Small Extracellular Vesicle piR-hsa-30937 Derived from Pancreatic Neuroendocrine Neoplasms Upregulates CD276 in Macrophages to Promote Immune Evasion Open Access

Pancreatic neuroendocrine neoplasms (PNEN) represent 30% of gastroenteropancreatic neuroendocrine neoplasms and the incidence of PNEN has been increasing in recent decades. PNEN are classified into functional (10%–30%) and nonfunctional (70%–90%; ref. 1). Surgical resection of localized PNEN is the only curative therapy, but the efficacy of treatments for advanced PNEN is limited. Immune checkpoint therapy provides high benefit in several cancers such as melanoma and non–small cell lung cancer. However, patients with PNEN show limited response to current immunotherapy options (2, 3), which could be partly attributed to low T-cell infiltration and low expression of immune checkpoint molecules including PD-1 and PD-L1 (4).

The immune checkpoint molecule CD276 (also known as B7H3), a member of the B7 ligand family, can inhibit T cells to drive tumor immune evasion (5). Previous studies have shown that CD276 is expressed in tumor cells, stromal cells and the vasculature in many cancers, and high CD276 expression correlates with poor prognosis (6). CD276 is expressed in most PNEN and CD276 expression is associated with a higher mitotic count in PNEN. High tumor-associated macrophage (TAM) infiltration is an independent predictor of poor prognosis for PNEN (7). TAMs can directly inhibit T-cell proliferation and recruit regulatory T (Treg) cells to suppress antitumor immunity (8). These studies indicate that CD276 and TAMs can regulate T-cell immunity in the tumor microenvironment (TME) of PNEN.

PIWI-interacting RNA (piRNA), which are 24–32 nucleotides (nt) in length and have a 2′-O-methylated 3′-end, is the largest class of small noncoding RNAs. piRNAs bind to PIWI proteins to form an RNA-induced silencing complex (RISC). PIWI protein/-piRNA can directly cut and degrade target RNA similar to Ago protein/miRNA, and mediate epigenetic regulation, such as DNA methylation and N6-methyladenosine (m6A) methylation (9). In breast cancer, piR-36712 can directly interact with SEPW1P to form a RISC, which suppresses SEPW1 expression by competition of SEPW1 mRNA with SEPW1P RNA for miR-7 and miR-324, and inhibits progression and chemoresistance (10). Small extracellular vesicles (sEV) transfer proteins, DNA and RNA and thereby mediate intercellular communication. piRNAs can be packaged into sEVs, enter the circulation and be taken up by recipient cells to exert biologic functions. The extensive distribution by sEVs and abundant regulatory function of piRNAs increase their potential as biomarkers and therapeutic targets in cancers. Many studies have focused on sEV miRNAs, long noncoding RNAs, and circular RNAs (11–13), but little is known about immune regulation in the TME by sEV piRNAs.

In this study, we identified that CD276 was mainly abundant in TAMs in PNEN TME and piR-hsa-30937 was enriched in serum sEVs of patients with PNEN. PNEN cell–derived sEV piR-hsa-30937 targeted PTEN to activate the AKT pathway and drive CD276 expression in macrophages. CD276⁺ macrophages inhibited T-cell proliferation and IFNγ production. piR-hsa-30937 knockdown and CD276 blockade restored T-cell immunity to inhibit PNEN growth and metastasis. Our study suggests that piR-hsa-30937 and CD276 could be promising targets for immunotherapy in PNEN.

Tissue samples were obtained from 22 patients with PNEN who underwent surgery at the First Affiliated Hospital of Nanjing Medical University between 2021 and 2023. Tumor and adjacent nontumor tissues were diagnosed by two pathologists independently. Paraffin-embedded tissue sections were collected and stored at 4°C. Blood samples were obtained from 18 patients with PNEN and 30 healthy volunteers. Serum samples were collected and stored at 80°C. Details of the patients with PNEN and healthy volunteers are in Supplementary Table S1. The study was approved by the Ethics Committee/Institution of The First Affiliated Hospital of Nanjing Medical University (2023-SR-121). The study was conducted in accordance with the Declaration of Helsinki. Written informed consent was obtained from all patients and volunteers before participation.

RNA-sequencing files of 84 PNEN specimens from EGAS00001005024 were included in the analysis. The ggplot2 R package and survival R package (median as cutoff) were used for statistical analysis and visualization in R v4.3.2. Data for CD276, CD3D and CD8A were analyzed.

The normal human pancreatic ductal cell line HPNE, human monocytic cell line THP-1 and HEK293T were obtained from ATCC in 2022. The PNEN cell line QGP-1 was obtained from the Japanese Collection of Research Bioresources cell bank in 2022. BON-1 was kindly gifted by Prof. Xianjun Yu from Fudan University Shanghai Cancer Center in 2023. HPNE, THP-1, and QGP-1 were maintained in RPMI1640 medium (Gibco), 293T was maintained in DMEM (Gibco), and BON-1 was maintained in DMEM/F-12 (1:1; Gibco). Medium was supplemented with 10% FBS (Yeasen). All cell lines were authenticated by short tandem repeat analysis within 2 years and tested for Mycoplasma (Vazyme). Cells were used within 10 passages for the study.

For the differentiation of THP-1 cells into macrophages, THP-1 cells were treated with 100 ng/mL phorbol 12-myristate 13-acetate (PMA; Sigma-Aldrich) for 24 hours. For human monocyte-derived macrophages (hMDM), peripheral blood monocytes were isolated from the blood of healthy volunteers using Lymphoprep (STEMCELL) and cultured at 37°C with 5% CO₂ for 2 hours. The nonadherent cells were discarded and the adherent cells were cultured in RPMI1640 medium supplemented with 10% FBS and 20 ng/mL MCSF (Pepro-Tech) for 7 days. The morphology of hMDMs at day 7 is shown in Supplementary Fig. S1A. For the coculture of macrophages with PNEN cells or HPNE, macrophages were seeded in the lower chamber in a 6-well plate, and PNEN cells or HPNE cells were seeded in the transwell chamber with a 0.4-μm pore size (Corning).

Plasmids for shpiR-hsa-30937, shPTEN, shNC, piR-hsa-30937 overexpression and piR-NC were constructed by Hanyin Biotechnology CO., Ltd. The shpiR-hsa-30937, shPTEN, and shNC plasmids were constructed using the U6-MCS-CMV-LUC-PGK-Puro vector and the shNC plasmid was an empty vector. The piR-hsa-30937 overexpression and piR-NC plasmids were constructed with the CMV-ZsGreen1-MCS-PGK-Puro vector. All constructs were verified by sequencing and the constructed sequences are shown in Supplementary Table S2. For lentiviral production, HEK293T cells were transfected with the indicated plasmids, pMD2.G (Addgene) and psPAX2 (Addgene) plasmids using PEI (yeasen). For stable cells with shpiR-hsa-30937 or shNC, PNEN cells were transfected with the above supernatant for 48 hours and selected using puromycin (MCE) for 7 days. THP-1 macrophages were transfected with the above supernatant for 48 hours and then used for the following assays.

For serum sEVs from patients with PNEN and healthy volunteers, 3 mL serum was diluted in PBS and centrifuged at 110,000 × g for 90 minutes at 4°C. For sEVs from the culture medium (CM) of HPNE, QGP-1, and BON-1, supernatant was centrifuged at 300 × g for 10 minutes, 3,000 × g for 15 minutes, and 15,000 × g for 30 minutes at 4°C to remove any cell and debris. Then, the supernatant was centrifuged at 110,000 × g for 70 minutes at 4°C to pellet the sEVs. The pellet was resuspended in PBS and centrifuged at 110,000 × g for 70 minutes at 4°C. The size of sEVs was analyzed using the NanoSight NS300 system (Malvern Instruments). sEVs were also examined by transmission electron microscopy (TEM) using a HT7800 RuliTEM (Hitachi).

GW4869 (MCE) was used to inhibit sEV secretion of PNEN cells. PNEN cells were cultured with 10 μmol/L GW4869 or DMSO (MCE) as control for 48 hours. Then, the CM of PNEN cells were collected for treatment of THP-1 macrophages.

To physically deplete sEVs in the CM of PNEN cells, CM was centrifuged at 110,000 × g for 70 minutes at 4°C and the supernatant was collected as sEV-depleted CM for treatment of THP-1 macrophages.

For treatment of RNase A (Tiangen) and Triton X-100 (MCE), 300 μL PNEN cell–derived sEVs were equally separated into three groups. To degrade RNA, sEVs were incubated by 2 mg/mL RNase A with or without Triton X-100 (0.1%) for 20 minutes at 37°C. RNasin (Tiangen) was used at 5 U/μL to stop the RNase digestion and then RNA was extracted (see RNA extraction and qRT-PCR).

To inhibit the AKT signaling pathway, THP-1 macrophages were incubated by PNEN cell–derived sEVs or piR-hsa-30937 overexpression virus with 10 μmol/L MK-2206 (Beyotime) for 48 hours.

To verify the internalization of sEVs by THP-1 macrophages, 100 µL sEVs derived from PNEN cells or HPNE were labeled with PKH-26 (Umibio) and then added into medium of THP-1 macrophages for 48 hours. Nuclei were stained with Hoechst 33342 (Beyotime). Images were acquired on Axio Observer (Zess) and merged using Zen Blue 2012 (Zess).

Peripheral blood lymphocytes (PBL) were extracted from the blood of healthy volunteers using Lymphoprep and labeled with carboxy-fluorescein succinimidyl ester (CFSE; Invitrogen) according to the manufacturer’s instructions. Cells were then plated in 6-well plates and cultured at 37°C with 5% CO₂ for 2 hours, and the nonadherent cells were collected. Before this, THP-1-derived macrophages were cocultured with HPNE or QGP-1, incubated with sEVs from PNEN cells (QGP-1 and BON-1) or HPNE, or incubated with the piR-hsa-30937 overexpression or piR-NC lentivirus for 48 hours. Then, PBLs were cocultured with above macrophages at a 20:1 ratio in 200 µL RPMI1640 medium containing anti-human CD3 (OKT3, eBioscience; 1 mg/mL) and 50 IU/ml IL2 (PeproTech) in 24-well plates for 4 days with or without the 10 μg/mL human neutralizing anti-CD276. After the coculture, PBLs were harvested for flow cytometric analysis of T-cell proliferation and IFNγ production. Antibodies are shown in Supplementary Table S3.

Four-week-old female NOD.Cg-Prkdcscid Il2rg^em1/Smoc (M-NSG) mice were purchased from Shanghai Model Organisms Center. For the generation of PBL humanized mice, 20 M-NSG mice were intravenously injected with 8 × 10⁶ PBLs isolated from the blood of a single healthy donor using Lymphoprep and then randomly divided into four groups (n = 5 each group). After 24 hours, 5 × 10⁶ QGP-1 cells were subcutaneously injected into the right flank of each mouse. Anti-mouse CD276 (BioXcell) or the isotype control IgG was administrated intraperitoneally at 100 μg per mouse on days 9, 12, 15, and 18. Tumor size was measured every 3 days by vernier caliper and tumor volume was calculated by volume = (length*width²)/2. All mice were sacrificed on day 21. All animal experiments were approved by the Institutional Animal Care and Use Committee of Nanjing Medical University (IACUC-2302036). Antibodies are shown in Supplementary Table S3.

Serum sEVs from 5 patients with PNEN and 5 healthy donors were collected (see Isolation and analysis of sEVs). Small RNA libraries were generated by Illumina TruSeq Small RNA Sample kit (Illumina). Total RNA was subjected to sequential 3′ and 5′ adapter ligations using T4 RNA Ligase 2 (Epicenter) followed by reverse transcription using SuperScript II Reverse Transcriptase (Invitrogen) to generate a cDNA library. The amplified cDNAs were purified by 6% Novex TBE PAGE gels (Invitrogen) and bands between 147 and 157 nt containing 22–30 nt fragments were cut out. Library quality was assessed in an Agilent Bioanalyzer 2100 system (Agilent). The qualified cDNA libraries were sequenced on the Novaseq 6000 platform in PE150 (Illumina). The piRNA sequencing and analysis were performed by OE Biotech Co., Ltd. Furthermore, to predict potential binding sites of piRNAs with PTEN mRNA, the sequence of each upregulated piRNA and the PTEN sequence were analyzed using miRanda.

RNA from cells and sEVs was extracted by TRIzol reagent (Invitrogen) and 1 mg RNA was converted into cDNA with stem-loop RT primers using 1st Strand cDNA Synthesis Kit (by stem-loop, Vazyme). Then, qRT-PCR was performed on a LineGene 9600 Plus (Bioer) using Universal SYBR qPCR Master Mix (Vazyme). U6 was used as an internal control for the qRT-PCR of cellular piRNAs and cel-miR-39 was used to normalize for technical variation among sEV piRNAs using the 2^−ΔΔCT method. Each expression analysis contained three independent biological replicates. Primers are shown in Supplementary Table S4.

Cells and sEVs were lysed in RIPA buffer (Beyotime). A total of 30 mg protein per sample was loaded onto SDS-PAGE gels and transferred onto nitrocellulose filter membranes (Millipore). Membranes were blocked with 5% nonfat milk (Beyotime) and incubated with primary antibodies at 4°C for 12 hours, and then incubated with the horseradish peroxidase–conjugated secondary antibody at 4°C for 6 hours. Finally, the membranes were visualized using the enhanced chemiluminescent by Tanon 5200 system (Tanon). Antibodies are shown in Supplementary Table S3.

Constructed wild-type (WT) or mutant (MUT) PTEN 3′ UTR (untranslated region) fragments were cloned into the pmirGLO (Promega) vector by Tsingke. All constructs were verified by sequencing. Reporter plasmids and piR-hsa-30937 or piR-NC plasmids were transfected into HEK293T cells using PEI (Yeasen). Cell lysates were harvested 24 hours after transfection. Then, firefly luciferase and renilla luciferase activities were measured on Synergy H1 (Biotek) using a Dual Luciferase Reporter Assay Kit (Vazyme).

Tissue samples from the patients with PNEN and experimental mice described above were fixed in 4% polyformaldehyde and embedded in paraffin. IHC staining was conducted in 5-μm serial sections from paraffin blocks using a REAL EnVision detection system (DAKO). Cells or areas stained with brown were defined as the positive staining. The fluorescent multiplex IHC (mIHC) was conducted using the tyramide signal amplification kit from Runnerbio Technology CO., Ltd. Random fields were taken from each section on Axio Observer (Zess) and merged using Zen Blue 2012 (Zess). Antibodies are shown in Supplementary Table S3.

Flow cytometry was performed on a CytoFLEX (Beckman) and FlowJo v10.8.1 was used for data analysis. For surface markers, cell suspensions were blocked using FcR Blocking Reagent (130-059-901, Miltenyi Biotec) and stained by corresponding antibodies. For IFNγ staining, PBLs after coculture were stimulated with 20 ng/mL PMA (Sigma-Aldrich) and the Protein Transport Inhibitor Cocktail (00-4980-03, Invitrogen) for 4 hours and then stained by anti-CD3 or anti-CD8. Next, stained cells were fixed with the Fixation Buffer (Invitrogen) for 20 minutes, and permeabilized with the Permeabilization Buffer (Invitrogen), and then stained with anti-IFNγ. Each experiment contained three independent biological replicates. Antibodies are shown in Supplementary Table S3.

Cy3-labeled piR-hsa-30937 probes and FISH Kit were provided by RiboBio. All operations were conducted according to the manufacturer’s instructions. Nuclei were stained with DAPI. Images were acquired on an Axio Observer (Zess) and merged using Zen Blue 2012 (Zess).

A total of 1.5 × 10⁷ THP-1 macrophages were incubated with 10 µg/mL PNEN cell–derived sEVs for 48 hours and then used for RNA immunoprecipitation (RIP) assay. All operations were in accordance with the instructions of the Magna RIP Kit (Millipore). A total of 5 μg of anti-PTEN and anti-IgG (Abmart) were incubated with 40 μL magnetic beads. The bound RNAs and input RNAs were extracted and reverse transcribed to cDNA for qRT-PCR (see RNA extraction and qRT-PCR). The relative enrichment was normalized to the input. Each group contained three biological replicates. Antibodies are shown in Supplementary Table S3.

Statistical analyses were performed in GraphPad Prism 9. Results are expressed as mean ± SD. Statistical significance was determined by Student t test or one-way ANOVA test. The correlation analysis was performed by Pearson correlation coefficient. A P value < 0.05 was considered statistically significant.

The data that support the findings of this study are available in the article and its Supplementary Data files or from the corresponding author upon reasonable request. Next-generation sequencing data were deposited in Sequence Read Archive (SRA; PRJNA1068778).

IHC of samples from patients with PNEN showed that CD276 expression in macrophages was higher in tumor tissues than in adjacent nontumor tissues (Fig. 1A). The presence of more CD276⁺ macrophages in tumor tissues than in nontumor tissues was confirmed by mIHC (Fig. 1B). Furthermore, CD3⁺ T-cell infiltration in nontumor tissues was significantly higher than that in tumor tissues (Supplementary Fig. S2), which indicated T-cell exclusion from the TME. Next, mIHC results of PNEN tumor tissues and corresponding nontumor tissues showed an inverse correlation between the infiltration of CD276⁺ macrophages and T cells (Fig. 1C). In addition, IHC analysis of tumor tissues indicated an inverse correlation of the infiltration of CD276⁺ macrophages with T cells, including CD8⁺ T cells (Fig. 1D; Supplementary Fig. S3A). Moreover, mIHC showed that a considerable proportion of CD276 proteins were located in macrophages (Fig. 1B and C), which suggested a crucial role for CD276⁺ TAMs. In addition, we found that CD276 expression in QGP-1 cells was higher than that in BON-1 cells and was not detected in their sEVs (Supplementary Fig. S4). Furthermore, we found more CD276⁺ macrophage infiltration in PNEN tissues with liver metastasis (Fig. 1E), but no similar association of CD3⁺ or CD8⁺ T cell-infiltration (Supplementary Fig. S3B). Then, we analyzed CD276, CD3D and CD8A mRNA expression in the European Genome-phenome Archive (EGA) database. High CD276 expression was associated with pathologic lymph node involvement, high histopathologic grade, high tumor–node–metastasis (TNM) stage, metastasis, and poor survival (Fig. 1F), but neither CD3D nor CD8A expression showed significant associations (Supplementary Fig. S3C and S3D).

To investigate the immunosuppressive role of CD276⁺ macrophages, THP-1 and primary monocyte-derived macrophages were first cocultured with HPNE cells or PNEN cells for 48 hours, and results of flow cytometry assay confirmed that CD276 expression was higher in macrophages cocultured with PNEN cells than HPNE-cocultured macrophages and untreated macrophages (Fig. 2A and B). Next, macrophages, after coculture with HPNE or QGP-1, were cocultured with CFSE-labeled PBLs from healthy donors for 72 hours. Compared with untreated and HPNE-cocultured macrophages, macrophages cocultured with PNEN cells inhibited T-cell proliferation and IFNγ production and neutralizing anti-CD276 alleviated CD3⁺ T-cell suppression (Fig. 2C). HPNE-cocultured macrophages exhibited no significant inhibition on CD3⁺ T-cell proliferation or IFNγ production compared with untreated macrophages.

Previous studies showed immunosuppressive macrophage phenotypes can be driven by tumor-derived sEVs (14). Therefore, we considered that PNEN cell–derived sEVs could drive CD276 expression in macrophages. sEVs derived from HPNE, QGP-1, and BON-1 were analyzed by TEM and nanoparticle tracking analysis (Fig. 3A). The presence of sEV markers (CD63, TSG101, HSP70, and calnexin) was confirmed by Western blot analysis (Fig. 3B). PKH26-labeled sEVs from the above-mentioned cells were added into the medium of THP-1–derived macrophage cultures and the fluorescent images showed the internalization of these sEVs by macrophages (Fig. 3C).

To investigate the role of PNEN cell–derived sEVs in CD276 expression of macrophages, THP-1 macrophages or primary monocyte-derived macrophages were incubated with 10 μg/mL PNEN cell–derived sEVs for 48 hours. Flow cytometry revealed that CD276 expression in macrophages was remarkably upregulated by PNEN cell–derived sEVs and HPNE-derived sEVs showed no such effects (Fig. 3D; Supplementary Fig. S1B). Furthermore, results showed the dose–response relationship of CD276 expression with the dose of PNEN cell–derived sEVs (Supplementary Fig. S5A). Next, we investigated the effects of macrophages treated by PNEN cell–derived sEVs on T cells. THP-1 and hMDMs cocultured with PNEN cell–derived sEVs significantly inhibited CD3⁺ T-cell proliferation and IFNγ production, compared with control macrophages (Fig. 3E; Supplementary Fig. S1D). Neutralizing anti-CD276 rescued the T-cell suppression. Macrophages treated by HPNE-derived sEVs showed no significant suppression on the proliferation or IFNγ production of CD3⁺ T cells (Fig. 3E; Supplementary Fig. S1D). Similar trends on CD8⁺ T-cell proliferation and IFNγ production were observed (Fig. 3E).

To investigate the underlying mechanisms for the induction of CD276 on macrophages by PNEN cell–derived sEVs, we performed piRNA-sequencing analysis of serum sEVs from patients with PNEN and healthy volunteers. The quality control of the sequencing is shown in Supplementary Fig. S6. As shown in Fig. 4A, 14 downregulated piRNAs and 24 upregulated piRNAs were detected in serum sEVs from patients with PNEN compared with healthy volunteers (Supplementary Table S5). Previous studies have shown that CD276 expression can be driven via the PI3K/AKT pathway and suppressed by the PTEN/AKT pathway (15, 16). Thus, we hypothesized that the PNEN cell–derived piRNA must drive CD276 expression through the PTEN/AKT pathway.

Among upregulated piRNAs, piR-hsa-30937 was predicted as the most potential piRNA that could target PTEN by miRanda. Thus, we hypothesized that piR-hsa-30937 in PNEN cell–derived sEVs was transferred to macrophages and drove CD276 expression. First, we confirmed higher serum sEV piR-hsa-30937 expression in patients with PNEN than healthy individuals (Fig. 4B). In addition, higher piR-hsa-30937 expression was detected in sEVs from PNEN cells (QGP-1 and BON-1) than HPNE (Fig. 4C). Higher levels of piR-hsa-30937 were detected in THP-1 and hMDMs treated with PNEN cell–derived sEVs for 48 hours than macrophages treated by HPNE-derived sEVs (Fig. 4D; Supplementary Fig. S1C) and piR-hsa-30937 expression increased along with the sEV dose (Supplementary Fig. S5B). Moreover, piR-hsa-30937 expression in the CM of PNEN cells was little changed by RNase A treatment, but significantly reduced following treatment of RNase A and Triton X-100 (Fig. 4E), suggesting that extracellular piR-hsa-30937 was encased in membrane vesicles instead of directly secreted by PNEN cells. Next, sEVs in the CM of PNEN cells were depleted by ultracentrifugation or GW4869 and macrophages were treated with the control supernatant or sEV-depleted supernatant. piR-hsa-30937 expression was significantly reduced in macrophages treated by sEV-depleted supernatant (physically and pharmacologically) than THP-1 macrophages treated by control supernatant (Fig. 4F and G). PNEN cells stably transfected with shpiR-hsa-30937 virus and shNC virus were constructed. And we confirmed downregulated piR-hsa-30937 expression in cells and sEVs of QGP-1-shpiR-hsa-30937 and BON-1-shpiR-hsa-30937 (Fig. 4H and I). piR-hsa-30937 expression in macrophages transfected with piR-hsa-30937 overexpression virus was confirmed (Fig. 4J). RNA FISH assay showed piR-hsa-30937 expression both in the cytoplasm and nucleus of THP-1 macrophages treated with PNEN cell–derived sEVs (Fig. 4K).

To investigate whether sEV piR-hsa-30937 can drive CD276 expression to inhibit T-cell immunity, THP-1 macrophages were treated by sEVs from piR-hsa-30937-knockdown and control PNEN cells. Flow cytometry showed that piR-hsa-30937 knockdown in PNEN cells significantly suppressed their ability to upregulate levels of CD276 in THP-1 macrophages (Fig. 5A). Moreover, CD276 expression was significantly upregulated in THP-1 macrophages with piR-hsa-30937 overexpression or PTEN knockdown (Fig. 5B and C). Next, to investigate the effect of sEV piR-hsa-30937 on T-cell immunity, the above THP-1 macrophages were cocultured with PBLs. Flow cytometry showed that piR-hsa-30937 knockdown in PNEN cells attenuated the inhibition of T-cell proliferation and IFNγ production by PNEN cell–derived sEVs (Fig. 5D). On the other hand, THP-1 macrophages with piR-hsa-30937 overexpression significantly suppressed T-cell immunity, and this was restored by neutralizing anti-CD276 (Fig. 5E).

The potential binding sites of piR-hsa-30937 in the 3′ UTR of PTEN mRNA were predicted by miRanda (Fig. 6A). Results of luciferase reporter assays showed that piR-hsa-30937 overexpression significantly suppressed the luciferase activity of reporters containing the WT PTEN 3′ UTR, but showed no significant inhibition on the luciferase activity of the reporters containing the MUT PTEN 3′ UTR (Fig. 6B). To investigate the direct interaction between piR-hsa-30937 and PTEN, the RIP assay was conducted in THP-1 macrophages incubated with 10 μg/mL PNEN cell–derived sEVs and results were negative (Supplementary Fig. S7). To investigate whether sEV piR-hsa-30937 could activate the AKT signaling pathway to drive CD276 expression by targeting PTEN, THP-1 macrophages were incubated with sEVs or transfected with plasmids. As shown in Fig. 6C, PNEN cell–derived sEVs significantly downregulated PTEN expression and significantly increased phosphorylated AKT (p-AKT) and CD276 expression, and piR-hsa-30937 knockdown significantly attenuated the sEV-induced PTEN downregulation and p-AKT and CD276 upregulation. Furthermore, the length of time that THP-1 macrophages need to be exposed to sEVs was analyzed, and it was found that effects on PTEN could be clearly observed after 12 hours (Supplementary Fig. S8). piR-hsa-30937 overexpression in THP-1 macrophages significantly reduced PTEN expression and increased p-AKT and CD276 expression, which was consistent to the effects of PTEN knockdown in THP-1 macrophages (Fig. 6C).

Next, we used an AKT inhibitor MK-2206 to further confirm whether PNEN cell–derived sEV piR-hsa-30937 drove CD276 expression of THP-1 macrophages via the AKT signaling pathway. Flow cytometry showed that MK-2206 significantly inhibited upregulated levels of CD276 mediated by PNEN cell–derived sEVs or piR-hsa-30937 overexpression (Fig. 6D). The above results suggested that PNEN cell–derived sEV piR-hsa-30937 drove CD276 expression by downregulating PTEN to activate the AKT signaling pathway (Supplementary Fig. S9).

To investigate the effects of CD276⁺ macrophages induced by sEV piR-hsa-30937 on T-cell immunity in vivo, a tumor formation assay was performed in PBL humanized M-NSG mice. After the humanized mice were established, stable QGP-1 cells transfected with shNC and shpiR-hsa-30937 were subcutaneously injected. Results showed that both volumes and weights of the piR-hsa-30937-knockdown group were significantly lower than the shNC group. Compared with the control IgG-treated and untreated shNC groups, the anti-CD276 group had significantly lower volumes and weights (Fig. 7A). Results of mIHC showed that piR-hsa-30937 knockdown reduced CD276⁺ TAM infiltration and increased CD3⁺ T-cell infiltration, and anti-CD276 increased CD3⁺ T-cell infiltration (Fig. 7B). In addition, higher Ki67 index and more spontaneous liver and lung metastatic nodules were found in the untreated shNC group than that in the piR-hsa-30937-knockdown group, and anti-CD276 significantly reduced the Ki67 index, and liver and lung metastatic nodules compared with the control IgG group and untreated group (Fig. 7C). These results suggest that piR-hsa-30937 from PNEN cell–derived sEVs can drive CD276 expression in macrophages to inhibit T-cell immunity and promote the proliferation and metastasis of PNEN, and CD276 blockade can restore T-cell immunity to inhibit the proliferation and metastasis of PNEN in vivo.

T cells play an important role in PNEN immunity. However, PNEN exhibit a “cold” TME with low T-cell infiltration, which contributes to immune evasion and insensitivity to immunotherapy. High intratumoral CD3⁺ T-cell infiltration is associated with improved recurrence free survival and high levels of FoxP3⁺ Treg cells within liver metastases predict poor survival (17). PNEN are classified into four transcriptomic subtypes: metastasis-like primary (MLP)-1, MLP-2, insulinoma-like and intermediate (18). The MLP-1 subtype, with the increased tumor size and poor prognosis, features an immunosuppressive gene expression phenotype including T-cell immune modulation. Single-sample gene set enrichment analysis confirms enrichment for coinhibition of T cells in the MLP-1 subtype, and mIHC shows more CD68⁺ cells in the MLP-1 subtype compared with intermediate and MLP-2 subtypes (19). Besides T cells, macrophages in PNEN TME play a crucial role in PNEN progression. High TAM infiltration is positively correlated with higher PNEN grade and stage, liver metastasis, and poor survival (7, 20–22). High expression of PD-L1 is found in TAMs in many cancers such as glioma, which mediates T-cell exhaustion and immunotherapy resistance (23). However, Cai and colleagues detected PD-L1 expression on TAMs only in three cases from 104 PNEN specimens, and CD276 is expressed in more PNEN than PD-L1 (78% vs. 53%; ref. 7). In addition, our study showed that CD276 was expressed considerably on macrophages in all clinical specimens and F4/80⁺ macrophages in tumors of PBL humanized M-NSG mice. The results support a crucial role for CD276⁺ macrophages in immunoregulation of T cells.

The binding partners of CD276 have not been identified and mechanisms of CD276-mediated immune evasion remain elusive. Inhibition of CD276 enhances IFNγ production of CD8⁺ and CD4⁺ T cells in a cell–cell contact manner, and intact IFNγ signaling is critical for a positive response to CD276 inhibition (6). Our study showed that CD276⁺ macrophages also inhibit IFNγ production of T cells, and that this could be restored by neutralizing CD276 antibodies. CD276 blockade eliminates cancer stem cells of head and neck squamous cell carcinoma in a CD8⁺ T cell–dependent manner (5). Besides a role in immunity, CD276 is overexpressed during pathologic tumor vasculature angiogenesis but not physiological angiogenesis, which indicates that CD276-targeted therapies might exhibit higher specificity for tumor vessels than current antiangiogenic agents, which also block normal physiologic angiogenesis (24). CD276-targeted therapies include mAbs, chimeric antigen receptor–modified T (CART) cells and antibody–drug conjugates (25). CD276 expression is generally low in human normal tissues, which may contribute to the absence of dose-limiting toxicity (26). Enoblituzumab, a CD276 antibody, is under evaluation in clinical trials (NCT02982941, NCT02381314, NCT02475213, and NCT04630769). A single-arm, phase II trial of enoblituzumab in prostate cancer has shown feasibility, general safety, and potential clinical activity (27). CD276-targeting CAR T cells evaluated for diffuse intrinsic pontine glioma suggest the feasibility of repeated intracranial CD276 CAR T-cell dosing and locoregional immune activation induced by intracranial delivery (28). These data suggest that CD276-targeted therapies could be feasible, safe, and promising for PNEN.

Recent studies indicate that piRNA can mediated epigenetic regulation, including DNA methylation, histone modification, and m6A RNA methylation (9). piRNA-14633 can increase m6A RNA methylation levels and METTL14 mRNA stability to promote cervical cancer progression (29). piRNA-30473 can enhance the m6A level by upregulation of WTAP and then increase HK2 expression to promote the progression of diffuse large B-cell lymphoma (30). piR-NAs can also be encased into sEVs, which are transferred to recipient cells in the TME, to achieve cell–cell communication. In several cancers, piRNAs are also enriched in serum sEVs of patients and sEV piRNAs serve as potential biomarkers (31). Multiple myeloma–derived sEV piRNA-823 can promote tumorigenesis by reeducation of endothelial cells in the TME (32). Senescent neutrophil–derived sEV piRNA-17560 can promote chemoresistance and epithelial–mesenchymal transition of breast cancer by upregulating ZEB1 expression via FTO-mediated m6A demethylation (33). However, the role of sEV piRNAs in the tumor immune microenvironment has not been revealed.

The application of piRNAs for tumor therapy may have potential advantages. piRNAs are expressed mainly in germ cells and cancer cells and more selective in identifying targets than miRNAs (34), contributing to reduced side effects, higher specificity and safety. In addition, RNAi-based therapies targeting PIWI/piRNAs could be feasible and safe because of the specific expression. Clinical trials of some RNAi therapies were withdrawn mainly because of safety concerns (35). Thus, we considered that therapies targeting piR-hsa-30937 with a safe, specific, and effective delivery system could be promising. Therapies targeting piR-hsa-30937 could reduce CD276⁺ macrophage infiltration into tumors and increase T-cell infiltration, which may improve selectivity to immunotherapy of PNEN.

Our study identifies aberrantly expressed serum sEV piRNAs in patients with PNEN and reveals the role of sEV piR-hsa-30937 on CD276⁺ macrophages and T-cell immunity in the PNEN TME. However, our study has several limitations. The number of clinical samples is limited and thus we are unable to explore the association of sEV piR-hsa-30937 expression and CD276⁺ macrophage infiltration with PNEN survival.

In conclusion, our study reveals an immunosuppressive role of CD276⁺ macrophages, induced by piR-hsa-30937 from PNEN cell–derived sEVs in the TME. CD276⁺ macrophages can inhibit T-cell proliferation and IFNγ production to promote PNEN growth and metastasis. Targeting sEV piR-hsa-30937 or CD276 can rescue T-cell immunity. sEV piR-hsa-30937 and CD276 serve as potential therapeutic targets for immunotherapy of PNEN.

No disclosures were reported.

Y. Zhong: Conceptualization, resources, data curation, software, formal analysis, validation, investigation, visualization, methodology, writing–original draft, project administration, writing–review and editing. Y. Tian: Data curation, software, formal analysis, funding acquisition, validation, investigation, visualization, methodology, writing–review and editing. Y. Wang: Data curation, formal analysis, validation, investigation, visualization, methodology, writing–review and editing. J. Bai: Writing–review and editing. Q. Long: Resources. L. Yan: Writing–review and editing. Z. Gong: Writing–review and editing. W. Gao: Supervision, methodology, writing–review and editing. Q. Tang: Resources, supervision, funding acquisition, validation, project administration, writing–review and editing.

We would like to thank the Core Facility of the First Affiliated Hospital of Nanjing Medical University for its help in the detection of experimental samples.

This work was supported by the Science Foundation Project of Ili & Jiangsu Joint Institute of Health (grant no. yl2020lhms05) and Wuxi “Taihu talent plan” for the excellent medical expert team (grant no. 2021-9).