piR-1170 drives brain metastasis and immune evasion via WTAP-mediated m6A methylation reprogramming in triple-negative breast cancer

Abstract

Therapeutic target for triple-negative breast cancer (TNBC) brain metastases remains a critical unmet clinical challenge. The roles of PIWI-interacting RNAs (piRNAs) in driving brain metastasis are poorly understood, despite their known dysregulation and oncogenic functions in cancer. Here, we identified piR-1170 as a clinically relevant driver of TNBC brain metastasis through multi-model validation. Analysis of the TNBC cohort from Sun Yat-sen University Cancer Center revealed significant piR-1170 upregulation in brain metastases correlating with poor patient survival. First, upstream analysis confirmed that hnRNPK binds to piR-1170 to maintain the piRNA's stability, thereby sustaining piR-1170 upregulation in TNBC. Then, Functional studies with metastasis models demonstrated the brain-specific metastatic activity of piR-1170, enhancing tumor cell adhesion to brain endothelia, vascular extravasation, and parenchymal invasion. Mechanistically, piR-1170 promotes WTAP expression to enhance m6A methylation of DGAT2 and CD274 transcripts, activating de novo lipid synthesis and PD-L1-driven immune suppression to promote tumor adaptation to lipid-scarce metastases and avoid immune surveillance. Our study defined the piR-1170-driven axis that operates through coordinated metabolic reprogramming and immunosuppression, thus revealing its potential as a therapeutic candidate for TNBC brain metastasis.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12943-026-02568-y.

Introduction

Triple-negative breast cancer (TNBC) has been characterized by the absence of hormones and HER2 receptors, rendering endocrine and HER2-targeted therapies ineffective. And TNBC is highly prone to early recurrence and metastasis, particularly to the brain, and disproportionately contributes to breast cancer mortality [1]. Approximately 30% of metastatic TNBC patients develop brain metastases, which correlate with the poorest overall survival outcomes [2]. The treatment of TNBC brain metastases remains a critical therapeutic challenge, highlighting an urgent need for translating novel molecular targets into clinical applications. The blood-brain barrier (BBB) and blood-tumor barrier (BTB) impede effective drug delivery to brain metastases, exacerbating the challenges in treating TNBC brain metastases [3].

Beyond impeding drug delivery, the BBB/BTB restricts nutrient influx from circulation, creating a hypoxic microenvironment depleted of essential metabolites, growth factors, and fatty acids [4]. Consequently, the distinct nutrient landscape in brain tissue, compared to primary tumors, necessitates metabolic adaptations for cancer cell proliferation in the cerebral microenvironment. The lipid-restricted brain microenvironment induces a metabolic shift in brain-metastatic tumors, rendering them dependent on de novo fatty acid synthesis for survival and proliferation [5]. SREBP-1, a master transcriptional regulator of lipogenic enzymes including FASN, ACLY, and SCD1, is essential for breast cancer brain metastasis [6]. This finding substantiates the critical role of activated fatty acid synthesis in supporting breast cancer progression within the brain microenvironment.

PIWI-interacting RNAs (piRNAs) are a class of small noncoding RNAs with 23–31 nucleotides in length [7]. piRNAs are characterized by two distinct structural features: a 5’ monophosphate group and 2’-O-methylation at their 3’ terminus [8, 9]. These modifications confer piRNAs with specific and precise target recognition capabilities while ensuring exceptional structural stability, representing unique molecular signatures that distinguish piRNAs from other small RNA species [10]. Initially identified in animal germline cells, piRNA plays a crucial regulatory role in germline development processes [11]. Emerging evidence demonstrates that piRNAs exhibit dysregulated expression patterns across various cancers, with numerous studies identifying specific piRNAs implicated in tumorigenesis and cancer progression [12]. Specifically, protein-coding genes are regulated by piRNA through several mechanisms: (1) H3K9 trimethylation induced by piRNAs involved in TE silencing can spread to neighboring regions, leading to epigenetic silencing of nearby protein-coding genes [13]. (2) piRNAs direct Aub to reach nanos mRNAs (post-coding determinant clusters) and induce specific mRNAs de-adenylation and decay by interacting with the RNA-binding proteins Smaug and CCR4-NOT de-adenylation complexes [14]. (3) Aub loaded with piRNA induces polyadenylation by recruiting the cytoplasmic poly(A) polymerase Wispy, which stabilizes germ cell mRNA in the germ plasmid [15].

Cancer cells acquire metastatic competence through genetic and epigenetic alterations [16]. As the predominant eukaryotic mRNA modification, m6A dynamics, governed by its “writers,” “erasers,” and “readers”, regulate critical cellular processes and contribute significantly to pathological mechanisms, particularly in cancer development [17]. However, the involvement of m6A methylation in TNBC brain metastasis remains elusive.

To elucidate molecular mechanisms underlying TNBC brain metastasis, we conducted piRNA microarray sequencing comparing primary tumors with brain metastases. Differential expression analysis revealed significant upregulation of piR-1170 in brain metastases. Functional studies demonstrated that piR-1170 knockdown markedly suppressed brain metastasis in vivo. Mechanistically, piR-1170 regulates Wilms tumor 1-associated protein (WTAP), a key mediator of TNBC brain metastasis, through m6A-dependent modulation of target transcript translation. Furthermore, piR-1170-WTAP axis promotes intracerebral adaptation by enhancing DGAT2-mediated fatty acid synthesis and CD274-driven immune evasion.

Materials and methods

Study cohorts

Breast cancer brain metastasis (BCBM) specimens were obtained from clinically diagnosed patients undergoing neurosurgery at Sun Yat-sen University Cancer Center (SYSUCC, Guangzhou) between June 2021 and August 2023. Immediately after resection, tissues were immersed in RNAlater™ (Invitrogen) at 4 °C for 24 h, flash-frozen in liquid nitrogen, and maintained at −80 °C until RNA isolation. This study was approved by the Institutional Review Board of SYSUCC (No. 2021 − 358) and conducted in accordance with the Declaration of Helsinki principles.

RNA extraction, library Preparation and deep sequencing

Total RNA was extracted from breast cancer tissues and BCBM specimens using TRIzol Reagent (Invitrogen, cat. NO 15596026)following the methods by Chomczynski et al. [18]. Following extraction, residual genomic DNA was removed by DNase I treatment. RNA purity was assessed by measuring the A260/A280 ratio on a NanoDrop™ One spectrophotometer (Thermo Fisher Scientific). RNA integrity was verified by 1.5% agarose gel electrophoresis. The concentration of qualified RNA samples was accurately determined using a Qubit™ 3.0 Fluorometer with the Qubit™ RNA Broad-Range Assay Kit (Life Technologies, Q10210).

For piRNA library construction, 3 µL of total RNA was used as input with the KC-Digital™ small RNA Library Prep Kit for Illumina^® (Wuhan Seqhealth Co., Ltd., Cat. No. DR08602), following the manufacturer’s protocol. To mitigate duplication bias during PCR amplification and sequencing, the kit employs an 8-base random unique molecular identifier (UMI) to label each pre-amplified small RNA molecule. The eluted cDNA library was size-fractionated on a 6% polyacrylamide gel. DNA fragments approximately 160 bp in length were excised, purified, and quantified using the Qubit™ 3.0 system. Finally, paired-end sequencing (PE150) was performed on an Illumina HiSeq X Ten platform.

Data analysis

Raw sequencing reads were initially processed with fastp (v0.23.2) to trim adapter sequences and remove low-quality reads. To eliminate potential amplification and sequencing duplication biases, clean reads were subjected to an in-house UMI-based deduplication pipeline. Briefly, reads were clustered by their UMI sequences. Within each UMI cluster, pairwise alignment was performed, and reads exhibiting 100% sequence identity were grouped into a sub-cluster, ensuring that only unique RNA molecules were retained. After deduplication, reads with lengths between 18 and 32 nucleotides were selected for subsequent analysis. piRNA identification and quantification were performed by aligning these reads to known and de novo predicted piRNA sequences in the piRNAdb database (https://www.pirnadb.org). Following initial annotation, all piRNA hits were cross-referenced and updated to their corresponding unique identifiers in the piRBase database (http://bigdata.ibp.ac.cn/piRBase/index.php) for subsequent analysis.

Tissue dissociation, organoid culture and passage

Fresh tissues from breast cancer and brain metastasis specimens were dissociated by mincing into 1–3 mm³ fragments in minimal digestion buffer: DMEM/F12 medium (GIBCO) supplemented with 300 U mL^− 1 collagenase type III (Worthington), 100 U mL^− 1 hyaluronidase (Sigma Aldrich), 5% FBS (GIBCO), 5 µg mL^− 1 insulin (Sigma), 500 ng mL^− 1 hydrocortisone (Sigma Aldrich), 10 ng mL^− 1 EGF (PeproTech), and 20 ng mL^− 1 cholera toxin (Sigma Aldrich). According to the size of tissues, 6–12 mL digestion buffer was added to the tissue pieces. The tissues were digested on a shaker at 37 °C for 1–2 h with occasional pipetting until the visible pieces disappeared.

Following dissociation, cell clusters were pelleted by centrifugation (400×g, 4 min), washed once with DMEM/10% FBS, and centrifuged again (400×g, 3 min). The pellet was resuspended in ice-cold 70% Matrigel (Corning) and plated as 20 µL droplets in pre-warmed 6-well plates (Corning). After 30 min of solidification at 37 °C under 5% CO2, 2.5 mL organoid culture medium was added per well and replenished every 2–3 days. The previously established culture conditions were used with slight modifications [19]. Briefly, the composition of the culture medium was Advanced DMEM/F12 medium supplemented with 30% Wnt3A conditional medium, 10% R-spondin 1 conditional medium, 10% Noggin conditional medium, 5 × 10^{− 6} M Y‐27,632, 0.5 × 10^{− 6} M SB202190, 0.5 × 10^{− 6} M A83‐01, 5 ng mL^{− 1} EGF, 5 × 10^{− 9} M neuregulin‐1, 500 ng mL^{− 1} hydrocortisone, 1.25 × 10^{− 3} M N‐acetyl‐L‐cysteine, 15 × 10^{− 3} M HEPES, 1× B27, 1× Glutamax, 5 × 10^{− 9} M β‐estradiol, 1× insulin‐transferrin‐selenium‐sodium pyruvate, 0.5 µg mL^{− 1} amphotericin B, 5 µg mL^{− 1} gentamicin, and 5 µg mL^{− 1} plasmocin.

Organoids were passaged every 1–3 weeks depending on confluency and morphology. Following medium aspiration, clusters were dissociated in ice-cold 0.25% Trypsin-EDTA (Gibco), with trituration and incubated at 37 °C for 4 min. Mechanical dissociation was repeated if necessary. Cells were then centrifuged (400×g, 4 min), washed with 6 mL DMEM, and pelleted again. Resuspended in fresh Matrigel at a 1:3 split ratio, organoids were replated and maintained under standard culture conditions.

Public data mining

piRNA expression data were obtained from piRNAQuest V.2 (http://dibresources.jcbose.ac.in/zhumur/pirnaquest2) and Gene Expression Omnibus (GEO, https://www.ncbi.nlm.nih.gov/geo/) and then under a series of independent bioinformatic analyses. piRNA-mRNA binding interactions were analyzed by publicly available algorithms. piRNA-mRNA binding interactions were analyzed by ENCORI/starBase (https://rnasysu.com/encori/).

RNA isolation and real-time quantitative PCR analysis

Total RNA was isolated from all clinical specimens and cell lines using TRIzol Reagent (Invitrogen), according to the manufacturer’s protocol. For reverse transcription, mRNA was converted to cDNA using the PrimeScript RT Master Mix Kit (RR036A, Takara Bio), while small non-coding RNAs (including piRNAs and miRNAs) were reverse transcribed using the PrimeScript RT reagent Kit (RR037A, Takara Bio). Quantitative PCR (qPCR) was performed in triplicate on a Bio-Rad CFX96 system using TB Green Premix Ex Taq II (RR420A, Takara Bio).

A critical and validated normalization strategy was employed. For piRNA and miRNA quantification, expression levels were normalized to the geometric mean of two small nuclear/nucleolar RNAs (U6 snRNA and RNU48), which were identified as the most stable reference genes across all sample sets using the geNorm algorithm. For mRNA quantification, expression was normalized to the geometric mean of two classical reference genes (β-actin and GAPDH), whose stability was confirmed under our experimental conditions. The use of multiple, validated reference genes for each RNA class ensures robust and biologically appropriate normalization. Relative expression levels were calculated using the 2 − ΔΔCt method, with the assumption that all primer pairs exhibited amplification efficiencies close to 100%. All primer and stem-loop RT primer sequences are provided in Supplementary Table 1 [20].

Cell transfection

siRNAs and miRNA inhibitors/mimics were synthesized by GenePharma (Shanghai, China). Breast cancer cells or 293 T cells (RRID: CVCL_0063) were cultured in 6-well plates (2 × 10⁵/well) overnight and transfected with Lipofectamine 3000 (Invitrogen, CA, USA) following the manufacturer’s instructions [21]. The medium was changed after 8–12 h, and the cells were cultured for another 48–72 h. The sequences of the siRNAs and miRNA inhibitors/mimics used are listed in Supplementary Table 2.

piRNA-1170 Silencing by Antagomir treatment

Cholesterol-conjugated antagomirs, incorporating 2’-O-methyl (2’-OMe) and phosphorothioate backbone modifications, demonstrate enhanced metabolic stability and cellular uptake efficiency for robust in vivo gene silencing. In this study, chemically modified piR-1170 antagomir (anta-1170) was synthesized through cholesterol conjugation of vivo-optimized oligonucleotides (GenePharma), with a scrambled-sequence antagomir (anta-NC) serving as the negative control. These structural modifications conferred prolonged circulatory longevity and improved nuclease resistance, enabling effective piR-1170 suppression. The sequence of anta-1170 was as follows: 5’-AsGsAGCUAAUAGAAAGGCUAGGACCAAACCsUsAsUsChol-3’.

Overexpression of piR-1170 via a specially modified mimic

Based on the structural requirements of functional piRNAs described in the background section - specifically a 5’ monophosphate group and 2’-O-methylation at the 3’ terminus - we collaborated with RiboBio to synthesize the pure nucleotide sequence of piR-1170 with these essential modifications (designated as piR-1170 mimic^mod). This engineered construct was designed to maximally mimic the biological effects of endogenous piRNA overexpression.

Animals’ experiments

Four-week-old female BALB/c nude (RRID: MGI:2683685) mice were purchased from Beijing Vital River Laboratories Animal Technology (Beijing, China). All experimental procedures were approved by the Institutional Animal Care and Use Committee of Sun Yat-sen University and conducted in compliance with NIH guidelines. Detailed methodologies are provided in Supplementary Methods.

Heterogeneous cell–cell adhesion assay

Human brain microvascular endothelial (HBMEC), mouse cerebellar astrocytes C8-D1A were seeded in a 6 well plate and allowed to form a confluent monolayer for the appropriate time. Then, the tops of the monolayers were seeded with cancer cells labeled with CellTracker Green CMFDA (Yeasen Biotech, Shanghai, China). Following 10–15 min incubation, adherent cells were PBS-washed twice to remove nonadherent populations. Fluorescent-labeled tumor cells were quantified through microscopic imaging and cell counting per field.

Trans-BBB migration assay

C8-D1A cerebellar astrocytes (5 × 10⁵) were seeded on inverted transwell membranes. Following media renewal every 20 min for 6 h, bEnd.3 cells (2.5 × 10⁵) were plated on the upper surface. After 72 h incubation at 37 °C, BBB formation was confirmed. Serum-depleted transwells were PBS-washed and transferred to 24-well plates containing conditioned media (24 h serum-starved). CMFDA-labeled tumor cells (4 × 10⁴) were added to the upper chamber, with transmigration assessed after 18–22 h.

In vitro angiogenesis assay

Brain endothelial cells were used for the in vitro angiogenesis assay. 5 × 10⁴ cells were plated on matrigel with tumor-conditioned media and monitored for tube formation up to 12 h. Branch points per field were quantitated.

RNA-FISH and subcellular fractionation

piR-1170 spatial distribution in TNBC cells was mapped through RNA fluorescence in situ hybridization (FISH) using Cy5-conjugated probes (GenePharma, Shanghai, China), with nuclear counterstaining by DAPI (Solarbio, Beijing, China). Confocal imaging (ZEISS, Oberkochen, German) confirmed cytoplasmic/nuclear partitioning. Parallel cellular fractionation (PARIS Kit, Invitrogen, USA) enabled compartment-specific RNA quantification via RT-qPCR, validating predominant cytoplasmic localization of piR-1170.

RNA pulldown assay and mass spectrometry

The RNA pull-down procedure used the RNA pulldown kit (Thermo Fisher Scientific Inc, Waltham, MA, USA) according to the manufacturer’s instructions. Biotinylated probes for piR-1170 and control sequences were custom-designed and obtained from GenePharma (Shanghai, China). The sequences of the biotin-labeled piR-1170 and control probes are as follows: piR-1170 probe: AGAGCTAATA GAAAGGCTAG GACCAAACCTAT; control probe: UUGUACUACACAAAAGUACUG. Moreover, the specific strips originated from RNA pull-down assay were cut for mass spectrometry, which was performed at Fitgene Corporation (Guangzhou, China).

Dual-luciferase reporter assay

Putative piR-1170 binding motifs on WTAP mRNA were mapped using piRNA Quest V.2 (http://dibresources.jcbose.ac.in/zhumur/pirnaquest2/start.php). Wild-type (WT) and mutant (MUT) sequences of the WTAP 3’UTR were PCR-amplified and subcloned into pmirGLO dual-luciferase vectors. MDA-MB-231 (RRID: CVCL_0062) and SUM159PT cells (RRID: CVCL_5423) (1 × 10⁵/well) in 24-well plates were co-transfected with WT/MUT constructs and piR-1170 inhibitors/mimics or negative controls. Luciferase activity was quantified 48 h post-transfection, with firefly signals normalized to Renilla controls.

RNA Immunoprecipitation (RIP) and methylated RNA Immunoprecipitation (meRIP)

These experiments were performed by using BersinBio™ RNA Immunoprecipitation (RIP) kit (Catalog Bes5101, BersinBio, Guangzhou, China) and BersinBio™ Methylated RNA Immunoprecipitation (meRIP) Kit (Catalog Bes5203-2). TRIzol (Invitrogen, CA, USA) was used to extract coprecipitated RNA and RT-qPCR was performed to analyze it. Antibodies used for RIP are also listed in Supplementary Table 3.

Actinomycin D treatment

TNBC cells that had been transfected with negative control, piR-1170 inhibitor or mimic were counted, and the same number of cells were plated down in 6-well plates. And then, cells were treated with 5 µg/mL actinomycin D (D23070, Sigma-Aldrich, St. Louis, MO, USA) after cell adhesion.

Western blotting assay

Western blot analysis was conducted using standard SDS-PAGE protocols. Proteins were separated electrophoretically, transferred to PVDF membranes, and blocked with 5% non-fat milk in TBST for 1 h. Membranes were incubated overnight at 4 °C with primary antibodies, followed by 1 h incubation with HRP-conjugated secondary antibodies at room temperature. Signals were detected using an enhanced chemiluminescence (ECL) kit (Yeasen Biotech, Shanghai, China). Antibody details are provided in Supplementary Table 3.

RNA m6A Dot blot

Total RNA was extracted, quantified, and diluted to concentrations of 250 ng/µl and 500 ng/µl. Following denaturation at 95 °C for 5 min, 1 µl aliquots of each dilution were spotted onto positively charged nylon membranes (Roche, Basel, Switzerland). The membranes were UV crosslinked twice, blocked with 5% non-fat milk in TBST for 1 h, and then probed with m6A antibody (RRID: AB_2918796) at 4 °C overnight. After incubation with HRP-conjugated secondary antibody, immunoreactive signals were visualized using an ECL detection system. RNA loading was verified by methylene blue staining of the membrane.

Immunofluorescence (IF)

Cells were plated on chamber slides, fixed with 4% formaldehyde, and permeabilized with 1% Triton X-100. After blocking, cells were co-incubated with anti-β-catenin (Cell Signaling Technology Cat# 8480, RRID: AB_11127855) and anti-hnRNPK (Proteintech Cat# 11426-1-AP, RRID: AB_2264314) primary antibodies at 4 °C overnight, followed by sequential incubation with CoraLite594-conjugated goat anti-rabbit (red) (Proteintech Cat# SA00013-4, RRID: AB_2810984) and CoraLite488-conjugated goat anti-mouse (green) (Proteintech Cat# SA00013-1, RRID: AB_2810983) secondary antibodies. Nuclear counterstaining was performed using DAPI. Antibodies used for immunofluorescence are also listed in Supplementary Table 3.

BODIPY in vitro staining

After transfection as describe, MDA-MB-231(RRID: CVCL_0062) and SUM159PT(RRID: CVCL_5423) cells were seeded at on chamber slides. Following treatment, cells were fixed in 4% paraformaldehyde for 20 min, washed three times in PBS, and incubated in PBS with BODIPY 493/503 (Beyotime Biotechnology, Shanghai, China) and DAPI for 10 min at room temperature. Slides were washed three times in PBS, and mounted with an anti-fade mounting medium. Three randomly selected visual fields per slides were photographed (63 × magnification) using a high-resolution laser scanning confocal microscope LSM880 (ZEISS, Oberkochen, Germany).

Immunohistochemistry (IHC)

Paraffin-embedded tissue sections were baked at 65 °C for 2 h, dewaxed in xylene, and rehydrated through an ethanol gradient. Antigen retrieval was performed by sequential treatment with sodium citrate buffer (pH 6.0) at 95 °C for 30 min and EDTA buffer (pH 9.5) using a pressure cooker for 10 min. After cooling to room temperature, endogenous peroxidase activity was blocked for 10 min, followed by PBS washing and blocking with normal goat serum at 37 °C for 30 min. Sections were incubated with primary antibody at 4 °C overnight, then with HRP-conjugated secondary antibody for 1 h at room temperature. Immunoreactivity was visualized using DAB substrate, followed by hematoxylin counterstaining and ethanol dehydration. All immunohistochemical reagents were obtained from ZSBIO (Beijing, China). Semiquantitative analysis was performed by multiplying the percentage of positive cells (0–4 scale) by staining intensity (1–3 scale) [22].

Statistical analyses

Statistical analyses were performed using R (version 4.3.3) and GraphPad Prism (version 8.0). Intergroup comparisons were conducted using Student’s t-test (two groups) or one-way ANOVA (multiple groups). Correlation analyses were performed using Spearman’s rank test. Data are presented as mean ± SEM from three independent experiments, with statistical significance denoted as *p < 0.05, **p < 0.01, ***p < 0.001, and ****p < 0.0001; ns indicates not significant (p > 0.05).

Results

Brain-specific piR-1170 expression is elevated in breast cancer brain metastases

We initiated our investigation by profiling piRNA expression in paired primary tumors and brain metastases from three TNBC patients at Sun Yat-sen University Cancer Center (SYSUCC) (Fig. 1A and Supplementary Table 4). Integrating these data with existing piRNA sequencing datasets from matched tumor-normal tissue pairs, we identified consistently dysregulated piRNAs across both cohorts [23] (Fig. 1B and Supplementary Table 5). Notably, piR-1170 (DQ571526, also named hsa_piR_001170, piR-hsa-1849, hsa-piR-1614, or hsa_piRNA_30433) emerged among the top ten significantly differentially expressed piRNAs in both datasets. Subsequent validation using GEO datasets, clinical specimens, and cell line models confirmed piR-1170 upregulation in breast cancer, with particularly pronounced expression in TNBC (Fig. S1A-C). To assess the clinical relevance of piR-1170, we performed fluorescence in situ hybridization (FISH) analysis on a tissue microarray comprising 193 TNBC specimens from SYSUCC. Survival analysis revealed that elevated piR-1170 expression significantly correlated with poorer patient outcomes (Fig. 1C-D). Pan-cancer analysis revealed predominant piR-1170 expression in breast cancer, indicating its potential functional significance in breast cancer (Fig. S1D). Notably, piR-1170 exhibited predominant expression in brain tissue among normal organs, suggesting its potential involvement in brain metastasis through pre-metastatic niche formation (Fig. S1E) [24]. We further established organoid biobanks derived from breast cancer primary lesions and brain metastases specimens from SYSUCC, and we observed heterogeneous organoid morphologies, including variably sized solid structures, cystic formations, clusters resembling grape-like architecture, and near-complete discohesive aggregates (Fig. 1E-F). Furthermore, comparative FISH analysis revealed consistently elevated expression of piR-1170 in brain metastasis-derived breast cancer organoids compared to primary tumor-derived counterparts (Fig. 1E-F). To elucidate the organotrophic metastasis pattern mediated by piR-1170, we established a spontaneous metastasis model through orthotopic implantation of MDA-MB-231 cells in BALB/c nude mice (Fig. 1G). FISH staining of metastatic lesions revealed brain-specific piR-1170 upregulation, with significantly higher expression in brain metastases compared to liver or pulmonary metastases (Fig. 1H-I). These findings provide initial evidence supporting the functional significance of brain-specific piR-1170 in breast cancer brain metastasis.

Fig. 1.

Brain-specific piR-1170 is highly expressed in breast cancer brain metastases. A Heatmap of top 10 differentially expressed piRNAs expression profiling in paired primary tumors and brain metastases derived from TNBC patients. B Heatmap illustrating the top 10 differentially expressed piRNAs in paired primary TNBC tumors and adjacent normal tissues. C Representative tissue microarray (TMA) images showing piR-1170 expression levels in 193 TNBC patient samples. D Kaplan-Meier analysis of overall survival (OS) in TNBC patients stratified by piR-1170 expression levels. E-F Bright-field images (top row) depicting brain metastasis-derived BC organoids (E) and primary tumor-derived BC organoid (F) phenotypes. Corresponding fluorescence imaging (the bottom row) demonstrating the expression of piR-1170 in brain metastasis-derived BC organoids (E) and primary tumor-derived BC organoid (F). G The pattern diagram for building a model of spontaneous metastasis of TNBC tumors in vivo. H Quantitative analysis of piR-1170 expression levels in metastatic sites from the spontaneous metastasis model (n = 10 per group). Brain metastasis (BM), liver metastasis (LM), and pulmonary metastasis (PM) in the studied samples. I Representative images of piR-1170 detection in different metastases from the spontaneous metastasis mice model

piR-1170 facilitates brain metastasis through enhanced endothelial adhesion, extravasation, and invasion of TNBC cells

To further explore the functional role of piR-1170 in TNBC brain metastasis, we depleted piR-1170 using antagomir-1170 (anta-1170) and overexpressed it using a synthetic mimic engineered with a 5’-monophosphate and 3’-terminal 2’-O-methylation, recapitulating the signature structural motifs of endogenous piRNA in 231-BrM and 159-BrM cells (generated using the selection procedure as previously [25]). Quantitative reverse transcription PCR was used to detect the efficiency of depletion and overexpression after transfection (Fig. S1F). And colony-forming capacity of 231-BrM and 159-BrM cells remained unchanged following piR-1170 knockdown or overexpression (Fig. S1G). Similarly, neither silencing nor forced expression of piR-1170 induced changes in proliferation (Fig. S1H). To determine the role of piR-1170 in brain metastasis, we established experimental brain metastasis models by performing intracardiac injections in BALB/c nude mice. Mice were injected with 231-BrM or 159-BrM cells that had been pretreated with either the piR-1170 antagonist (anta-1170) or a negative control antagonist (anta-NC) (Fig. 2A). Bioluminescence imaging (BLI) at 10 min post-injection confirmed the uniformity of cell delivery via left ventricular injection and successful model establishment. Subsequent imaging at 14, 21, and 28 days revealed that suppression of piR-1170 significantly attenuated brain metastatic burden (Fig. 2B), as quantified by luciferase activity (Fig. 2C). HE staining further confirmed reduced brain metastasis formation in piR-1170-deficient groups (Fig. 2D). The presence of GFAP-positive reactive astrocytes, a hallmark of brain metastasis, was significantly reduced in metastases derived from piR-1170-knockdown cells compared to controls (Fig. 2E). Given the crucial role of angiogenesis in metastasis, we assessed vascular density in xenograft-derived brain metastases. piR-1170 depletion significantly reduced microvessel density compared to control metastases (Fig. 2F).

Fig. 2.

Upregulated piR-1170 facilitates TNBC brain metastasis. A Schematic of the brain metastasis model establishment in BALB/c nude mice. B-C In vivo bioluminescence imaging (B) and corresponding fluorescence intensities (C) at 10 min, 14 days, 21 days and 28 days post-implantation (n = 6 per group). D Representative images of HE staining in brain metastatic lesion formed by anta-NC and anta-1170 cells. E Tissues from the brain-metastatic lesion were used for IF staining to observe cancer cell-astrocyte interaction. Tumor cells, astrocyte cells and nuclei were stained with pan-CK (green), GFAP (red) or DAPI (blue), respectively. F Representative images of blood vessel densities in brain metastases formed by anta-NC and anta-1170 cells. Vascular endothelial cells and nuclei were stained with CD31 (red) or DAPI (blue), respectively. G-H In vivo bioluminescence imaging (G) and corresponding fluorescence intensities (H) at 0–72 h post-intracardiac injection (n = 6 per group). I Tumor cell extravasation analysis at 48 h post-injection, showing CMFDA-labeled tumor cells (green), CD31 + vasculature (red), and DAPI + nuclei (blue), with quantitative extravasation data. J Quantitative analysis of tumor cell adhesion to astrocyte monolayers after alteration of piR-1170 in vitro. K Quantitative assessment of tumor cell adhesion to human brain microvascular endothelial cells (HBMECs) after alteration of piR-1170 in vitro. The error bars represent the SDs of triplicate experiments

Brain metastasis formation involves sequential steps: circulatory survival, capillary arrest, extravasation, and proliferative colonization [26]. To delineate piR-1170’s role in this cascade, we performed intracranial implantation and observed that piR-1170 suppression significantly impaired the localization of 231-BrM and 159-BrM cells within the brain parenchyma at 24 h post-injection compared to controls (Fig. 2G-H). We next assessed piR-1170’s impact on transendothelial migration using CMFDA-labeled 231-BrM and 159-BrM cells. And we found that fewer tumor cells in the piR-1170 knockdown group showed signs of extravasation within 48 h (Fig. 2I). To investigate piR-1170-mediated tumor-stroma interactions, we evaluated its role in tumor cell adhesion to astrocytes and HBMECs in vitro. piR-1170 knockdown significantly impaired 231-BrM cells adhesion to both astrocyte (Fig. 2J) and HBMEC monolayers, whereas overexpression of piR-1170 enhanced 231-BrM cells adhesion to both cell types (Fig. 2K). These findings align with our in vivo observations, demonstrating that piR-1170 suppression attenuates early tumor cell adhesion, angiogenesis, and reactive astrocyte activation in brain metastases. This can be attributed, at least in part, to the fact that knockdown of piR-1170 reduced the interaction of tumor cells with astrocytes and cerebrovascular endothelial cells.

To test if the 231-BrM and 159-BrM cells had a piR-1170-dependent survival advantage in the brain, we utilized stereotactic injection to implant TNBC cells intracranially (Fig. 3A). piR-1170 knockdown significantly suppressed intracranial tumor growth (Fig. 3B-D). Subsequently, we established organoid cultures from dissociated brain metastasis tissues. We further demonstrated that tumor/astrocyte cell intermixture was preserved in spheroids by immunofluorescence microscopy (Fig. S2A). The growth of brain metastasis-derived organoids was markedly inhibited in the piR-1170 knockdown group under three-dimensional (3D) culture conditions. Notably, the BrM group exhibited a predominantly invasive clonal morphology, in contrast to the piR-1170 knockdown group, which was primarily characterized by dormant clones — an observation further confirmed by Ki67 immunostaining (Fig. 3E-F and Fig. S2B). Angiogenesis represents a crucial process in metastatic lesion development [27]. To elucidate the role of piR-1170 in this process, we assessed its impact on angiogenesis. Conditioned medium from piR-1170-knockdown 231-BrM and 159-BrM cells significantly impaired brain endothelial tube formation compared to control medium, whereas conditioned medium from piR-1170-overexpressing 231-BrM and 159-BrM cells promoted cerebrovascular angiogenesis (Fig. 3G). Furthermore, employing an in vitro blood-brain barrier (BBB) model [28] (Fig. 3H), we demonstrated that piR-1170 knockdown markedly attenuated 231-BrM and 159-BrM cells transmigration across the BBB, whereas piR-1170 overexpression markedly enhanced this process (Fig. 3I). And transwell assay (Fig. 3J) and the wound healing assay (Fig. 3K) consistently revealed that piR-1170 potentiates 231-BrM and 159-BrM cells migration and invasion. These comprehensive in vitro findings substantiate that piR-1170 facilitates TNBC brain metastasis through multifaceted regulation of metastatic processes.

Fig. 3.

piR-1170 confers survival advantage in TNBC brain metastasis. A Schematic of intracranial injection-based brain metastasis model. B-C In vivo bioluminescence imaging (B) and corresponding fluorescence intensities (C) at 1 day, 3 days, 7 days and 11 days post-implantation (n = 6 per group). D Representative HE staining images of metastatic foci in murine brain sections across experimental groups. E Representative bright-field images of brain metastasis-derived organoids from anta-NC and anta-1170 groups. F 3D confocal reconstruction of brain metastasis organoids stained with pan-CK (green, tumor cells) and Ki67 (red, proliferative cells). G Representative images and quantitation of HBMEC tube formation assay using conditioned media from experimental groups. H Schematic of in vitro blood-brain barrier (BBB) transendothelial migration assay. I Representative images (left) and statistical analysis (right) of migrated tumor cells in different groups. Tumor cells were labeled with CMFDA (green) and red arrows point to migrated tumor cells. J Representative images (left) and comparative analysis (right) of TNBC cell migratory capacity between groups. K Representative images (left) and quantitative analyses (right) of the migration area in the scratch wound assay. The error bars represent the SDs of triplicate experiments

hnRNPK-mediated piR-1170 sustenance promotes TNBC progression

To elucidate the molecular mechanism underlying piR-1170 upregulation in TNBC, we performed RNA pull-down assays followed by silver staining and mass spectrometry analysis using biotin-labeled piR-1170 and control probes (Fig. 4A). Comparative analysis revealed hnRNPK as the predominant RNA-binding protein, exhibiting the highest match score and protein abundance in the 55 kDa band, suggesting its potential role in regulating piR-1170 expression (Fig. 4B-C and Fig. S3A). The 3D structures of piR-1170 and hnRNPK were obtained from RNAfold web server and the Protein Data Bank to further investigate the interaction between them. The molecular docking between piR-1170 and hnRNPK was analyzed using HDOCK [29], which predicted multiple high-confidence binding models for piR-1170 with hnRNPK (Fig. 4D). The Western blotting analysis of proteins captured by biotin-labeled piR-1170 probes in RNA pull-down assays demonstrated direct interaction with hnRNPK (Fig. 4E), which was reciprocally confirmed through RNA immunoprecipitation (RIP) experiments (Fig. 4F and Fig. S3B). To investigate their spatial relationship, we first localized piR-1170 primarily in the perinuclear region through RNA-FISH (Fig. S3C) and subcellular fractionation assays (Fig. 4G). Subsequent confocal microscopy revealed cytoplasmic co-localization of hnRNPK and piR-1170 in the perinuclear region (Fig. 4H). Notably, while modulation of piR-1170 levels did not affect hnRNPK expression at either mRNA or protein levels (Fig. S3D), hnRNPK knockdown significantly reduced piR-1170 expression, implying the predominant role of hnRNPK in the regulation of piR-1170 expression (Fig. 4I). The regulatory role of hnRNPK was further corroborated by accelerated degradation of piR-1170 following actinomycin D treatment in hnRNPK‑depleted cells (Fig. 4J). Moreover, immunohistochemical analysis of matched primary tumors and brain metastases revealed consistently higher expression of hnRNPK in brain metastatic lesions (Fig. S3E). Comparative assessment of hnRNPK expression across breast cancer metastases from different organs further demonstrated that its level was highest in brain metastases, supporting the notion that hnRNPK upregulates piR‑1170 specifically in the brain microenvironment (Fig. S3F). Collectively, these findings establish hnRNPK as a critical regulator of piR‑1170 expression during TNBC brain metastasis.

Fig. 4.

hnRNPK stabilizes piR-1170 through direct interaction. A Schematic showing the RNA pulldown assay using a biotinylated piR-1170 probe. B Silver staining image of the RNA pulldown assay with biotinylated piR-1170 and the control probe. C The hnRNPK peptide fragment was identified by mass spectrometry (MS). D The interaction between the piR-1170 and hnRNPK proteins was predicted using the HDOCK server. The piR-1170 secondary structure was predicted with the RNAfold web server according to the minimum free energy. The 3D structure of hnRNPK was obtained from the Protein Data Bank. E Detection of hnRNPK in RNA pull-down complexes by Western blot. Input lysate (10% load). F RNA Immunoprecipitation (RIP)-qPCR analysis of piR-1170 enrichment by hnRNPK. G Subcellular distribution of piR-1170 analyzed by qPCR following nuclear-cytoplasmic fractionation. H Representative confocal images of dual fluorescence in situ hybridization (FISH) revealing spatial co-localization of hnRNPK (green) and piR-1170 (red). I qRT-PCR detecting the expressions of piR-1170 under hnRNPK knockdown in TNBC cells. J RNA stability assay demonstrating hnRNPK-dependent piR-1170 maintenance under actinomycin D treatment. The error bars represent the SDs of triplicate experiments

piR-1170 stabilizes WTAP mRNA through competitive sequestration of targeting miRNAs

Guided by the established principle that piRNA-mRNA interactions require strict nucleotide complementarity at positions 2–11 of the 5’ end [30], we identified 108 candidate genes whose mRNAs may be able to interact with piR-1170 through gene-wide transcriptome comparisons (Supplementary Table 6). Subsequently, we employed piRNAQuest V.2 to quantitatively assess piRNA binding potential among candidate genes [31]. Finally, we selected the ten genes with the highest probability of binding to piR-1170 and found that only WTAP showed a corresponding decrease or increase in mRNA level after knockdown or overexpression of piR-1170 (Fig. 5A-B and Fig. S4A), with parallel changes observed at the protein level (Fig. 5C). RNA decay assays showed that overexpression of piR-1170 increased the stability of WTAP mRNA, whereas knockdown of piR-1170 had the opposite effect (Fig. 5D). Consistent with this regulatory pattern, hnRNPK depletion similarly decreased WTAP mRNA stability (Fig. 5E) and correspondingly reduced WTAP protein levels in 231-BrM and 159-BrM cells (Fig. 5F). Moreover, the dual-luciferase reporter assay demonstrated that the binding of piR-1170 to the 3’UTR of WTAP mRNA (Fig. 5G). Once the sequence in 3’UTR region of WTAP mRNA was mutated, there will no longer be a difference in the binding of piR-1170 and WTAP mRNA in response to a change of piR-1170 expression (Fig. S4B).

Fig. 5.

piR-1170 stabilizes WTAP mRNA through sequence-specific interactions. A Predicted base-pairing interaction between piR-1170 and WTAP mRNA 3’UTR. B RT-qPCR analysis of WTAP mRNA levels following piR-1170 modulation. C Immunoblot analysis of WTAP protein expression in TNBC cells with piR-1170 knockdown or overexpression. D mRNA quantification of WTAP in TNBC cells with piR-1170 knockdown or overexpression in response to actinomycin D treatment. E mRNA quantification of WTAP in TNBC cells with or without hnRNPK inhibition in response to actinomycin D treatment. F hnRNPK knockdown-induced WTAP protein reduction by immunoblotting. G Dual-luciferase reporter assay of WTAP 3’UTR activity with piR-1170 manipulation. H Schematic depicting shared binding motif for piR-1170 and miR-133a/b in WTAP 3’UTR. I Relative luciferase activity of psiCHECK2 vector bearing WTAP 3’UTR in 231-BrM and 159-BrM cells co-transfected with piR-1170 inhibitor or mimic and miR-133a/miR-133b inhibitor or mimic. The error bars represent the SDs of triplicate experiments

MicroRNAs typically bind to complementary sequences within the 3’UTR of target mRNAs [32]. Through comprehensive bioinformatic analysis using multiple prediction databases (PITA, miRanda, PicTar, and TargetScan), we identified two microRNAs, miR-133a-3p and miR-133b, that share a common binding site with piR-1170 in the 3’UTR of WTAP mRNA (Fig. 5H and Fig. S4C). Subsequently, miR-133a and miR-133b were confirmed to bind to the 3’UTR of WTAP mRNA and to degrade WTAP mRNA by dual luciferase reporter assays (Fig. S4D). Even more importantly, both miR-133a and miR-133b were in competition with piR-1170 for binding to the 3’UTR area of WTAP mRNA (Fig. 5I). Thus, stabilization of WTAP mRNA by piR-1170 was achieved by competing with miR-133a/miR-133b for binding to the 3’UTR of WTAP mRNA.

piR-1170 enhances m6A methylation modification by targeting WTAP

As an m6A writer, WTAP serves as an important component of the methyltransferase complex that is essential for the regulation of m6A methylation. Consistent with this function, we demonstrated that piR-1170 knockdown significantly reduced global m6A levels, as evidenced by both dot blot analysis (Fig. 6A) and m6A-specific immunofluorescence staining (Fig. 6B). To further identify the downstream molecules regulated by piR-1170 through m6A modification, we performed m6A MeRIP-seq in 231-BrM cells with piR-1170 knockdown (Fig. 6C), and sequencing homogeneity assays demonstrated that the RNA used for sequencing was of good quality (Fig. 6D). Meanwhile, the correlation within these samples for examination was close to 1, which proved that the immunoprecipitation was successfully conducted (Fig. 6E). piR-1170 depletion led to a significant reduction in m6A peak numbers, confirming its role in maintaining m6A methylation levels (Fig. 6F). Specifically, piR-1170 knockdown in 231-BrM cells resulted in decreased percentage of m6A peak distribution in the prIntron and npIntron region (Fig. 6G).And in line with the findings of a previous study [33], m6A sites were enriched in GGAC-rich sequences, as visualized by sequence logo analysis (Fig. 6H).

Fig. 6.

piR-1170 mediates m6A methylation through WTAP mRNA interaction. A Dot blot assays for determining m6A abundance in mRNA extracted from the indicated TNBC cells. MB: Methylene Blue. B Immunofluorescence staining of mRNA m6A in 231-BrM cells with piR-1170 knockdown or overexpression. C Workflow for MeRIP-seq. Methylated RNAs selected by an anti-m6A antibody are subjected to high-throughput sequencing. D Sequencing homogeneity assays for distribution of reads across genes in different groups. E Correlation testing between samples for detecting similarities in protein binding patterns in each group. Each point represents log2 (reads count) on each exon. F Histogram of the number of hypermethylated m6A peaks in each group. G Transcriptome-wide distribution of hypermethylated m6A peaks in 231-BrM cells with or without piR-1170 knockdown. Pie chart presenting the proportions of m6A sites within distinct mRNA regions: CDS, 5’ UTR, 3’ UTR, npExon (the exonic region of non-coding genes), npIntron (the intronic region of non-coding genes) and prIntron (the intronic region of coding genes). H Logo plot of the sequences adjacent to mRNA m6A sites identified through HOMER motif analysis. p values were calculated by the HOMER algorithm

piR-1170 induces fatty acid metabolic reprogramming and immune evasion with upregulation of DGAT2 and CD274 via m6A methylation

To definitively identify piR-1170-regulated targets, we integrated a multi-omics approach and applied a concerted filtering strategy. Putative targets were selected based on the concurrent fulfillment of the following three criteria: (i) direct binding to the m6A methyltransferase complex component WTAP, as identified by WTAP CLIP-seq [34]; (ii) a significant reduction in m6A peak occupancy upon piR-1170 knockdown, as measured by m6A MeRIP-seq (Supplementary Table 7); and (iii) a consequent significant alteration in mRNA expression levels following piR-1170 depletion, as assessed by RNA-seq(Supplementary Table 8). Transcripts meeting all three criteria were classified as high-confidence functional targets of piR-1170 via the m6A pathway. This comprehensive approach identified DGAT2 and CD274 as key downstream targets (Fig. 7A). And overage plot for the DGAT2 and CD274 gene region showed that m6A read peaks were significantly reduced when piR-1170 knockdown (Fig. 7B and Fig. S5A). Further, the RIP assays confirmed WTAP’s binding capacity to DGAT2 and CD274 mRNAs (Fig. 7C and Fig. S5B), while m6A MeRIP assays demonstrated m6A modifications on these transcripts (Fig. 7D and Fig. S5C). These data also demonstrate piR-1170’s essential role in governing WTAP-RNA interactions and m6A deposition on DGAT2 and CD274 mRNAs. Both gain- and loss-of-function experiments demonstrated that piR-1170 bidirectionally modulates DGAT2 and CD274 mRNA abundance in TNBC cells (Fig. 7E and Fig. S5D).

Fig. 7.

piR-1170 facilitates fatty acid biosynthesis in TNBC through m6A methylation modification. A Integrative omics analysis depicting overlapping transcripts from RNA-seq, MeRIP-seq (RNAs that m6A peak number reduced) and WTAP CLIP-seq (obtained from GSE46705). B Metagene profiling analysis of m6A distribution patterns across DGAT2 transcriptional units following piR-1170 depletion. C-D RIP-qPCR (C) and MeRIP-qPCR (D) were used to evaluate the interactions between WTAP and the transcripts of DGAT2. E RT‒qPCR was used to measure the RNA levels of DGAT2 upon piR-1170 knockdown or overexpression in 231-BrM cells. F The half-life of DGAT2 mRNA was evaluated in 231-BrM cells upon piR-1170 knockdown or overexpression. G Representative immunofluorescent images of Nuclear-cytoplasmic partitioning of SREBP1 in 231-BrM cells upon piR-1170 knockdown or overexpression. H Relative mRNA levels of genes involved in lipogenesis in 231-BrM cells upon piR-1170 knockdown with or without WTAP overexpression. I Relative protein levels of genes involved in lipogenesis in TNBC cells upon piR-1170 knockdown with or without WTAP overexpression. J Representative fluorescence imaging of lipid droplets stained with BODIPY 493/503 in TNBC cells upon piR-1170 knockdown with or without WTAP overexpression. The error bars represent the SDs of triplicate experiments

Next, we examined the mRNA content of both DGAT2 and CD274 in the indicated cells after treatment with actinomycin D. Unsurprisingly, the results showed that piR-1170 knockdown facilitated the mRNA degradation of both DGAT2 and CD274, whereas overexpressed piR-1170 stabilized them (Fig. 7F and Fig. S5E). Recent evidence has confirmed that m6A readers exert regulatory effects on m6A-modified transcripts [35]. In contrast to the decay-promoting roles of YTHDF2, YTHDF3, a newly discovered family of m6A readers—namely IGF2BP1, IGF2BP2, and IGF2BP3—primarily enhances the stability and translation of their target mRNAs [36]. We next sought to identify the reader mediating piR-1170’s effect on DGAT2 and CD274 mRNA stability. Accordingly, we conducted correlation analyses between m6A reader expression and DGAT2/CD274 levels in TNBC patients from the TCGA-BRCA cohort, revealing statistically significant associations for all examined readers except YTHDF1 and IGF2BP1, suggesting their potential involvement in m6A-mediated regulation of DGAT2 and CD274 (Fig. S5F). Further experimental validation demonstrated that IGF2BP3 enhances DGAT2 mRNA stability, whereas IGF2BP2 promotes CD274 mRNA stability (Fig. S5G). Collectively, piR-1170 stabilized DGAT2 and CD274 mRNAs and regulated their expression through WTAP-dependent m6A modification.

Emerging evidence suggests dual regulatory mechanisms of lipid metabolism in cancer progression. DGAT2 inhibition has been shown to attenuate SREBP-1 proteolytic processing, thereby suppressing lipogenic gene transcription [37]. Notably, clinical observations reveal marked enhancement of fatty acid synthesis (FAS) in breast cancer brain metastases, highlighting the critical role of this metabolic pathway in metastatic progression [5]. Building on these mechanistic and clinical insights, we hypothesized that piR-1170 might regulate FAS in TNBC cells through DGAT2-mediated activation of SREBP-1 proteolytic processing. Immunofluorescence analysis showed that piR-1170 knockdown decreased nuclear SREBP-1 levels, whereas its overexpression increased SREBP-1 nuclear translocation (Fig. 7G). These findings suggest that piR-1170 may facilitate SREBP-1 maturation through proteolytic cleavage in the endoplasmic reticulum, ultimately leading to enhanced nuclear accumulation of the active transcription factor. To elaborate whether piR-1170 regulates FAS through WTAP-dependent mechanisms, we analyzed the expression profiles of SREBPs and key FAS enzymes in 231-BrM cells. The results showed that the expression of SREBPs and key FAS enzymes decreased after piR-1170 knockdown and were restored following WTAP re-overexpression (Fig. 7H-I). Further, neutral lipid staining with BODIPY 493/503 revealed that piR-1170 depletion markedly reduced intracellular lipid droplet accumulation in 231-BrM and 159-BrM cells, which was effectively rescued by WTAP overexpression (Fig. 7J). Similarly, the downregulation of CD274 at both mRNA and protein levels, that occurs following piR-1170 depletion, could also be rescued by overexpression of WTAP (Fig. S5H-I). Taken together, we investigated the important role of piR-1170 in driving brain metastasis of TNBC and uncovered that piR-1170 exerted certain effects on FAS and immune evasion through WTAP-mediated m6A modification.

To further validate our findings, we analyzed primary and brain metastatic lesions from TNBC patients. Notably, piR-1170 expression was significantly upregulated in brain metastatic lesions compared to primary tumors, accompanied by increased expression of downstream targets (WTAP, CD274, DGAT2, SREBP1, FASN, and SCD1) (Fig. S6A). Importantly, WTAP overexpression rescued the impaired invasiveness and migration capacity of TNBC cells induced by piR-1170 knockdown (Fig. S6B-C).

Collectively, through integrated analyses of paired specimen sequencing and systematic validation, we demonstrated that piR-1170 is selectively upregulated in brain metastases and functionally contributes to multiple stages of metastatic progression. Mechanistically, targeting piR-1170 suppresses WTAP-mediated m6A modification of DGAT2 and CD274 transcripts, thereby inhibiting de novo fatty acid synthesis and tumor immune evasion. This multifaceted regulation ultimately attenuates brain metastasis in triple-negative breast cancer (Fig. 8).

Fig. 8.

Graphical abstract of the study. piR-1170 upregulation confers triple-negative breast cancer cells with brain metastatic competence by enhancing WTAP-mediated m6A modifications of DGAT2 and CD274 mRNAs, thereby co-activating de novo lipogenesis and PD-L1-dependent immune evasion to circumvent lipid scarcity and immune surveillance in metastatic niches. The graphical abstract was created in BioRender (https://BioRender.com)

Discussion

Despite the established roles of piRNAs in diverse cancer types and tumor progression stages, their functional involvement and mechanistic contributions to brain metastasis in TNBC remain poorly understood [38]. In this study, we identify piR-1170 as a critical oncogenic driver in TNBC brain metastasis. Although the mature sequence of piR-1170 is identical to a region of the mitochondrial genome, its annotation in the piRBase database and its specific detection via a small RNA-targeted qRT-PCR assay—with signals confirmed to be derived from the processed RNA molecule and not from contaminating mitochondrial genomic DNA by negative -RT controls—support its classification as a nucleus-encoded piRNA. Mechanistically, piR-1170 promotes fatty acid de novo synthesis and immune evasion through WTAP regulation. This finding is particularly significant given the heightened dependence on fatty acid metabolism during TNBC brain metastasis [39–41], and the role of piRNAs in enhancing fatty acid synthesis is crucial for the survival/growth of metastatic TNBC. Compared with the regulation of lipid metabolic process reported previously [42–45], our study identified a novel modulation of fatty acid metabolism in TNBC. In addition, emerging evidence have highlighted the critical role of tumor immune escape in the formation and progression of tumor metastases [46–49], aligning with our discovery that piR-1170 modulates the tumor microenvironment through CD274 stabilization in TNBC. Our study elucidated a direct mechanism underlying the suppression of brain metastasis in piR-1170-deficient TNBC cells. The dual functions of piR-1170 in fatty acid synthesis and immune evasion make it a pivotal oncogene in TNBC malignancies, as it not only fuels survival of TNBC cells but also facilitates their immune evasion, thereby providing a sort of “two-hit” effect during tumour metastasis.

piRNAs and their associated PIWI proteins are frequently upregulated across various tumor tissues and cell lines. In breast cancer, piRNAs have been demonstrated to facilitate key oncogenic processes such as epithelial-mesenchymal transition (EMT) [50] and tumor angiogenesis [51]. However, the canonical understanding of piRNA function faces challenges, as piRNA biogenesis requires numerous germline-specific proteins [52], and a substantial fraction of piRNAs have not been definitively linked to specific PIWI partners [12]. This has raised questions about their precise roles and mechanisms in somatic cancer contexts. Our study addresses this gap by demonstrating that a specific piRNA can exert a potent prometastatic function, driving brain metastasis in a manner independent of canonical PIWI protein association. This finding not only challenges the traditional PIWI-centric framework but also solidifies the evidence for a non-canonical, PIWI-independent oncogenic role of piRNAs in cancer progression.

piRNAs have been implicated in tumorigenesis through the modulation of genetic and epigenetic modifications, including transcriptional regulation [53]. Certain piRNAs regulate the methylation status of oncogene promoter regions by recruiting histone methyltransferases [54]. Some piRNAs were found to recruit the DNA methylase to mediate the methylation of tumor-associated genes [55, 56]. Moreover, some piRNAs were also reported their functions of participating in m6A modification [57, 58], which is consistent with our findings that piR-1170 regulates m6A methylation via WTAP. The primary mechanism of piRNAs in tumorigenesis involves their interaction with mRNAs or lncRNAs, leading to their degradation through base pairing [59–61]. Consistent with limited prior reports [50, 62, 63], we demonstrated that piR-1170 enhances WTAP mRNA stability via base pairing, a previously uncharacterized regulatory mechanism. Interestingly, we identified shared binding sites on WTAP 3’UTR for both miR-133a/miR-133b and piR-1170, revealing that piR-1170 competitively inhibits microRNA-mediated WTAP degradation. Nevertheless, the generalizability of this phenomenon remains uncertain, necessitating additional empirical investigations to validate its prevalence. Additionally, while hnRNPK is a well-characterized RNA-binding protein with diverse roles in splicing, stability, and translation [64], its identification as a key stabilizer of a specific piRNA uncovers a previously unrecognized layer of piRNA regulation in cancer pathogenesis, moving beyond the canonical PIWI-centric mechanisms. Our study revealed a distinct mechanism: the piR-1170-hnRNPK interaction upregulated piR-1170 expression without altering hnRNPK protein levels.

The integration of piRNA-based therapeutics with conventional pharmacological interventions represents a promising therapeutic strategy. Similar to circRNA, piRNA exhibits exceptional stability and target specificity, positioning it as a viable candidate for non-coding RNA vaccine development, analogous to mRNA vaccine platforms, with potential applications in combinatorial cancer therapy [65]. Clinical evidence from the phase IIb KEYNOTE-942 trial demonstrates that the mRNA-4157/V940 vaccine combined with pembrolizumab reduces disease recurrence risk by 44% in high-risk melanoma patients without significant immune-mediated toxicities, highlighting the potential of RNA-based combination therapies [66]. Our in vivo data has proved therapeutic efficacy of piR-1170 knockdown in suppressing multiple stages of TNBC brain metastasis, including cerebral localization, extravasation, proliferation and angiogenesis. These findings underscore the translational potential of developing piR-1170-targeted agents, particularly in combination with immunotherapies, for managing advanced TNBC.

In summary, our findings demonstrate that piR-1170 is overexpressed in TNBC and coordinates pivotal oncogenic functions in metastasis, fatty acid metabolism, and immune evasion. We further elucidate that piR-1170 achieves this by modulating DGAT2 and CD274 expression through WTAP-dependent m6A modification. To translate these discoveries, future work should focus on the clinical validation of piR-1170 as a biomarker, the detailed dissection of its mechanisms in immune suppression, and the development of effective combination therapeutic strategies.

Conclusions

We concluded that piR-1170 is a clinically relevant oncogenic driver and a promising therapeutic target for TNBC brain metastasis. Through regulating WTAP-dependent m6A modification of DGAT2 and CD274, piR-1170 integrates the dual hallmarks of metabolic reprogramming and immune suppression, enabling brain colonization of tumor cells. Given its significant upregulation in tumor metastases and its essential functional role, targeting piR-1170 presents a compelling and previously unexplored therapeutic vulnerability to combat brain metastasis of TNBC.

Authors’ contributions

Hailin Tang, Weidong Wei and Jun Tang conceived and designed the research. Yongzhou Luo, Wenwen Tian performed most of the biochemical and molecular experiments, with assistance from Ye Feng, Xiaofang He. Animal experiments and IHC analyses were performed by Xuefang Huang, Min-Yi Situ. Data analyses were performed by Yuanliang Yan. Xudong Zhu wrote the manuscript. Yanan Kong revised the manuscript and established patient-derived organoid models. All authors have read and approved the article.

Funding

This research was funded by the National Natural Science Foundation of China (Nos. 82473036, Hailin Tang; Nos. 82403554, Wenwen Tian; No. 82303738, Xiaofang He; No. 82303557, Xudong Zhu) and Natural Science Foundation of Guangdong Province (2023A1515030040, Feng Ye).

Data availability

No datasets were generated or analysed during the current study.

Declarations

Ethics approval and consent to participate

This study was approved by Institutional Research Ethics Committee of Sun Yat-sen University Cancer Center (No. 2021 − 358) and conducted under the guidance of the Declaration of Helsinki.

Consent for publication

Consent was obtained from each patient.

Competing interests

The authors declare no competing interests.