Main

The Ewing family of tumors including Ewing sarcoma and primitive neuroectodermal tumor (PNET) arise in bone or soft tissue of children and adolescents and have a high propensity for recurrence and distant metastases.1, 2, 3 Ewing sarcoma/PNETs are characterized by chromosomal translocations4, 5, 6, 7, 8, 9 that fuse the EWSR1 gene (22q12) to a subset of ETS-transcription factor gene family members, most commonly FLI1 (11q24)10 and less frequently ERG (21q22),11 ETV1 (7p22),12 E1A-F (17q21),13, 14 or FEV (2q35–36).15 The ETS gene family members are defined by an approximately 85 amino acid (AA) long DNA-binding domain that binds to a core GGAA/T nucleotide sequence and further specificity in binding is defined by a flanking DNA core motif.16, 17, 18 EWSR1 is an ubiquitously expressed protein with an RNA-binding domain within its C-terminal region and a strong transactivation domain at the N-terminal.19, 20, 21 The transactivation domain exerts its activity when juxtaposed to a DNA-binding domain. The fusion of the EWSR1 N-terminal region to the C-terminus of an ETS protein results in an aberrant transcription factor that is presumed to be the initiating oncogenic event in an Ewing sarcoma/PNET.

Although most EWSR1 rearrangements in Ewing sarcoma result in fusion with an ETS-transcription factor gene family member, there are also rare reports of EWSR1 fusions with genes encoding a member of the zinc-finger family of proteins.22, 23 In addition, a fusion between EWSR1 and a gene from a transcription factor family other than ETS, the NFATc2 gene (encodes for a member of the NFAT-transcription factor family), has recently been described.24

In this study, cytogenetic analysis of an extraskeletal Ewing sarcoma/PNET arising in the lumbosacral region of a 5-year-old female revealed a t(4;22)(q31;q12) resulting in a novel fusion between EWSR1 and a member of the WSTF-SNF2h chromatin-remodeling complex family of genes, SMARCA5.

Materials and methods

Clinical

The patient was a 5-year-old female who presented with a 5-week history of low-back pain and increasing right lower extremity weakness. MRI studies with and without contrast revealed a large enhancing mass of the lumbosacral spinal canal (L4-5 through S2-3) with evidence of cortical destruction of the posterior sacral spine and anterior extension to the colonic wall at the rectosigmoid junction.

Following admission for pain management, the patient subsequently underwent an L4 through midsacral laminoplasty and excision of the epidural tumor. Intraoperatively, it was acknowledged that not all of the tumor could be removed (particularly anteriorly and laterally on the right side). Grossly, the excised specimen was composed of red to tan fragments of bone and soft tissue measuring 6.0 × 5.5 × 4.6 cm in aggregate.

Cytogenetic Analysis

Cytogenetic analysis was performed on a representative sample of the excised neoplasm using standard tissue culture and harvesting procedures. Briefly, the tissue was disaggregated mechanically and enzymatically, then cultured in RPMI 1640 media supplemented with 20% fetal bovine serum and 1% penicillin/streptomycin-L-glutamine (Irvine Scientific, Santa Ana, CA, USA) for 3–8 days. Two to four hours before harvest, cells were exposed to colcemid (0.02 μg/ml). Following hypotonic treatment (0.074 M KCl for 30 min for flasks and 0.8% Na citrate for 25 min for coverslips), the preparations were fixed three times with methanol:glacial acetic acid (3:1). Metaphase cells were banded with Giemsa trypsin, and the karyotypes were described according to the International System for Human Cytogenetic Nomenclature (ISCN, 2009).25

Cosmid and Bacterial Artificial Chromosome Probes

For analysis of the EWSR1 (22q12) and FLI1 (11q24) gene loci, cosmid probes26 (obtained from O Delattre, Institut Curie, Paris, France) flanking these two gene loci were selected in combination with an α-satellite probe for the centromeric region of chromosome 4 (Oncor, Gaithersburg, MD, USA). In an effort to further define the chromosomal breakpoint on chromosome 4, the following 4q31-specific bacterial artificial chromosome (BAC) clones were identified from the NCBI Map Viewer (http://www.ncbi.nlm.nih.gov/mapview) and obtained from BAC/PAC Resources Center (Children's Hospital Oakland Research Institute, Oakland, CA, USA): RP11-83A24, RP11-308D13, RP11-739C17, RP11-54P19, RP11-481K16, RP11-578N3, RP11-269F11, RP11-318C13, RP11-269F11, and RP11-557J10. In addition, RP11-222M10, a BAC clone spanning the EWSR1 locus, was also used.

Fluorescence In Situ Hybridization

Bicolor fluorescence in situ hybridization (FISH) studies were performed on 4;22 translocation-positive metaphase cells and/or cytologic touch preparations of the tumor tissue. Cosmid and BAC probes were directly labeled by nick translation with either Spectrum Green or Spectrum Orange-dUTP utilizing a modified protocol of the manufacturer (Vysis, Downers Grove, IL, USA). Hybridization was conducted as previously described.27 Hybridization signals were assessed in five metaphase cells or 200 interphase nuclei with strong, well-delineated signals by two different individuals. As a negative control, FISH studies were simultaneously conducted on karyotypically normal peripheral-blood lymphocytes. Images were acquired using the Cytovision Image Analysis System (Applied Imaging, Santa Clara, CA, USA).

Identification of EWSR1–SMARCA5 Fusion Transcript

Total RNA was isolated using the RNeasy Mini kit (Qiagen, Valencia, CA, USA). The 3′ RACE (3′-rapid amplification of cDNA ends) was performed using the SMART-RACE cDNA amplification kit and protocol (Clontech, Palo Alto, CA, USA). Briefly, first-strand cDNA was reverse transcribed from 0.3 μg total RNA using Superscript II and the 3′-RACE cDNA synthesis primer (5′-CDS) from the kit. For construction of 5′-RACE-Ready cDNA, 0.3 μg of the RNA sample was mixed with 5′-CDS primer A and SMART II A oligo, and reverse transcribed as recommended by the supplier. An aliquot of each first-strand cDNA reaction was then amplified using an EWSR1 or SMARCA5 gene-specific forward or reverse primer (Table 1) and a universal primer mix (SMART-RACE kit). A total of 5 μl of 100-fold dilution of the first PCR products was then reamplified using AmpliTaq in a second PCR reaction with heminested primers consisting of the universal anchor primer (SMART-RACE kit) as the reverse primer and EWSR1-2 (internal to EWSR1-1; Table 1) as the forward primer. Second-round PCR products were electrophoresed, purified, and subcloned into pCR4/TOPO vector (Invitrogen). Colonies with recombinant plasmids containing the RACE-PCR products were cleaved with EcoRI restriction enzyme and analyzed by gel electrophoresis to estimate the size of the insert. Plasmid clones from the 5′ and 3′ RACE reactions with the longest inserts were selected for sequencing. Two plasmids, EWSR1–RACE.52 and SMARCA5–RACE.33 overlapped with EWSR1–SMARCA5 (421–1793) and extended the nucleotide sequences into 5′ and 3′ direction yielding a cDNA contig of 3820 nucleotides.

Table 1 Oligonucleotide primers used in this study
Full size table

Reverse Transcription–Polymerase Chain Reaction for EWSR1–SMARCA5

RT–PCR analysis was performed using the Advantage one-step RT–PCR kit (Clontech) with sense EWSR1 and antisense SMARCA5 primers (Table 1) in the following combination; EWSR1-421/SMARCA5-1973, EWSR1-615/SMARCA5-1450, and EWSR1-421/SMARCA5-1954. The PCR thermal cycling protocol was performed according to the manufacturer's instructions with an annealing temperature of 58°C. The RT–PCR products of all reactions were subcloned and sequenced.

Construction of Full-Length EWSR1–SMARCA5 cDNA

The EWSR1–SMARCA5 (421–1793) RT–PCR product in the PCR4/TOPO vector was cut with BamHI and MluI restriction enzymes, ligated to an NotI/BamHI fragment of EWSR1–RACE.52 and an MluI/BsrGI fragment of SMARCA5–RACE.33, and introduced into the EagI/BsrGI sites of LITMUS38i vector (New England Biolabs, Ipswich, MA, USA).

Generation of Retroviral Vectors and Expression in NIH3T3 Cells

The EWSR1–SMARCA5 chimeric cDNA in the LITMUS38i vector was amplified by PCR to add the epitope tag FLAG and the Kozak consensus translation initiation sequences into the N-terminal region. The primers used in this construction step were 5′-CCACCATGGATTACAAGGATGACGACGATAAGGCGTCCACGGATTACA GTACC-3′ and 5′-TCATAGTTTCAGCTTCTTTTTTCTTCCTCGACCATCAGGTGCGCC-3′. The amplified fragments were then cloned into pCR4/TOPO plasmid (Invitrogen) and sequenced multiple times. Subsequently in another PCR amplification, the EagI site was added to the 5′ and the XhoI site was added to the 3′ end of the EWSR1–SMARCA5 fusion cDNA. The sequence-confirmed, FLAG-tagged EWSR1–SMARCA5 was digested with EagI and XhoI and then transferred into the mouse retroviral vector MIEG328 (obtained from DA Williams, Children's Hospital Medical Center, Cincinnati, USA). MIEG3 is a bicistronic murine stem cell virus-based retroviral vector, containing an encephalomyocarditis virus IRES element (internal ribosome entry site) immediately preceding the gene encoding eGFP (enhanced green fluorescent protein). The co-expression of eGFP with the gene of interest enables detection of infected cells by flow cytometry or fluorescent microscopy. Generation of retroviral supernatants from Phoenix Ampho (ATCC product# SD 3443) packaging cells was achieved as previously described.29 The retroviral vector was transfected into the Phoenix-Ampho packaging cell line by calcium phosphate, and the viral supernatants were collected 48 h after transfection by centrifugation.30 The supernatants, containing the EWSR1–SMARCA5-MIEG3 retrovirus particles, were used to infect NIH3T3 cells. NIH3T3 cells were infected twice with the MIEG3 retrovirus containing the fusion gene, wild-type EWSR1, SMARCA5, and the various deletion mutants of EWSR1–SMARCA5. GFP-positive cells were sorted using fluorescence-activated cell sorting (FACS), as described previously.29

Construction of EWSR1–SMARCA5 Deletion Mutants

EW-SMd270-351 deletion mutants in pCR4–TOPO were created to delete the SNF2_N region utilizing the QuickChange kit (Stratagene, La Jolla, CA, USA) with primers 5′-CTACGGGCAGCAGAATGTAAAATGGGGTAAA-3′ and 5′-CCCAAGAAGAGAAATTGTTTGAAGAGTCTTTCCTAGGCCC-3′ in the first step, and for the second step, primers 5′-GGGCCTAGGAAAGACTCTTCAAACAA TTTCTCTTCTTGGG-3′ and 5′-ATAGGAGATAAAGAACAAAGAGCTGCTTTTG TCAGAGACGTTTTATTACCGGGAGAATGGG-3′. Deletion mutants EW-SM1-870 and EW-SM1-989 were prepared by PCR using the following primer pairs: 5′-CCACCATGGATTACAAGGATGACGACG-3′ and 5′-TGGAG GAAAGAACTGGAAATCCTGAACATTGG-3′, and 5′-CCACCATGGATTACAA GGATGACGACG-3′ and 5′-AATATCATCACGACCCCACTTCTC ATTAGC-3′, respectively.

Sequence Analysis

Plasmid clones were sequenced using ABI PRISM® BigDye™ Terminator Cycle Sequencing kit version 2 and ABI PRISM® 3730 DNA Analyzer, a capillary electrophoresis system (ABI). Sequence analysis was performed using the MacVector with Assembler Version 10.4.11 (MacVector, Inc, Cary, NC, USA) sequence analysis program.

Immunoprecipitation and Western Blotting

Forty-eight hours after infection, the transfected NIH3T3 cells were harvested. Cytoplasmic and nuclear extracts were prepared using NE-PER® reagents (Pierce Biotechnology, Inc, Rockford, IL, USA) according to the manufacturer's instructions. Precleared nuclear and cytoplasmic extracts were immunoprecipitated using EWSR1 and hSNF2H antibodies. Antibody/protein complexes were collected by Protein-A Sepharose, and subsequently subjected to SDS–polyacrylamide gel electrophoresis and transferred to Immobilon-P polyvinylidine difluoride membranes (Millipore, Billerica, MA, USA). Membranes were blocked with 3% bovine serum albumin, incubated with M2 anti-FLAG antibody (Sigma, St Louis, MO, USA), washed and treated by secondary antibodies conjugated with peroxidase, and visualized by enhanced chemiluminescence (Pierce Biotechnology, Inc). For immunoblotting, transduced cells were lysed in M-PER (Pierce Biotechnology, Inc). Protein concentrations were assessed utilizing the BCA Protein Assay kit (Pierce Biotechnology, Inc) and 25 μg of total protein lysates/lane were run on denaturing SDS–polyacrylamide gels, transferred to nitrocellulose membranes. Membranes were blocked with 3% bovine serum albumin, incubated with the indicated primary antibodies and horseradish peroxidase-conjugated secondary antibodies, and developed using enhanced chemiluminescence (Pierce Biotechnology, Inc).

Primary antibodies used in these studies were goat polyclonal anti-hSNF2H (sc-8760, C-16), goat polyclonal anti-EWSR1 (sc-6533, N-18), or goat polyclonal anti-human EAT-2 (sc-21572), M2 anti-FLAG antibody (Sigma), and horseradish peroxidase-conjugated secondary antibodies obtained from Santa Cruz Biotechnology (Santa Cruz, CA, USA).

Soft Agar Assay

Transformation of NIH3T3 cells by EWSR1, SMARCA5, EWSR1–SMARCA5, and EWSR1–SMARCA5 deletion variants was examined using a colony formation in soft agar culture assay. SeaKem GTG agar in H2O was prepared, autoclaved, and mixed 2 × DMEM with 20% FBS to reach a final concentration of 0.9% and allocated at 1 ml/35 mm dish. A second agar containing 0.34% agar was prepared in a similar process, but also included NIH3T3-transformed cells at a final concentration of 2 × 104/ml. A total of 2 ml of this second agar mixture was layered on top of the base layer, and allowed to solidify for 30 min–1 h, before incubating at 37°C in 5% CO2. Plates were cultured for 3–4 weeks or until colonies were visible. The colonies were stained with 1 ml/well of 0.05% nitroblue tetrazolium in PBS and colonies larger than 100 mm were counted for each plate.

Results

Histology

Histopathologically, relatively solidly packed, uniformly small-sized neoplastic cells in a background of necrosis were identified (Figure 1a). Individual cells exhibited an increased nuclear to cytoplasmic ratio with hyperchromatic nuclei surrounded by smooth nuclear membranes and occasional clear to slightly vacuolated cytoplasms. Mitotic figures were readily identifiable. Immunohistochemical studies demonstrated that the neoplastic cells were reactive for vimentin, CD99 (membranous, Figure 1a inset), synaptophysin (focal), and neuron-specific enolase (focal). The neoplastic cells were negative for chromogranin, cytokeratin (AE1/3), muscle-specific actin, desmin, and leukocyte common antigen. The diagnosis of extraskeletal Ewing sarcoma/PNET was rendered.

Figure 1
figure 1

(a) Representative histology of the tumor is showing a relatively monotonous round cell population with occasional mitotic figures. The neoplastic cells exhibit membranous immunoreactivity for CD99 (inset). (b) Partial G-banded karyotype demonstrating the t(4;22)(q31;q12). (c) FISH analysis conducted on destained metaphase cells exhibiting the 4;22 translocation with an alphoid sequence probe specific for chromosome 4 and EWSR1 breakpoint flanking cosmid probes confirm the presence of an EWSR1 rearrangement with translocation of the green probe signal distal to EWSR1 on the der(22) (red arrow) to the der(4) (white arrow). The green and yellow arrows indicate the normal chromosome 4 and normal chromosome 22 homologs, respectively. (d) FISH analysis with the RP11-418K16 BAC clone spanning the proximal (centromeric) portion of the SMARCA5 locus in green and the RP11-222M10 BAC clone spanning the EWSR1 locus in red confirmed the presence of a fusion between these two clones on the der(4), yellow arrow. The smaller red signal represents the remainder of the EWSR1 spanning probe on the der(22) and the single green and larger red signals represent the normal chromosome 4 and 22 homologs, respectively.

Full size image

Cytogenetic and FISH Data

A t(4;22)(q31;q22) was observed as the sole anomaly in nine cells (Figure 1b). One cell demonstrated extra copies of chromosomes 4, 8, and 22, in addition to the 4;22 translocation. Another cell was karyotypically normal. The complete chromosomal complement is 46,XX,t(4;22)(q31;q11.2)[9]/49,idem,+4,+8,+22[1]/46,XX [1].

FISH analysis of metaphase cells exhibiting the 4;22 translocation with an alphoid sequence probe specific for chromosome 4 and EWSR1 breakpoint flanking cosmid probes confirmed the presence of a rearrangement of the EWSR1 locus with translocation of the probe signal distal to EWSR1 on the derivative chromosome 22 homolog [der(22)] to the derivative chromosome 4 homolog [der(4)] (Figure 1c). However, FISH analysis of these abnormal metaphase cells with cosmid probes proximal and distal to EWSR1 and FLI1, respectively, were negative for an EWSR1–FLI1 fusion, suggesting involvement of a gene other than FLI1 (data not shown).

FISH interphase cell positional cloning studies of the der(4) narrowed the breakpoint region to a single BAC clone, RP11-481K16. Bicolor FISH analysis using the RP11-481K16 and RP11-222M10 (spanning the EWSR1 locus) BACs confirmed the presence of a fusion between these two clones in 25% of the cells analyzed (Figure 1d). Two genes located within the region covered by the RP-11481K16 BAC clone include FREM and SMARCA5. Note, only a proximal portion (centromeric) of SMARCA5 is covered by RP-11481K16.

Characterization of EWSR1–SMARCA5 Transcript and Fusion Protein

Sequence analysis of the RACE products demonstrated an in-frame fusion of EWSR1 and SMARCA5. SMARCA5 maps to 4q3131 and like EWSR1 is oriented with its 3′ end telomeric. In contrast, the other gene overlapping with the RP-11481K16 BAC clone sequence, FREM, is in the opposite transcriptional orientation.

The presence of EWSR1–SMARCA5 fusion transcripts was subsequently confirmed by RT–PCR analysis (Figure 2). A reciprocal SMARCA5–EWSR1 fusion product was not identified. The constructed full-length EWSR1–SMARCA5 cDNA was found to result from the fusion of the first 7 exons of EWSR1 to the last 19 exons of SMARCA5 (Figure 3). This fusion transcript is 3986 nucleotides long and encodes a 1143 AA chimeric protein composed of the 264 NH2-terminal AAs of the transcriptional activation domain/regulatory domain of EWSR1 and the 878 COOH-terminal AAs of hSNF2H (corresponding to AA position of hSNF2H from 175 to 1052) plus a single AA (N, Asn) at the breakpoint. The EWSR1–hSNF2H fusion protein, in addition to the serine–tyrosine–glutamine–glycine-rich (SNYG) N-terminal domain of EWSR1, contains five conserved domains, SNF2_N, SrmB, Hand, Slide, and SANT of hSNF2H.31, 32

Figure 2
figure 2

Ethidium-bromide agarose gel electrophoresis of the reverse-transcription polymerase chain reaction (RT–PCR) products. Lanes M1 and M2 represent the 1.0-kb and 100-bp DNA molecular weight marker ladders, respectively. Lanes 1, 2, and 3 demonstrate the EWSR1–SMARCA5 fusion transcripts as detected by the gene-specific primer combinations indicated in the upper left hand corner (primer combinations also described in Table 1).

Full size image
Figure 3
figure 3

Schematic diagrams of EWSR1, SMARCA5, and EWSR1–SMARCA5 fusion gene (a) and fusion protein (b). The nucleotide and AA sequence of the fusion gene and its protein product around the breakpoint are illustrated in part (c).

Full size image

Western blot analyses conducted on the nuclear and cytosolic extracts of NIH3T3 cells transfected with wild-type EWSR1, SMARCA5, and EWSR1–SMARCA5 cDNAs in MIEG3 vector revealed that only the nuclear extracts exhibited bands of the expected size for all of the proteins (Figure 4). This finding suggests that EWSR1–hSNF2H can be synthesized in vitro and that the chimeric protein enters the nucleus with a similar efficiency to EWSR1 and hSNF2H.

Figure 4
figure 4

Immunoblot analyses of EWSR1, hSNF2H and EWSR1–hSNF2H proteins demonstrate the subcellular localization of EWSR1–hSNF2H fusion protein (cytoplasmic (C) and nuclear (N) extracts).

Full size image

Transforming Properties of EWSR1–hSNF2H

To assess for functional similarities or differences between EWSR1–FLI1 and EWSR1–hSNF2H, the ability for both fusions to modulate expression of a common target gene was investigated. EAT-2 (SH2 domain-containing 1B, SH2D1B) was selected because its expression is induced in EWSR1–FLI1, but not in EWSR1 or FLI1-transfected NIH3T3 cells.33

A comparison of NIH3T3 cells infected with MIEG3–EWSR1–FLI1 and MIEG3–EWSR1–SMARCA5 fusion constructs revealed comparable levels of EWSR1–FLI1 and EWSR1–hSNF2H in cell lysates (Figure 5a). However, in contrast to EWSR1–FLI1-transformed NIH3T3 cells, EAT-2 expression was not induced in the NIH3T3 cells by the EWSR1–SMARCA5 construct (Figure 5b), suggesting that EWSR1–FLI1 and EWSR1–hSNF2H may not regulate the same repertoire of target genes.

Figure 5
figure 5

Effect of the EWSR1–SMARCA5 on EAT-2 expression. (a) Immunoblot demonstrating equivalent levels of protein expression of the FLAG-tagged EWSR1–FLI1 and EWSR1–hSNF2H proteins in NIH3T3 cells. (b) Immunoblot analysis of EAT-2 in NIH3T3 cells expressing EWSR1–FLI1 or EWSR1–hSNF2H proteins.

Full size image

The failure of EWSR1–hSNF2H to induce EAT-2 expression implies that there might be functional differences between EWSR1–FLI1 and EWSR1–hSNF2H fusion proteins. To determine whether these differences might be explained by NIH3T3 cell transformation ability, NIH3T3 populations expressing either fusion were tested in liquid medium for anchorage-dependent growth. As illustrated in Figure 6, both EWSR1–FLI1 and EWSR1–hSNF2H support anchorage-independent growth of NIH3T3 cells.

Figure 6
figure 6

Cell morphology of NIH3T3 cells infected with MIEG3, MIEG3–EWSR1–FLI1, and MIEG3–EWSR1–SMARCA5 retroviruses. NIH3T3 cells infected by the MIEG3, MIEG3–EWSR1–FLI1, and MIEG3–EWSR1–SMARCA5 retroviruses were plated on six-well plates without prior FACS selection. Colonies shown are representative of the cell population.

Full size image

For assessment of the role of the various hSNF2H motifs present in the chimeric EWSR1–hSNF2H protein and for determination of which region(s) is/are essential for transformation, epitope-tagged deletion mutants of EWSR1–hSNF2H producing retrovirus were constructed in which three different motifs of the hSNF2H protein were deleted. The transforming properties of the mutated EWSR1–SMARCA5 constructs were evaluated by infecting NIH3T3 cells with either the EWSR1–SMARCA5 wild-type virus or one of the deletion mutated versions (Figure 7). Agar assays conducted on primary infectants revealed that NIH3T3 cells expressing any of the deleted fusion cDNA constructs displayed lowered colony forming activity or failed to form colonies in soft agar. Western blot analyses of the polyclonal primary infected cells of all three deleted constructs demonstrated the presence of the mutant protein (Figure 7). The expression level of the deletion mutants was equal to or greater than those seen in full-length EWSR1–SMARCA5-transformed clones. Therefore, the lack of transforming activity by the mutated constructs appears to be due to loss of biologic function and not from underexpression or instability of the deleted products.

Figure 7
figure 7

(a) Schematic structure of EWSR1–hSNF2H and the deletion mutants. (b) Immunoblot analyses of EWSR1–hSNF2H and the deletion mutants. (c) Soft agar colony assay of EWSR1–hSNF2H and the deletion mutants-expressing NIH3T3.

Full size image

Discussion

A remarkable diversity in the EWSR1 C-terminal fusion gene partners that provide essential DNA-binding capacity to the chimeric protein have been observed in Ewing sarcoma/PNET. FLI1, a member of the ETS family of transcription factors, dominates as the most common translocation gene partner in Ewing sarcoma/PNET, followed by ERG, and less commonly other ETS gene family members.34 In all Ewing sarcoma/PNETs with EWSR1–ETS fusions, the DNA-binding domain of the ETS factor is included, resulting in aberrant transcription factors that can regulate genes mediating the oncogenic phenotype of this neoplasm.33, 35 The EWSR1 protein contains an N-terminal serine–tyrosine–glutamine–glycine-rich region that when fused to a heterologous DNA-binding domain, potently stimulates gene transcription. The ETS factors bind purine-rich sequences with GGAA/T core consensus sequence, surrounded by sequences that contribute to the specificity of each EWSR1–ETS fusion variant.

In this study, the identification, cloning, and functional analysis of a novel fusion oncoprotein in an extraskeletal Ewing sarcoma/PNET is described. The t(4;22)(q31;q12) chromosomal translocation juxtaposes EWSR1 to SMARCA5. SMARCA5 encodes hSNF2H, a protein with remarkable similarity in AA sequence to ISWI, a key component of chromatin-remodeling factors in Drosophila.31, 32 The packaging of DNA into chromatin limits the accessibility of transcription factors to regulatory regions of genes. Eukaryotic cells contain two principal modules of chromatin-modifying activities: histone-modifying complexes and ATP-dependent chromatin-remodeling complexes.36 The protein, hSNF2H, is one of the components of various ATP-dependent chromatin-remodeling complexes37 including ACF/BAZ-like32, 38, 39 WICH,40 CHRAC,41 NoRC,42 RSF,43 and NuRD complexes.44 While there are some differences between complexes, all feature a conserved repositioning specificity in vivo and also the ability of hSNF2H to mediate DNA accessibility for the interacting DNA-binding factors by sliding the histone octamer. In vitro, hSNF2H is able to interact with DNA regardless of the presence of core histones.45

Divergent biological functions of the chromatin-remodeling complexes may largely arise from other properties conferred by complex-specific subunits. hSNF2H, as a component of the WICH complex, mobilizes nucleosomes, regulates the transcription of various genes, and has a function in RNA polymerase I and RNA polymerase III transcription.46

In RSF, one of the chromatin remodeling and spacing factor systems, hSNF2H interacts with Rsf-1 and promotes the formation of competent RNA polymerase II transcription initiation complexes on chromatin, as a result of its ability to mobilize nucleosomes.47 Recently, Rsf-1 has been linked to cancer-specific gene amplification in ovarian, breast, bladder, esophageal, and head and neck cancers.48 It has been suggested that the interaction between Rsf-1 and hSNF2H may represent a survival signal for tumors, which overexpress Rsf-1.

Translocations involving a chromatin-modifying gene have also been described in leukemia.49 Specifically, in t(9;11)(p22;q23)49 and t(11;19)(q23;p13.3)50 leukemias, MLL (mixed-lineage leukemia gene; 11q23) is fused to MLLT3 (AF9) and MLLT1 (ENL), respectively. MLLT3 and MLLT1 are highly homologous with each other and are also homologous to proteins associated with the SWI/SNF complex. MLLT1 encodes for one of the subunits of the SWI/SNF family of chromatin-remodeling complexes.51 The MLL–MLLT1 fusion protein is associated with the chromatin-remodeling complex and synergistically activates transcription of various genes. MLL–MLLT1 fusion protein recruits the chromatin-remodeling complex to genes, such as HoxA7, which are normally not controlled by MLL.52

Our report is the first to describe an EWSR1 fusion partner gene encoding for a chromatin-remodeling and spacing factor in Ewing sarcoma/PNET. The EWSR1–hSNF2N fusion protein, in addition to the serine–tyrosine–glutamine–glycine-rich (SNYG) N-terminal domain of EWSR1, contains five conserved domains and motifs, SNF2_N, SrmB, Hand, Sant, and Slide of hSNF2N. SNF2_N domain is found in proteins involved in a variety of processes including transcriptional regulation, DNA repair, DNA recombination, and chromatin unwinding. The SrmB domain is found in proteins with DNA and RNA helicase activity; it maintains one of the two ATP-binding sites of hSNF2H. The Hand domain confers DNA and nucleosome-binding properties to the protein. Sant and Slide have DNA-binding activity.

The EWSR1–hSNF2H fusion protein may function as part of a chromatin-remodeling complex. The EWSR1–hSNF2H chimeric protein could directly interact with DNA and regulate genes. The SNF2_N domain, maintained in the EWSR1–hSNF2H fusion protein, might serve as the ATPase component of the SNF2/SWI multi-subunit complex, and utilize energy derived from ATP hydrolysis to disrupt histone–DNA interactions, resulting in an increased accessibility of DNA to transcription factors and transcriptional regulation of genes not normally regulated by EWSR1. The SLIDE domain has a role in DNA binding, contacting DNA target sites similar to c-Myb. The SANT domain is found in regulatory transcriptional repressor complexes in which it also binds DNA. The observation of deletion of the SNF2-N domain in EWSR1–hSNF2H in addition to loss of SLIDE and SANT domains with DNA-binding capacity suggests that the hSNF2N portion of the chimeric oncoprotein functions as an aberrant transcriptional regulator.

In spite of their structural differences, both EWSR1–FLI1 and EWSR1–ETV1 fusion proteins induce the expression of SH2D1B (EAT-2), an EWSR1–FLI1 target gene.33 Considering that many ETS proteins can bind to the same DNA sites suggests that EWSR1–ETS fusions promote oncogenesis via similar biologic pathways. In contrast, the EWSR1–hSNF2H fusion protein does not induce SH2D1B (EAT-2) expression in transfected NIH3T3 cells, suggesting that it may pursue a different biological pathway than EWSR1–ETS fusions in order to promote a similar Ewing sarcoma/PNET phenotype. Notably though, EWSR1–hSNF2H-expressing NIH3T3 cells do exhibit anchorage-independent growth and form colonies in soft agar, findings demonstrative of tumorigenic potential.

In summary, although most Ewing sarcoma/PNET EWSR1 fusions have involved members of the ETS gene family, there are rare reports of EWSR1 fusions with zinc-finger and NFAT-transcription factor family members.22, 23, 24 To date, however, there have been no reports of a fusion between EWSR1 and a chromatin-remodeling gene. In this study, we cloned the fusion gene produced by a t(4;22)(q31;q12) in an extraskeletal Ewing sarcoma/PNET and demonstrated the tumorigenic activation of SMARCA5, a gene coding for a chromatin-reorganizing protein, by chimera formation with EWSR1. Future studies based on the structure–function relationship of the EWSR1–hSNF2H protein and functional analysis of the chimeric protein will provide insight into the mechanism of sarcomagenesis induced by this novel class of oncoproteins. Considering the structural and functional disparity between ETS genes and SMARCA5, the novel EWSR1–hSNF2H can serve as an excellent model system for examining alternative tumorigenic pathways in Ewing sarcoma/PNET.