Fig 2: Identification of proteins bound to recombinant FET-NTD baits The three normal RNA-binding FET proteins consist of N-terminal repetitive and structurally disordered domains (NTD: N-terminal domain), central RNA-binding domains (RRM: RNA recognition motifs), zinc finger domains (ZF), and degenerated repeat regions (RGG: RGG repeat regions). The GST-tagged FET-NTD baits shown were used for pulldown experiments and represent the shortest parts commonly present in the FET fusion oncoproteins. The FET-binding motif is a conserved sequence required for complex formation between the three FET proteins. Amino acid numbers are indicated.Schematic illustration of a representative FET fusion protein (FUS-DDIT3 type II) and its parental proteins. DBD: DNA-binding domain, LZ: leucine zipper domain. Type I and type II show locations of the two most common MLS fusion breakpoints in FUS.Coomassie staining of SDS–PAGE-separated pulldown samples with/without cell extracts (C.E) using GST-tagged FET-NTDs as baits. Sepharose with bound GST was included as control. Background from recombinant protein baits (shown by white arrowheads) as well as smaller partial recombinant products is visualized in the left panel (C.E -). Several high-molecular-weight proteins are retained by the FET-NTDs (C.E +). Black arrowheads indicate protein bands and gel parts analyzed by mass spectrometry (see also Table EV1).Immunoblot analysis (IB) of pulldown samples with FET-NTD baits with/without cell extracts (C.E). Antibodies against ARID1A, BRG1, and BAF60A were used for detection of SWI/SNF components. Input samples diluted 1:5 were included as a control. In situ proximity ligation assays (PLA) using antibodies against BRG1, ARID1A, DDIT3, and C-terminal parts of normal FUS show protein complexes containing FUS/BRG1, FUS/ARID1A, FUS-DDIT3/BRG1, and FUS-DDIT3/ARID1A as red fluorescent spots in nuclei of MLS cell lines 2645-94 and 402-91. C-terminal parts of FUS are not present in the FUS-DDIT3 fusion protein, and normal DDIT3 is not expressed in these cell lines. Merged images also include DAPI nuclear counterstain in blue. Combinations of primary antibodies are used to detect interactions (left panel). In control experiments (right panel), one primary antibody is omitted to evaluate the background fluorescent signals. Scale bars = 10 µm. Source data are available online for this figure.

Fig 3: Morphologic effects of siRNA-mediated knockdown of SMARCD1, SMARCD2 and SMARCD3 in LnCAP cells. Immunostaining of β–tubulin and nuclear pore complex (NPC) proteins was performed in LnCAP cells treated with (A) a scrambled control or siRNAs targeting (B) SMARCD1, (C) SMARCD2, (D) SMARCD3, (E) SMARCD1 and SMARCD2, (F) SMARCD1 and SMARCD3 and (G) SMARCD2 and SMARCD3. Arrows indicate binucleated cells.

Fig 4: mRNA and protein levels of SMARCD1, SMARCD2 and SMARCD3 in benign and malignant prostate cell lines. (A) Expression levels of SMARCD1, SMARCD2 and SMARCD3 in various prostate cell lines were measured by qPCR. RNA was isolated from several biological replicates (n = 3). Asterisks indicate the statistical significance of differences in gene expression between the respective cell line and non-malignant RWPE-1 cells. Error bars indicate the standard error of the mean (SEM). Standard deviations were calculated using the relative-fold expression determined in various technical replicates (n = 6). * p ≤ 0.05; ** p ≤ 0.01; *** p ≤ 0.001; **** p ≤ 0.0001. (B) SMARCD1, SMARCD2 and SMARCD3 protein levels in various prostate cell lines were determined by Western blotting.

Fig 5: Morphologic effects of siRNA-mediated knockdown of SMARCD1, SMARCD2 and SMARCD3 in RWPE-1 cells. Immunostaining of β–tubulin and nuclear pore complex (NPC) proteins was performed in RWPE-1 cells treated with (A) a scrambled control or siRNAs targeting (B) SMARCD1, (C) SMARCD2, (D) SMARCD3, (E) SMARCD1 and SMARCD2, (F) SMARCD1 and SMARCD3 and (G) SMARCD2 and SMARCD3. Arrows indicate binucleated cells.