Chromosomal rearrangements often result in active regulatory regions juxtaposed upstream of an oncogene to generate an expressed gene fusion. Repeated activation of a common downstream partner–with differing upstream regions across a patient cohort–suggests a conserved oncogenic role. Analysis of 9,638 patients across 32 solid tumor types revealed an annotated long noncoding RNA (lncRNA), Breast Cancer Anti-Estrogen Resistance 4 (BCAR4), was the most prevalent, uncharacterized, downstream gene fusion partner occurring in 11 cancers. Its oncogenic role was confirmed using multiple cell lines with endogenous BCAR4 gene fusions. Furthermore, overexpressing clinically prevalent BCAR4 gene fusions in untransformed cell lines was sufficient to induce an oncogenic phenotype. We show that the minimum common region to all gene fusions harbors an open reading frame that is necessary to drive proliferation.

Implications:

BCAR4 gene fusions represent an underappreciated class of gene fusions that may have biological and clinical implications across solid tumors.

Chromosomal rearrangements are common somatic aberrations in cancer genomes, often leading to the juxtaposition of two genes, creating gene fusions. In addition to their biological roles in oncogenesis, some gene fusions are clinically relevant diagnostic markers, prognostic indicators, and therapeutic targets (1). Advances in next-generation sequencing coupled with improved gene fusion detection tools have accelerated the discovery of novel gene fusions in solid tumors. E26 transformation-specific (ETS) transcription factor family members (2), ALK and ROS fusions in lung cancer (3), and RAF kinase gene fusions in melanoma, gastric, and prostate cancer (4) represent important gene fusion discoveries across cancers. These exemplify a biological pattern called “functional recurrence”: various 5′ regulatory regions join a particular 3′ oncogene across samples and cancer types. While a single gene fusion event may be infrequent and overlooked, functionally recurrent fusions activating the same downstream partner can accentuate highly relevant oncogenes and provide molecular insight into their function. Some gene fusions are low frequency (<5%) events in a single cancer (even with functionally recurrent events), but accumulation across solid tumors expands the potential biological and clinical relevance to a significantly broader patient population.

In this study, we applied INTEGRATE (5), our highly sensitive gene fusion discovery tool, in a pan-cancer strategy to detect functionally recurrent gene fusions. From our analysis of 9,638 patients across 32 different cancer types within The Cancer Genome Atlas (TCGA) consortium, we prioritized a novel class of Breast Cancer Anti-Estrogen Resistance 4 (BCAR4) gene fusions because: (i) they are the most prevalent gene fusions not previously characterized across solid tumors; (ii) full-length BCAR4 is implicated in cancer (6, 7); and (iii) all patients express a minimum common region of BCAR4 that contains an open reading frame (ORF), suggesting a conserved oncogenic role. Silencing the two most common fusions (LITAF-BCAR4 and ZC3H7A-BCAR4) decreased proliferation in cancer cell lines; conversely, overexpressing the fusions in benign models increased it. Mutating the ORF abated BCAR4 gene fusion-driven proliferation. Collectively, our pan-cancer analysis discovered a functionally recurrent class of BCAR4 gene fusions that act through a protein to alter cell-cycle and proliferation across solid tumors.

TCGA datasets

We downloaded the aligned BAM files at https://portal.gdc.cancer.gov/ for the cancer types included in the TCGA Pan-Cancer Analysis listed in Supplementary Table S1. When available, RNA sequencing (RNA-seq) BAM files for matched adjacent normal tissue were also downloaded. Following TCGA practices, COAD and READ were merged to form a colorectal cancer cohort.

Gene fusion discovery

INTEGRATE is an open-source gene fusion discovery tool designed to map gene fusions using aligned RNA-seq reads and whole genome sequencing (WGS) paired-end sequencing reads, if available (5). INTEGRATE version 0.2.6 was run using default parameters in “RNA only” mode on the aligned RNA-seq reads (8). Analysis was based on hg38. Functionally recurrent 3′ fusion partners were identified if a gene was only found as a 3′ partner in somatic gene fusions across patients. Summary figures of novel and previously reported functionally recurrent 3′ fusion partners were generated using R (version 4.0.5).

Cell culture

SNU308 and TUHR14TKB were selected for their endogenous expression of BCAR4 gene fusions with no expression of full length BCAR4. The TUHR14TKB cell line was purchased from RIKEN BRC Cell Engineering Division (RCB1383; RRID_CVCL_5953). SNU308 gallbladder cell line was purchased from the Korean Cell Line Bank (RRID:CVCL_5048). Cell lines were grown in RPMI-1640 (Genesee Scientific) supplemented with 10% FBS (Sigma) and 1% penicillin/streptomycin (Thermo Fisher Scientific). hTER1-HME1 were purchased from ATCC (ATCC, catalog no, CRL-10317, RRID:CVCL_0598). MCF10a were a gift from Dr. Ron Bose (Washington University; St. Louis, MO). Cell lines were cultured in MEGM Mammary Epithelial Cell Growth Medium BulletKit (Lonza) or following this protocol: https://brugge.med.harvard.edu/protocols. Cell line authentication and validation were performed by ATCC, RIKEN, and Korean Cell Line Bank. Cells were passaged less than 10 times and monitored for Mycoplasma by PCR.

Transfections and overexpression models

Cells were transfected with 10 nmol/L of custom silencer select siRNA (Supplementary Table S2) and Lipofectamine RNAiMax (Thermo Fisher Scientific) following the manufacturer's protocol and used for assays 72 hours later. Fusion constructs were synthesized by Invitrogen and Infusion cloned (Takara) into the pCFG5-IEGZ plasmid (a gift from Dr. Ron Bose). A terminal FLAG-tag was added to the BCAR4 ORF and Infusion cloned. Sequences were confirmed by Sanger sequencing at Genewiz (South Plainfield). 293T cells were transfected with 3.75 μg of expression plasmid and an 8:1 ratio of pUMVC: VSVG. The next day media was exchanged for cell-specific complete media and virus collected at 48 and 72 hours. 2 mL of virus with 8 μg/mL Polybrene (Sigma) was added to mammalian cells, centrifuged at 2,500 RPM for 75 minutes, and fresh media exchanged after 6 hours. Cells were incubated with virus/polybrene for 6 hours the next day and used for assays 48 hours later. Point mutations within the ATG start site were introduced with Q5 site-directed mutagenesis Kit (NEB) to create the LITAF-BCAR4 L-B mutant construct.

RNA isolation and cDNA synthesis

Total RNA was isolated with NucleoSpin RNA plus with DNA removal column (Macherey-Nagel). cDNA was synthesized with High Capacity cDNA Reverse Transcription Kit (Invitrogen) or 1-Step TB Green PrimeScript qRT-PCR kit (Takara).

qRT-PCR

Gene expression was confirmed with qRT-PCR using PowerSyBr Green (Invitrogen) or 1-Step TB Green PrimeScript (Takara). The comparative CT (ΔΔCT) method was used with values normalized to the housekeeping gene, RPL32, and to control samples. All primers (Supplementary Table S2) were obtained from Integrated DNA Technologies and determined to have 90% to 110% primer efficiency.

FITC Annexin V apoptosis detection

Cells were seeded at 200,000 cells/well and Annexin V staining determined 2 days later according to the manufacturer's protocol (BD Pharmingen).

Cellular nuclear and cytoplasmic isolation

Cells were isolated according to the PARIS kit protocol (Thermo Fisher Scientific) and gene expression determined by qPCR. Nuclear and cytoplasmic isolations were calculated by normalizing respective genes to total RNA expression.

Cell counting

Cells were seeded at 100,000 cells/well and counted every 2 days using the Countess II FL Automated Cell Counter (Thermo Fisher Scientific).

EdU proliferation assay

Cells were seeded at 250,000 cells/well and treated 2 days later with 10 μmol/L EdU (Thermo Fisher Scientific) for 2 hours. Cells were processed following the manufacturer's instructions and stained for DNA content with FxCycle Violet (Thermo Fisher Scientific). Analysis was performed on a FACScan flow cytometer (Becton Dickinson) at the Siteman Cancer Center Flow Cytometry Core (St. Louis, MO). A minimum of 10,000 events per sample were collected. FlowJo Version10 (RRID:SCR_008520; Becton Dickinson) was used to analyze data.

Western blotting

Antibodies against FLAG (2368S; RRID:AB_2217020, 1:1000; Cell Signaling Technology), actin (3700S; RRID:AB_2242334, 1:10,000; Cell Signaling Technology), Goat anti-mouse or anti-rabbit peroxidase-conjugated secondary antibodies (Thermo Fisher Scientific) were used for experiments. Protein samples were prepared with ice-cold RIPA lysis buffer (25 mmol/L Tris pH7.5, 1% NP40, 0.1% SDS, 150 mmol/L NaCl, 0.5% sodium deoxycholate, and 1X Halt Protease Inhibitor Cocktail; Thermo Fisher Scientific). Lysate was subjected to a short sonication and then clarified by centrifugation at maximum speed, 4°C for 10 minutes. Protein concentration was determined with the DC Protein Assay (BioRad). Samples were diluted in loading buffer, boiled, and equal protein concentrations (20–30 μg) loaded and resolved by SDS-polyacrylamide gel electrophoresis in 12% or 4% to 12% Bolt Bis-Tris precast gels (Invitrgoen). Gels were transferred at 60°C for 1 hour to a nitrocellulose membrane (BioRad). Proteins were detected with specific antibodies and visualized and quantified on the ChemDoc MP Imaging System (BioRad) using secondary antibodies and Clarity Western ECL Substrate (Thermo Fisher Scientific).

Small peptide prediction and validation

For the proteogenomic search, we used ORFs predicted in transcripts of lncRNAs annotated in LNCipedia (9) appended to canonical proteins from Uniprot (RRID:SCR_002380; ref. 10) to ensure that the proteogenomic database is not biased towards the discovery of the BCAR4 ORF. We used an equal number of reversed decoys to utilize the target-decoy strategy setting with a threshold of 0.01 for the FDR. Raw mass spectrometry files in mzML format were downloaded from the Clinical Proteomic Tumor Analysis Consortium data resource (CPTAC; RRID:SCR_017135, https://cptac-data-portal.georgetown.edu/datasets; ref. 11). Sequences between start codons (AUG, CUG, UUG) and stop codons (UAG, UGA, UAA) in each of the forward translated frames, with a minimum length of 100 nucleotides, were used as putative ORFs in lncRNAs for the construction of the proteogenomic database. This minimum threshold was selected to maximize signal to noise ratio: encompassing the discovery of known small ORFs (HOXB-AS3/LOC100507537) without increasing false positives from shorter peptides. Supplementary Table S3 shows the variable and fixed modifications used for each protein labeling protocol. The MSFragger search engine (12) was used allowing semitryptic peptides, two missed cleavage sites, 12C/13C isotope errors, and a precursor-ion mass tolerance of 20 ppm. PeptideProphet (RRID:SCR_000274) and ProteinProphet (RRID:SCR_000286; https://github.com/Nesvilab/philosopher; refs. 13, 14) were used to process the search engine results and infer protein groups, respectively. To further validate the novel peptides from the proteogenomic search, we followed the global FDR filtering step by peptide-centric validation as implemented in PepQuery (http://www.pepquery.org; ref. 15) and verified that the identified peptides from the BCAR4 ORF are indeed statistically significant (PepQuery P ≤ 0.01). PSIPRED Protein Analysis Workbench (RRID:SCR_010246; ref. 16) was used to predict protein secondary structure and InterProScan (RRID:SCR_005829; ref. 17) to identify protein domains.

Data availability

The data generated in this study are publicly available at https://github.com/ChrisMaherLab.

Pan-cancer analysis discovers recurrent BCAR4 fusions

We used INTEGRATE to analyze RNA-seq data from 9,638 patients across 32 different cancer types as part of TCGA consortium. To identify functionally recurrent gene fusion candidates, we prioritized genes that were recurrent 3′ gene fusion partners despite having different 5′ partners. As shown in Fig. 1A, the two most prevalent recurrent 3′ gene fusion partners are ERG (202 patients) and TACC3 (55 patients). This is expected as ERG fusions with androgen-sensitive 5′ partners, including TMPRSS2, are highly recurrent gene fusions in prostate cancer. Further, FGFR3-TACC3 is a highly recurrent gene fusion observed across multiple solid tumors. The third most prevalent class of gene fusions resulted in the expression of an annotated lncRNA–BCAR4–in 47 patients across 11 cancer types. BCAR4 gene fusions are more prevalent than known clinically actionable recurrent gene fusions with immediate translational impact such as ABL1, BRAF, and ALK fusions.

Figure 1.

Pan-cancer discovery of BCAR4 gene fusions. A, Dot plot of recurrent gene fusions by number of patients across cancer types. Callout boxes list detected 5′ partners of known gene fusions. B, Left, Structure of expressed BCAR4 fusion transcripts with the 5′ gene fusion partners represented in various colors and BCAR4 represented in green. Fusions are sorted by descending prevalence. B, Right, Representation of patients across cancer types expressing BCAR4 fusion transcripts. Rows correspond to the various BCAR4 isoforms. A cell is colored blue if the patient expresses a particular BCAR4 fusion, and the intensity corresponds to the number of RNA-seq reads supporting evidence of the fusion expression. STAD, stomach adenocarcinoma; CESC, cervical squamous cell carcinoma and endocervical adenocarcinoma; BRCA, breast invasive carcinoma; ESCA, esophageal carcinoma; OV, ovarian serous cystadenocarcinoma; SKCM, skin cutaneous melanoma; BLCA, bladder urothelial carcinoma; LUAD, lung adenocarcinoma; UCEC, uterine corpus endometrial carcinoma; CRC, colorectal cancer; PRAD, prostate adenocarcinoma.

Figure 1.

Pan-cancer discovery of BCAR4 gene fusions. A, Dot plot of recurrent gene fusions by number of patients across cancer types. Callout boxes list detected 5′ partners of known gene fusions. B, Left, Structure of expressed BCAR4 fusion transcripts with the 5′ gene fusion partners represented in various colors and BCAR4 represented in green. Fusions are sorted by descending prevalence. B, Right, Representation of patients across cancer types expressing BCAR4 fusion transcripts. Rows correspond to the various BCAR4 isoforms. A cell is colored blue if the patient expresses a particular BCAR4 fusion, and the intensity corresponds to the number of RNA-seq reads supporting evidence of the fusion expression. STAD, stomach adenocarcinoma; CESC, cervical squamous cell carcinoma and endocervical adenocarcinoma; BRCA, breast invasive carcinoma; ESCA, esophageal carcinoma; OV, ovarian serous cystadenocarcinoma; SKCM, skin cutaneous melanoma; BLCA, bladder urothelial carcinoma; LUAD, lung adenocarcinoma; UCEC, uterine corpus endometrial carcinoma; CRC, colorectal cancer; PRAD, prostate adenocarcinoma.

Close modal

The most common BCAR4 gene fusion events are intrachromosomal rearrangements with breakpoints between the initial untranslated exons of either ZC3H7A or LITAF and the fourth exon of BCAR4 (Fig. 1B). This results in the regulatory regions of ZC3H7A or LITAF minimally activating the expression of the fourth exon of BCAR4. Similarly, most expressed BCAR4 gene fusion transcripts include early exons of 5′ partners spliced upstream of exon 4 of BCAR4. This functionally recurrent structure of BCAR4 gene fusions across patients suggests exon 4 is the minimum region necessary to drive its function. Importantly, BCAR4 gene fusions are tumor specific with no detection in adjacent normal tissues available from 718 TCGA patients.

Silencing BCAR4 gene fusions decreases cell-cycle progression and proliferation in cancer cells

BCAR4 is a known oncogene promoting tumor growth (7), suggesting a cancer relevance for BCAR4 gene fusions. To identify cancer cell lines harboring BCAR4 fusions, we analyzed RNA-seq data from the Cancer Cell Line Encyclopedia (CCLE) using INTEGRATE and found two cell lines harboring the most common 5′ gene fusion partners: SNU308 gallbladder cancer cells (expressing LITAF-BCAR4) and TUHR14TKB renal carcinoma cells (expressing ZC3H7A-BCAR4; Supplementary Fig. S1A). These patient-derived lines, endogenously expressing only BCAR4 fusions, are ideally suited to study the effects of BCAR4 fusions on cancerous phenotypes.

To determine whether the BCAR4 fusions drive cell-cycle progression in cancer cells, the fusions were transiently silenced with siRNAs and the cell-cycle profile assessed with flow cytometry to monitor EdU incorporation and DNA content. Knockdown of fusion expression resulted in significantly fewer S-phase cells and increase in G1 phase cells (Fig. 2A and B). Indeed, these fusion-targeting siRNAs consistently reduced the proportion of S-phase cells by 10% to 25% across these cell lines and increased the proportion of G1 phase cells by 5% to 19%. Knockdown of BCAR4 gene fusions did not alter cell viability as determined by annexin staining (Supplementary Fig. S2A). These data show that endogenously expressed BCAR4 gene fusions alter cell-cycle, specifically S-phase entry, in two different cancer cell lines.

Figure 2.

BCAR4 gene fusions alter cell-cycle and proliferation. Representative dot plots of EdU-DNA stain (FxCycle) flow cytometry analysis and quantification of G1, S, G2–M cell populations after siRNA-mediated silencing of BCAR4 fusions in SNU308 (n = 3; A) or TUHR14TKB (n = 4; B) cells. A and B, Right, qRT-PCR analysis confirmed knockdown of BCAR4 fusions. Representative dot plots of EdU-DNA stain (FxCycle) flow cytometry analysis and quantification of G1, S, G2–M cell populations overexpressing EV or gene fusion (L-B or Z-B fusion) in HME1 (n = 4; C) or MCF10a cells (n = 6; D). Bar graphs present normalized mean ± SEM. Paired ratiometric two-tailed t tests were performed. C and D, Right, Cell growth curve analysis of HME1 (n = 5) or MCF10a (n = 6) cells overexpressing EV, L-B, or Z-B fusions. Graphs present mean ± SEM. Paired two-tailed t tests were performed. *, P < 0.05; **, P < 0.01; ***, P < 0.001.

Figure 2.

BCAR4 gene fusions alter cell-cycle and proliferation. Representative dot plots of EdU-DNA stain (FxCycle) flow cytometry analysis and quantification of G1, S, G2–M cell populations after siRNA-mediated silencing of BCAR4 fusions in SNU308 (n = 3; A) or TUHR14TKB (n = 4; B) cells. A and B, Right, qRT-PCR analysis confirmed knockdown of BCAR4 fusions. Representative dot plots of EdU-DNA stain (FxCycle) flow cytometry analysis and quantification of G1, S, G2–M cell populations overexpressing EV or gene fusion (L-B or Z-B fusion) in HME1 (n = 4; C) or MCF10a cells (n = 6; D). Bar graphs present normalized mean ± SEM. Paired ratiometric two-tailed t tests were performed. C and D, Right, Cell growth curve analysis of HME1 (n = 5) or MCF10a (n = 6) cells overexpressing EV, L-B, or Z-B fusions. Graphs present mean ± SEM. Paired two-tailed t tests were performed. *, P < 0.05; **, P < 0.01; ***, P < 0.001.

Close modal

BCAR4 gene fusions increase cell-cycle progression and proliferation in benign cells

Our data show that endogenous BCAR4 fusions influence the behavior of cancer cells; next, we evaluated whether BCAR4 gene fusions can drive proliferation in normal epithelial cell lines. We hypothesized that expression of full length BCAR4 or the common BCAR4 fusions would drive cell-cycle progression resulting in more proliferating (S-phase) cells and fewer G1-phase cells. Full length BCAR4, the LITAF-BCAR4 (L-B fusion), or the ZC3H7A-BCAR4 fusion (Z-B fusion) were introduced into untransformed cell lines (HME1 and MCF10a; Supplementary Fig. S1B) and expression levels were monitored by qRT-PCR (Supplementary Fig. S3); the cell-cycle profile and proliferation were then assessed. Cell-cycle analysis confirmed that HME1 cells expressing full length BCAR4 had 29% more S-phase cells than empty vector (EV) controls (Supplementary Fig. S4A). Interestingly, the L-B and Z-B fusions more efficaciously drove HME1 cells into S-phase (56% and 65% increases, respectively; Fig. 2C). Similarly, expression of BCAR4 gene fusions in MCF10a cells significantly increased the proportion of S-phase cells (greater than 80% relative to EV; Fig. 2D). There were no concurrent changes in cell viability (Supplementary Fig. S2B). These results demonstrate BCAR4 fusion expression can promote cell-cycle progression in benign cell lines.

To determine whether the observed increase in cell-cycle progression leads to an increase in cellular proliferation, the growth of full length BCAR4- and fusion-overexpressing cells were monitored with cell counting. Significant increases in the number of HME1 cells expressing the L-B and Z-B fusions were observed starting at 48 hours, and at 96 hours there were approximately 3-fold more cells relative to EV controls (Fig. 2C, right). Expression of full length BCAR4 yielded more modest results (<2-fold more than EV at 96 hours; Supplementary Fig. S4B). MCF10a BCAR4-fusion cells also had increased cellular proliferation (Fig. 2D, right). These data indicate that BCAR4 gene fusion expression is sufficient to increase cell growth in two normal cell lines.

BCAR4 encodes a small peptide responsible for the oncogenic phenotypes

Since the fourth exon of BCAR4 is the minimum common region found in all gene fusion events across cancers, we speculated that it was functionally important. Our sequence analysis of BCAR4 identified a previously reported small peptide (18). The BCAR4 ORF is 366 nucleotides in length and produces a 121 amino acid peptide chain predicted to have structure and multiple transmembrane domains (Supplementary Fig. S5). Evidence of the BCAR4 peptide in cancer patient samples was found by mining publicly available mass spectrometry data. Two different semi-tryptic digestion fragments corresponding to the BCAR4 protein were reliably detected in at least four cancer types (Fig. 3A; Supplementary Fig. S6A and S6B; Supplementary Table S4). In addition, ribosome profiling (Ribo-seq) analysis from public datasets (19) shows a focal enrichment in the predicted ORF within exon 4 of the BCAR4 transcript (Supplementary Fig. S6C). These data are strong evidence that BCAR4 encodes an expressed protein in solid tumors.

Figure 3.

BCAR4 encodes a small protein that alters cell-cycle and proliferation. A, Left, Schematics depicting the BCAR4 gene, its predicted open reading frame in exon 4, and the tryptic peptides of BCAR4 detected by mass spectrometry. A, Right, BCAR4 peptide support in lung adenocarcinoma (LUAD), uterine corpus endometrial carcinoma (UCEC), ovarian serous cystadenocarcinoma (OV), and breast invasive carcinoma (BRCA). Representative dot plots of EdU-DNA stain (FxCycle) flow cytometry analysis and quantification of G1, S, G2–M cell populations of B, HME1 (n = 6) and C, MCF10a (n = 3) overexpressing EV, L-B, or mutant L-B (mut L-B) fusions. Bar graphs present normalized mean ±SEM. Paired ratiometric two-tailed t tests were performed. B and C, Right, Cell growth curve analysis of HME1 (n = 6) or MCF10a cells (n = 4) overexpressing EV, LB, or mut L-B. Graphs present mean ±SEM. Paired two-tailed t tests were performed. D, Western blot of FLAG-tagged BCAR4 ORF expression in HME1 and MCF10a BCAR4-fusion–overexpressing cells, n = 4. *, P < 0.05; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001.

Figure 3.

BCAR4 encodes a small protein that alters cell-cycle and proliferation. A, Left, Schematics depicting the BCAR4 gene, its predicted open reading frame in exon 4, and the tryptic peptides of BCAR4 detected by mass spectrometry. A, Right, BCAR4 peptide support in lung adenocarcinoma (LUAD), uterine corpus endometrial carcinoma (UCEC), ovarian serous cystadenocarcinoma (OV), and breast invasive carcinoma (BRCA). Representative dot plots of EdU-DNA stain (FxCycle) flow cytometry analysis and quantification of G1, S, G2–M cell populations of B, HME1 (n = 6) and C, MCF10a (n = 3) overexpressing EV, L-B, or mutant L-B (mut L-B) fusions. Bar graphs present normalized mean ±SEM. Paired ratiometric two-tailed t tests were performed. B and C, Right, Cell growth curve analysis of HME1 (n = 6) or MCF10a cells (n = 4) overexpressing EV, LB, or mut L-B. Graphs present mean ±SEM. Paired two-tailed t tests were performed. D, Western blot of FLAG-tagged BCAR4 ORF expression in HME1 and MCF10a BCAR4-fusion–overexpressing cells, n = 4. *, P < 0.05; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001.

Close modal

The BCAR4 protein contributes to cancer phenotypes (18); however, there is also evidence that BCAR4 functions as a lncRNA (7). For insight into the function of BCAR4, we assayed the localization of BCAR4 fusion transcripts: cytoplasmic localization would indicate translation potential while nuclear localization would indicate a noncoding function of BCAR4. Endogenously expressed BCAR4 fusion transcripts localized predominantly to the cytoplasm: 89% of LITAF-BCAR4 RNA in SNU308 cells and 62% of ZC3H7A-BCAR4 RNA in TUHR14TKB cells (Supplementary Fig. S7). These data are consistent with translation of the BCAR4 ORF.

To directly test the functional significance of the BCAR4 ORF, we mutated the start codon in the LITAF-BCAR4 fusion plasmid, thus abrogating potential translation of the BCAR4 ORF (Supplementary Fig. S1B). HME1 and MCF10a cells expressing the mutant L-B fusion had significantly fewer S-phase cells than wild-type L-B fusion expressing cells, indicating that translation of the ORF is at least partly responsible for the cell-cycle phenotype (Fig. 3B and C). Over time these differences in cell-cycle progression led to large changes in cell number: 96 hours after plating, there were twice as many HME1 cells expressing the L-B fusion than the mutant L-B fusion (Fig. 3B, right). Changes in cell-cycle and cell number were also observed in MCF10a cells (Fig. 3C, right). To directly detect BCAR4 protein in our models, we FLAG-tagged the constructs and observed robust expression in cells expressing the L-B fusion but not the mutant L-B fusion (Fig. 3D; Supplementary Fig. S8). Some residual signal was observed from the mutant L-B construct, potentially caused by use of a second in-frame ATG codon just downstream of the canonical start codon. Expression of the BCAR4 ORF alone (without additional noncoding sequence) similarly promoted cell-cycle progression and proliferation in MCF10a cell models (Supplementary Fig. S9). Overall, these data show that the fourth exon of BCAR4 can produce a small peptide capable of inducing cell growth.

Using our highly sensitive gene fusion discovery tool, INTEGRATE, we discovered a novel gene fusion activating a known oncogene, BCAR4. This is the first unbiased pan-cancer study to systematically discover functionally recurrent BCAR4 gene fusions across cancer types. These findings have clinical significance as (i) full-length BCAR4 is implicated in poor clinical outcome and treatment resistance (6); (ii) full-length BCAR4 has oncogenic potential in breast cancer (7, 18) and NSCLC (20); and (iii) focal amplifications of BCAR4 were found in 20% of cervical patients in the TCGA dataset (21). Our findings extend the clinical significance of BCAR4 into new cancer types–including highly aggressive cancers–to provide an important mechanism of how BCAR4 induces oncogenic properties.

Our findings indicate that BCAR4 chimeras are oncogenic driver events across cancer types. We discovered 47 BCAR4 gene fusions with 19 different 5′ partners–mostly involving only the first untranslated regulatory exon of the 5′ gene. A previous study in NCSLC also found BCAR4 gene fusions with the 5′ partner CD63 (22). Our pan-cancer study expands beyond lung cancer to discover BCAR4 gene fusion events in 11 cancers, which surpasses the prevalence of well characterized functionally recurrent gene fusions such as EVT and ROS (two cancers), ERG (four cancers), BRAF and ALK (six cancers). The consistent expression of exon 4 of BCAR4 suggests functional recurrence and implicates a biological and clinical importance of BCAR4 gene fusions in solid tumors.

It is challenging to confidently annotate RNAs as coding or noncoding with growing evidence suggesting RNAs may encode small ORFs producing stable, functional peptides. Complicating this process is the possibility that RNAs can have both coding and noncoding functions (23–26). Full length BCAR4 was first classified as a protein with expression in human oocytes and placenta (27); however, more recently is it described as a lncRNA (7), raising the possibility that BCAR4 may function as a noncoding RNA and/or a protein-coding gene. Here, we show that the minimum common region of BCAR4 across all gene fusions is the fourth exon, which harbors an ORF. Our computational and experimental data provide evidence for translation of the ORF to produce a functional protein capable of influencing oncogenic phenotypes. This is supported by detecting BCAR4 peptide in patient tumors and FLAG-tagged BCAR4 protein in cell culture models. Fusion and ORF-only constructs had significantly increased proliferation in benign cell lines while a construct with a mutated ATG start site had significantly diminished protein expression and proliferation. Genomic editing could provide additional evidence for the importance of the peptide relative to the lncRNA function. These results establish BCAR4 fusions and their resulting peptide products as potential oncogenic drivers and possible therapeutic targets.

A. Nickless reports a patent for BCAR4 as a predictive biomarker and therapeutic target across cancers pending. J. Zhang reports a patent for BCAR4 as a predictive biomarker and therapeutic target across cancers pending. C.A. Maher reports grants from Emerson Collective; and grants from The Alvin J. Siteman Cancer Center Siteman Investment Program, The Foundation for Barnes-Jewish Hospital Cancer Frontier Fund, and Barnard Trust during the conduct of the study; other support from Tempus outside the submitted work; in addition, C.A. Maher has a patent for BCAR4 as a predictive biomarker and therapeutic target across cancers pending. N.M. White reports a patent for BCAR4 as a predictive biomarker and therapeutic target across cancers pending. No disclosures were reported by the other authors.

A. Nickless: Conceptualization, data curation, formal analysis, validation, investigation, writing–original draft, writing–review and editing. J. Zhang: Software, formal analysis, visualization. G. Othoum: Formal analysis, investigation, visualization, writing–review and editing. J. Webster: Formal analysis, investigation, writing–review and editing. M.J. Inkman: Software, writing–review and editing. E. Coonrod: Investigation, methodology, writing–review and editing. S. Fontes: Investigation. E.B. Rozycki: Investigation. C.A. Maher: Conceptualization, formal analysis, supervision, funding acquisition, visualization, writing–review and editing. N.M. White: Conceptualization, formal analysis, supervision, validation, visualization, writing–original draft.

C.A. Maher received funding from The Alvin J. Siteman Cancer Center Siteman Investment Program through The Foundation for Barnes-Jewish Hospital Cancer Frontier Fund (grant no. GR0012692) and Barnard Trust (grant no. GR0008791), and The Emerson Collective (grant no. GR0017807). We thank the Alvin J. Siteman Cancer Center at Washington University School of Medicine and Barnes-Jewish Hospital (St. Louis, MO), for the use of the Siteman Flow Cytometry. The Siteman Cancer Center is supported in part by an NCI Cancer Center Support Grant (grant no. P30 CA091842).

Note: Supplementary data for this article are available at Molecular Cancer Research Online (http://mcr.aacrjournals.org/).

1.
Tomlins
SA
,
Rhodes
DR
,
Perner
S
,
Dhanasekaran
SM
,
Mehra
R
,
Sun
X-W
, et al
.
Recurrent fusion of TMPRSS2 and ETS transcription factor genes in prostate cancer
.
Science
2005
;
310
:
644
8
.
2.
Maher
CA
,
Kumar-Sinha
C
,
Cao
X
,
Kalyana-Sundaram
S
,
Han
B
,
Jing
X
, et al
.
Transcriptome sequencing to detect gene fusions in cancer
.
Nature
2009
;
458
:
97
101
.
3.
Govindan
R
,
Ding
L
,
Griffith
M
,
Subramanian
J
,
Dees
ND
,
Kanchi
KL
, et al
.
Genomic landscape of non-small cell lung cancer in smokers and never-smokers
.
Cell
2012
;
150
:
1121
34
.
4.
Palanisamy
N
,
Ateeq
B
,
Kalyana-Sundaram
S
,
Pflueger
D
,
Ramnarayanan
K
,
Shankar
S
, et al
.
Rearrangements of the RAF kinase pathway in prostate cancer, gastric cancer and melanoma
.
Nat Med
2010
;
16
:
793
8
.
5.
Zhang
J
,
White
NM
,
Schmidt
HK
,
Fulton
RS
,
Tomlinson
C
,
Warren
WC
, et al
.
INTEGRATE: gene fusion discovery using whole genome and transcriptome data
.
Genome Res
2016
;
26
:
108
18
.
6.
Godinho
MFE
,
Sieuwerts
AM
,
Look
MP
,
Meijer
D
,
Foekens
JA
,
Dorssers
LCJ
, et al
.
Relevance of BCAR4 in tamoxifen resistance and tumour aggressiveness of human breast cancer
.
Br J Cancer
2010
;
103
:
1284
91
.
7.
Xing
Z
,
Lin
A
,
Li
C
,
Liang
K
,
Wang
S
,
Liu
Y
, et al
.
lncRNA directs cooperative epigenetic regulation downstream of chemokine signals
.
Cell
2014
;
159
:
1110
25
.
8.
Lau
JW
,
Lehnert
E
,
Sethi
A
,
Malhotra
R
,
Kaushik
G
,
Onder
Z
, et al
.
The cancer genomics cloud: collaborative, reproducible, and democratized-a new paradigm in large-scale computational research
.
Cancer Res
2017
;
77
:
e3
6
.
9.
Volders
P-J
,
Helsens
K
,
Wang
X
,
Menten
B
,
Martens
L
,
Gevaert
K
, et al
.
LNCipedia: a database for annotated human lncRNA transcript sequences and structures
.
Nucleic Acids Res
2013
;
41
:
D246
251
.
10.
UP
Consortium
.
UniProt: a hub for protein information
.
Nucleic Acids Res
2015
;
43
:
D204
212
.
11.
Edwards
NJ
,
Oberti
M
,
Thangudu
RR
,
Cai
S
,
McGarvey
PB
,
Jacob
S
, et al
.
The CPTAC data portal: a resource for cancer proteomics research
.
J Proteome Res
2015
;
14
:
2707
13
.
12.
Kong
AT
,
Leprevost
FV
,
Avtonomov
DM
,
Mellacheruvu
D
,
Nesvizhskii
AI
.
MSFragger: ultrafast and comprehensive peptide identification in mass spectrometry-based proteomics
.
Nat Methods
2017
;
14
:
513
20
.
13.
Keller
A
,
Nesvizhskii
AI
,
Kolker
E
,
Aebersold
R
.
Empirical statistical model to estimate the accuracy of peptide identifications made by MS/MS and database search
.
Anal Chem
2002
;
74
:
5383
92
.
14.
Nesvizhskii
AI
,
Keller
A
,
Kolker
E
,
Aebersold
R
.
A statistical model for identifying proteins by tandem mass spectrometry
.
Anal Chem
2003
;
75
:
4646
58
.
15.
Wen
B
,
Wang
X
,
Zhang
B
.
PepQuery enables fast, accurate, and convenient proteomic validation of novel genomic alterations
.
Genome Res
2019
;
29
:
485
93
.
16.
Buchan
DWA
,
Jones
DT
.
The PSIPRED Protein Analysis Workbench: 20 years on
.
Nucleic Acids Res
2019
;
47
:
W402
7
.
17.
Quevillon
E
,
Silventoinen
V
,
Pillai
S
,
Harte
N
,
Mulder
N
,
Apweiler
R
, et al
.
InterProScan: protein domains identifier
.
Nucleic Acids Res
2005
;
33
:
W116
120
.
18.
Godinho
M
,
Meijer
D
,
Setyono-Han
B
,
Dorssers
LCJ
,
van Agthoven
T
.
Characterization of BCAR4, a novel oncogene causing endocrine resistance in human breast cancer cells
.
J Cell Physiol
2011
;
226
:
1741
9
.
19.
Michel
AM
,
Fox
G
,
M Kiran
A
,
De Bo
C
,
O'Connor
PBF
,
Heaphy
SM
, et al
.
GWIPS-viz: development of a ribo-seq genome browser
.
Nucleic Acids Res
2014
;
42
:
D859
864
.
20.
Yang
H
,
Yan
L
,
Sun
K
,
Sun
X
,
Zhang
X
,
Cai
K
, et al
.
lncRNA BCAR4 increases viability, invasion, and migration of non-small cell lung cancer cells by targeting glioma-associated oncogene 2 (GLI2)
.
Oncol Res
2019
;
27
:
359
69
.
21.
Cancer Genome Atlas Research Network
,
Albert Einstein College of Medicine
,
Analytical Biological Services
,
Barretos Cancer Hospital
,
Baylor College of Medicine
,
Beckman Research Institute of City of Hope
, et al
Integrated genomic and molecular characterization of cervical cancer
.
Nature
2017
;
543
:
378
84
.
22.
Bae
K
,
Kim
JH
,
Jung
H
,
Kong
S-Y
,
Kim
YH
,
Kim
S
, et al
.
A fusion of CD63-BCAR4 identified in lung adenocarcinoma promotes tumorigenicity and metastasis
.
Br J Cancer
2021
;
124
:
290
8
.
23.
Slavoff
SA
,
Mitchell
AJ
,
Schwaid
AG
,
Cabili
MN
,
Ma
J
,
Levin
JZ
, et al
.
Peptidomic discovery of short open reading frame-encoded peptides in human cells
.
Nat Chem Biol
2013
;
9
:
59
64
.
24.
Chen
J
,
Brunner
A-D
,
Cogan
JZ
,
Nuñez
JK
,
Fields
AP
,
Adamson
B
, et al
.
Pervasive functional translation of noncanonical human open reading frames
.
Science
2020
;
367
:
1140
6
.
25.
Othoum
G
,
Coonrod
E
,
Zhao
S
,
Dang
HX
,
Maher
CA
.
Pan-cancer proteogenomic analysis reveals long and circular noncoding RNAs encoding peptides
.
NAR Cancer
2020
;
2
:
zcaa015
.
26.
Prensner
JR
,
Enache
OM
,
Luria
V
,
Krug
K
,
Clauser
KR
,
Dempster
JM
, et al
.
Noncanonical open reading frames encode functional proteins essential for cancer cell survival
.
Nat Biotechnol
2021
;
39
:
697
704
.
27.
Angulo
L
,
Perreau
C
,
Lakhdari
N
,
Uzbekov
R
,
Papillier
P
,
Freret
S
, et al
.
Breast-cancer anti-estrogen resistance 4 (BCAR4) encodes a novel maternal-effect protein in bovine and is expressed in the oocyte of humans and other non-rodent mammals
.
Hum Reprod Oxf Engl
2013
;
28
:
430
41
.

Supplementary data