX2, int(X;18)(p11.2;q11.2)-positive synovial sarcomas. Hum. Mol. Genet., 4,1097–1099.

4. de Bruijn, D.R.H., Baats, E., Zechner, U., De Leeuw, B., Balemans, M., OldeWeghuis, D., Hirning-Folz, U. and van Kessel, A. (1996) Isolation andcharacterization of the mouse homolog of SYT, a gene implicated in thedevelopment of human synovial sarcoma. Oncogene, 13, 643–648.

5. Brett, D., Whitehouse, S., Antonson, P., Shipley, J., Cooper, C. and Goodwin,G. (1997) The SYT protein involved in the t(X;18) synovial sarcomatranslocation is a transcriptional activator localised in nuclear bodies. Hum.Mol. Genet., 6, 1559–1564.

6. Witzgall, R., O’Leary, E., Leaf, A., Onaldi, D. and Bonventre, J.V. (1994) TheKruppel-associated box-A (KRAB-A) domain of zinc finger proteinsmediates transcriptional repression. Proc. Natl Acad. Sci. USA, 91,4514–4518.

7. dos Santos, N.R., de Bruijn, D.R., Balemans, M., Janssen, B., Gartner, F.,Lopes, J.M., de Leeuw, B. and Geurts van Kessel, A. (1997) Nuclearlocalization of SYT, SSX and the synovial sarcoma-associated SYT–SSXfusion proteins. Hum. Mol. Genet., 6, 1549–1558.

8. Reyes, J.C., Muchardt, C. and Yaniv, M. (1997) Components of the humanSWI/SNF complex are enriched in active chromatin and are associated withthe nuclear matrix. J. Cell Biol., 137, 263–274.

9. Muchardt, C. and Yaniv, M. (1993) A human homologue of S.cerevisiaeSNF2/SWI2 and Drosophila brm genes potentiates the transcriptionalactivation by the glucocorticoid receptor. EMBO J., 12, 4279–4290.

10. Chiba, H., Muramatsu, M., Nomoto, A. and Kato, H. (1994) Two humanhomologues of Saccharomyces cerevisiae SWI2/SNF2 and Drosophilabrahma are transcriptional coactivators cooperating with the estrogenreceptor and the retinoic acid receptor. Nucleic Acids Res., 22, 1815–1820.

11. Peterson, C.L. and Tamkun, J.W. (1995) The SWI–SNF complex: achromatin remodeling machine? Trends Biochem. Sci., 20, 143–146.

12. Tamkun, J.W., Deuring, R., Scott, M.P., Kissinger, M., Pattatucci, A.M.,Kaufman, T.C. and Kennison, J.A. (1992) Brahma: a regulator of Drosophilahomeotic genes structurally related to the yeast transcriptional activatorSNF2/SW12. Cell, 68, 561–571.

13. Wang, W., Cote, J., Xue, Y., Zhou, S., Khavari, P.A., Biggar, S.R., Muchardt,C., Kalpana, G.V., Goff, S.P., Yaniv, M., Workman, J.L. and Crabtree, G.R.(1996) Purification and biochemical heterogeneity of the mammalianSWI–SNF complex. EMBO J., 15, 5370–5382.

14. Wang, W., Xue, Y., Zhou, S., Kuo, A., Cairns, B.R. and Crabtree, G.R. (1996)Diversity and specialization of mammalian SWI/SNF complexes. GenesDev., 10, 2117–2130.

15. Fryer, C.J. and Archer, T.K. (1998) Chromatin remodelling by the glucocorti-coid receptor requires the BRG1 complex. Nature, 393, 88–91.

16. Delattre, O., Zucman, J., Plougastel, B., Desmaze, C., Melot, T., Peter, M.,Kovar, H., Joubert, I., de Jong, P., Rouleau, G. and Thomas, G. (1992) Genefusion with an ETS DNA-binding domain caused by chromosome transloca-tion in human tumours. Nature, 359, 162–165.

17. Ryan, M.P., Jones, R. and Morse, R.H. (1998) SWI–SNF complexparticipation in transcriptional activation at a step subsequent to activatorbinding. Mol. Cell. Biol., 18, 1774–1782.

18. Rozenblatt-Rosen, O., Rozovskaia, T., Burakov, D., Sedkov, Y., Tillib, S.,Blechman, J., Nakamura, T., Croce, C.M., Mazo, A. and Canaani, E. (1998)The C-terminal SET domains of ALL-1 and TRITHORAX interact with theINI1 and SNR1 proteins, components of the SWI/SNF complex. Proc. NatlAcad. Sci. USA, 95, 4152–4157.

19. Strober, B.E., Dunaief, J.L., Guha, S. and Goff, S.P. (1996) Functionalinteractions between the hBRM/hBRG1 transcriptional activators and thepRB family of proteins. Mol. Cell. Biol., 16, 1576–1583.

20. Goodwin, G.H. (1997) Isolation of cDNAs encoding chicken homologues ofthe yeast SNF2 and Drosophila Brahma proteins. Gene, 184, 27–32.

21. Lim, F.L., Soulez, M., Koczan, D., Thiesen, H.J. and Knight, J.C. (1998) AKRAB-related domain and a novel transcription repression domain inproteins encoded by SSX genes that are disrupted in human sarcomasOncogene, 17, 2013–2018.

Dow

nloaded from https://academ

ic.oup.com/hm

g/article-abstract/8/4/585/2896768 by guest on 03 April 2019