Background

T cell exhaustion, caused by prolonged exposure to tumor antigens and the overexpression of immune checkpoint proteins like programmed cell death protein 1 (PD-1), TIM3, and TIGIT, is a major challenge for the immune system to effectively surveil and combat cancer.1 2 Activation of these checkpoint pathways leads to immune suppression, inhibiting T cell responses and enabling cancer cells to evade immune surveillance.3 Overcoming Tcell exhaustion is crucial for improving the efficacy of cancer immunotherapies and developing innovative therapies.4 5

There are multiple approaches to target immune checkpoints.6 The most common and clinically approved approach is the use of antibodies targeting checkpoint complexes on the cell surface.7 However, these approaches have demonstrated limited effectiveness in certain populations because multiple checkpoint proteins are involved in the programming of T cell exhaustion, single-gene targeting approaches, despite their specificity, may not yield sustainable impacts.8 9 To this end, using microRNAs (miRNAs) which can simultaneously target multiple checkpoint genes would be a promising strategy to expand targeting efficacy.10 11 This approach is expected to more effectively reduce T cell exhaustion and restore their effector functions, providing a more sustainable therapeutic effect. MiRNAs are critical regulators of transcriptional networks, orchestrating key cellular events such as T cell memory formation and exhaustion. MiR-31 and miR-155 modulate inhibitory receptors and cytokine sensitivity, 12 13while miR-29a promotes memory-like T cell states and the Let-7 family inhibits the PI3K/AKT/mTOR pathway to prevent exhaustion.14 15 Targeting miRNAs holds potential to reverse T cell exhaustion, restore robust T cell-mediated antitumor responses, and significantly enhance the efficacy of cancer immunotherapies.

With an unbiased bioinformatics interrogation of the human transcriptome and small RNA sequencing of CD8+ T cells from an in vitro exhaustion model, we identified miR-379-5p as a novel modulator of the T cell checkpoint program and suppressor of T cell exhaustion by targeting multiple checkpoint pathways. MiR-379-5p restores T cell effector function, maintains the memory phenotype, and resists the immunosuppressive tumor microenvironment. Moreover, we demonstrate the potential of miR-379-5p in enhancing immunotherapy using tumor organoids and paired peripheral cytotoxic T lymphocytes derived from patients with cancer. These results advance our understanding of how miRNAs regulate T cell exhaustion and offer a potential strategy for anticancer immunotherapy.

MethodsPrimary CD8+ T cell isolation and culture

Peripheral blood mononuclear cells (PBMCs) from blood donors were isolated from fresh whole blood samples using Ficoll-Paque Premium gradients (GE HealthCare). Red blood cells were lysed with ACK lysis buffer (Gibco) and incubated on ice for 5 min and PBMCs were washed with phosphate-buffered saline (PBS) twice. Cells were pelleted and then resuspended in magnetic-activated cell sorting (MACS) buffer (PBS+0.5% bovine serum albumin+2 µM EDTA). CD8+ T cells were purified using the CD8+ T cell Isolation Kit (Miltenyi Biotec, 130-096-495). CD8+ T cells were cultured in high glucose Roswell Park Memorial Institute (RPMI) 1640 medium supplemented by 10% fetal bovine serum (Hyclone, GE HealthCare Life Sciences), 1% Antibiotic-Antimycotic (Thermo Fisher), and 50 nM of B-mercaptoethanol (Thermo Fisher, 31350010). For CD8+ T cell stimulation, cells were treated with 1:200 T Cell TransAct microbeads (Miltenyi Biotec, 130-111-160) and 100 U/mL interleukin (IL)-2. Cells were passaged every 2 days and maintained at 1–2 million cells/mL.

Small RNA sequencing by Illumina NovaSeq

A total of 1 µg of RNA was used for library preparation. First, the 3' SR Adaptor for Illumina was ligated to the small RNA using a 3' Ligation Enzyme. To prevent adaptor-dimer formation, excess 3' SR Adaptor was hybridized with the SR RT Primer for Illumina. Next, the 5' SR Adaptor for Illumina was ligated to the small RNA using a 5' Ligation Enzyme, and first-strand complementary DNA (cDNA) was synthesized using ProtoScript II Reverse Transcriptase. Each sample was then amplified by PCR using P5 and P7 primers. The PCR products were purified using DNA clean beads, and the purified products ranging from 140 to 160 bp were recovered and cleaned up using PAGE. The libraries were validated using an Agilent 2100 Bioanalyzer. Finally, libraries with different indexes were multiplexed and sequenced using paired-end PE150 on the Illumina NovaSeq.

Data analysis of small RNA sequencing

To remove technical sequences, pass-filter data in FASTQ format were processed using Trimmomatic (V.0.30) to obtain high-quality clean data. The data processing included the following steps: removing adapter sequences, trimming 5’ or 3’ end bases that contain N’s or have quality values below 20, removing bases with an average quality score below 20 using a sliding window of 4 bp, and discarding reads that are less than 18 bp long after trimming. We used miRDeep2 to identify miRNAs and their expression levels. Differential expression analysis was conducted using the edgeR Bioconductor package (V.3.19). MiRNAs with a false discovery rate (FDR) below 0.001 were considered differentially expressed after statistical testing with the negative binomial distribution. GOSeq (V.1.34.1) was used to identify Gene Ontology (GO) terms that annotate a list of miRNA target genes with a significant adjusted p value of <0.05. The topGO package was used to plot Directed Acyclic Graphs. KEGG (Kyoto Encyclopedia of Genes and Genomes), a collection of databases dealing with genomes, biological pathways, diseases, drugs, and chemical substances, was used for pathway enrichment analysis. In-house scripts were employed to enrich significantly differentially expressed genes in KEGG pathways.

Data collection and preparation

The processed data sets of miRNA sequencing (miRNA-seq) and RNA sequencing (RNA-seq) were retrieved from DriverDB, miR-TV, and YM500.16–23 Data in DriverDB and miR-TV were sourced from The Cancer Genome Atlas (TCGA) data portal (https://portal.gdc.cancer.gov/), including annotated primary tumor, normal, and metastatic tissues. miRNA-seq data in YM500 were acquired from CGHub (https://cghub.ucsc.edu/), processed through the YM500 miRNA-seq pipeline, and annotated using the miRBase and the DASHR database V.1.0. The experiments validated relations between miRNA and the immune checkpoint genes based on miRTarBase. The predicted relationship between miRNAs and the immune checkpoint genes was assessed by YM500 and defined by 12 computational tools.

Identification of differentially expressed miRNAs

For the breast invasive carcinoma data from TCGA (TCGA-BRCA), differential expression analysis was performed using DESeq2, with a p value threshold of <0.05. miRNAs with a log2 fold change >1 were classified as upregulated, and fold change <−1 as downregulated.

Immune infiltration analysis

Tumor infiltration indices for exhausted CD8+ T cells and naïve CD8+ T cells from the TCGA-BRCA data set using ImmuCellAI.24 Correlations between miRNA expression and tumor infiltration indices were assessed using Spearman’s correlation analysis, with a significance threshold of α=0.001.

Survival analysis

MiRNA expression and patient overall survival were assessed in the TCGA-BRCA cohort using the survival package in R. Cox regression analysis was employed to estimate HRs, and differences between high-risk and low-risk patient groups were assessed via log-rank tests, with significance determined at a p value<0.01.

Cell lines and culture conditions

The BRCA cell line, MDA-MB-231 (HTB-26), the cervical squamous cell carcinoma and endocervical adenocarcinoma (CESC) cell line, HeLa (CCL-2), the skin cutaneous melanoma (SKCM) cell line, A2058 (CRL-3601), and the murine melanoma cell line, B16-F10 (CRL-6475), were obtained from American Type Culture Collection (ATCC). Cells were cultured in Dulbecco's Modified Eagle Medium/Nutrient Mixture F-12 (DMEM/F12) medium supplemented by 10% fetal bovine serum (Thermo Fisher) and 1% penicillin/streptomycin (Thermo Fisher) in a 5% CO2 incubator at 37°C. The acute T cell leukemia cell line, Jurkat, was purchased from ATCC. The cells were cultured in high glucose RPMI-1640 medium supplemented by 10% fetal bovine serum (Thermo Fisher), 1% Antibiotic-Antimycotic (Thermo Fisher) in a 5% CO2 incubator at 37°C.

Lentivirus production and transduction

For generating red fluorescent protein (RFP) and luciferase-expressing cell lines, the pLAS2w.RFP-C.Ppuro plasmid was obtained from the RNA Technology Platform and Gene Manipulation Core Facility (RNAi core) of the National Core Facility for Biopharmaceuticals at Academia Sinica in Taiwan. The pLenti CMV Puro LUC were purchased from Addgene. The plasmids were transfected into HEK-293FT cells (Thermo Fisher, R70007) at roughly 80% confluency in 10 cm tissue culture plates. Viral supernatant was collected at 48 and 72 hours post-transfection, filtered via a 0.45 µm filtration unit (Millipore). Filtered virus was concentrated using the PEG-8000 solution (Merck, 89510) at 1,600×g for 1 hour. The concentrated supernatant was subsequently aliquoted, flash frozen, and stored in −80°C until use. Cells were transduced with concentrated lentivirus 24 hours after isolation. 8 µg/mL of polybrene was added to each well. Plates were sealed and then spun at 800×g at 32°C for 90 min. 24 hours after infection, the cells were selected with complete media with 2 µg/mL puromycin.

Cell transfection with miRNA mimics

Hsa-miR-379-5p and hsa-miR-scramble controls were purchased from Dharmacon. The miRNA mimics were transfected into cells using Celetrix LE+electroporation following the manufacturer’s instructions (voltage 450 V, pulse time 20 ms, one pulse in the 20 µL size electroporation). The cells and total RNA were collected 48 hours after the transfection. The expression levels of the miRNAs were verified by quantitative reverse transcription PCR (qRT-PCR) analysis, and the cells were processed for further study.

Quantitative reverse transcription PCR

Total RNA was isolated using the TRIzol reagent (Life Technologies) following the manufacturer’s instructions. cDNAs were synthesized with the MMLV Reverse Transcription Kit (Invitrogen). miRNA expression levels were analyzed by qRT-PCR using iQ SYBR Green Supermix (Bio-Rad). The fold changes were determined using the comparative cycle threshold method and normalized to 18 s. All experiments were performed in biological triplicate. Primer sequences were obtained from PrimerBank.

Flow cytometry analysis

Cells were harvested and adjusted cell number to a concentration of 1×106 cells per mL in an ice-cold fluorescence activated cell sorting (FACS) buffer (PBS, 0.5% BSA, 0.1% sodium azide). Fixation and permeabilization for intracellular staining were performed using the True-Nuclear Transcription Factor Buffer Kit (BioLegend, 424401). The fluorochrome-labeled antibodies were added to the cell suspension and incubated for 30 min at 4°C in the dark. Cells were washed twice by centrifugation at 500 g for 5 min and resuspended in 600 µL ice-cold FACS buffer. Flow cytometry analyses were acquired on a BD FACSVerse and analyzed with BD FACSuite software. For the gating strategy, dead cells were excluded by staining with 7AAD (BioLegend, 420404). The following monoclonal antibodies from BioLegend were used: FITC anti-human CD3 (1:40 dilution, 317306), APC/Fire 750 anti-human CD8a (1:40 dilution, 300932), PE/Cyanine7 anti-human CD4 (1:40 dilution, 300512), FITC anti-human CD279 (1:20 dilution, 379206), PE anti-human CD366 (1:20 dilution, 364806), APC anti-human TIGIT (1:20 dilution, 372706), PE anti-human CD44 (1:20 dilution, 397504), and APC anti-human CD62L (1:20 dilution, 385106). APC anti-mouse Nur77 REAfinity (1:50 dilution, 130-111-418) was obtained from Miltenyi Biotec.

In vitro killing assay

RFP-expressing tumor cells or B16F10-Luc cells were seeded in a 96-well plate (3×103 cells per well) and cultured overnight. Human CD8+ T cells or OT-I T cells were co-cultured with the tumor cells at the indicated effector-to-target (E:T) ratios for 48 hours. Tumor cell death was assessed using either RFP signal and BioTracker NucView 488 Green Caspase-3 Dye (Sigma-Aldrich, SCT101) monitored using the IncuCyte S3 live imaging system (Sartorius) or via crystal violet staining and luminescence assay.

Luciferase reporter assay of miRNA-messenger RNA interactions

The 3’ untranslated region (3’-UTR) regions of TIM3 and TIGIT messenger RNA (mRNAs) were synthesized (Integrated DNA Technologies). The nucleotide fragments were inserted in the 3’-UTR of the luciferase gene in the pmirGLO Dual-Luciferase vector (Promega). Mutations of the putative 3’-UTR were generated using the QuikChange Mutagenesis Kit (Agilent Technologies). The luciferase reporter with the miR-379-5p mimic or the control miRNA were co-transfected into Jurkat cells using Celetrix LE+electroporation following the manufacturer’s instructions (1.5×106 cells, 800 nM miRNA mimic, voltage 400 V, pulse time 20 ms, one pulse in the 20 µL size electroporation). Luciferase activities were measured 48 hours after transfection using the Luciferase Assay Kit (Promega).

Luciferase reporter assay of MIR-379 gene promoter

pCMV6-Entry-NR4A1-Myc-DDK plasmid was purchased from OriGene. The plasmid and control vector were transfected into Jurkat cells using Celetrix LE+electroporation following the manufacturer’s instructions (voltage 450 V, pulse time 20 ms, one pulse in the 20 µL size electroporation). The pGL3 promoter luciferase reporters with the pCMV6-Entry-NR4A1-Myc-DDK plasmid and control vector were co-transfected into Jurkat cells using Celetrix LE+electroporation following the manufacturer’s instructions (1.5×106 cells, reporter plasmid 1 µg, expression plasmid 1 µg, voltage 400 V, pulse time 20 ms, one pulse in the 20 µL size electroporation). pRL-TK was co-transfected into cells as normalization for transfection efficiency. Luciferase activities were measured 48 hours after transfection using the Luciferase Assay Kit (Promega).

Chromatin immunoprecipitation

The cells were harvested and fixed with 1% formaldehyde, followed by chromatin immunoprecipitation (ChIP) assay according to the manufacturer’s instructions (HighCell# ChIP Kit, Diagenode). Chromatin fragmentation was achieved using a Bioruptor Pico chromatin shearing sonicator (Diagenode). Cell pellets were isolated and lysed, and chromatin was sonicated to an average size of 200 bp. A small aliquot of the supernatant was used as the input control, while the remaining sonicated chromatin was incubated with an anti-nuclear receptor subfamily 4 group A member 1 (NR4A1) antibody (Santa Cruz, sc-365113) and mouse IgG antibody (Santa Cruz, sc-2025) as a negative control, respectively. Transcription factor binding elements were analyzed using specific primer sets through qRT-PCR.

Generation of human and murine-derived tumor organoids

MDA-MB-231-Luc tumors and B16F10-OVA-Luc tumors were isolated from tumor-bearing mice and cut into small pieces (1–3 mm³). The minced tissues were washed with AdDF+++medium (advanced DMEM/F12 medium containing GlutaMAX (1x), 10 mM 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid (HEPES), and 1% antibiotics), and the tissue pellets were digested with collagenase (1–2 mg/mL; Sigma, C9407) while incubating at 37°C for 1 hour with gentle shaking. The suspension was collected and filtered through a 100 µm diameter pore size filter. The retained tissue blocks were collected by washing with 5 mL of AdDF+++medium containing 2% fetal bovine serum and centrifuged at 400 g. The tissue pellet was washed again using the same buffer, resuspended, and seeded in ultra-low attachment plates. The culture was then incubated in complete AdDF+++medium in a humidified incubator at 37°C with 5% CO2.

Organoid killing assay

Tumor organoids were washed once with PBS and then plated at a density of 1×104 cells per well. The cells were subsequently cultured for 1 week to allow for growth. 1×105 OT-I T cells or human CD8+ T cells, which had been acutely or chronically stimulated for 7 days as previously described, were then co-cultured with the tumor organoids. After 24 hours of co-culture, the cells were lysed and luciferase activity was measured using a Luciferase Assay Kit (Promega) following the manufacturer’s instructions. Luciferase activity was normalized to cells cultured in the absence of T cells.

Generation of patient-derived tumor organoid

Patient-derived tumor organoids (PDTOs) were cultured as described previously.25 Fresh breast cancer tissues were minced into pieces measuring 1–3 mm3. These minced tissues were then washed using AdDF+++medium (advanced DMEM/F12 medium containing GlutaMAX ×1, HEPES 10 mM, and 1% antibiotics), and the tissue pellets were subsequently digested using collagenase (1–2 mg/mL; Sigma, C9407) while incubating at 37°C for 1 hour with gentle shaking. Following digestion, the suspension was collected and strained through a filter with a pore size of 100 µm in diameter to remove any remaining solid tissue fragments. The retained tissue blocks were then collected by washing with 5 mL of AdDF+++medium containing 2% fetal bovine serum and centrifuged at 400 g. The tissue pellet was washed again using the same buffer, resuspended, and embedded in 40 µL of reduced growth factor cold Cultrex BME type 2 (Trevigen, 3533-010-02) in each well of a 24-well plate, allowing it to solidify at 37°C for 20 min. Subsequently, the cultures were incubated in 400 µL of BC organoid medium in a humidified incubator at 37°C with 5% CO2.

PDTO killing assay

CD8+ T cells from patients were isolated and stimulated as previously described. The miRNA was transfected into these cells using Celetrix LE+electroporation, following the manufacturer’s instructions (voltage 450 V, pulse time 20 ms, one pulse in the 20 µL size electroporation). PDTOs were stained with CellTracker Red CMTPX dye according to the manufacturer’s instructions and seeded into a 96-well plate coated with 30 µL of 50% BME2 matrix. Subsequently, T cells were co-cultured with the tumor organoids for 48 hours. The organoids were monitored using the IncuCyte S3 live imaging system (Sartorius), and their size was analyzed by red fluorescence using ImageJ software V.1.50e.

OT-I–B16-F10-OVA-Luc tumor model in NOD-SCID mice

NOD/SCID (NOD.CB17-Prkdcscid/NcrCrl) female mice were injected subcutaneously with 1×104 B16-F10-OVA-Luc cells in a 1:1 mix of PBS and Matrigel (Corning). Mice were included in the study if they developed a successful tumor model, defined by a tumor volume reaching a predetermined threshold, as measured by In Vivo Imaging System (IVIS) imaging (Xenogen). Mice were excluded if the tumor implantation procedure resulted in complications, such as perforation of surrounding tissues (evidenced by hemorrhage or infection), if the tumor implant became dislodged or failed to grow, or if the animal died prematurely, preventing the collection of required behavioral and histological data. Each experimental group consisted of seven mice. Mice were randomized after successful tumor establishment, using a computer-based random order generator. 4 days later, 2×106 OT-I T cells that had been chronically stimulated as described previously and transfected with miR-379-5p or miR-Scr by Celetrix LE+electroporation following the manufacturer’s instructions (1.5×107 cells, voltage 1,080 V, pulse time 20 ms, one pulse in the 200 µL size electroporation) were adoptively transferred to mice via intravenous injection every 7 days. Tumor growth was monitored by IVIS. No mice were excluded from any group, and all data were included in the analysis without removing any outliers.

OT-I–B16-F10-OVA-Luc tumor model in C57BL/6 mice

C57BL/6 female mice were subcutaneously injected with 1×10⁴ B16-F10-OVA-Luc cells in a 1:1 PBS/Matrigel mix. Mice were included once tumors reached a predefined size (measured by length and width). Those with unsuccessful tumor growth were excluded. Each group (n=6–7) was randomized after tumor establishment. 7 days later, 2×10⁶ OT-I T cells (chronically stimulated and transfected with miR-379-5p or miR-Scr) were electroporated (Celetrix LE+, 1.5×10⁷ cells, 1,080 V, 20 ms, single pulse, 200 µL) and injected intravenously every 7 days for four doses. Mice received 100 µg anti-programmed death-ligand 1 (PD-L1) or control IgG via intraperitoneal injection every 3 days for eight doses. Tumor growth was monitored by length and width measurements. Mice were considered deceased if they died naturally or if tumors exceeded 20 mm. No mice were excluded, and all data were analyzed without removing outliers.

Statistical analysis

Data analysis was performed using Microsoft Excel 2019, GraphPad Prism V.9, and R Studio (V.3.6.1). The normality of the data was assessed using the Shapiro-Wilk test and the Kolmogorov-Smirnov test. Parametric statistical analyses were conducted for normally distributed data, including Student’s t-test and two-way analysis of variance (ANOVA). For non-parametric data, the Mann-Whitney U test was applied. The correlation analysis was evaluated using the Spearman’s correlation coefficient. Statistical methods and significance are indicated in each result.

ResultsIdentification of miRNAs regulating CD8+ T cell exhaustion

CD8+ T cells were isolated and purified from whole blood from healthy donors and subjected to chronic stimulation with anti-CD3/anti-CD28 antibodies and IL-2 (figure 1A). This resulted in the upregulation of exhaustion markers, including PD-1, TIM-3, and TIGIT (figure 1B and C). RNA-seq of the naïve and exhausted T cells revealed that the exhaustion phenotype in vitro was associated with distinct gene expression profiles of T lymphocytes (figure 1D), and was further confirmed by reduced proliferative capacity and cytotoxic activity (figure 1E, F).

Figure 1

Identification of miRNA signature in CD8+T cell exhaustion. (A) Schematic depiction of experimental procedure of in vitro human CD8+ T cell exhaustion model. The figure was created with BioRender.com. (B) Quantitative reverse transcription PCR analysis of major immune checkpoint genes from in vitro CD8+ T cell exhaustion model. Data are the means±SD (n=3). Statistical significance was determined using Student’s t-test. *p<0.05; **p<0.01. (C) FACS analysis of cell-surface PD-1, TIM3, and TIGIT expression in exhausted CD8+ T cells. (D) GSEA enrichment for exhaustion signature genes in vitro exhausted CD8+ T cells compared with naïve CD8+ T cells. (E) Representative histogram plots of CD8+ T cell proliferation response to anti-CD3/CD28 stimulation in the exhausted CD8+ T cell compared with naïve CD8+ T cells using CellTrace Far Red dye for 5 days. (F) Exhausted and naïve CD8+ T cells co-cultured with MDA-MB-231-RFP cells at effector-to-target ratio of 10:1 in the presence of the CD3-/CD28-stimulating antibodies in IncuCyte for 48 hours. Cytotoxicity was assessed using caspase-3/7 apoptosis dye staining. The results represent the mean±SD (n=3). Statistical significance was determined by two-way analysis of variance with Tukey’s multiple comparisons test. ***p<0.001. (G) Heatmap of top 25 differentially expressed miRNAs from miRNA sequencing of exhausted CD8+ T cells. Color indicates z-score value calculated using the edgeR Bioconductor package. Scale bar=log2 fold change. (H) Top Kyoto Encyclopedia of Genes and Genomes (KEGG) terms after functional enrichment analysis of miRNA target genes in exhausted CD8+ T cells. (I) Identification strategy for significantly differentially expressed miRNAs in CD8+ T cell exhaustion. (J) Volcano plot showing statistical significance (p value) versus differential expression of the miRNAs in tumors and normal tissues. X-axis values are log2 transformation of fold change of expression, and y-axis values are negative log10 transformation of p values estimated by DESeq. (K) Heatmap correlating candidate miRNAs and both naïve and exhausted TIL levels in BRCA data sets estimated by Immune Cell Abundance Identifier. (L) Volcano plot of p values versus HRs of miRNAs in patients with breast cancer. X-axis values are log2 transformation of HRs, and y-axis values are −log10 p values estimated by log-rank test. (M) Kaplan-Meier curve of patients with BCRA subdivided by hsa-miR-379-5p expression. Statistical significance was determined using the log-rank test and the HR was calculated using the Cox proportional hazards regression model. BRCA, breast invasive carcinoma; FACS, fluorescence activated cell sorting; miRNA, microRNA; mRNA, messenger RNA; PD-1, programmed cell death protein 1; TIGIT, T cell immunoreceptor with Ig and ITIM domains; TIL, tumor-infiltrating lymphocyte; TIM3, T cell immunoglobulin and mucin-domain containing-3; Tna, naïve CD8 T cells; Tex, exhausted CD8 T cells.

To assess miRNA expression, comparative miRNA-seq of the exhausted and unstimulated T cells was conducted. We found 114 differentially expressed miRNAs (|log2FC|>1, FDR<0.001), 23 upregulated and 91 downregulated (figure 1G, online supplemental table 1). KEGG enrichment analysis revealed several crucial targeted pathways related to T cell differentiation and function, including WNT, TNF, MAPK, and JAK-STAT (figure 1H).

To identify key miRNAs involved in T cell exhaustion and antitumor immunity, we analyzed differentially expressed miRNAs in exhausted T cells using an integrated bioinformatics strategy (figure 1I). Differential miRNA expression levels were first assessed in the TCGA-BRCA data set. 4 of the 23 upregulated miRNAs were significantly elevated in tumor tissues, whereas 23 out of 91 downregulated miRNAs were considerably reduced in these tissues (figure 1J, online supplemental table 2).

In addition, the Immune Cell Abundance Identifier (ImmuneCellAI) was used to estimate the proportion of different T cell subtypes in the data set, including exhausted T cells, and to assess the relationship between these miRNAs and the infiltration of early-stage and exhausted CD8+ T cells in tumors. 24 We identified two upregulated miRNAs that exhibited a positive correlation with T cell exhaustion, while 11 downregulated miRNAs displayed a progressive loss of expression in association with T cell exhaustion in tumors (figure 1K, online supplemental table 3). Finally, the correlation between miRNA expression and patient outcomes was assessed. Only 1 out of 11 miRNAs, hsa-miR-379-5p, was correlated with favorable survival in patients with breast cancer (figure 1L–M, online supplemental table 4).

miR-379-5p was associated with promoted antitumor immunity and favorable prognosis

Using TCGA data sets of multiple cancer types, miR-379-5p abundance was found to be inversely correlated with the presence of more differentiated or exhausted T cells in the tumor microenvironment (figure 2A). MiR-379-5p expression positively correlated with early activation stages (naïve CD8+ and central memory T cells) and negatively correlated with later stages of T cell differentiation (effector memory and exhausted T cells), particularly in BRCA and CESC (figure 2B, online supplemental figure 1A). A similar trend was observed in SKCM, although certain results did not reach statistical significance (online supplemental figure 1B).

Figure 2

miR-379-5p is associated with promoted antitumor immunity and favorable prognosis. (A) A landscape of the correlation of miR-379-5p expression in the differentiation stages of CD8+ T cell infiltration across various cancer types. Positive correlations are indicated in shades of red, negative correlations in shades of blue, and the scale of coefficient ranges from −1.0 to 1.0. Non-significant correlations are indicated in shades of gray. (B) The correlation between miR-379-5p and differentiation stages of CD8+ T cell infiltration in BRCA. (C) DE analysis showed miR-379-5p was significantly downregulated in BRCA. Statistical significance was evaluated by Student’s t-test. (D) Kaplan-Meier analysis illustrated that low expression of miR-379-5p and CD8+ T cell infiltration was significantly associated with poor overall survival in BRCA. Statistical significance was determined using the log-rank test. The HR was calculated using the Cox proportional hazards regression model. BLCA, bladder urothelial carcinoma; BRCA, breast invasive carcinoma; CESC, cervical squamous cell carcinoma and endocervical adenocarcinoma; COAD, colon adenocarcinoma; HNSC, head and neck squamous cell carcinoma; KIRC, kidney renal clear cell carcinoma; KIRP, kidney renal papillary cell carcinoma; LUAD, lung adenocarcinoma; LUSC, lung squamous cell carcinoma; miRNA, microRNA; OV, ovarian serous cystadenocarcinoma; PAAD, pancreatic adenocarcinoma; PRAD, prostate adenocarcinoma; RPM, reads per million; SARC, sarcoma; SKCM, skin cutaneous melanoma; STAD, stomach adenocarcinoma; THCA, thyroid carcinoma; TIL, tumor-infiltrating lymphocyte; UCEC, uterine corpus endometrial carcinoma.

We examined the impact of miR-379-5p and CD8+ infiltration on patient outcomes in BRCA, CESC, and SKCM. We found significantly lower miR-379-5p expression in tumor tissues compared with normal tissues in BRCA and CESC. In SKCM, no significant difference was observed (p=0.78), likely due to the limited normal tissue sample size (n=2) (figure 2C, online supplemental figure 2). Moreover, high miR-379-5p expression combined with CD8+ T cell infiltration was significantly associated with improved overall survival in BRCA (HR=0.533, p=0.00463) and CESC (HR=0.513, p=0.00484), with a weak trend toward significance in SKCM (HR=0.61, p=0.304) (figure 2D, online supplemental figure 3), suggesting that miR-379-5p-mediated T cell infiltration enhances antitumor immunity and improves outcome.

miR-379-5p directly targets the immune checkpoints TIM3 and TIGIT

To investigate the mechanism linking miR-379-5p with T cell-mediated antitumor immunity, we analyzed the interaction between miR-379-5p and 23 major immune checkpoints using miRTarBase annotations and 12 prediction tools for potential miRNA-target gene interactions. An interaction between miR-379-5p and the 3’-UTR of HHLA2, TIM3, and TIGIT was supported by at least six prediction tools (online supplemental table 5). HHLA2 was excluded from further consideration because it is primarily expressed on antigen-presenting cells and tumor cells.26 Consistent with its role in T cell exhaustion (figure 2A), miR-379-5p expression negatively correlated with the expression of its predicted targets, TIM3/HAVCR2 and TIGIT, across various cancer types (figure 3A, B, online supplemental figure 4).

Figure 3

miR-379-5p directly binds and downregulates key immune checkpoints TIM3 and TIGIT. (A) A landscape of the correlation of miR-379-5p, TIM3, and TIGIT expression in CD8+ T cell infiltration across various cancer types. Positive correlations are indicated in shades of red, negative correlations in shades of blue, and the coefficient ranges from −1.0 to 1.0. Non-significant correlations are indicated in shades of gray. (B) The correlation between miR-379-5p and TIM3 and TIGIT expression of CD8+ T cell infiltration in BRCA. (C–E) The luciferase vectors that contain the human wild-type (WT) and mutant (MUT) targeting sites of TIM3 and TIGIT 3’-UTR regions were co-transfected into Jurkat cells with miR-379-5p or the miR-Scr control microRNA. The relative luciferase/Renilla activities were analyzed in the cells 24 hours after the transfection. The results represent the mean±SD (n=3). Statistical significance was tested by t-test. *p<0.05; **p<0.01. (F) Quantitative reverse transcription PCR analysis of the expression of TIM3 and TIGIT in miR-379-5p transfected CD8+ T cells comparing with the miR-Scr control group. Data are the means±SD (n=3). Statistical significance was determined using Student’s t-test. **p<0.01. (G) FACS analysis of surface expression of TIM3, TIGIT, and PD-1 in miR-379-5p transfected CD8+ T cells comparing with the miR-Scr control group. (H) Quantification of the frequency of TIM3+TIGIT+population in CD8+ T cells transfected with miR-379-5p comparing with the miR-Scr control group. Data are the means±SD (n=3). Statistical significance was determined using Student’s t-test. **p<0.01. (I–K) miR-Scr and miR-379-5p transfected CD8+ T cell were co-cultured with MDA-MB-231 cells expressing RFP (MDA-MB-231-RFP) at E:T ratio of 10:1 in the presence of the CD3-/CD28-stimulating antibodies in IncuCyte for 48 hours. Cytotoxicity was assessed using caspase-3/7 apoptosis dye staining. The results represent the mean±SD (n=3). Statistical significance of the time-lapse cell killing curve was determined by two-way ANOVA with Tukey’s multiple comparisons test. ***p<0.001. Statistical significance of relative apoptotic cell fold changes at 48 hours was determined by Student’s t-test. ***p<0.001. (L) miR-Scr and miR-379-5p transfected CD8+ T cell were co-cultured with MDA-MB-231-RFP cells at varying E:T ratios in the presence of the CD3-/CD28-stimulating antibodies for 48 hours. Tumor cell viability was evaluated by crystal violet staining. The heatmap plots represent the mean cytotoxicity (n=3). Statistical significance was determined by Student’s t-test. *p<0.05. (M) miR-Scr and miR-379-5p transfected OT-I T cell were co-cultured with B16F10-OVA-Luc or B16F10-Luc cells at varying E:T ratios for 24 hours. Tumor cell viability was evaluated by luminescence assay. The heatmap plots represent the mean cytotoxicity (n=3). Statistical significance was determined by Student’s t-test and two-way ANOVA with Tukey’s multiple comparisons test. (N, O) The miR-Scr- and miR-379-5p-transfected CD8+ T cell were co-cultured with the organoids derived from MDA-MB-231 cells expressing luciferase (MDA-MB-231-Luc) for 48 hours. The schematic representation was created with BioRender.com. Tumor cell proliferation was determined with chemiluminescence signals. The results represent the mean±SD (n=3). Statistical significance was determined by Student’s t-test. *p<0.05; **p<0.01; ***p<0.001. ANOVA, analysis of variance; BLCA, bladder urothelial carcinoma; BRCA, breast invasive carcinoma; CESC, cervical squamous cell carcinoma and endocervical adenocarcinoma; COAD, colon adenocarcinoma; E:T, effector-to-target; FACS, fluorescence activated cell sorting; HNSC, head and neck squamous cell carcinoma; KIRC, kidney renal clear cell carcinoma; KIRP, kidney renal papillary cell carcinoma; LUAD, lung adenocarcinoma; LUSC, lung squamous cell carcinoma; mRNA, messenger RNA; OV, ovarian serous cystadenocarcinoma; PAAD, pancreatic adenocarcinoma; PD-1, programmed cell death protein 1; PRAD, prostate adenocarcinoma; RFP, red fluorescent protein; SARC, sarcoma; SKCM, skin cutaneous melanoma; STAD, stomach adenocarcinoma; THCA, thyroid carcinoma; TIGIT, T cell immunoreceptor with Ig and ITIM domains; TIM3, T cell immunoglobulin and mucin-domain containing-3; UCEC, uterine corpus endometrial carcinoma; 3’-UTR, 3’ untranslated region.

To evaluate whether miR-379-5p directly binds to TIM3 and TIGIT, the 3’-UTR of the TIM3 and TIGIT genes, containing either wild-type or mutated sequences of the predicted binding motifs, were cloned at the 3’ end of a luciferase reporter gene (figure 3C). Ectopic expression of miR-379-5p inhibited the luciferase activity only in the wildtype miR-379-5p-binding motifs of the TIM3 and TIGIT 3’-UTRs (figure 3D, E), suggesting that miR-379-5p downregulates TIM3 and TIGIT by directly binding to their 3’-UTR.

Transduction of miR-379-5p significantly reduced TIM3 and TIGIT expression in activated primary human CD8+ T cells, both RNA expression, as assessed using qRT-PCR (figure 3F) and cell-surface protein expression with flow cytometry (figure 3G and H). To further investigate the function of miR-379-5p in CD8+ T cell-mediated antitumor immunity, primary human CD8+ T cells were isolated from healthy individuals and transduced with miR-379-5p. These modified T cells (miR-379-5p OE) were then co-cultured with RFP-expressing human cancer cell lines in the IncuCyte system for 48 hours, and cytotoxicity was assessed by real-time imaging. The results demonstrate that miR-379-5p OE CD8+ T cells exhibit stronger cytotoxic activity against breast cancer cells (MDA-MB-231) (figure 3I–K), as well as cervical cancer (HeLa) and melanoma (A2058) cells (online supplemental figure 5), compared with control T cells (miR-Scr). A dose-dependent increase in tumor cell lysis was observed with rising E:T ratios, confirming a cytotoxic response (figure 3L, online supplemental figure 6). Notably, at all E:T ratios, miR-379-5p OE CD8+ T cells consistently exhibited enhanced cytotoxic activity compared with control cells, highlighting the activity of miR-379-5p in strengthening CD8+ T cell-mediated antitumor function.

To determine whether miR-379-5p enhances antigen-specific, major histocompatibility complex (MHC)-I-dependent cytotoxicity, we assessed OT-I T cell-mediated killing using an autologous killing assay against B16F10-Luc and B16F10-Luc cells expressing chicken ovalbumin (OVA)-derived peptide SIINFEKL (B16F10-OVA-Luc) (figure 3M). MiR-379-5p expression significantly enhanced OT-I T cell cytotoxicity against B16F10-OVA-Luc compared with miR-Scr, with killing efficiency increasing with rising E:T ratios. In contrast, no significant difference was observed when targeting B16F10-Luc cells. These findings suggest that miR-379-5p promotes antigen-specific and MHC-I-dependent T cell cytotoxicity, highlighting its potential to enhance tumor-reactive T cell responses. Further testing using tumor organoid cultures derived from xenograft tumors of the MDA-MB-231-Luc cancer cell line showed that miR-379-5p expression significantly increased the tumor-killing ability of both activated and exhausted CD8+ T cells derived from a normal donor (figure 3N–O). These findings suggest that miR-379-5p promotes the antitumor activity of CD8+ T cells undergoing chronic stimulation-induced exhaustion.

miR-379-5p maintains the memory program in CD8+ T cell differentiation

Given that the reduction in miR-379-5p was strongly associated with exhausted CD8+ tumor-infiltrating lymphocyte (TIL) across various cancer types and accompanied by the induction of exhaustion genes such as TIM3 and TIGIT, we investigated the effect of miR-379-5p expression on T cell differentiation during chronic stimulation. Before examining its functional effects, we first assessed the duration of miR-379-5p expression following electroporation. Quantification of miR-379-5p levels over time revealed that expression peaked at 24 hours post-transduction, reaching a 1521.5±259.4-fold increase relative to baseline. Under continuous activation and in vitro culture conditions, miR-379-5p levels remained detectable in exhausted CD8+ T cells even after 10 days, although at a reduced level of 32.9±12.3-fold, suggesting that transient electroporation-mediated miR-379-5p delivery provides sustained expression for an extended period despite ongoing T cell activation (online supplemental figure 7). Naïve CD8+ T cells from a healthy donor were stimulated with anti-CD3/anti-CD28 for 7 days, after which exogenous miR-379-5p was introduced via electroporation and followed by a 3-day incubation. Compared with cells transduced with a scrambled control miRNA (miR-Scr), miR-379-5p-transduced T cells exhibited an increased population with a central-memory phenotype (CD44+CD62L+) rather than effector phenotype (CD44+CD62L−) (figure 4A,B),27 suggesting that miR-379-5p expression helps maintain the memory program in CD8 T cells. Similarly, expression of miR-379-5p resulted in reduced cell population of CD45RA−CD45RO+, the marker associated with differentiated phenotype of T cells, and increased population of CD45RA+CD45RO+ cells shifting toward the memory-like phenotype (figure 4C,D).28