MiR-139 modulates CD8+ T cells exhaustion by targeting CD28 in breast cancer Authors: Jianrong Chen, Haiyong Zhang, Yanting Lv, Xiufang Li, Mengjun Hu Published: 2026 ABSTRACT SummarymiR-139 acts as a tumor suppressor in breast cancer cells, yet its role within the tumor immune microenvironment (TIME) remains unclear. This study explores the dual functions of miR-139 in both tumor and immune cells, particularly its effect on CD8+ T cell function and chemokine-mediated immune recruitment. Bioinformatics analysis using TCGA and GEO data identified miR-139 as differentially expressed in breast cancer. Using samples from 32 patients, we detected elevated miR-139 levels in tumor-infiltrating CD8+ T cells compared with adjacent normal tissues. Further experiments confirmed that miR-139 directly targets CD28, leading to its downregulation. Concurrently, increased expression of exhaustion markers PD-1 and TIGIT was observed. Moreover, tumors with high miR-139 expression showed upregulation of T-cell-recruiting chemokines CX3CL1, CXCL12, and CXCL14. These results demonstrate that miR-139 promotes CD8+ T cell exhaustion via suppression of CD28 and facilitates chemokine-mediated T cell recruitment, highlighting its immunosuppressive role in breast cancer. Introduction According to the recent Global Cancer Observatory (GLOBOCAN) 2020 data, the incidence of breast cancer (BC) in women is the leading cause of cancer incidence worldwide, with an estimated 2.3 million new cases, accounting for 11.7% of all cancer cases.1 Despite advancements in early detection and therapeutic strategies, BC continues to be the leading cause of cancer-related mortality among women globally.1 The etiology of BC involves a complex interplay of genetic predisposition and environmental factors.2 Although significant progress has been made in screening, diagnostic accuracy, and treatment modalities, disease relapse and therapeutic resistance remain major clinical obstacles, contributing to poor patient outcomes.3 Given these challenges, elucidating the molecular mechanisms driving BC progression is critical for the identification of novel therapeutic targets and the development of more effective treatment strategies. MicroRNAs (miRNAs) are small non-coding RNA molecules, typically 18–25 nucleotides in length, which play crucial roles in post-transcriptional gene regulation. Current genomic studies have identified over 1,900 miRNAs in humans, which are predicted to regulate approximately 60% of all protein-coding genes.4 Among these, miR-139 has emerged as a significant regulatory molecule that modulates critical cellular processes including proliferation, apoptosis, and metastasis across various cancer types.5 In the context of BC , accumulating evidence demonstrates that miR-139 is frequently downregulated and functions as a potent tumor suppressor. Mechanistic studies have revealed that miR-139 exerts its anti-tumor effects by targeting multiple oncogenic factors, including RAB1A,6 COL11A1,7 ZEB1,8 and SOX8.9 These findings highlight the multifaceted tumor-suppressive role of miR-139 in breast carcinogenesis. Despite these advances, the specific functions of miR-139 within the tumor microenvironment (TME), particularly its interactions with stromal components and immune cells, remain poorly understood and warrant further investigation. The CD28 costimulatory receptor, a 44 kDa transmembrane glycoprotein, is constitutively expressed on nearly all human T lymphocytes during ontogeny.10 In naive T cells, CD28-mediated co-stimulation serves three critical functions: (1) promoting cell cycle progression, (2) potently inducing IL-2 production, (3) activating anti-apoptotic pathways, and (4) facilitating immune synapse formation.11,12 These mechanisms collectively enable optimal T cell activation and clonal expansion. CD8+ cytotoxic T lymphocytes (CTLs) represent the principal effectors of tumor-specific adaptive immunity.13 Their anti-tumor activity is critically dependent on CD28 co-stimulation, which provides the necessary secondary signal for full activation. Notably, CD28 downregulation has emerged as a biomarker of T cell senescence and a feature of immunosuppressive T cell populations in cancer.14 Clinical observations reveal that diminished CD28 expression correlates with T cell exhaustion phenotypes in acute myeloid leukemia patients and impaired anti-tumor responses in various malignancies.15 However, the functional consequences of CD28 downregulation in BC-infiltrating T lymphocytes remain poorly understood. In this study, we analyzed the differentially expressed miRNAs using three datasets downloaded from TCGA portal and GEO database. We noticed that the tumor suppressor, miR-139, is downregulated in BC samples compared with adjacent non-tumor tissues or normal breast tissues. However, patients with higher miR-139 level have poorer prognosis but increased number of tumor infiltrated CD8+ T cells. This research explored the roles of miR-139 in modulating the TME in BC. Results Comprehensive bioinformatics analysis uncovers miR-139 dysregulation patterns and immune correlations in breast cancer To explore the roles of miRNAs in BC, we analyzed the sequencing data from TCGA portal including 1,090 BC and 103 control samples, and two datasets (GEO database: GSE37963 and GSE154255) downloaded from GEO portal. We observed that 12 miRNAs (miR-100, miR-10b, miR-21, miR-139, miR-99a, miR-130b, miR-486, miR-182, miR-195, miR-335, miR-145, and miR-301a) are dysregulated in all these three datasets (Figures 1A–1C). After analyzing the prognosis of patients with BC, we found that higher levels of miR-100, miR-10b, miR-182, miR-195, and miR-99a are related to longer overall survival (Figure S1). In contrast, higher levels of miR-130b, miR-139, miR-145, and miR-301a relate to shorter overall survival (Figure S1). Notably, miR-139 was downregulated in tumor tissues, implying a potential antitumor role. However, it remains unclear as to why this antitumor molecule is correlated with poorer overall survival (Figure S1; Figure 1D). Subsequently, we overexpressed miR-139 in MCF-7 and MDA-231 cells and observed increased apoptosis (Figure S2A) and reduced viability (Figure S2B), which was consistent with the findings from other researchers. Therefore, we proposed that miR-139 may be a multi-functional miRNA that has different roles in tumor cells and tumor-infiltrating lymphocytes (TILs). We analyzed the correlations between miR-139 and different subpopulations of TILs. Results indicated that tumors expressing higher levels of miR-139 have increased B cell, CD8+ T cells, resting dendritic cells, and monocytes and reduced resting natural killer (NK) cells, M0 macrophages, M2 macrophages, and neutrophils (Figure 1E).Figure 1Comprehensive bioinformatics analysis reveals miR-139 dysregulation patterns and immune associations in breast cancer(A) Volcano plots display differentially expressed miRNAs from three independent breast cancer cohorts. Red dots denote significantly upregulated miRNAs (p < 0.05, log2|FC|>1), while blue dots represent downregulated miRNAs (same thresholds). Gray dots indicate non-significant miRNAs. Dashed lines demarcate statistical significance cutoffs.(B) Three-way Venn diagram identifies 12 differentially expressed miRNAs across TCGA-BRCA, GSE97811, and GSE110835 datasets.(C) miR-139-5p expression comparison in paired tumor/normal specimens (n = 32 pairs) from our institutional cohort (∗∗∗p < 0.001, paired t test).(D) Prognostic significance of miR-139-5p in TCGA breast cancer (n = 1,088). Kaplan-Meier curves show poorer overall survival in high expressers (median cutoff; log rank p = 0.021).(E) CIBERSORT deconvolution of immune infiltrates in TCGA cases stratified by miR-139-5p expression (high, n = 549; low, n = 548). Bar plot shows significantly altered immune cell fractions (ANOVA with Benjamini-Hochberg false discovery rate [FDR] correction; ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001). Data are represented as mean ± SD. Comprehensive bioinformatics analysis reveals miR-139 dysregulation patterns and immune associations in breast cancer (A) Volcano plots display differentially expressed miRNAs from three independent breast cancer cohorts. Red dots denote significantly upregulated miRNAs (p < 0.05, log2|FC|>1), while blue dots represent downregulated miRNAs (same thresholds). Gray dots indicate non-significant miRNAs. Dashed lines demarcate statistical significance cutoffs. (B) Three-way Venn diagram identifies 12 differentially expressed miRNAs across TCGA-BRCA, GSE97811, and GSE110835 datasets. (C) miR-139-5p expression comparison in paired tumor/normal specimens (n = 32 pairs) from our institutional cohort (∗∗∗p < 0.001, paired t test). (D) Prognostic significance of miR-139-5p in TCGA breast cancer (n = 1,088). Kaplan-Meier curves show poorer overall survival in high expressers (median cutoff; log rank p = 0.021). (E) CIBERSORT deconvolution of immune infiltrates in TCGA cases stratified by miR-139-5p expression (high, n = 549; low, n = 548). Bar plot shows significantly altered immune cell fractions (ANOVA with Benjamini-Hochberg false discovery rate [FDR] correction; ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001). Data are represented as mean ± SD. miR-139 inhibits CD28 expression in CD8+ T cells To explore the role of miR-139 in the tumor immune microenvironment (TIME) in BC, we first checked the expression landscape of miR-139 across different types of cells in TIME. T cells, B cells, NK cells, and endothelial cells were isolated from 5 tumors using flow cytometry (Figure S3A), and the relative levels of miR-139 were detected by RT-qPCR. We observed that miR-139 has relatively higher levels in immune cells, especially CD8+ T cells and NK cells, and lower levels in endothelial cells, suggesting a role of miR-139 in immune cells (Figure S3B). To investigate the mechanism by which miR-139 controls the TIME in BC, the potent targets of miR-139 were predicted using TargetScan 8.0 (https://www.targetscan.org/vert_80/). A total of 3,486 predicted targets were subjected to functional annotation using DAVID (https://davidbioinformatics.nih.gov/). We noticed that three genes (TGFB1, CD28, and CCR7) that modulate T cell homeostasis and T cell differentiation are potent miR-139 targets that were not verified by other researchers (Figure 2A). We then employed dual luciferase assay to examine the interaction between miR-139 and the 3′ UTRs from TGFB1, CD28, or CCR7 in HEK293T cells. We observed that miR-139 could inhibit the luciferase activity by targeting CD28 (Figure 2B). Meanwhile, miR-139 inhibitor significantly upregulated the luciferase activity (Figure 2C). When 5 nucleotides in the 3′ UTR of CD28 were mutated (Figure 2D), miR-139 mimic could not repress the dual luciferase activity (Figure 2E). These results indicated that miR-139 inhibited firefly luciferase expression by targeting the 3′ UTR of CD28.Figure 2miR-139 directly suppresses CD28 expression via 3′ UTR targeting(A) Predicted binding sites of miR-139-5p within the 3′ UTRs of CD28, CCR7, and TGFB1, generated using TargetScan.(B and C) Luciferase reporter assays in HEK293T cells co-transfected with miR-139-5p mimic (B) or inhibitor (C) and reporter vectors containing wild-type 3′ UTRs of candidate genes. Relative luciferase activity was measured 48 h post-transfection (mean ± SEM, n = 3; ∗∗p < 0.01, ∗∗p < 0.001, two-tailed t test).(D) RT-qPCR validation of miR-139-5p overexpression or knockdown in primary human CD8+ T cells after electroporation (mean ± SEM, n = 5 donors; ∗∗p < 0.01, ∗∗∗p < 0.001).(E) Flow cytometry analysis of CD28 surface expression in miR-139-modulated CD8+ T cells (representative histograms and quantification; mean fluorescence intensity (MFI) ± SEM, n = 5; ∗p < 0.05, ∗∗p < 0.01). miR-139 directly suppresses CD28 expression via 3′ UTR targeting (A) Predicted binding sites of miR-139-5p within the 3′ UTRs of CD28, CCR7, and TGFB1, generated using TargetScan. (B and C) Luciferase reporter assays in HEK293T cells co-transfected with miR-139-5p mimic (B) or inhibitor (C) and reporter vectors containing wild-type 3′ UTRs of candidate genes. Relative luciferase activity was measured 48 h post-transfection (mean ± SEM, n = 3; ∗∗p < 0.01, ∗∗p < 0.001, two-tailed t test). (D) RT-qPCR validation of miR-139-5p overexpression or knockdown in primary human CD8+ T cells after electroporation (mean ± SEM, n = 5 donors; ∗∗p < 0.01, ∗∗∗p < 0.001). (E) Flow cytometry analysis of CD28 surface expression in miR-139-modulated CD8+ T cells (representative histograms and quantification; mean fluorescence intensity (MFI) ± SEM, n = 5; ∗p < 0.05, ∗∗p < 0.01). To understand whether miR-139 controls CD28 expression in T cells, primary CD8+ T cells were isolated from peripheral blood and then transfected with miR-139 mimic or inhibitor by electroporation. As shown in Figure 2F, miR-139 level increased 11.4-fold in T cells transfected with miR-139 mimic and reduced by 68.1% in those transfected with miR-139 inhibitor. Meanwhile, the cell surface CD28 level was reduced in T cells with miR-139 overexpression and upregulated in miR-139-knockdown T cells (Figure 2G). CD8+ T cells in BC tumors express increased miR-139, PD-1, and TIGIT To further investigate the roles of miR-139 in BC, peripheral blood, tumors, and paired adjacent non-tumor tissues were collected from 32 patients with BC. The clinical characteristics of all enrolled patients are presented in Table 1. As shown in Figure 3A, the levels of miR-139 were reduced in BC tissue samples. Subsequently, CD8+ T cells were isolated from tumor, adjacent non-tumor tissue, and peripheral blood of 13 patients with BC. Analysis of CD8+ T cells by flow cytometry was carried out for memory T cell subsets (Figure 3B). Cells were identified as naive (CCR7+CD45RA+), central memory (CM, CCR7+CD45RA−), effector memory (EM, CCR7−CD45RA−), or effector memory RA+ (EMRA, CCR7−CD45RA+). Tumor CD8+ TILs are mainly composed of EM cells (48.8%) (Figure 3B). The frequency of naive CD8+ T cells was sharply diminished in tumors compared with peripheral blood and adjacent non-tumor tissues (Figure 3B). These data indicate the preferential accumulation of antigen-experienced EM cells in tumor tissue. Subsequently, CD8+ T cells were isolated from peripheral blood, adjacent non-tumor tissues, and BC samples followed by miR-139 detection. miR-139 was upregulated in tumor infiltrated CD8+ T cells compared with the peripheral CD8+ T cells and adjacent non-tumor samples (Figure 3C). Meanwhile, tumor infiltrated CD8+ T cells with higher miR-139 level expressed less cell surface CD28 (Figure 3D).Table 1Clinical characteristicsClinicopathological parametersPatients (N)Age (years)<5011≥5021SexFemale32Male0RaceAsian32Other0EthnicityHan31Hmong1Tumor size (mm3)>2012≤2020T stageT120T27T3441N stageN04N122N25N31M stageM028M14Histological gradeI7II19III6Molecular subtpeER positive21PR positive16HER2 postive18Triple negative7ER, estrogen receptor; PR, progesterone receptor; HER2, human epidermal growth factor receptor 2. ER, PR, or HER2 status was classed as positive when more than 30% of cells were stained.Figure 3Elevated miR-139 expression in tumor-infiltrating CD8+ T cells correlates with impaired CD28 signaling in breast cancer(A) miR-139-5p expression in matched tumor and adjacent normal tissues from 32 patients with breast cancer, quantified by RT-qPCR (∗∗∗∗p < 0.0001, paired t test).(B) Memory subset distribution of CD8+ T cells from PBMCs (N = 13 donors), tumor tissues (N = 13 patients), and paired adjacent non-tumor tissues (N = 13 patients), analyzed by flow cytometry. Gating strategy identifies naive (CCR7+CD45RA+), central memory (CM; CCR7+CD45RA−), effector memory (EM; CCR7−CD45RA−), and terminally differentiated EMRA (CCR7−CD45RA+) populations (∗p < 0.05, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001, one-way ANOVA).(C) miR-139-5p levels in flow-sorted CD8+ T cells from paired tumors, adjacent non-tumor tissues, and peripheral blood (n = 13; ∗p < 0.05, ∗∗∗p < 0.001, one-way ANOVA).(D) CD28 surface expression (MFI) on tumor-infiltrating vs. circulating CD8+ T cells (n = 13; ∗∗∗∗p < 0.0001, one-way ANOVA).Data represent mean ± SEM. Table 1Clinical characteristicsClinicopathological parametersPatients (N)Age (years)<5011≥5021SexFemale32Male0RaceAsian32Other0EthnicityHan31Hmong1Tumor size (mm3)>2012≤2020T stageT120T27T3441N stageN04N122N25N31M stageM028M14Histological gradeI7II19III6Molecular subtpeER positive21PR positive16HER2 postive18Triple negative7ER, estrogen receptor; PR, progesterone receptor; HER2, human epidermal growth factor receptor 2. ER, PR, or HER2 status was classed as positive when more than 30% of cells were stained. Clinical characteristics ER, estrogen receptor; PR, progesterone receptor; HER2, human epidermal growth factor receptor 2. ER, PR, or HER2 status was classed as positive when more than 30% of cells were stained. Elevated miR-139 expression in tumor-infiltrating CD8+ T cells correlates with impaired CD28 signaling in breast cancer (A) miR-139-5p expression in matched tumor and adjacent normal tissues from 32 patients with breast cancer, quantified by RT-qPCR (∗∗∗∗p < 0.0001, paired t test). (B) Memory subset distribution of CD8+ T cells from PBMCs (N = 13 donors), tumor tissues (N = 13 patients), and paired adjacent non-tumor tissues (N = 13 patients), analyzed by flow cytometry. Gating strategy identifies naive (CCR7+CD45RA+), central memory (CM; CCR7+CD45RA−), effector memory (EM; CCR7−CD45RA−), and terminally differentiated EMRA (CCR7−CD45RA+) populations (∗p < 0.05, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001, one-way ANOVA). (C) miR-139-5p levels in flow-sorted CD8+ T cells from paired tumors, adjacent non-tumor tissues, and peripheral blood (n = 13; ∗p < 0.05, ∗∗∗p < 0.001, one-way ANOVA). (D) CD28 surface expression (MFI) on tumor-infiltrating vs. circulating CD8+ T cells (n = 13; ∗∗∗∗p < 0.0001, one-way ANOVA). Data represent mean ± SEM. CD28 receptor on T cells provides a critical costimulatory signal for T cell activation. Reduced CD28 expression on T cells relates to T cell exhaustion in patients with acute myeloid leukemia.15 To explore the role of miR-139 in infiltrated CD8+ T cells in BC, the cell surface levels of checkpoint molecules (PD1, TIGIT, and CD244) that represent exhausted phenotypes were detected by flow cytometry in antigen-experienced CD8+ T cells (CD45RA-CD8+) and peripheral CD8+ T cells. The frequencies of PD-1+, TIGIT+, or CD244+ cells in tumor infiltrated CD8+ T cells were significantly higher than that in peripheral blood mononuclear cells (PBMCs) (Figure 4A) and adjacent non-tumor tissues (Figure S4), indicating an exhausted phenotype. Antigen experienced CD8+ T cells with higher levels of miR-139 expressed lower levels of CD28 but higher levels of PD-1, TIGIT, and CD244 (Figure 4B). Furthermore, reduced CD28 level relates to upregulated PD-1, TIGIT, and CD244 (Figure 4C). Furthermore, CD8+ TILs transfected with miR-139 mimic expressed increased PD-1, TIGIT, and CD244 (Figure 4D),while CD8+ TILs transfected with miR-139 inhibitors expressed reduced PD-1, TIGIT, and CD244 (Figure 4E).Figure 4Elevated miR-139 in CD8+ TILs is associated with an exhausted phenotype characterized by co-inhibitory receptor upregulation(A) Flow cytometric analysis of immune checkpoint expression on non-naive (CD45RA-) CD8+ T cells from PBMCs and tumor tissues. Left: representative contour plots showing gating strategy. Right: quantification of checkpoint-positive cells (n = 32, ∗∗∗∗p < 0.0001, t test).(B) Comparative analysis of checkpoint molecule expression in CD45RA- CD8+ T cells stratified by intratumoral miR-139 levels. Data show significantly reduced CD28 but elevated PD-1 and TIGIT expression in the miR-139-high group (miR-139-high group n = 16, miR-139-low group n = 16, ∗∗p < 0.05, ∗∗∗p < 0.001, t test).(C) CD28-low CD45RA- CD8+ T cells exhibited significantly higher expression of exhaustion markers (PD-1, TIGIT, and CD244) compared with CD28-high counterparts (∗∗p < 0.01, ∗∗∗p < 0.001, t test).(D) miR-139 overexpression significantly upregulated PD-1, TIGIT, and CD244 mRNA levels (mimic vs. negative control [NC]; 72 h post-transfection) (n = 3 repeats, ∗p < 0.05, t test).(E) miR-139 inhibition, conversely, reduced these markers (inhibitor vs. NC) (n = 3 repeats, ∗p < 0.05, t test).Data represent mean ± SEM. Elevated miR-139 in CD8+ TILs is associated with an exhausted phenotype characterized by co-inhibitory receptor upregulation (A) Flow cytometric analysis of immune checkpoint expression on non-naive (CD45RA-) CD8+ T cells from PBMCs and tumor tissues. Left: representative contour plots showing gating strategy. Right: quantification of checkpoint-positive cells (n = 32, ∗∗∗∗p < 0.0001, t test). (B) Comparative analysis of checkpoint molecule expression in CD45RA- CD8+ T cells stratified by intratumoral miR-139 levels. Data show significantly reduced CD28 but elevated PD-1 and TIGIT expression in the miR-139-high group (miR-139-high group n = 16, miR-139-low group n = 16, ∗∗p < 0.05, ∗∗∗p < 0.001, t test). (C) CD28-low CD45RA- CD8+ T cells exhibited significantly higher expression of exhaustion markers (PD-1, TIGIT, and CD244) compared with CD28-high counterparts (∗∗p < 0.01, ∗∗∗p < 0.001, t test). (D) miR-139 overexpression significantly upregulated PD-1, TIGIT, and CD244 mRNA levels (mimic vs. negative control [NC]; 72 h post-transfection) (n = 3 repeats, ∗p < 0.05, t test). (E) miR-139 inhibition, conversely, reduced these markers (inhibitor vs. NC) (n = 3 repeats, ∗p < 0.05, t test). Data represent mean ± SEM. To further investigate whether miR-139 overexpression induces a functional exhaustion phenotype in CD8+ T cells, we examined cell proliferation following miR-139 transfection. We found that miR-139 significantly inhibited CD8+ T cell proliferation (Figure S5A). Meanwhile, miR-139 reduced the percentages of granzyme B-positive and CD107a-positive cells (Figure S5B). In vitro cytotoxicity assays were performed using MCF-7 and MDA-MB-231 cells as targets at effector-to-target (E:T) ratios ranging from 0.5:1 to 5:1. The results demonstrated that miR-139 overexpression impaired the cytolytic activity of T cells (Figure S5C). Furthermore, ELISA analysis revealed that miR-139 suppressed the secretion of IFN-γ and TNF-α into the culture supernatant (Figure S5D). These results indicated that increased miR-139 in CD8+ T cells contributed to exhaustion phenotype. To identify whether miR-139 induced CD8+ T cell exhaustion through targeting CD28, CD8+ T cells were transfected with miR-139 mimic with or without CD28 overexpression vector. The cell surface CD28 level was detected by flow cytometry (Figure S6A). The mRNA levels of TCF7, PD-1, TIGIT, TIM3, and LAG3 were examined by RT-qPCR. We observed that TCF7 level was repressed by miR-139, which was rescued totally by CD28 (Figure S6B). Meanwhile, the levels of PD-1, TIGIT, TIM3, and LAG3 were increased by miR-139 mimic, the phenotype of which was recovered by CD28 (Figure S6B). Furthermore, CD8+ T cells transfected with miR-139 mimic had reduced viability, while cell viability showed no significant difference between control cells and cells co-transfected with miR-139 mimic and CD28 expression vector (Figure S6C). These data indicate that increased miR-139 in CD8+ T cells relate to exhaustion phenotype via repressing CD28 expression. BC tumors with higher miR-139 levels recruit CD8+ T cells via increased production of CX3L1, CXCL12, and CXCL14 Exhausted T cells have reduced proliferation capacity, which could not explain the increased proportion of CD8+ T cells in tumors with higher levels of miR-139. Chemokines are a group of key factors controlling T cell movement, which can be expressed by both tumor cells and immune cells.16 Therefore, we hypothesized that miR-139 dysregulation in tumor cells might upregulate some chemokines, promote T cell infiltration, and then induce their exhaustion. We analyzed the main chemokine expression in tumors with low or high levels of miR-139. As shown in Figure 5A, BC tumors with higher miR-139 level had higher levels of CX3CL1, CXCL12, CXCL14, CXCL2, XCL1, and XCL2, but lower levels of CXCL8, CXCL10, CXCL11, and CXCL17. Since CX3CL1, CXCL12, and CXCL14 were reported to be positively regulated to T cell infiltration in solid tumors, we detected their expression by immunohistochemistry. We observed that tumors with higher level of miR-139 expressed higher levels of CX3CL1, CXCL12, and CXCL14 (Figure 5B). However, the main receptors for these chemokines in CD8+ T cells were not significantly altered (Figure 5C). Meanwhile, CX3CL1, CXCL12, and CXCL14 could individually promote CD8+ T cell migration in a dose dependent manner (Figure 5D). Furthermore, the cell culture medium from BC cells (MCF7 or MDA-MB-231) promoted CD8+ T cell migration (Figure 5E), while the cell culture medium collected from cells overexpressing miR-139 demonstrated the most potent chemotaxis-promoting activity (Figure 5E).Figure 5High miR-139 expression in CD8+ T cells correlates with distinct chemokine profiles and enhanced T cell migration in breast cancer(A) Differential expression analysis of 17 chemokines in TCGA breast cancer cohort (N = 1,088) stratified by miR-139 levels in CD8+ T cells. Heatmap shows significantly upregulated chemokines (red) in miR-139-high tumors (∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001, Wilcoxon test).(B) Immunohistochemical validation of top candidate chemokines (CX3CL1, CXCL12, and CXCL14) in an independent cohort (N = 32). Left: representative images (scale bars, 100 μm). Right: quantification by H-score. (∗p < 0.05, ∗∗∗p < 0.001, t test).(C) Flow cytometric analysis of chemokine receptor expression on non-naive (CD45RA-) CD8+ T cells. Boxplots show median with IQR (miR-139-high group n = 15, miR-139-low group n = 16, t test).(D and E) In vitro chemotaxis assay (n = 3 repeats, ∗p < 0.05, ∗∗p < 0.01, t test).Data represent mean ± SEM. High miR-139 expression in CD8+ T cells correlates with distinct chemokine profiles and enhanced T cell migration in breast cancer (A) Differential expression analysis of 17 chemokines in TCGA breast cancer cohort (N = 1,088) stratified by miR-139 levels in CD8+ T cells. Heatmap shows significantly upregulated chemokines (red) in miR-139-high tumors (∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001, Wilcoxon test). (B) Immunohistochemical validation of top candidate chemokines (CX3CL1, CXCL12, and CXCL14) in an independent cohort (N = 32). Left: representative images (scale bars, 100 μm). Right: quantification by H-score. (∗p < 0.05, ∗∗∗p < 0.001, t test). (C) Flow cytometric analysis of chemokine receptor expression on non-naive (CD45RA-) CD8+ T cells. Boxplots show median with IQR (miR-139-high group n = 15, miR-139-low group n = 16, t test). (D and E) In vitro chemotaxis assay (n = 3 repeats, ∗p < 0.05, ∗∗p < 0.01, t test). Data represent mean ± SEM. To explore the possible mechanisms by which miR-139 modulates the levels of CX3CL1, CXCL12, and CXCL14, we screened the predicted targets of miR-139 after functional annotation. We noticed that 13 genes are functionally enriched in regulating gene expression, and 15 genes were enriched in chromatin remodeling (Figure S7). Among these candidate targets of miR-139, GLI2, a transcriptional effector of the Hedgehog signaling pathway, was identified to be related to immune infiltration and tumor growth. Scales et al. generated GLI2 and GLI3 knockout mice and observed that GLI2 and GLI3 knockout fibroblasts change cytokines expression and promote T cell migration.17 However, the roles of GLI2 in human BC remain unclear. We subsequently knocked down GLI2 in MCF7 and MDA-231 by using siRNA (Figure 6A) and detected the expression of CX3CL1, CXCL12, and CXCL14 by RT-qPCR. The secretory level of these chemokines in cell culture medium was detected by ELISA. We found that both mRNA (Figure 6B) and protein (Figure 6C) levels of CX3CL1, CXCL12, and CXCL14 were upregulated in GLI2 knockdown cells, indicating that GLI2 suppressed these chemokines’ expressions. To confirm whether miR-139 controls GLI2 expression, we constructed a reporter vector by cloning the full length of GLI2 3′ UTR containing two predicted miR-139 target sites (Figure 6D). The results of dual luciferase assay indicated that miR-139 could directly target GLI2 3′ UTR and repress its expression (Figure 6E). When the two predicted miR-139 target sites were mutated, the luciferase activity was not inhibited by miR-139 (Figure S8). These data indicated that tumors with higher levels of miR-139 produced more CX3CL1, CXCL12, and CXCL14 to recruit CD8+ T cells. When T cells infiltrated into the TME, some factors in the TME induced miR-139 expression in CD8+ T cells, which further repressed CD28 expression and induced exhaustion.Figure 6MiR-139 promotes chemokines production in breast cancer cells through targeting GLI2(A–C) siRNAs targeting GLI2 were transfected into MCF7 and MDA-MB-231 cells. The mRNA level of GLI2 was detected by RT-qPCR 48 h after transfection (A) (n = 3 repeats, ∗∗p < 0.01, t test). The mRNA levels of CX3CL1, CXCL12, and CXCL14 were examined by RT-qPCR (B) (n = 3 repeats, ∗p < 0.05, ∗∗p < 0.01, One-way ANOVA). The secretory levels of CX3CL1, CXCL12, and CXCL14 in cell culture medium were detected by ELISA 96 h after transfection (C) (n = 3 repeats, ∗p < 0.05, ∗∗p < 0.01, one-way ANOVA).(D) The schematic diagram of predicted interaction between miR-139 and GLI2 3′ UTR.(E) Dual luciferase assay (n = 3 repeats, ∗p < 0.05, ∗∗p < 0.01, t test).Data represent mean ± SEM. MiR-139 promotes chemokines production in breast cancer cells through targeting GLI2 (A–C) siRNAs targeting GLI2 were transfected into MCF7 and MDA-MB-231 cells. The mRNA level of GLI2 was detected by RT-qPCR 48 h after transfection (A) (n = 3 repeats, ∗∗p < 0.01, t test). The mRNA levels of CX3CL1, CXCL12, and CXCL14 were examined by RT-qPCR (B) (n = 3 repeats, ∗p < 0.05, ∗∗p < 0.01, One-way ANOVA). The secretory levels of CX3CL1, CXCL12, and CXCL14 in cell culture medium were detected by ELISA 96 h after transfection (C) (n = 3 repeats, ∗p < 0.05, ∗∗p < 0.01, one-way ANOVA). (D) The schematic diagram of predicted interaction between miR-139 and GLI2 3′ UTR. (E) Dual luciferase assay (n = 3 repeats, ∗p < 0.05, ∗∗p < 0.01, t test). Data represent mean ± SEM. To explore the mechanism by which GLI2 represses the expression of candidate chemokines, we performed ChIP assays followed by qPCR to determine whether GLI2 binds to their promoter regions. As shown in Figure S9A, endogenous GLI2 was successfully recruited using a GLI2-specific antibody. Meanwhile, two segments of the CXCL14 promoter were significantly enriched by the GLI2 antibody (Figures S9B and S9C). In contrast, the promoter regions of CX3CL1 and CXCL12 showed no obvious enrichment (Figures S9B and S9C). These results indicate that GLI2 inhibits CXCL14 expression by directly binding to its promoter, whereas GLI2 suppresses CX3CL1 and CXCL12 expression through an indirect mechanism. T cell exhaustion in cancer is driven by various extrinsic factors, including tumor cells, inflammatory cells, and immunosuppressive cytokines. Additionally, metabolites accumulating within the TME can shape T cell fate.18,19 Hypoxia, a common metabolic stress in the TME, has been shown to preferentially affect terminally exhausted intratumoral T cells.19,20 To identify factors that induce miR-139 expression, we cultured CD8+ T cells under hypoxic conditions and observed a 2.61-fold upregulation of miR-139 compared with normoxic controls (Figure S10A). Furthermore, treatment of CD8+ T cells with the cytokine IL-10 also upregulated miR-139 expression (Figure S10B). These results suggest that miR-139 expression in CD8+ T cells can be induced by both hypoxia and immunosuppressive cytokines, such as IL-10. We also compared the miR-139 expression levels between CD8+ T cells and primary tumor cells isolated from five BC tumors. MiR-139 was detectable in both cell types, with relatively higher levels observed in CD8+ T cells compared with tumor cells (Figure S11). Discussion miRNAs are a group of short non-coding RNAs that usually have multiple targets and their functions are always cell-type-specific. It is reported that miR-139 is frequently downregulated and functions as a tumor suppressor. In this study, we analyzed the high throughput sequencing data from TCGA portal and two GEO datasets and detected miR-139 levels in 32 BC tumor samples. We confirmed the downregulation of miR-139 in BC tumors. Meanwhile, we noticed that higher miR-139 level was related to shorter overall survival time, which is not consistent with the tumor suppressor role of miR-139. After analyzing the sequencing data from TCGA portal, we found that miR-139 may regulate T cell compositions in the TME, especially CD8+ T cells, which have the capacity to kill tumor cells. We analyzed the isolated CD8+ T cells from BC tumors and found an increased expression of miR-139, which induced CD8+ T cell exhaustion by inhibiting CD28 expression. This study dissects the role of miR-139 in T cells in the BC TME, providing a potential target for BC immunotherapy. Chemokines are tiny, secreted proteins that can induce immune cells infiltration. We compared the chemokines expression between BC tumors that differentially expressed miR-139 and tried to explore why CD8+ T cells increased in BC tumors that expressed higher levels of miR-139. We found miR-139 level is related to different chemokine expression patterns. BC tumors with higher levels of miR-139 expressed higher levels of CX3CL1, CXCL12, and CXCL14. CX3CL1 is a multifunctional inflammatory chemokine with a single receptor CX3CR1. Emerging evidence indicates that CX3CL1 level is positively correlated with CD8+ T cells in BC, and high CX3CL1 level is related to a patient’s good prognosis, suggesting its important role in recruiting CD8+ T cell.21,22 However, Su et al. reported that lactate/GPR81 recruits regulatory T cells (Tregs) by modulating CX3CL1 without influencing CD8+ T cells in gastric cancer.23 In this study, we isolated CD8+ T cells from PBMCs and examined the chemotaxis of CX3CL1 in vitro. Meanwhile, as reported by Su et al., CX3CL1 may also recruit Tregs, and that might contribute to the poor prognosis of patients. CXCL12 and CXCL14 were also identified to promote activated CD8+ T cell chemotaxis in vitro and recruited tumor infiltrating CD8+ T cells in vivo in multiple solid tumors.24,25 Meanwhile, the main chemokine receptors (CX3CR1, CXCR3, and CXCR4) were not significantly altered by miR-139, suggesting that high miR-139 level BC tumors recruited CD8 T cells through upregulating chemokines. However, we noticed that some chemokines, such as CXCL10, could also contribute to the generation of a “hot” TME, which was downregulated in tumors with high miR-139 level. These data indicated that the CD8+ T cells recruitment in BC tumors is a result of multiple dysregulated chemokines, the mechanism of which may be complex and need to be further investigated. CXCR4, the receptor for CXCL12 and CXCL14 that expressed on T cells, was identified to be a direct target of miR-139 in T cell acute lymphoblastic leukemia (T-ALL) cell lines.26 However, in this study, we analyzed the CXCR4 expression on CD8+ T cells and did not find a negative correlation between miR-139 and CXCR4 levels (Figure S12). These results indicated that miR-139 function is cell-type-specific, and CXCR4 expression is not repressed by miR-139 in CD8+ T cells. Alternatively, we identified that CD28 is one of the direct targets in T cells. Emerging evidence indicates that CD28 co-stimulation might be critical for active T cell response in tumor.27,28 Loss of CD28 expression is found to be related to severe T cell exhaustion in acute myeloid leukemia.15 In this study, we confirmed that miR-139 upregulation in CD8+ T cells is one of the key factors inducing T cell exhaustion in BC and provides a candidate target for immunotherapy. In the TME, exhausted T cells fail to eliminate cancer due to a functional decline characterized by the overexpression of inhibitory receptors (e.g., PD-1, CTLA-4, TIM-3, LAG-3, BTLA, and TIGIT) and impaired production of effector cytokines and cytolytic molecules. While transcription factors like Blimp-1, T-bet, NFATc1, and BATF have been established as critical regulators of T cell exhaustion in chronic infection,29,30,31,32 the intracellular signaling pathways that govern this dysfunctional state in cancer remain poorly understood. In this study, we identified that miR-139 overexpression in CD8+ T cells contributes to exhaustion by targeting CD28, shedding light on of intracellular mechanism inducing T cell exhaustion. The potential extrinsic factors contributing to T cell exhaustion in cancer include tumor cells, inflammatory cells, and immunosuppressive cytokines. In addition, metabolites accumulating in the TME can also shape T cell fate.18,19 Hypoxia, a common metabolic stress in tumors, has been shown to be more prevalent in terminally exhausted intratumoral T cells.19,20 To identify factors that induce miR-139 expression, we cultured CD8+ T cells under hypoxic conditions and observed a 2.61-fold upregulation of miR-139 compared with normoxic controls. Furthermore, treatment of CD8+ T cells with the cytokine IL-10 also upregulated miR-139 expression. These results suggest that miR-139 in CD8+ T cells may be induced by both hypoxia and immunosuppressive cytokines. However, given the complexity of the tumor immunosuppressive microenvironment, miR-139 expression is likely regulated by multiple factors. Further studies employing in vivo models and high-throughput approaches will help elucidate the systematic regulatory mechanisms of miR-139. Furthermore, to explore the relationship between miR-139 expression and BC subtypes, we utilized the online bioinformatics tool UALCAN (https://ualcan.path.uab.edu/) to analyze TCGA_BRCA data. The results showed that miR-139 is broadly downregulated in all subtypes of BC (Figure S13), indicating that the immunoregulatory mechanism we proposed is widely applicable across different BC subtypes. This broad downregulation pattern of miR-139 across BC subtypes highlights its potential as a universal regulatory molecule in BC immune microenvironment remodeling, which is consistent with the functional roles of miR-139 we observed in in vitro experiments. Further analysis of specific BC subtypes revealed that triple-negative BC (TNBC) and HER2-positive tumors exhibit relatively lower miR-139 expression levels compared with luminal tumors (Figure S13A). Additionally, TP53 mutant tumors show relatively lower miR-139 levels than TP53 wild-type tumors (Figure S13B). However, the biological implications of these expression differences remain unclear and require further investigation. It has been identified that miR-139 could repress tumor cell growth in BC. This study confirmed the tumor-suppressive function of miR-139 in BC cells, while also revealing its immunosuppressive role in CD8+ T cells. Given this dual role of miR-139 in both tumor cells and immune cells, cell-type-specific delivery is crucial for its potential clinical application. In recent years, various innovative drug delivery systems have been developed for cancer treatment. A wide range of nanoscale materials—including synthetic polymers, proteins, lipids, and organic or inorganic particles—have been utilized in the design of cancer therapeutics.33 We propose that nanoparticles conjugated with tumor-antigen-targeting antibodies and carrying miR-139 may directly induce tumor cell death. Meanwhile, nanoparticles functionalized with antibodies against CD8+ T cell surface proteins and loaded with miR-139 antagonists may help alleviate T cell exhaustion and enhance immunotherapeutic efficacy. In conclusion, we identified that the levels of miR-139 increased in CD8+ T cells in BC and induced T cell exhaustion by targeting CD28. Tumors with higher levels of miR-139 expressed upregulated CX3CL1, CXCL12, and CXCL14 that recruited T cells, inducing their exhaustion by upregulating miR-139. Limitations of the study There are some limitations of this study: (1) although hypoxia and IL-10 were identified as inducers of miR-139 upregulation in CD8+ T cells in vitro, the immunosuppressive TME is far more complex. This complexity arises from factors such as metabolic alterations, interactions with immunosuppressive cells and cytokines, acidic pH, and tumor-derived exosomes. Therefore, a comprehensive identification of the regulators governing miR-139 expression in vivo remains an important area for future investigation. (2) We identified that high miR-139 expression correlates with abnormal chemokine levels in BC. We further confirmed that miR-139 regulates the expression of CX3CL1, CXCL12, and CXCL14 by targeting GLI2. Using ChIP-qPCR, we demonstrated that GLI2 directly binds to the promoter of CXCL14 but not to those of CX3CL1 and CXCL12, suggesting that GLI2 modulates the expression of these chemokines through distinct mechanisms. Nevertheless, the detailed molecular mechanisms underlying GLI2-mediated chemokine regulation in BC cells remain to be further elucidated; (3) Through analyzing the high-throughput sequencing data, we found that high levels of miR-139 are related to altered immune cells compositions in BC. This study only elucidates the role of miR-139 in CD8+ T cells. The mechanism by which miR-139 modulates other immune cells infiltration needs to be investigated in the future. (4) We have identified the dual role of miR-139 in both BC cells and CD8+ T cells. However, the proposed mechanism by which miR-139 promotes CD8+ T cell exhaustion requires further validation in animal models. (5) Given that all enrolled participants were female and predominantly of the Han Chinese ethnicity, we were unable to fully evaluate the potential impacts of sex, race, and ethnicity on our findings. Resource availability Lead contact Requests for further information and resources should be directed to and will be fulfilled by the lead contact, Mengjun Hu (hu-mengjun@outlook.com). Materials availability All unique/stable reagents generated in this study are available from the lead contact without restriction. Data and code availability •This paper analyzes existing, publicly available data, accessible at Genomic Data Commons Data Portal; GEO database: GSE37963 and GSE154255.•This paper does not report original code.•Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request. •This paper analyzes existing, publicly available data, accessible at Genomic Data Commons Data Portal; GEO database: GSE37963 and GSE154255. This paper analyzes existing, publicly available data, accessible at Genomic Data Commons Data Portal; GEO database: GSE37963 and GSE154255. •This paper does not report original code. This paper does not report original code. •Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request. Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request. Acknowledgments We acknowledge all the help received from the staff at the central laboratory of Zhuji People’s Hospital. Author contributions Conceptualization, J.C. and M.H.; methodology, J.C., H.Z., Y.L., and X.L.; investigation, J.C., H.Z., Y.L., and X.L.; writing – original draft, J.C. and H.Z.; writing – review and editing, M.H.; funding acquisition, M.H.; resources, J.C. and M.H.; supervision, M.H. Declaration of interests The authors declare no competing interests. STAR★Methods Key resources table REAGENT or RESOURCESOURCEIDENTIFIERAntibodiesmouse anti-GLI2 monoclonal antibodySanta Cruz BiotechnologyRRID: AB_10708124; Cat.No. sc-271786BV510 Mouse Anti-Human CD45BD BiosciencesRRID:AB_2738067; Cat No: 563204FITC Mouse Anti-Human CD3BD BiosciencesRRID:AB_3686061; Cat No: 570821PE Mouse Anti-Human CD8BD BiosciencesRRID:AB_3686103; Cat No: 570894PerCP-Cy™5.5 Mouse anti-Human CD279 (PD-1)BD BiosciencesRRID:AB_10645786; Cat No: 561273RB780 Mouse Anti-Human TIGITBD BiosciencesRRID:AB_3685420; Cat No: 569940APC Mouse Anti-Human CD244BD BiosciencesRRID:AB_11153502; Cat No: 562350Mouse Anti-Human CD45RABD BiosciencesRRID:AB_3685186; Cat No: 569620BV605 Mouse Anti-Human CD28BD BiosciencesRRID:AB_2737927; Cat No: 562976Biological samplesTreatment-naïve breast cancer patientsZhuji People’s Hospital of Zhejiang ProvinceN/AChemicals, peptides, and recombinant proteinsCD8+ T cell Isolation KitMiltenyi BiotecCat.No.130-094-156Fetal bovine serumN/APenicillin-Streptomycin (10,000 U/mL)Thermo Fisher ScientificCat.No. 15140122Dispase IIRocheCat.No. 4942078001Collagenase Type IVSigma-AldrichCat.No. C4-22-1GDnase IRoche11284932001EDTA (0.5 M)Thermo Fisher ScientificR1021TRIzol reagentThermo Fisher ScientificCat.No. 15596026TaqMan™ Advanced miRNA AssayThermo Fisher ScientificCat.No. A25576Collagenase ARocheCat.No. 101035780013,3′-Diaminobenzidine (DAB) Liquid SubstrateSigma-AldrichCat.No. D7304-1SETX-VIVO 15 serum-free mediumLonzaCat.No. 02-060QHuman IL-2 Recombinant ProteinPeproTechCat.No. 200-02-50UGHuman Fractalkine/CX3CL1 ELISA KitThermo Fisher ScientificCat.No. EHCX3CL1Human SDF-1 beta/CXCL12B ELISA KitThermo Fisher ScientificCat.No. EHCXCL12BHuman BRAK/CXCL14 ELISA KitThermo Fisher ScientificCat.No. EHCXCL14Human TNF alpha ELISA KitThermo Fisher ScientificCat.No. ECH009Human IFN gamma ELISA KitThermo Fisher ScientificCat.No. KHC4021Protein A AgaroseThermo Fisher ScientificCat.No. 20334Deposited dataGSE37963NCBI Gene Expression Omnibushttps://www.ncbi.nlm.nih.gov/gds/?term=GSE37963+GSE154255NCBI Gene Expression Omnibushttps://www.ncbi.nlm.nih.gov/gds/?term=GSE154255Experimental models: Cell linesMCF-7American Type Culture CollectionRRID:CVCL_0031MDA-MB-231American Type Culture CollectionRRID:CVCL_0062Primary human CD8+ T cellsZhuji People’s Hospital of Zhejiang ProvinceN/AOligonucleotidesmiR-139 mimic and inhibitorRIBOBIOCat.No. miR20000250-1-5;Cat.No. miR10000250-1-5Software and algorithmsR softwareR Foundationhttps://www.r-project.org/FlowJoBD bioscienceshttps://www.flowjo.com/SPSSIBMhttps://www.ibm.com/support/pages/downloading-ibm-spss-modeler-160 REAGENT or RESOURCESOURCEIDENTIFIERAntibodiesmouse anti-GLI2 monoclonal antibodySanta Cruz BiotechnologyRRID: AB_10708124; Cat.No. sc-271786BV510 Mouse Anti-Human CD45BD BiosciencesRRID:AB_2738067; Cat No: 563204FITC Mouse Anti-Human CD3BD BiosciencesRRID:AB_3686061; Cat No: 570821PE Mouse Anti-Human CD8BD BiosciencesRRID:AB_3686103; Cat No: 570894PerCP-Cy™5.5 Mouse anti-Human CD279 (PD-1)BD BiosciencesRRID:AB_10645786; Cat No: 561273RB780 Mouse Anti-Human TIGITBD BiosciencesRRID:AB_3685420; Cat No: 569940APC Mouse Anti-Human CD244BD BiosciencesRRID:AB_11153502; Cat No: 562350Mouse Anti-Human CD45RABD BiosciencesRRID:AB_3685186; Cat No: 569620BV605 Mouse Anti-Human CD28BD BiosciencesRRID:AB_2737927; Cat No: 562976Biological samplesTreatment-naïve breast cancer patientsZhuji People’s Hospital of Zhejiang ProvinceN/AChemicals, peptides, and recombinant proteinsCD8+ T cell Isolation KitMiltenyi BiotecCat.No.130-094-156Fetal bovine serumN/APenicillin-Streptomycin (10,000 U/mL)Thermo Fisher ScientificCat.No. 15140122Dispase IIRocheCat.No. 4942078001Collagenase Type IVSigma-AldrichCat.No. C4-22-1GDnase IRoche11284932001EDTA (0.5 M)Thermo Fisher ScientificR1021TRIzol reagentThermo Fisher ScientificCat.No. 15596026TaqMan™ Advanced miRNA AssayThermo Fisher ScientificCat.No. A25576Collagenase ARocheCat.No. 101035780013,3′-Diaminobenzidine (DAB) Liquid SubstrateSigma-AldrichCat.No. D7304-1SETX-VIVO 15 serum-free mediumLonzaCat.No. 02-060QHuman IL-2 Recombinant ProteinPeproTechCat.No. 200-02-50UGHuman Fractalkine/CX3CL1 ELISA KitThermo Fisher ScientificCat.No. EHCX3CL1Human SDF-1 beta/CXCL12B ELISA KitThermo Fisher ScientificCat.No. EHCXCL12BHuman BRAK/CXCL14 ELISA KitThermo Fisher ScientificCat.No. EHCXCL14Human TNF alpha ELISA KitThermo Fisher ScientificCat.No. ECH009Human IFN gamma ELISA KitThermo Fisher ScientificCat.No. KHC4021Protein A AgaroseThermo Fisher ScientificCat.No. 20334Deposited dataGSE37963NCBI Gene Expression Omnibushttps://www.ncbi.nlm.nih.gov/gds/?term=GSE37963+GSE154255NCBI Gene Expression Omnibushttps://www.ncbi.nlm.nih.gov/gds/?term=GSE154255Experimental models: Cell linesMCF-7American Type Culture CollectionRRID:CVCL_0031MDA-MB-231American Type Culture CollectionRRID:CVCL_0062Primary human CD8+ T cellsZhuji People’s Hospital of Zhejiang ProvinceN/AOligonucleotidesmiR-139 mimic and inhibitorRIBOBIOCat.No. miR20000250-1-5;Cat.No. miR10000250-1-5Software and algorithmsR softwareR Foundationhttps://www.r-project.org/FlowJoBD bioscienceshttps://www.flowjo.com/SPSSIBMhttps://www.ibm.com/support/pages/downloading-ibm-spss-modeler-160 Experimental model and study participant details Ethics approval statement This study was performed in accordance with the Declaration of Helsinki and approved by the Ethics Committee of Zhuji People’s Hospital (Approval No. 2023-1221). All methods were carried out in accordance with relevant guidelines and regulations. Patients and samples collection A cohort of 32 treatment-naïve BC patients was enrolled at Zhuji People’s Hospital of Zhejiang Province between January 2018 and November 2019. All participants were of Asian ethnicity, including 31 Han Chinese and one Hmong individual. All patients were female; among them, 21 were aged 50 years or older, and 11 were younger than 50 years. Patients were stratified into low and high miR-139 expression groups based on intratumoral miR-139 levels. All participants met the following inclusion criteria: (1) newly diagnosed BC without prior treatment, (2) no history of preoperative chemotherapy or radiotherapy, and (3) willingness to provide paired tumor and adjacent normal tissue samples (collected ≥2 cm from tumor margins). Exclusion criteria include: (1) prior history of any anti-tumor treatment, including but not limited to chemotherapy, radiotherapy, targeted therapy, immunotherapy, endocrine therapy, (2) Inability to provide qualified paired tumor and adjacent normal tissue samples, (3) Comorbidity with other malignant tumors, and (4) Pregnancy or lactation status. Following surgical resection, all specimens underwent immediate histopathological verification of tumor type and grade by two independent pathologists. The specimens were rapid cryopreservation in liquid nitrogen within 30 min post-resection and then long-term storage at −80°C in a monitored biobank until analysis. The study protocol was approved by the Human Basic and Clinical Research Ethics Committee of Zhuji People’s Hospital (Approval No. 2023-1221) and conducted in full compliance with the Declaration of Helsinki. Prior to enrollment, all participants provided written informed consent after receiving complete information about the study objectives and procedures. Access to identifiable patient information was strictly limited to authorized research personnel following institutional data protection protocols. Comprehensive clinicopathological characteristics, including tumor stage, histological grade, and molecular subtype, were systematically recorded for all cases (Table 1). In vitro cell culture Human breast cancer MCF-7 and MDA-MB-231 cells were obtained from American Type Culture Collection (ATCC) and cultured in Dulbecco’s Modified Eagle Medium containing 10% fetal bovine serum (Hyclone, Logan, UT, USA),100 IU/mL penicillin and 10 mg/mL streptomycin. All cells were maintained at 37°C under an atmosphere of 5% CO2. All breast cancer cell lines used in this study were authenticated via short tandem repeat (STR) profiling before experiments. Routine mycoplasma testing was also performed throughout the culture period, and no mycoplasma contamination was detected. Primary human CD8+ T cells were divided into two conditions: in 1.5% hypoxia chamber or regular incubator. Cells were cultured in 20 IU/mL recombinant human IL-2 in RPMI-1640 medium containing 10% fetal bovine serum (Hyclone, Logan, UT, USA), 100 IU/mL penicillin and 10 mg/mL streptomycin. Routine mycoplasma testing was also performed throughout the culture period, and no mycoplasma contamination was detected. Method details Databases This study included gene expression data from 1,088 breast cancer patients sourced from The Cancer Genome Atlas (TCGA). The comprehensive dataset, accessed through the cBioPortal platform (https://www.cbioportal.org/) on February 19, 2021. Protein expression patterns in both normal and malignant breast tissues were retrieved from the Human Protein Atlas (HPA) database (https://www.proteinatlas.org/). T cell isolation from peripheral blood and electroporation CD8+ T lymphocytes were purified from peripheral blood mononuclear cells (PBMCs) using the CD8+ T cell Isolation Kit (Miltenyi Biotec, catalog #130-094-156) through magnetic-activated cell sorting (MACS) with negative selection. Following isolation, cell concentration was determined by trypan blue exclusion counting, and cells were resuspended in electroporation buffer containing either miR-139 mimic or inhibitor (20 μM final concentration). For cell transfection, we employed the Amaxa 4D-Nucleofector System (Lonza) using program FI-115 with the Human T cell Nucleofector Kit according to manufacturer’s specifications. Immediately post-transfection, cells were transferred to pre-equilibrated complete RPMI-1640 medium supplemented with 10% fetal bovine serum and 1% penicillin-streptomycin.Transfected CD8+ T cells were maintained at 37°C in a humidified incubator with 5% CO2 atmosphere for subsequent functional assays. Chemotaxis assay For chemotaxis evaluation, 2 × 105 CD8+ T cells in serum-free medium were seeded into the upper chamber of 5 μm-pore Transwell inserts (Corning, USA). The lower chamber contained either: conditioned medium from breast cancer cell cultures, or chemokine solutions consisting of recombinant human CX3CL1, CXCL12, or CX3CL14 (all from PeproTech) diluted in complete medium. The total volume in each lower chamber was maintained at 600 μL. Following 6 h of incubation at 37°C under 5% CO2, migrated CD8+ T cells in the lower chamber were collected and quantified using a hemocytometer. All migration experiments were performed in triplicate. Isolation of tumor-infiltrating CD8+ T lymphocytes (TILs) Fresh tumor specimens were enzymatically digested in RPMI-1640 medium containing a triple-enzyme cocktail consisting of 3 mg/mL Dispase II (Sigma-Aldrich), 1 mg/mL Collagenase IV (Sigma-Aldrich) and 1 mg/mL DNase I (Roch). The tissue suspension was incubated at 37°C for 60 min with gentle mechanical disruption every 15 min using a 10 mL serological pipette. Enzymatic digestion was terminated by adding EDTA (Thermo Fisher) to a final concentration of 10 mM followed by 5 min incubation at room temperature. The resulting single-cell suspension was then filtered through a 70 μm nylon mesh (Falcon), washed three times with ice-cold PBS (centrifugation at 400 × g, 5 min, 4°C) and then subjected to CD8+ T cell enrichment using the human CD8+ T cell Isolation Kit (Miltenyi Biotec, Cat# 130-094-156) through negative selection according to the manufacturer’s protocol. Purified CD8+ T cells were either processed for flow cytometric analysis or cultured in X-VIVO 15 serum-free medium (Lonza) supplemented with 6000 IU/mL recombinant human IL-2 (PeproTech) at 37°C in a humidified 5% CO2 incubator. Tumor cell isolation Fresh tumor specimens (n = 5) were enzymatically digested in RPMI-1640 medium containing a triple-enzyme cocktail consisting of 3 mg/mL Dispase II (Sigma-Aldrich), 1 mg/mL Collagenase IV (Sigma-Aldrich) and 1 mg/mL DNase I (Roch). The tissue suspension was incubated at 37°C for 60 min with gentle mechanical disruption every 15 min using a 10 mL serological pipette. Enzymatic digestion was terminated by adding EDTA (Thermo Fisher) to a final concentration of 10 mM followed by 5 min incubation at room temperature. The resulting single-cell suspension was then filtered through a 70 μm nylon mesh (Falcon), washed three times with ice-cold PBS (centrifugation at 400 × g, 5 min, 4°C) and then subjected to tumor cells isolation by using Tumor Cell Isolation Kit (Miltenyi Biotec, Germany, cat. # 130-108-339). Briefly, the non-tumor cells are magnetically labeled with a cocktail of monoclonal antibodies conjugated with MACS MicroBeads. Then, the cell suspension is loaded onto a MACS Column, which is placed in the magnetic field of a MACS Separator. The magnetically labeled non-tumor cells are retained within the column. The unlabeled human tumor cells run through. Analysis of miRNA expression Total RNA extraction was performed using TRIzol Reagent (Thermo Fisher Scientific) following the manufacturer’s standardized protocol. RNA quality control was conducted using a NanoDrop One/OneC spectrophotometer (Thermo Fisher Scientific), with samples required to meet stringent quality thresholds: A260/A280 ratio of 1.8–2.1 and A260/A230 ratio ≥1.9. For miR-139-5p quantification, we employed the TaqMan Advanced miRNA Assay system (Thermo Fisher Scientific) with the levels of U6 snRNA as endogenous control. Comparative ΔΔCt method (2ˆ-ΔΔCt) was used for relative quantification. Flow cytometry analysis Tissue specimens were meticulously minced into 1–2 mm3 fragments and subjected to enzymatic digestion in DMEM (Gibco) supplemented with 2 mg/mL collagenase A (Roche) and 50 U/mL DNase I (Sigma-Aldrich). Digestion was carried out at 37°C for 1 h under gentle agitation to ensure efficient dissociation. The resulting cell suspension was sequentially filtered through a 70 μm nylon mesh cell strainer (Falcon) to eliminate undigested tissue debris. Following centrifugation at 400 × g for 5 min at 4°C, the cell pellet was washed twice with ice-cold PBS (pH 7.4) to remove enzymatic residues. Single-cell suspensions were incubated for 30 min at room temperature (20°C–25°C) in PBS containing 2% heat-inactivated FBS (Gibco) with the following antibody panels: Lymphocyte identification: CD45, CD3, and CD8; Checkpoint marker analysis: PD-1, TIGIT, and CD244. Samples were acquired on a BD LSRFortessa flow cytometer (BD Biosciences) using a sequential gating strategy: FSC-A vs. SSC-A: Gating on live cell populations. FSC-H vs. FSC-A: Exclusion of doublets to isolate single cells. CD45+: Selection of leukocytes. CD3+CD8+: Identification of cytotoxic T cells. CD45RA−: Enrichment of memory/effector subsets. CD45RA−CD8+ T cells were isolated via fluorescence-activated cell sorting (FACS) using a BD FACSAria III (4°C) and promptly processed for RNA extraction to ensure sample integrity. Immunohistochemistry Formalin-fixed, paraffin-embedded (FFPE) tissue sections were deparaffinized and subjected to heat-induced antigen retrieval in 10% citrate buffer using microwave irradiation (10 min, two cycles). After cooling to room temperature, endogenous peroxidase activity was quenched, and nonspecific binding was blocked with 5% bovine serum albumin (BSA) for 30 min at 25°C. Sections were then incubated with primary antibodies overnight at 4°C. Following three washes with Tris-buffered saline containing 0.1% Tween 20 (TBST), slides were treated with an HRP-conjugated secondary antibody. Immunoreactivity was visualized using a 3,3′-diaminobenzidine (DAB) chromogenic substrate kit according to the manufacturer’s instructions. Stained sections were imaged under bright-field microscopy (Leica DM2000, Leica Microsystems GmbH). For signal quantification, positively stained cells were enumerated in ≥10 randomly selected fields per section, with two sections analyzed per sample. The percentage of stained tumor cells (0–100%) was recorded, and an H-score (range: 0–300) was calculated using the formula: H-score = (Intensity × % positive cells). To minimize bias, automated measurements were independently validated by two blinded investigators. Dual-luciferase assay To generate the luciferase reporter vectors, segments of the 3′UTRs of CD28, CCR7, TGFB1, and GLI2 that contain the predicted miR-139 binding sites were individually cloned into the pmirGLO vector downstream of the firefly luciferase gene. Additionally, vectors carrying mutant versions of the CD28 or GLI2 3′UTRs were constructed to validate the specific miR-139 target sites. For luciferase reporter assays, MCF7 or MDA-MB-231 cells were seeded in 48-well plates. The reporter vector was co-transfected with miR-139 mimic or inhibitor by using lipofectamine 2000 (Thermo Fisher Scientific). Two days later, cells were harvested and assayed with the Dual-Luciferase Assay (Promega). Each treatment was performed in triplicate in three independent experiments. The results were expressed as relative luciferase activity (Firefly LUC/Renilla LUC). Enzyme-linked immunosorbent assay (ELISA) Cell culture media were collected into a centrifuge tube and centrifuge at 1,500 rpm for 10 min at 4°C. The aliquot supernatant was stored at −80°C. For cell lysates preparation, the cells were washed by prechilled PBS for 3 times and then resuspended by extraction buffer followed by incubation on ice for 10min. After centrifugation at 12000g for 10 min at 4°C, the aliquot supernatants were collected and stored at −80°C until use. The concentration of CX3CL1, CXCL12, CXCL14, TNF-α and IFN-γ were measured by using the Human Fractalkine/CX3CL1 ELISA Kit, Human SDF-1 beta/CXCL12B ELISA Kit, Human BRAK/CXCL14 ELISA Kit, Human TNF alpha ELISA Kit or Human IFN gamma ELISA Kit respectively (Thermo Fisher Scientific). In vitro killing assay CD8+ T cells were transfected with miR-139 mimic or miR-control for 48 h, and then co-cultured with GFP labeled target cells at effector to target (E:T) ratio of 0.5:1, 1:1, 2.5:1 and 10:1 in a 24-well plate. After 12 h, cells were collected and stained with the tumor cell surface marker to evaluate the cytotoxicity of T cells by flow cytometry. Chromatin immunoprecipitation (ChIP) assay For each assay, 5 × 106 cells were cross-linked with 1% formaldehyde at 37 °C for 5 min. The cells were then resuspended in SDS lysis buffer [1% SDS, 10 mM EDTA, 50 mM Tris-HCl (pH 8.1)] and incubated on ice for 15 min. The lysate was sonicated 10 times (15 s each) on ice, followed by centrifugation at 13,000 rpm for 10 min at 4 °C. The supernatant was diluted 10-fold in dilution buffer [0.01% SDS, 1.1% Triton X-100, 1.2 mM EDTA, 167 mM NaCl, 16.7 mM Tris-HCl (pH 8.1)] and precleared by incubation with Protein A-Sepharose for 1 h on a rotating platform at 4 °C. Precleared chromatin was incubated overnight at 4 °C with 1 μg of anti-GLI2 antibody (Santa Cruz Biotechnology, cat. sc-271786) or 1 μg of mouse IgG under constant rotation. Immune complexes were collected using Protein A-Sepharose (Sigma-Aldrich, cat. GE17-0780-01) and washed three times with LiCl buffer [0.25 M LiCl, 1% Triton X-100, 1% deoxycholic acid, 1 mM EDTA, 10 mM Tris-HCl (pH 8.1)] and twice with TE buffer [1 mM EDTA, 10 mM Tris-HCl (pH 8.0)], followed by phenol-chloroform extraction and ethanol precipitation. Quantitative real-time PCR was performed to examine the enrichment of genes promoter regions. For each primer set, Ct values were determined from the input DNA. Immunoblotting The protein samples were denatured by boiling in SDS sample buffer supplemented with β-mercaptoethanol. Equal amounts of protein (20 μg per lane) were resolved by 12% SDS-PAGE and subsequently transferred to a polyvinylidene fluoride (PVDF) membrane (Amersham Pharmacia Biotech, St. Albans, UK) via electroblotting. GLI2 levels were detected using mouse anti-GLI2 monoclonal antibody (Santa Cruz Biotechnology, cat. sc-271786). Quantification and statistical analysis All statistical analyses were performed using SPSS version 16.0 (SPSS Inc., Chicago, IL, USA). Continuous variables were compared between paired breast cancer tissues and adjacent normal tissues using paired Student’s t-tests. The Wilcoxon test was adopted for non-normally distributed data. Survival outcomes were evaluated using Kaplan-Meier analysis with Log Rank tests to compare differences between patient groups. A two-tailed p-value <0.05 was considered statistically significant for all analyses. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001.