The BRCA1 Pseudogene Negatively Regulates Antitumor Responses through Inhibition of Innate Immune Defense Mechanisms

Innate antiviral immunity is a significant mechanism in cancer (1). It mediates intrinsic antitumor responses through several mechanisms, which include triggering of apoptosis of cancer cells, stimulating spontaneous DNA damage, and increasing antitumor efficacy of radiotherapy and chemotherapy. Both cancer cells and virus-infected cells threaten the host by expressing neo-antigens and by evading control mechanisms of host immune surveillance. Cancer cells can actually mimic a viral infection process by activating RNA-sensing pattern recognition receptors (PRR) and by stimulating cytokine production and activating cytotoxic immune cells, the latter killing infected cells via externally induced cell lysis and apoptosis. RIG-I and MDA5 of the RIG-I–like receptor (RLR) family are important PRRs involved in the detection of RNA viruses (2, 3). RLRs initiate host antiviral responses that induce type I IFNs and other cytokines, leading to the transcription of hundreds of IFN-stimulated genes (ISG; ref. 4). Activation of RLRs by dsRNA ligands not only triggers host immune responses but can also directly induce apoptosis of cancer cells, in an IFN-dependent or independent manner (5). Consequently, cancer cells are highly susceptible to RLR-induced cell death via intrinsic and extrinsic apoptosis and immune activation, indicating that the signaling pathway driven by RLRs is a promising molecular pathway to target in cancer immunotherapy.

Interestingly, recent evidence suggests that antiviral innate immunity is regulated not only by exogenous non–self-RNA but also by host-derived pseudogene RNAs. Pseudogenes have been considered non-functional artifacts of evolutionary processes due to degenerative features such as the accumulation of disruptive mutations or their lack of regulatory elements (6, 7). However, a growing body of evidence indicates biological roles for pseudogenes as gene expression regulators or immune modulators. 5S ribosomal RNA pseudogene transcripts, in particular RNA5SP141, were shown to bind to RIG-I and induce the expression of antiviral cytokines during infection with herpes simplex virus type 1 or the related herpes virus Epstein–Barr virus (8). Lethe, a pseudogene long-noncoding RNA (lncRNA), is selectively induced by proinflammatory cytokines or glucocorticoid receptor agonists, and serves as a functional regulator of inflammatory signaling (9). These data suggest that pseudogenes may play a role in regulating antiviral defense and inflammatory signaling pathways.

The chromosome 17q21 region containing BRCA1 has a partially duplicated pseudogene, BRCA1P1 (Gene ID: 394269, HUGO ID: 28470), which contains only three of the 24 exons of BRCA1 (10–12). It also includes an insertion of the acidic ribosomal phosphoprotein P1 pseudogene (RPLP1P4) in exon 1a, displaying unique features of a chimeric pseudogene derived from the two parent genes, BRCA1 and RPLP1. The presence of BRCA1P1 on the same chromosome close to BRCA1 appears to create a hotspot for homologous recombination (13), leading to genomic rearrangements between BRCA1P1 and BRCA1 in families with a high risk of breast and ovarian cancers. BRCA1 deficiency is frequently observed in breast, ovarian, and other cancers (14, 15). Although the roles of BRCA1 in regulating homologous recombination and DNA damage repair have been extensively studied (16), the biological relevance of BRCA1P1 pseudogene in breast cancer has not been elucidated.

In this study, we discovered an important role for BRCA1P1 in regulating antiviral program-like responses in breast cancer cells. In contrast with BRCA1′s involvement in homologous recombination repair, BRCA1P1-depleted cancer cells resemble virus-infected cells, and express high amounts of ISGs and cytokines, which ultimately make them more susceptible to cell death. In a breast cancer xenograft mouse model, BRCA1P1 depletion suppresses tumor growth and stimulates proinflammatory cytokines and local immunity. To our knowledge, this is the first study that demonstrates an important role of BRCA1P1 pseudogene in antitumor responses through regulation of antiviral innate immunity and tumor growth.

We obtained written informed consent from the patients for anonymous use of banked tumor tissues for research. The studies were conducted in accordance with recognized ethical guidelines of the Declaration of Helsinki, and were approved by an institutional review board at the University of Chicago Medical Center. Methods for case selection, tumor RNA extraction, microarray and RNA sequencing were described previously (17). In brief, we selected female patient cases with invasive ductal carcinoma from our Breast Program Bio-specimen Bank. Patients who had received neoadjuvant chemotherapy were excluded, as we were interested in pretreatment gene expression. Areas of malignant tissue were isolated from frozen tissues and homogenized by the Tissue Lyzer LT (Qiagen). RNA was extracted using the Qiagen AllPrep DNA/RNA/Protein Mini Kit protocol (Qiagen). The integrity of RNAs was validated using a 2100 Bioanalyzer (Agilent) at the University of Chicago Genomics Facility. RNAs with a minimum RNA integrity number of 8 were used for cDNA synthesis and qRT-PCR experiments. Molecular subtypes of the breast tumors were determined by mRNA expression of the PAM50 intrinsic classifier as previously described (17).

All animals were humanely handled and monitored for health conditions according to the Institutional Animal Care and Use Committee–approved protocols. Eight-week-old female Foxn1nu/nu (Harlan) mice were anesthetized via inhalation with 2% vaporized isoflurane and were unilaterally injected with 4 × 10⁶ MDA-MB-231 wild-type (WT) or BRCA1P1-knockout (KO) cells (100 μL, 50% Matrigel, 8–9 animals per model) into the fourth inguinal mammary gland at the base of the nipple. Tumor measurements were performed weekly using calipers to calculate tumor volume using the formula: 1/2 (Length × Width²). The assessment was blind to the animal model and was performed by the same person throughout the study. Animal weight was monitored twice weekly. Spleens were collected from animals in each model at the end of the study for downstream experiments.

Primary human mammary epithelial cells (HMEC) were purchased from Lonza. Breast cancer cells were obtained from the ATCC. Cells were authenticated for species and unique DNA profile using short tandem repeat (STR) analysis by the provider (ATCC). Cells were cultured in media recommended by the ATCC, and tested negative for Mycoplasma contamination using the MycoAlert Kit (Lonza).

Cells were seeded at 3–5 × 10⁵ cells/well on 6-well plate and transfected with 20–50 nmol/L of BRCA1P1-ASO (LNA GapmeRs, Exiqon) or 2 μg of poly(I:C) (InvivoGen) using DharmaFECT 1 or 4 (Dharmacon), or Lipofectamine 3000 transfection reagent (Invitrogen). One day after transfection, cells were transferred to 96-well cell culture plates at 1–1.5 × 10⁴ cells/well. Chemotherapy drugs (5 μmol/L doxorubicin or 4 μmol/L camptothecin; Sigma-Aldrich) were added to cells one day after transfer, and incubated for 24 hours. The Caspase-Glo 3/7 Assay (Promega) was used to screen apoptotic cells by measuring luminescent signals using a Synergy H1 Plate Reader (BioTek).

Guide RNAs (gRNA) were designed using the chopchop tools (http://chopchop.cbu.uib.no/), IDT Alt-R CRISPR-Cas9 guide RNA (https://www.idtdna.com/site/order/designtool/index/CRISPR_CUSTOM) or sgRNA designer (https://portals.broadinstitute.org/gpp/public/analysis-tools/sgrna-design) and synthesized by Integrated DNA Technologies. Two pairs of gRNAs were designed to delete 1,147 bp (gRNAs 1 and 4) or 1,469 bp (gRNAs 2 and 3) of the BRCA1P1 sequence. For CRISPR activation (CRISPRa), three gRNAs were designed to target the BRCA1P1 promoter. DNA sequences of gRNAs will be provided upon request. The KO experiment was performed as previously described in the literature (18). Briefly, gRNAs were cloned into pSpCas9 (BB)-2A-GFP (Addgene #48138) and pSpCas9 (BB)-2A-Puro (Addgene #62988) using the BbsI restriction enzyme site. For CRISPRa, we used SP-dCas9-VPR (Addgene #63798). MDA-MB-231 cells were cotransfected with gRNAs using Lipofectamine 3000 (Invitrogen) and incubated for two days. For the isolation of BRCA1P1-KO clones, cells were treated with puromycin (1 μg/mL) for one week, diluted to one cell per 100 μL media and plated into each well of a 96-well plate. Single colony cells were expanded and subjected to PCR validation and DNA sequencing.

T47D and MDA-MB-231 cells were transfected with control or BRCA1P1-ASOs, incubated for 24 hours, and processed for RNA purification. MDA-MB-231 cells with BRCA1P1-KO or WT genotype were also subjected to RNA purification. Total RNAs were isolated with the RNeasy kit (Qiagen) following the protocol provided by the manufacturer. To avoid any possibility of DNA contamination, total RNAs were treated with RNase-free DNase I (Qiagen). Quality control was performed with the Agilent 4200 TapeStation system (Agilent). RNAs with RNA integrity numbers greater than eight were selected and subjected to the Illumina Human HT12 microarray platform. The microarray data were analyzed by significance analysis of microarrays (SAM) between control or BRCA1P1-ASO–treated cells or between BRCA1P1-KO and WT cells. The significant gene expression in ASO-treated cells in microarray experiments was identified and subjected to DAVID functional annotation bioinformatics microarray analysis. The microarray datasets are available through GEO (GSE112572 and GSE112573).

Sendai virus (SeV, Cantell strain) was purchased from Charles River Laboratories. Cells were infected with SeV at 5 HAU/mL and lysed at 8, 16, and 24 hours after infection for quantitative real-time PCR (qRT-PCR). VSV-eGFP was kindly provided by Sean Whelan (Harvard). Cells were infected with VSV-eGFP at MOI of 0.005 and analyzed for eGFP expression using flow cytometry (LSR Fortessa flow cytometer, BD Biosciences) after 8, 16, and 24 hours later. For flow cytometry, cells were harvested and then pelleted by centrifugation at 300 × g for 5 minutes. Cells were washed with PBS and fixed with 4% (v/v) paraformaldehyde in PBS at room temperature for 30 minutes, followed by washing cells twice with PBS. The main cell population was gated in the FCS/SSC blot and then analyzed in the FITC channel. Positive cells were above the cutoff value, which is set to 98% of measurable events of the negative control.

Cells were plated at 5,000 cells/well in 96-well plates and allowed to adhere for 24 hours. IncuCyte Caspase-3/7 Green reagent was added to the media at a 1:1,000 dilution (Essen Bioscience No. 4440). Images of live cells were acquired using a 10× objective every 2 hours (four images/well) with the IncuCyte S3 live-cell analysis system. Data were analyzed using IncuCyte analysis software. Cell density was quantified as a measure of cell proliferation. Apoptosis was quantified as the total integrated intensity of green fluorescent signal from cells with activated caspase-3/7 (apoptotic cells) normalized to cell density calculated from phase-contrast images. The experiment was performed twice with 7–10 technical replicates.

Each experiment was repeated at least three times and the mean ± standard deviation (SD) or mean ± standard error (SE) is shown with t test (unpaired, two-tailed) values. Statistical significance was set at *, P < 0.05; **, P < 0.01; ***, P < 0.001; and ****, P < 0.0001. We performed one-way ANOVA to compare the expression level across the breast tumor subtype groups. Statistical analysis was carried out using Microsoft Excel and GraphPad Prism 6.0 (GraphPad Software, Inc.). Plots were generated using GraphPad Prism 6.0.

The microarray datasets are available through GEO (GSE112572 and GSE112573). There are no restrictions on availability of data. The data that support the findings of this study are available from the corresponding author upon request.

More detailed Materials and Methods are included in Supplementary Information. The oligonucleotides and antibodies used in the experiments are listed in Supplementary Tables S1 and S2, respectively.

The chromosome 17q21 region comprising the BRCA1P1 pseudogene shows parallel elements of genomic structure, as BRCA1 is located head-to-head with NBR2, and BRCA1P1 is located head-to-head with NBR1 (Fig. 1A). Cross-species comparative analysis among chicken, mouse, and human revealed that the BRCA1P1 pseudogene is present only in the human genome. A comparison of DNA sequences between BRCA1 and BRCA1P1 showed that the pseudogene contains only three exons (exons 1a, 1b, and 2), with three major insertions that include a 343 bp insertion of the RPLP1 pseudogene (RPLP1P4) in exon 1a (10) and two ALU element insertions in exon 1b and intron 1b of BRCA1P1 (Fig. 1B). Notably, exon 2 contains a point mutation that alters the translation initiation codon (ATG) of BRCA1P1 to ATA. On the basis of the degenerative mutations occurring in the pseudogene (insertions and a start codon mutation), we predicted no protein coding potential for BRCA1P1 transcripts.

To determine whether the BRCA1P1 pseudogene is expressed through the bidirectional promoter between BRCA1P1 and NBR1, we cloned the 1,243 bp region of the BRCA1P1 promoter into a luciferase reporter gene construct and assessed the promoter activity in T47D cells (Fig. 1C). Sequence homology analysis between the BRCA1 and BRCA1P1 promoters showed that 85.7% of the promoter regions were identical. The BRCA1P1 promoter exhibits a 9.6 ± 1.0-fold increase in luciferase activity compared with a pGL₃B empty vector. Using this active promoter, the BRCA1P1 pseudogene is transcribed from exon 1a to exon 1b (including introns) with a resulting RNA of at least 1,629 nucleotides [details in Supplementary Methods (characterization of BRCA1P1 transcripts) and Supplementary Fig. S1].

We surveyed BRCA1P1 expression in breast cell lines and breast tumors of different molecular subtypes and genetic backgrounds (Fig. 1D–F). Three pairs of primers were designed to quantify the pseudogene transcripts from exon 1a, exon 1b, and intron 1a to exon 1b (unspliced form), respectively (Supplementary Fig. S2A). Because a spliced form of the BRCA1P1 transcript was previously reported in HeLa cells (12), we first compared the expression between the spliced and unspliced (intron 1a to exon 1b) forms. This analysis showed that the majority of the BRCA1P1 transcripts (98%) are not spliced in T47D and MDA-MB-231 cells (Supplementary Fig. S2B).

Using primers detecting unspliced transcripts of BRCA1P1, we assessed BRCA1P1 expression in several breast cancer cell lines. A comparison of BRCA1P1 expression between nonmalignant and malignant breast cells showed an average 2.8-fold increased expression in breast cancer cell lines compared with primary mammary epithelial cells (HMEC; Fig. 1D). Expression of BRCA1P1 in MDA-MB-231 and T47D cells was 2.86 ± 0.96 and 3.63 ± 1.10-fold higher than that in HMEC, respectively. Next, we examined BRCA1P1 expression in 36 frozen primary breast tissues of different molecular subtypes (Fig. 1E and F). Molecular subtypes of the breast tumors were determined by mRNA expression of the PAM50 intrinsic classifier (19). Expression of BRCA1P1 increased by 2.1-fold in breast tumors compared with normal breast tissues, with no subtype-specific pattern of expression. Together, these data indicate that breast cancer cells and tumors exhibit elevated BRCA1P1 expression compared with normal breast epithelial cells or normal breast tissues.

To determine the biological role of the pseudogene in breast cancer, we inhibited expression of BRCA1P1 in breast cancer cells using two different approaches. First, antisense oligonucleotides (ASO, LNA GapmeRs) were custom designed to target exon 1a of BRCA1P1. The efficiency of knockdown by ASO was validated at the RNA expression level using qRT-PCR (Supplementary Fig. S2C). The expression of BRCA1P1 was significantly dampened by BRCA1P1-specific ASO, with no silencing effect on the expression of the parent genes (BRCA1 and RPLP1) or adjacent genes (NBR1) in T47D and MDA-MB-231 cells. Second, we deleted the BRCA1P1 pseudogene from the genome of MDA-MB-231 cells using CRISPR-Cas9 genome-editing tools (Supplementary Fig. S3A). Cotransfection of two guide RNAs (gRNA1 and 4) resulted in a partial deletion (1,147 bp) of the BRCA1P1 pseudogene (Supplementary Fig. S3B), with no inhibitory effects on the parent gene (BRCA1) expression (Supplementary Fig. S3C).

To identify genes and pathways influenced by BRCA1P1 loss or depletion, we performed genome-wide gene expression profiling (Fig. 2A). This analysis identified 3,981 upregulated genes in BRCA1P1-ASO–treated groups compared with control-ASO groups both in T47D and MDA-MB-231 cells, as well as 723 upregulated genes in BRCA1P1-KO cells compared with the WT cells. One hundred and thirteen genes were upregulated in both BRCA1P1-KO and knockdown (ASO-treated) cells. Gene ontology enrichment analysis of the 113 overlapping genes revealed the antiviral defense response as one of the most upregulated pathways.

Validation of microarray data by qRT-PCR confirmed increased expression of antiviral ISGs [IFIT3, MDA5 (IFIH1), RIG-I (DDX58), HERC6, and APOBEC3B] in two clones of BRCA1P1-KO compared with WT control cells (Fig. 2B). Protein expression of IFIT3, MDA5, and RIG-I was also significantly increased in the KO clones compared with control cells (Fig. 2C). The data suggested that BRCA1P1 deficiency triggered an innate immune defense program characterized by transcriptional upregulation of many antiviral genes.

Because the replication and spread of viral pathogens are restricted by innate immune mechanisms and most prominently cytokine signaling, we next examined whether loss of BRCA1P1 stimulated cytokine expression. We used qRT-PCR to analyze cytokine expression because basal cytokine expression levels in breast cancer cells are extremely low, and thus often undetectable by microarrays. We found a dramatic increase in TNF expression in BRCA1P1-depleted cells compared with control cells (Fig. 2D). Expression of TNF was 10.7 ± 1.1- and 11.0 ± 0.8-fold greater in two BRCA1P1-KO clones, respectively, compared with the WT cells (left, Fig. 2D). TNF expression also gradually increased in T47D cells treated with increasing concentrations of BRCA1P1-ASOs (right, Fig. 2D). Of note, we did not observe any expression of type I IFNs (IFNA1 and IFNB1) in MDA-MB-231 cells using qRT-PCR (Supplementary Fig. S4A and S4B). Copy-number alteration data in the Cancer Cell Line Encyclopedia revealed a frequent deletion of type I IFN genes in cancer cell lines (Supplementary Fig. S4D). In accord, MDA-MB-231 cells carry homozygous (deep) deletions of 17 subtypes of IFNα and IFNβ genes. In contrast, the expression of type III IFN (IFNL1) was markedly increased in MDA-MB-231 cells, compared with nonmalignant HMEC cells (Supplementary Fig. S4C). Furthermore, loss of BRCA1P1 led to a significant increase in IFNL1 expression in MDA-MB-231 cells compared with the WT cells (left, Fig. 2E). Knockdown of BRCA1P1 using ASO also increased IFNL1 expression, compared with control-ASO–treated cells (right, Fig. 2E). Collectively, our data indicate that loss of BRCA1P1 stimulates the expression of cytokines (TNFα and IFNλ) and a range of antiviral ISGs.

We next determined the effect of BRCA1P1 loss on virus-induced antiviral gene expression and virus replication in MDA-MB-231 cells (Fig. 3A–C). SeV (Paramyxoviridae) and eGFP-expressing vesicular stomatitis virus (VSV-eGFP, Rhabdoviridae) are well known to trigger innate immune responses via activation of the RLR signaling pathway. The expression of cytokine/chemokine (TNF and CCL5) and ISGs (IFIT1 and RSAD2) during SeV infection was significantly higher in BRCA1P1-KO cells than in infected WT cells (Fig. 3A). In particular, the expression of TNF was greatly increased in BRCA1P1-KO cells compared with the WT cells at 24 hours after SeV infection (a 2,568.1 ± 431-fold vs. 361.7 ± 16.1-fold increase). The expression of IFNL1 and IFNL2/L3 was also greater in the KO cells at 8 hours after infection (a 14,527 ± 1,520-fold vs. 5,754 ± 320-fold increase; Fig. 3B). In accord, cells depleted of BRCA1P1 suppressed the replication of VSV-eGFP more effectively than did the WT cells [2.36 vs. 38.3 eGFP-positive cells (relative units)] (Fig. 3C). Collectively, these findings indicated that loss of BRCA1P1 triggered a cytokine-mediated antiviral program.

Previous reports showed that poly(I:C), a synthetic analogue of double-stranded RNAs, binds to cytosolic RIG-I and MDA5 receptors (20, 21) and induces innate immune signaling that leads to the cytosol-to-nuclear translocation of several key transcription factors [e.g., NF-κB, IFN-regulatory factors (IRF), and AP-1; ref. 22]. To determine whether BRCA1P1 regulates cytokine expression through activation of transcription factors, we first examined poly(I:C)-driven nuclear translocation of NF-κB in breast cancer cells (Supplementary Fig. S5A). In MDA-MB-231 cells, as in fibroblast cells, nuclear localization of the NF-κB subunit RelA increased upon poly(I:C) transfection, starting from 4 to 8 hours after transfection. Next, as MDA-MB-231 cells express type III IFNs (Supplementary Fig. S4C), we examined nuclear transport of IRF1, which plays a major role in type III IFN induction. We observed an increase in IRF1 nuclear localization at 4 to 8 hours after poly(I:C) transfection. A rapid increase in TNF and IFNL expressions was also observed starting at 4 to 8 hours after treatment, followed by increases in expression for ISGs (IFIT3, IFIH1, and DDX58), ILs (IL8, IL6, IL1A, and IL1B), and the transcription factors (IRF1, STAT1, and STAT2; Supplementary Fig. S5B and S5C).

We then compared poly(I:C)-induced gene expression between BRCA1P1-WT and KO cells (Fig. 3D). This analysis showed significant increases in the expression of TNF (a 4,660 ± 136-fold vs. 575 ± 15-fold increase), IFNL1 (a 244,948 ± 8,898-fold vs. 111,222 ± 1,845-fold increase) and IFNL2/3 (a 219,251 ± 6,429-fold vs. 33,140 ± 1,105-fold increase) in BRCA1P1-KO cells compared with the WT cells at 8 hours after treatment. It is of note that we were able to measure gene expression only up to 8 hours after treatment because poly(I:C)-transfected BRCA1P1-KO cells were severely apoptotic and not able to survive for longer times. The expression of ILs (IL1A, IL1B, and IL6) and antiviral genes [IFIH1 (MDA5) and STAT1] was also significantly higher in BRCA1P1-KO cells than in WT cells (Supplementary Fig. S6). These data suggested that depletion of BRCA1P1 sensitized breast cancer cells to apoptosis by stimulating expression of cytokines.

We also investigated the effect of BRCA1P1 overexpression on TNF and IFNL expression in breast cancer cells (Fig. 3E). To stimulate BRCA1P1 expression using its own promoter and to express BRCA1P1-lncRNA in the proper subcellular location (nucleus), we used the CRISPR activation system (CRISPRa), which led to a 4.5-fold increase in BRCA1P1 expression in MDA-MB-231 cells (left). Overexpression of BRCA1P1 significantly decreased expression of TNF and IFNL1 compared with the control vector-transfected cells (middle and right). The overexpression data together with the depletion data indicate that BRCA1P1 regulates TNF and IFNL expression in breast cancer cells.

To understand the mechanism whereby BRCA1P1 regulates cytokine expression, we first determined the subcellular localization of BRCA1P1-lncRNA using two different methods (Fig. 4A). Fractionation of cellular compartments showed a strong enrichment of BRCA1P1-lncRNA in the nuclei of T47D, CAMA-1 and HCC-1937. Single-molecule RNA-FISH (smRNA-FISH) revealed that the majority of BRCA1P1 signals in cancer cells were restricted to the nucleus, whereas most GAPDH signals were detected in the cytoplasm (Fig. 4A). Together, these data showed nuclear localization of BRCA1P1 transcripts.

We thus tested whether BRCA1P1 interacts with transcription factors important for innate immune signaling in the nucleus of MDA-MB-231 cells. Specifically, we focused on NF-κB, IRF1 and STAT-1 because these transcription factors are known to regulate expression of TNF, IFNL and/or ISGs in response to viral infection or dsRNA stimulation. RNA immunoprecipitation data revealed an association of BRCA1P1-lncRNA with the NF-κB subunit RelA, but not with IRF1 or STAT-1 (Fig. 4B). RNA18S5 (18S rRNA), which served as a negative control, exhibited no interaction with any of them. These data suggested a specific interaction of BRCA1P1-lncRNA with RelA in the nucleus.

To determine whether the interaction of BRCA1P1-lncRNA with RelA influences the enrichment of RelA at the target promoters, we conducted chromatin immunoprecipitation of RelA proteins on the promoters of TNF, IFNL1, and IFIH1 genes (Fig. 4C). We observed greater abundance of the RelA transcription factor at the promoters in BRCA1P1-KO cells compared with WT cells. Inversely, overexpression of BRCA1P1 decreased RelA enrichment at the promoters of TNF, IFNL1, and IFIH1 (Fig. 4D).

We also performed an electrophoretic mobility shift assay to determine whether BRCA1P1-lncRNA inhibits the association of RelA with the target promoters (Fig. 4E). As expected, we observed a strong association of recombinant RelA proteins (p65) with an NF-κB cis–acting regulatory element at the TNF promoter, indicated by the retardation in electrophoretic mobility of the probe on PAGE (lane 2, Fig. 4E). In contrast, the strength of this interaction decreased with increasing amounts (1 to 10 fmols) of BRCA1P1-lncRNA in a dose-dependent manner (lanes 3–5, Fig. 4E). The specificity of RelA's interaction with the motif was confirmed by further retardation in mobility (super-shift) resulting from interaction with anti-RelA antibodies (lane 6). These data further strengthen the deduction that BRCA1P1-lncRNA interferes with RelA's binding to target promoters.

Next, we generated truncated mutants of BRCA1P1-lncRNA and conducted RNA pull-down assays with recombinant RelA proteins (Fig. 4F and G). Whereas there was a strong association of BRCA1P1-lncRNA (1–1107 nt) with RelA (lane 1, Fig. 4G), deletion of exon 1a in BRCA1P1-lncRNA strongly decreased its interaction with RelA (lanes 3 and 4). In contrast, deletion of exon 1b and intron 1a had only a minor effect on the interaction (lane 2). These results suggest that exon 1a is important for the interaction of BRCA1P1-lncRNA with RelA. An RNA structure prediction showed that exon 1a forms a long stretch of small hairpin loops, which might play a role in its interaction with RelA (Supplementary Fig. S7A), although this remains to be determined. Collectively, these data showed that BRCA1P1-lncRNA physically interacts with RelA in part through exon 1a, inhibits NF-κB binding to promoters, and negatively regulates target gene expression.

Finally, we determined whether depletion of NF-κB abrogates BRCA1P1 loss-driven TNF expression (Fig. 4H). Depletion of BRCA1P1 increased TNF expression (a 57.8 ± 13.9-fold increase) in control siRNA-treated cells, which was reduced to a 23.2 ± 8.9-fold increase in RelA siRNA-treated cells. The expression of IFNL1 was also reduced by RelA silencing (a 43.9 ± 9.2 vs.62.6 ± 6.2-fold increase). Collectively, the data reinforce the observation that BRCA1P1 regulates expression of TNF and IFNL1 partly through NF-κB.

A radical defense mechanism to restrict the spread of a viral infection is apoptosis of the infected cell. We found that BRCA1P1-deficient cells are apoptotic, with an increased frequency of caspase-3/7–positive cells in BRCA1P1-KO and BRCA1P1-ASO–treated cells compared with the WT cells and control-ASO–treated cells, respectively (left, Fig. 5A and B). The increases in apoptosis appeared to restrict the proliferation of these cells, demonstrated by clear decreases in cell density of BRCA1P1-KO cells and BRCA1P1-ASO–treated cells, compared with the respective control cells (right, Fig. 5A and B). To determine whether the observed phenotypes (apoptosis and proliferation) are driven by BRCA1P1 specifically, we re-expressed BRCA1P1 in BRCA1P1-ASO–treated cells using CRISPRa (Fig. 5B). Re-expression of BRCA1P1 reduced apoptosis and restored proliferation of BRCA1P1-ASO–treated cells to the same levels as those observed in control-ASO–treated cells, demonstrating the specificity of BRCA1P1-driven regulation of apoptosis and proliferation in breast cancer cells. Interestingly, we did not observe any change in cell proliferation or apoptosis driven by BRCA1P1 depletion in HMEC cells (Fig. 5C), indicating that BRCA1P1 regulates apoptosis in cancer cells but not in normal breast cells.

We next tested whether cytokines induced apoptosis in BRCA1P1-KO cells. This is based on our data showing increased cytokine expression in BRCA1P1-KO cells (Fig. 2D and E), as well as previous literature reporting the pivotal role of TNFα in the regulation of apoptosis (23, 24). We treated BRCA1P1-KO and WT cells with poly(I:C) or TNFα, and examined the effects on apoptosis and cytokine expression. Treatment of cells with poly(I:C) induced higher TNF expression (Fig. 3D) and more severe apoptosis in BRCA1P1-KO cells than it did in control cells (Fig. 5D). Treatment with TNFα clearly increased apoptosis of BRCA1P1-KO cells with an average 3.9-fold increase compared with WT cells at 52 hours after treatment (Fig. 5E). The increase in apoptosis was accompanied by strongly elevated expression of TNF and IFNL1 in BRCA1P1-KO cells (Fig. 5F).

Because previous reports showed that TNFα induces apoptosis through reactive oxygen species (ROS) and DNA damage (23–27), we determined whether apoptosis driven by BRCA1P1 depletion is partly due to accumulation of ROS and DNA damage. In line with this, we observed significant increases in ROS and H2AX phosphorylation (γH2AX), a well-known DNA damage marker, in BRCA1P1-KO cells compared with the WT cells (Fig. 5G and H). Our data indicate that TNFα induces apoptosis of BRCA1P1-KO cells, partly through accumulation of ROS and DNA damage. Furthermore, we observed that BRCA1P1-depleted cells were more sensitive to genotoxic drugs, with increased apoptosis after doxorubicin and camptothecin treatment in BRCA1P1-KO cells (Fig. 5I), compared with the WT cells. Together, our data suggest that BRCA1P1 depletion increases apoptosis and sensitizes breast cancer cells to genotoxic drug treatments.

To determine the physiological role of BRCA1P1 in regulating tumor growth and antitumor immunity in vivo, we injected MDA-MB-231 cells of either the BRCA1P1 WT or KO genotype into the mammary glands of female athymic nude mice [(Crl:NU(NCr)-Foxn1nu]. In control mice that were injected with BRCA1P1-WT cells (n = 8), tumor volumes increased over time, up to a volume of 494.9 ± 241.4 mm³ at day 54 after injection. In contrast, tumor volumes in mice with BRCA1P1-KO cells (n = 9) remained below 50 mm³ for all mice throughout the study. Tumors for this group of mice were below measurable limits after day 30, except for one mouse with a tumor that reached 42 mm³ by day 54 (Fig. 6A). Because our in vitro studies showed increased cytokine expression in BRCA1P1-depleted cells compared with WT cells (Fig. 2D and E), we also assessed immune responses regulated by BRCA1P1 in our xenograft mouse model. Because athymic nude mice lack T cells, we measured cytokine expression in spleen tissues harvested from mice with BRCA1P1-WT and BRCA1P1-KO tumors. The transcript amounts of the cytokines TNFα, IL2, IFNγ, and IL12a in the spleen were significantly higher in mice with BRCA1P1-KO tumors compared with control mice with WT tumors (Fig. 6B). Collectively, these data suggest that BRCA1P1 depletion suppresses tumor growth partly through stimulation of local immunity, ultimately leading to reduced proliferation and enhanced apoptosis of cancer cells.

Innate immune defense pathways are critical for antitumor responses and the induction of apoptosis of cancer cells (1). Although the importance of antiviral mechanism is well known in cancer, the roles of host RNAs in antiviral defense mechanism have just begun to be elucidated. In this study, we discovered an immunogenic role for BRCA1P1-lncRNA in regulating antiviral defense mechanisms in breast cancer cells. Spread of a viral infection is restricted by three major defense mechanisms (22). The first involves the upregulation and activation of a set of antiviral proteins, which allow for virus sensing and attenuation of virus replication. The second, more drastic mechanism is the apoptosis of infected cells. The third is based on paracrine IFN signaling, which alerts noninfected cells. BRCA1P1 is potentially involved in all three mechanisms. BRCA1P1-deficient cells were prone to apoptosis and highly expressed antiviral immune genes, including IFNs, which are typical markers of virus-infected cells. Depletion of BRCA1P1 suppressed tumor growth and stimulated local immunity in a breast cancer xenograft mouse model. Our findings thus reveal a novel mechanism of regulation of antiviral responses by a host pseudogene RNA in breast cancer cells.

Mechanistically, BRCA1P1-lncRNA binds the NF-κB subunit RelA, inhibits the activity of RelA at its target promoters, and thereby negatively regulates transcription of antiviral genes. It is of note that BRCA1P1-lncRNA appears to interact with RelA irrespective of whether RelA dissociates (Fig. 4E–G) or associates with chromatin (Fig. 4B; Supplementary Fig. S7B). It has been reported that the NF-κB complex binds chromatin and recruits a chromatin-modifying complex (28–30). Specifically, the RelA subunit of NF-κB was reported to interact with histone deacetylases on chromatin (30). Thus, when BRCA1P1-lncRNA associates with RelA, it is likely to interact with histones indirectly through chromatin-bound RelA. In support of this, the level of BRCA1P1 enrichment on H2B (0.59 ± 0.02, relative unit) or H3 (0.47 ± 0.03) is similar to that of RelA (0.54 ± 0.08; Fig. 4B; Supplementary Fig. S7B). Considering the vast numbers of nucleosome-associated histones on every chromosome, the data indicate that only a very small portion of histones associates with BRCA1P1-lncRNA, likely through RelA bound on chromatin. Although further investigation will be required to determine the detailed mechanisms of BRCA1P1-regulated RelA activity, our data revealed a strong association of BRCA1P1-lncRNA with RelA. Interestingly, recent discoveries showed that cellular noncoding transcripts bind to RIG-I and/or its activator, the E3 ligase TRIM25, and promote RIG-I–mediated antiviral innate immune responses (8, 31, 32). Therefore, it is conceivable that there may be additional mechanisms through which BRCA1P1 activates antiviral signaling, such as a direct effect of BRCA1P1-lncRNA on immunostimulatory dsRNAs or their sensors, or by regulating the activity of dsRNA-activated kinases, which warrants future investigation.

Our data also showed that BRCA1P1 depletion stimulates the induction of proinflammatory cytokines in the spleen of our mouse model, suggesting an important role for BRCA1P1 in regulating local immunity. However, because athymic nude mice lack T cells, we were unable to evaluate the effects of BRCA1P1-deficiency on T cells or other immune cells. Therefore, future studies using humanized mice will be needed to fully understand the role of BRCA1P1-lncRNA in modulating local immunity and the tumor microenvironment. Along these lines, as NF-κB plays multiple roles in innate immunity and pro-inflammatory signaling, further investigation will be required to examine how NF-κB regulates BRCA1P1 depletion-driven antiviral immunity in a context-dependent manner. It also remains to be determined whether the interaction of BRCA1P1-lncRNA with RelA influences the RelA-mediated transcription of apoptosis genes, which may contribute to increased apoptosis in BRCA1P1-KO cells. In addition, future studies on the detailed mechanisms of how BRCA1P1 expression is regulated by upstream signals may increase our understanding of the regulatory circuits in BRCA1P1-mediated responses.

Despite some limitations, our study revealed a novel mechanism of innate immunity that is regulated by a pseudogene lncRNA. Regulation of innate immunity by the BRCA1P1 pseudogene might open up new possibilities for multilayered immunotherapy for cancer control (33). Tumor immunity and immunotherapy have become increasingly important in treatment strategies for a variety of tumors. They also point to etiology and progression pathways that could be targeted for prevention. Clinical trials evaluating the safety and efficacy of intratumoral injection of STING agonists, TLR agonists, or poly(I:C) derivatives with innate immune modulators are currently completed or are ongoing (34–37). Oncolytic viruses have also emerged as important agents in cancer treatment as they offer the attractive therapeutic combination of tumor-specific cell lysis together with immune stimulation (33, 38). Considering the importance of innate immunity and antiviral sensing in tumor immunity, our findings on the regulation of innate immunity by BRCA1P1 may have clinical relevance in the development of immunotherapies and should be further explored.

J. Zhang reports grants from NCI P20CA233307, as well as other funding from Breast Cancer Research Foundation, Kiphart Global Health Research Fund, and grants from NIH R01 AI087846 during the conduct of the study. O. Karginova reports grants from Breast Cancer Research Foundation during the conduct of the study. Q. Hao reports grants from University of Illinois at Urbana-Champaign during the conduct of the study. A. Prat reports personal fees from Nanostring Technologies, Astrazeneca, Roche, Novartis, Foundation Medicine, BMS, Daiichi Sankyo, Oncolytics Biotech, and Guardant Health outside the submitted work. K.V. Prasanth reports grants from NIA/NIH, NSF/EAGER, and NIGMS/NIH during the conduct of the study. O.I. Olopade reports other funding from CancerIQ, Tempus, and 54gene outside the submitted work. No other disclosures were reported.

Y.J. Han: Conceptualization, data curation, formal analysis, supervision, funding acquisition, validation, investigation, visualization, methodology, writing–original draft, project administration. J. Zhang: Formal analysis, validation, investigation, methodology, writing–review and editing. J.-H. Lee: Formal analysis, investigation, visualization, methodology. J.M. Mason: Formal analysis, investigation, visualization, methodology, writing–review and editing. O. Karginova: Formal analysis, investigation, visualization, writing–review and editing. T.F. Yoshimatsu: Formal analysis, investigation, methodology, writing–review and editing. Q. Hao: Formal analysis, investigation, visualization, methodology. I. Hurley: Investigation, methodology, writing–review and editing. L.P. Brunet: Formal analysis, investigation. A. Prat: Formal analysis, investigation. K.V. Prasanth: Formal analysis, funding acquisition, investigation, methodology, writing–review and editing. M.U. Gack: Conceptualization, data curation, formal analysis, supervision, funding acquisition, investigation, methodology, writing–review and editing. O.I. Olopade: Conceptualization, resources, data curation, supervision, funding acquisition, investigation, project administration, writing–review and editing.

We thank Drs. Fengyi Wan and Yifan Lei at the Department of Biochemistry and Molecular Biology, Johns Hopkins University, for their generous providing of truncated mutant constructs of RelA. We appreciate Lise Sveen's careful review on the article. We also thank Mariann Coyle and Maria Gomez-Vega for their assistance with cell culturing and fluorescence in situ hybridization. This work was supported by NCI P20CA233307 (to O.I. Olopade), Breast Cancer Research Foundation BCRF-20-071 (to O.I. Olopade and Y.J. Han), and the Kiphart Global Health Research Fund (to O.I. Olopade). This study was further supported in part by NIH R01 AI087846 (to M.U. Gack). The K.V. Prasanth laboratory is funded by grants from Cancer Center at Illinois seed grant and Prairie Dragon Paddlers, NIH R21AG065748, NIH R01GM132458, and NSF EAGER grant (MCB1723008). O.I. Olopade is an American Cancer Society Clinical Research Professor.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.