Abstract
MicroRNAs (miRNA), small noncoding RNAs that regulate gene expression, exist not only in cells but also in a variety of body fluids. These circulating miRNAs could enable intercellular communication. miRNAs are packaged in membrane-encapsulated vesicles, such as exosomes, or protected by RNA-binding proteins. Here, we report that miRNAs included in human melanoma exosomes regulate the tumor immune response. Using microscopy and flow cytometry, we demonstrate that CD8+ T cells internalize exosomes from different tumor types even if these cells do not internalize vesicles as readily as other immune cells. We explored the function of melanoma-derived exosomes in CD8+ T cells and showed that these exosomes downregulate T-cell responses through decreased T-cell receptor (TCR) signaling and diminished cytokine and granzyme B secretions. The result reduces the cells' cytotoxic activity. Using mimics, we found that miRNAs enriched in exosomes—such as Homo sapiens (hsa)-miR-3187-3p, hsa-miR-498, hsa-miR-122, hsa-miR149, and hsa-miR-181a/b—regulate TCR signaling and TNFα secretion. Our observations suggest that miRNAs in melanoma-derived exosomes aid tumor immune evasion and could be a therapeutic target.
Introduction
In the last decade, noncoding RNAs have emerged as a class of regulators of cellular functions and differentiation. MicroRNAs (miRNA) are endogenous noncoding RNAs about 19 to 22 nucleotides (NT) that modulate gene expression through mRNA silencing or degradation (1). A single miRNA has the capacity to inhibit several different mRNA targets. More than 60% of human genes are regulated by miRNAs (2). miRNAs may also act in cancer as tumor suppressors (TS-miR) or oncogenes (oncomiR), depending on tissue, cellular context, and target genes (3, 4).
miRNAs are detectable both in tumor tissues and in body fluids in extracellular vesicles, such as exosomes (5) where they participate in intercellular communication. The loading of miRNAs into exosomes is a selective process (6). Thus, miRNAs can be loaded into exosomes by (i) a neural sphingomyelinase 2 (nSMase2)–dependent pathway (7), (ii) an miRNA motif and the hnRNP (heterogeneous nuclear ribonucleoproteins) proteins (8, 9), (iii) the 3′ end of the miRNA sequence (10), or (iv) the miRISC (miRNA-induced silencing complex) pathway, also implicated in the miRNA biogenesis (11). Accordingly, exosomal miRNA content may not reflect the miRNA content of the parent cell, but rather is composed of a selected set of miRNAs (12–14). Some miRNAs seem to be specialized for packaging into exosomes. Exosomal miRNA could be then internalized by recipient cells where they play their functional roles (15).
In immune cells, miRNAs regulate cell development, cell differentiation and the production of inflammatory mediators. For example, upon activation by an external signal, the miRNA repertoire of T cells changes (16). Thus, Homo sapiens (hsa)-miR-214 expression was increased after activation, which inhibits PTEN (phosphatase and tensin homolog) and enhances T-cell proliferation (17). Other studies show a decrease of CD69 expression concomitant with the upregulation of hsa-miR-130/hsa-miR-301 (18) and a downregulation of PI3KR1 (phosphoinositide-3-kinase regulatory subunit 1) by hsa-miR-132-3p (19). Modifications in how miRNAs regulate T-cell functions could lead to an abnormal or ineffective immune response against tumors.
Cutaneous melanoma is an aggressive cancer that has increased in incidence worldwide. Excessive sun exposure remains the major risk factor for melanoma. People of all ages can be affected, although susceptibility increases with age, and women are more often affected than men (20). The prognosis is generally poor given the propensity of melanoma cells to spread to distant sites while evading immune system control (21). Even though melanoma is one of the most immunogenic tumors, mechanisms of the tumor escape from immune surveillance remain unclear. Here, we investigated whether miRNAs included in melanoma-derived exosomes could aid tumor escape from immune surveillance by downregulating the CD8+ T-cell response against tumor cells.
Materials and Methods
Cell lines and primary melanocytes
The melanoma cell lines M113, M117, M28, M45, and M6 were established from metastatic tumor fragments in our Inserm Unit or in the Unit of Cell therapy of Nantes and are registered in the Biocollection PC-U892-NL (CHU Nantes). Melanoma cell lines were cultured in RPMI-1640 (Gibco) supplemented with 10% exosome-depleted fetal bovine serum (Thermo Fisher Scientific), L-glutamine and penicillin–streptomycin (10 μg/mL, Gibco). HEK-293 cells (ATCC) were cultured in DMEM (Gibco) supplemented with 10% fetal bovine serum (Thermo Fisher Scientific), L-glutamine and penicillin–streptomycin (10 μg/mL, Gibco). Mesothelioma cell line Meso34 and colon colorectal cancer cell line HCT116 were cultured in RPMI-1640 (Gibco) supplemented with 10% exosome-depleted fetal bovine serum, L-glutamine and penicillin–streptomycin. Cell lines were tested each week for mycoplasma contamination using the PlasmoTest-Mycoplasma Detection kit (Invivogen) according to the manufacturer's protocol. Contaminated cells are discarded. Fresh cell lines were thawed every 2 months at approximately passage 10. Melanoma cell line authentications were performed by the expression of the specific antigens MLANA/MART-1 and meloe by RT-qPCR.
Melanocytes (PCS-200-013) were purchased from ATCC and cultured in dermal cell basal medium supplemented with one adult melanocyte growth kit (ATCC PCS-200-042). Culture medium was changed every 48 hours and kept in culture for 6 weeks.
CD8+ T cells
Peripheral blood mononuclear cells (PBMC) were obtained from healthy donors (ethics agreement CPDL-PLER-2018 021) and CD8+ fraction was immediately isolated using the EasySep Human CD8+ T-cell Enrichment Kit (Stemcell) according to the manufacturer's protocol. To ensure the purity of T-cell populations (>95% purity), they were characterized by flow cytometry [anti-human CD8 (0.4 μg/mL, BioLegend, cat. #301014)]. CTL03.1, a CD8+ T-cell clone specific for Melan-A, was derived and cultured as previously described (Vignard and colleagues, 2005). TILs were previously obtained from tumor-invaded lymph nodes of two melanoma patients (M113 and M117). CD8+ T cells isolated from PBMCs, CTL03.1, and TILs were grown in RPMI-1640 supplemented with 8% human serum (local production), L-glutamine and penicillin–streptomycin (10 μg/mL, Gibco) and 150 IU/mL of IL2 (Chiron).
Exosome purification
Melanoma-derived exosomes were isolated from approximately 20 × 106 of M113 or M117 tumor cells cultured in exosome-depleted media for 24 hours. Differential centrifugation was performed to isolate exosomes from media. Initial spins consisted of a 5-minute spin at 300 g, a 2,000 g spin for 10 minutes and a 10,000 g spin for 30 minutes. The supernatant was retained each time. The supernatant was then spun at 100,000 g for 120 minutes and the pellet was resuspended in PBS, to dilute remaining soluble factors, followed by another centrifugation at 100,000 g for 120 minutes. The final pellet contained the exosomes, which were resuspended in culture media. This protocol is based on previous exosome isolation methods (22). We used a Beckman ultracentrifuge Optima L-80XP with a SW28 rotor.
ExoQuick-TC exosome precipitation solution (System Biosciences) was also used to precipitate exosomes from media. Briefly, media were centrifuged at 3,000 g for 15 minutes to remove cell debris. Then, 1 mL of ExoQuick exosome precipitation solution to 10 mL media was added and mixed. After refrigerating at 4°C overnight, we centrifuged the mixture at 1,500 g for 30 minutes and removed the supernatant. Another centrifugation of 5 minutes to spin down residual fluid was done, and exosome pellets were resuspended in culture media.
Exosome quantification
Exosome diameter and concentration distribution were quantified using Tunable Resistive Pulse Sensing (TRPS) technology with qNano system (IZON). “NP100” nanopores were used to detect particles in the range of 50 to 330 nm. The stretch applied to the pore was 46 mm with an appropriate voltage (Ω0.80–1.00 V) to reach a 120 nA baseline with noise <10 pA. 110 nm beads were used as calibrator particles. Exosome protein concentration was also determined using the BCA Protein Assay Reagent kit (Interchim) according to the manufacturer's instructions.
Western-blotting analysis of melanoma-derived exosomes
Exosomes were lysed with NE-PER Nuclear and Cytoplasmic Extraction reagents (Pierce), and protein concentration was determined by BCA (Interchim). Proteins (4 μg) were electrophoresed by SDS-PAGE and transferred onto a PVDF membrane. The membrane was blocked for 2 hours at room temperature with 5% nonfat milk. The membrane was incubated with primary antibody (0.04 μg/mL, anti-CD63 (Abcam, cat. #ab59479)) or anti-CD81 (Invitrogen Life Technologies, cat. #10630D) overnight at 4°C, followed by a secondary antibody (0.8 μg/mL, goat anti-mouse IgG, HRP-conjugated, Interchim, cat. #115-036-072) for another 1 hour. Protein expression was assessed using a ChemiDocMP Imaging System (Bio-Rad).
Total RNA extraction
Total RNA from cells and exosomes was extracted with QIAzol reagent (QIAGEN) and the miRNeasy kit (QIAGEN) according to the manufacturer's protocols. RNA was screened for purity and concentration in a Nanodrop-1000 spectrophotometer (ThermoFisher Scientific).
miRNA array and analysis
The Agilent 2100 Bioanalyzer (Agilent) for total RNA (RNA nanochips) and for small RNA (small RNA chips) were used to assess the large and small RNA profiles in cells and exosomes. miRNA profiling was performed using Affymetrix microRNA 4.0 Array, which covers 2,578 human miRNAs annotated in miRBase V2.0. Briefly, 1 μg of each sample was labeled with Biotin using the FlashTag Biotin HSR RNA Labeling Kit (Affymetrix) and then hybridized overnight with the array according to the manufacturer's protocols. After washing and staining, the hybridized slides were read by a GeneChip Scanner 3000 7G (Affymetrix). The raw data were exported by GeneChip Command Console Software Version 4.0 (Affymetrix). miRNA data are analyzed using the Limma Package implemented in R3.4.2. miRNA target genes were predicted by miRNAtap and topGO packages implemented in R3.4.2. All original microarray data were deposited in the NCBI's Gene-Expression Omnibus database under the reference GSE125030.
MEME motif analysis
MEME Suite 4.12.0, a web-based tool for studying sequence motifs in RNA, was used to determine a common motif to miRNAs exported in exosomes (23). We applied a Markov model of order 0 to identify this motif.
cDNA synthesis and quantitative RT-PCR
RNAs were reverse transcribed using RevertAid H Minus Reverse Transcriptase (Thermo Fisher Scientific) and the RT product was used for expression analysis using Maxima SYBR Green/ROX qPCR Master Mix (Thermo Fisher Scientific). RPLP0 (Ribosomal Protein Lateral Stalk Subunit P0) and PPIA (Peptidylprolyl Isomerase A) genes were used as reference genes. Each reaction sample was run in duplicate. To circumvent any issue of nonspecific amplification, melting curve analysis was performed with a temperature gradient of 70°C–95°C. miRNAs were reverse transcribed using the MystiCq microRNA cDNA Synthesis Mix (Sigma-Aldrich) and quantified on a Mx3005p system (Stratagene) using the MystiCq microRNA SYBR green Readymix and primers (Sigma-Aldrich). Each reaction sample was run in duplicate. Small nucleolar RNA 44 (snoRNA44) was used as the internal reference gene. Primers are available in Supplementary Table S1.
miRNA transfection
One hundred μmol/L of a fluorescent Alexa-488–labeled miR mimic (IDT) was transfected into 50,000 M113 cells that had reached 80% confluence using Attractene Reagent (QIAGEN) on a 24-well plate, in triplicates. Following 24-hour transfection, supernatants were collected and exosomes were purified as described above. Purified exosomes were then cocultured with T cells and Alexa-488 fluorescence was measured by flow cytometry 24 hours later. Fifty μmol/L of each miRNA mimic (Sigma-Aldrich, cat. #HMI0614, #HMI1002, #HMI0241, #HMI0266, #HMI0270, #HMI1138) or antimir (antisense oligonucleotide; Sigma-Aldrich, cat. #HSTUD1138) was transfected into 100,000 CTL03.1 cells that had reached 80% confluence using TransIT-TKO Transfection Reagent (Mirus) on a 96-well plate, in triplicates. A miRNA mimic negative control (Ambion, cat. #4464058) was also transfected. Following 24-hour transfection, supernatants were collected to measure cytokine concentration, and cells were stained and analyzed by flow cytometry. Fifty μmol/L of hsa-antimiR-3187-3p (Sigma-Aldrich, cat. #HSTUD1138) was transfected into 12 μg of exosome M117 using TransIT-siQUEST transfection reagent (Mirus). Following 2-hour transfection, 100 μL of ExoQuick-TC reagent was added and incubated for 30 minutes at 4°C. After centrifugation, exosome pellet was resuspended in medium and coincubated with LT CD8+ as described below.
Plasmid miRNA cotransfection
Twenty nanograms of firefly luciferase reporter constructs containing the 3′ UTR of PTPRC, TNFα (S808846 and S806309 LigtSwitch Active Motif) or mutated TNFα was transfected in 10,000 HEK-293 cells along with miRNA mimics (10 μmol/L) using Attractene Transfection Reagent (QIAGEN). Cell extracts were prepared 24 hours after transfection, and luciferase activity was measured using the Dual Luciferase Reporter Assay System (Promega).
Site-directed mutagenesis
To generate mutations in the 3′UTR of the TNFα reporter construct, interaction site with the hsa-miR-498 was predicted using miRWalk, RNAhybrid, and TargetScan. Primers used are described in Supplementary Table S1. Mutants were generated by PCR using Pfu DNA Polymerase according to the manufacturer's protocol. Mutations were confirmed by DNA sequencing.
Coculture exosome/cell experiments
CD8+ T cells were cocultured with melanoma-derived exosomes for 24 hours at a concentration of 20 μg/mL when no other concentrations are mentioned in a complete medium containing 50 IU/mL of IL2. We also tested 1 hour and 5 hours of exposure to evaluate the uptake of melanoma-derived exosomes by CD8+ T cells and used different concentrations of exosomes for dose-dependent experiments. During coculture, CTL03.1 cells and TILs were activated for 24 hours with coated anti-CD3 (0.2 μg/mL, P2R Platform) and CD8+ T cells sorted from PBMC were stimulated also during 24 hours by Dynabeads Human T activator CD3/CD28 (Thermo Fisher Scientific). In some cases, melanoma-derived exosomes were labeled with PKH67 dye. Isolated exosomes (up to 28 μg protein) were stained in 200 μL of PKH67 (10−6 M in diluent buffer) and after 2 minutes, the staining reaction was stopped by adding 600 μL of RPMI-1%BSA. Exosomes were then precipitated using 200 μL of ExoQuick-TC incubated for 30 minutes at 4°C. After refrigeration, we centrifuged mixture at 14,000 rpm for 3 minutes and removed the supernatant. Exosomes were then resuspended in culture medium and used for coincubation with CD8+ T cells. To evaluate mechanisms of T-cell uptake, CTL03.1 cells were exposed to PKH67-labeled exosomes and incubated with cytochalasin D at 37°C, an actin cytoskeleton inhibitor (Sigma-Aldrich), at concentrations ranging from 0 to 20 μmol/L for 24 hours. Coculture was also done at 4°C to evaluate if exosome uptake is an energy-dependent process.
Confocal microscopy
After coculture, uptake of PKH67-labeled exosomes by recipient CD8+ T cells was visualized by a Nikon A1 confocal microscope equipped with a 60x/1.4 oil immersion objective (Nikon Instruents). Plasma membrane was stained with WGA-647 (5 μg/mL, Thermo Fisher Scientific) and nucleus with Hoesch 33342 (50 μg/mL, Thermo Fisher Scientific).
Transmission electron microscopy
CD8+ T-cell suspension cultures exposed to melanoma-derived exosomes were collected, centrifuged in pellets, and fixed with 1.6% glutaraldehyde (Fluka) in 0.1 M phosphate buffer for 2 hours at room temperature. Cells were rinsed three times (5 minutes) in 0.1 M phosphate-buffered saline, post-fixed 1 hour in 2% aqueous OsO4 (Agar Scientific), rinsed six times (5 minutes) in distillate water at room temperature. The cell pellets are dehydrated, at room temperature, in increasing concentrations of acetone: 30%, 60%, 90%, three times (each for 5 minutes), respectively, and 100% acetone six times (5 minutes). Acetone in the sample was exchanged with an epoxy resin (Agar 100 Resin Kit). Cell pellets were embedded in epoxy resin followed by polymerization at 37°C for 24 hours and 60°C for 72 hours. Embedded blocks were sectioned ultrathin with thickness of 70 nm or 80 nm using an ultramicrotome (Leica EM UC7). These sections were mounted on copper grids, stained with uranyless, and observed with a Jeol JEM-1010 electron microscope equipped with camera (Gatan Ametek) and operating at 80 kV. Exosome uptake was then detected by the melanin content of CD8+ T cells (24).
Flow cytometry
To document the expression of T-cell surface markers, T cells were labeled in PBS and 0.1% BSA with fluorochrome-conjugated anti-human CD25 (5 μg/mL, BD Biosciences, cat. #564033) and CD45 (0.3 μg/mL, BioLegend, cat. #304012) for 30 minutes at 4°C in the dark. Isotype controls were used for each experiment. After incubation, cells were washed and resuspended in PBS and analyzed using a BD LSRFortessa or a BD FACSCanto II. Ten thousand cell events were recorded. Dead cells and debris were excluded based upon lower forward scatter (FSC) and side scatter (SSC) signals. Analyses were performed using BD FACSDiva software (BD Biosciences). To analyze fluorescence of exosomes derived from Alexa-488-miR mimic transfected M113 cells, exosomes were labeled in PBS with anti-human CD63-APC (0.1 μg/mL, BD Biosciences, cat. #550956) for 30 minutes at 4°C in the dark. Exosomes were then analyzed using a BD Influx flow cytometer using an optimized configuration. Light scattering detection was performed in log mode and thresholding was based on measuring 0.22 μm filtered PBS.
Imaging flow cytometry
After 5 or 24 hours of coculture with PKH67-labeled melanoma-derived exosomes, PBMCs and CTL03.1 cells were harvested, washed, and stained for surface markers with fluorochome-conjugated anti-human CD56 (0.3 μg/mL, BD Biosciences, cat. #555518), CD19 (0.83 μg/mL, BD Biosciences, cat. #562947), CD14 (0.125 μg/mL, BD Biosciences, cat. #563743), CD3 (3 μg/mL, BD Biosciences, cat. #562426), and CD8 (0.4 μg/mL, BioLegend, cat. #301014). Dead cells were identified by Zombie NIR staining [BioLegend, channel 12 (745–800 nm, laser 642 nm)] and excluded. Analyses were performed using an ImageStreamX Mark II Imaging Flow Cytometer (Amnis Corporation) equipped with the INSPIRE software. A 40× magnification was used for all samples. Data analysis was performed using the IDEAS software (Amnis Corporation). The number of acquired single events was fixed to 30,000, allowing us to analyze at least 1,000 in-focus live single cells for each condition. Unmarked, Single staining and Fluorescence Minus One (FMO) controls were included for each experiment. Several masks were applied. First, the Erode Mask was used on the brightfield channel 1 and eroded for 3 pixels to generate the standard cytosol mask. Next, to analyze intracellular exosome spots, a Peak Mask on channel 2 (480–560 nm, laser 488 nm) was generated using bright mode and a spot to cell background ratio equal to 7.0. To determine the intensity of the internalized PKH67-marked exosomes on immune cells, an Intensity Mask on channel 2 (480–560 nm, laser 488 nm) was also created using the range 60 to 4,095 to eliminate autofluorescence signals (this threshold was determined on unmarked cells). These masks were then combined altogether using the boolean logic operator « AND » and used as an input in the Spot Count Feature to determine the number of spots per cell. Exosome-negative or -positive cells were respectively defined as cells without or with at least one spot in their cytosolic compartment.
Cytokine titration
Supernatants from cocultures of CD8+ T cells and tumor exosomes were harvested. TNFα was then quantified using the enzyme-linked immunosorbent assay (ELISA) kits (eBioscience) according to the manufacturer's instructions.
Cytotoxicity assay
CytoTox-Glo Cytotoxicity Assay (Promega) was performed accordingly to the manufacturer's instructions. In brief, 1,000 target cells, i.e., M113 cells, were added to CTL03.1 cells, preincubated or not with melanoma-derived exosomes M113 or M117 for 24 hours, at the 1:1, 1:10, and 1:40 effector (E) target (T) ratios in a final volume of 100 μL. The plates were incubated for 4 hours. Fifty microliters of supernatant was transferred to a fresh plate after spin down and 50 μL substrate was added to each well and mixed, and the plate was sealed. Digitonin was added in control wells to obtain the maximal cytotoxicity value. The mixture was incubated for 15 minutes at room temperature in the dark. Luminescence was read and specific lysis was calculated.
Availability of supporting data
All original microarray data are deposited in the NCBI's Gene Expression Omnibus database (GSE125030).
Statistical analysis
Error bars indicate ± SEM between biological replicates. Technical as well as biological triplicates of each experiment were performed. Statistical significance was determined using nonparametric Wilcoxon test. NS, nonsignificant; *, P < 0.05; **, P < 0.01; ***, P < 0.001. All statistical analyses were conducted using R3.4.2.
Results
Global analysis of miRNAs in melanoma-derived exosomes
We investigated the miRNA contents of melanoma-derived exosomes using Affymetrix miRNA 4.0 Arrays. First, exosomes were isolated using a staged centrifugation protocol or ExoQuick-TC reagent from cell lines' conditioned media (Supplementary Fig. S1). The sizes of extracellular vesicles were uniformly distributed with similar diameters among exosomes purification technologies (centrifugation: 102 ± 58 nm; ExoQuick-TC: 97 ± 42 nm; Supplementary Fig. S1A and S1B) and the detection of the common markers CD63 and CD81 in our purified microvesicles indicated that these were exosomes (Supplementary Fig. S1C).
Profiling of RNA isolated from exosomes and their donor cells indicates that exosomes were enriched in small RNA species (50% vs. 5% in the parental cell line; Fig. 1A). Then, the hierarchical clustering of the array data groups the samples according to their cellular or exosomal origin (Supplementary Fig. S2). We found that the miRNA content of melanoma-derived exosomes was highly correlated independently of their cell line origin (r2 = 0.98). Overall, there was lower similarity between the miRNA content in exosomes and their corresponding parent cells than between exosomes of different melanoma cell lines (Fig. 1B). We next analyzed the 2,578 human miRNAs located in the array by two approaches (Fig. 1C). The first one identified the differentially expressed miRNAs between melanoma-derived exosomes and their parent cells. In this way, we found 198 miRNAs upregulated in melanoma cell lines compared with their exosomes and 206 miRNAs upregulated in exosomes compared with their donor cells (Fig. 1D; Supplementary Table S2). To assess whether some miRNAs were specifically sorted into exosomes, we conducted an unbiased search for specific motifs overrepresented in exosomal-enriched miRNAs. This analysis by MEME identified a conserved minimal G-rich motif (Fig. 1E) suggesting that miRNAs containing this motif could be specifically sorted into exosomes.
miRNA profiling of melanoma-derived exosomes. A, One representative RNA content analysis of melanoma exosome and paired melanoma cell using Agilent RNA nano kit and small RNA kit (Bioanalyzer, Agilent; n = 10). B, Scatter plots of the exosome versus cell averaged array data or versus exosome averaged data. The correlation is shown. C, Flowchart of the miRNA analysis of melanoma exosomes and their respective donor cells. D, Heat map of the normalized data for miRNAs differentially expressed between cells and exosomes (P < 0.01). E, Overrepresented motif in miRNAs enriched in melanoma exosomes. MEME was used to discover this motif with a Markov model of order 0 assumed for the background sequences.
The second analytic approach identified the most commonly expressed miRNAs in the melanoma-derived exosomes, without making assumptions about their expression in parent cells. We defined two criteria: the first one is a Robust Multi-Array Average (RMA, an alternative to gene-expression value) above 5, corresponding to the mean expression of the array, and the second is an SD ≤0.2 between all samples to discover miRNAs expressed uniformly between melanoma-derived exosomes M113 and M117. We identified 44 miRNAs under these criteria (Table 1).
List of the most represented miRNAs in melanoma exosomes.
To conclude, we identified 250 miRNAs enriched or overexpressed in melanoma-derived exosomes that could modulate gene expression in recipient cells able to internalize them.
CD8+ T cells internalize melanoma-derived exosomes
To further investigate which cells in the tumor microenvironment can be modified by the miRNA content of melanoma-derived exosomes, we labeled melanoma-derived exosomes with PKH67 dye and monitored their uptake by healthy immune (PBMC) cells. Using live imaging flow cytometer Amnis Image Stream, we found that monocytes, T and NK (natural killer) cells can internalize melanoma-derived exosomes whereas B cells cannot (Fig. 2A; Supplementary Fig. S3).
Uptake of melanoma exosomes by immune cells. A, Representative images of PKH-67–labeled exosomes in PBMCs (Amnis Image Stream). Cells were stained respectively with anti-CD14, anti-CD56, anti-CD19, and anti-CD3 prior to acquisition. B, Transmission electronic microscopy of melanoma exosomes captured by CD8+ T cells. Melanoma exosomes are naturally stained by melanin. White arrows indicate pigment melanin. C, Confocal microscopy detection of PKH-67–labeled exosomes (green) in CD8+ T cells after coculture for 5 hours. Cell membranes were stained with WGA-647 (red), and nuclei were stained with Hoescht (blue). The image shows orthogonal views of two lymphocytes (scale bar, 5 mm). D, Measurement of the uptake of PKH-67–labeled exosomes by T cells (Amnis Image Stream). Representative images of melanoma exosomes internalized by CD8+ T cells. Cells were stained with anti-CD3 and anti-CD8 before acquisition. Histogram represents the percentage of PKH-67+ cells (n = 5) after incubation or not (in white) with melanoma exosomes (M113 exosomes in black, M117 exosomes in gray) for 5 hours. E, Melanoma exosomes internalization by CD8+ T cells is an active process. T cells were incubated with melanoma exosomes at 4°C (during 1 hour in light pink, 5 hours in violet, and 24 hours in pink) or at 37°C with cytochalasin D (0 μmol/L in red, 2 μmol/L in orange, 5 μmol/L in green, 10 μmol/L in blue, and 20 μmol/L in yellow) for 24 hours (n = 9). In gray, the control without exosomes. F, One representative flow cytometry plot of Alexa-488–labeled miR mimic internalized by CD8+ T cells (n = 3).
Because of their antitumor response ability, we focused next on CD8+ T cells. Resting or activated CD8+ T cells internalized melanoma-derived exosomes as soon as 5 hours after exosomes exposure (Fig. 2). We monitored melanoma-derived exosome uptake by CD8+ T cells because melanocyte-derived exosomes contain melanin (25), melanin can be used as a marker for electron microscopy. CD8+ T cells exposed to melanoma-derived exosomes showed black spots in their cytoplasm that were not visible in nonexposed cells (Fig. 2B). Confocal microscopy also showed that green-labeled melanoma-derived exosomes were internalized by CD8+ T cells (Fig. 2C). Image analysis, obtained by the resolutive and live imaging flow cytometer Amnis Image Stream, showed that more than 70% of CD8+ T cells exposed to melanoma-derived exosomes have internalized at least one exosome (Fig. 2D). CD8+ T-cell capacity to internalize exosomes was abolished when culture conditions were switched from 37°C to 4°C, suggesting that uptake was energy-dependent (Fig. 2E, top; Supplementary Fig. S4A). Cytochalasin D, a disruptor of cytoskeletal organization, also reduced exosome uptake in a dose-dependent manner (Fig. 2E, bottom).
We also investigated whether uptake of tumor exosomes by CD8+ T cells was specific to melanoma vesicles or could be extended to other tumor types. Thus, we purified and stained exosomes from mesothelioma and colon colorectal cancer cell lines. We observed that exosomes from these sources were also internalized by CD8+ T cells (Supplementary Fig. S4B).
We next validated that miRNAs were efficiently transferred from melanoma-derived exosomes to CD8+ T cells through the use of a fluorescent Alexa-488–labeled miR mimic transfected in melanoma cell line M113, prior to exosomes purification and cocultured with CD8+ T cells. We first validated by high-resolution flow cytometry that exosomes derived from these transfected cells were labeled with Alexa-488 (Supplementary Fig. S4C). Then, the appearance of green fluorescent dye in T cells demonstrated that the Alexa-488-miR mimic was delivered into CD8+ T cells from melanoma-derived exosomes (Fig. 2F). The content transfer from tumor to immune cells through exosomes was also validated by the presence of melanin in T cells (Fig. 2B).
Taken together, these results suggest that melanoma-derived exosomal uptake by the CD8+ T cells is an energy-dependent process that requires a functional cytoskeleton, which are both indicative of endocytic pathways. Melanoma-derived exosomes could thus modulate CD8+ T-cell activity through the release of their miRNA content.
hsa-miR-498 in melanoma-derived exosomes diminishes TNFα secretion by T cells
To evaluate the impact of miRNAs in melanoma-derived exosomes on the CD8+ T-cell function, we first measured cytokine productions by CD8+ T cells exposed or not to exosomes derived from two melanoma cell lines (M113 and M117) for 24 hours. For that purpose, we first activated T cells with anti-CD3 or anti-CD3/CD28 beads. We used CD8+ T cells sorted from healthy PBMCs, a Melan-A-specific CD8+ T-cell clone, called CTL03.1, and CD8+ T cells sorted from tumor-infiltrated lymphocytes from M113 and M117 melanoma patients. We found a significant decrease of TNFα secretion when CD8+ T cells were exposed to melanoma-derived exosomes compared with nonexposed cells (Fig. 3A; Supplementary Table S3), whereas IL2 and IFNγ secretions were not affected. Melanoma-derived exosomes reduced TNFα secretion by around 12% in CTL03.1 cells (P = 0.0002), by almost 20% in TIL M113 (P = 0.032) and 10 to 15% in TIL M117 (P = 0.032). The same tendency was observed in CD8+ T cells purified from healthy donors, comprising naïve and memory T cells, exposed to tumor exosomes (P = 0.069). Melanoma-derived exosomes decreased TNFα secretion in a dose-dependent manner (Supplementary Fig. S5A); this reduction was specific to the melanoma-derived exosomes because melanocyte-derived exosomes had no effect (Supplementary Fig. S5B). We observed a 2-fold decrease of the TNFα transcript 5 hours after melanoma-derived exosomes exposure (Fig. 3B).
Melanoma-derived exosomes decrease TNFα secretion by CD8+ T cells. A, TNFα protein concentrations in supernatant of CD8+ T cells activated and exposed for 24 hours to melanoma-derived exosomes (20 μg/mL), expressed as percentage of reduction (percentage of reduction relative to nonexposed CD8+ T cells). Concentrations of TNFα secreted by nonexposed T cells are 6,600 pg/mL for sorted CD8+ T cells, 39,049 pg/mL for the CTL03.1 clone, and 743 and 1565 pg/mL, respectively, for TILs M113 and M117. Results are expressed as mean ± SEM (n = 9). Wilcoxon tests are applied. B, TNFα RNA expression in CTL03.1 cells exposed to M113- or M117-derived exosomes determined by qPCR after 5 hours of coculture. The 2−ΔΔCt method was used to calculate relative changes in expression (n = 3). Results were normalized to the level of RPLP0 and PPIA gene expressions. C, TNFα secretion by CTL03.1 cells transfected with 50 μmol/L of candidate miRNAs (percentage of reduction relative to the nontransfected CTL03.1 cells). Wilcoxon tests are applied. Results are expressed as mean ± SEM (n = 3). D, Percentage of relative luminescence unit of HEK-293 cells transfected with candidate miRNAs (10 μmol/L) and our TNFα 3′UTR constructs (20 ng). All assays were conducted in triplicate.
To identify miRNAs from melanoma-derived exosomes implicated in this reduction, we screened the 206 miRNAs enriched in melanoma-derived exosomes and the 44 miRNAs highly expressed in exosomes using target prediction analyses (miRNAtap and topGO) implemented in R. We found that hsa-miR-122, hsa-miR-3187-3p, hsa-miR-498, hsa-miR-149, and hsa-miR-181 family potentially target genes implicated in the immune response in T cells (26, 27). Using the public GSE35387 data set, we confirmed that hsa-miR-122, hsa-miR-149, and hsa-miR-498 were enriched in melanoma-derived exosomes compared with melanoma cell lines and also compared with healthy exosome from melanocytes (28). hsa-miR-3187-3p was not investigated in this data set because it was discovered later. Moreover, hsa-miR-181 family was also overrepresented in melanoma-derived exosome in the GSE35387 data set (28). By RT-qPCR, we found that these candidate miRNAs were also enriched in three additional melanoma cell lines (M6, M28, and M45) confirming that some miRNAs are specifically sorted into melanoma-derived exosomes (Supplementary Fig. S6).
These candidate miRNAs were transfected in CTL03.1 cells and TNFα secretion was measured in cell culture supernatant. We found a significant decrease of TNFα secretion when CTL03.1 cells were transfected with all candidate miRNAs. The greatest reduction was observed for hsa-miR-3187-3p, leading to a decrease of more than 50% of the secretion (P = 6.10−5; Fig. 3C). To test the direct interaction between the TNFα 3′UTR and miRNAs, this 3′UTR sequence was inserted downstream of the RenSP luciferase reporter gene and transiently cotransfected with miRNA mimics into HEK-293 cells. We found that only members of the hsa-miR-181 family and hsa-miR-498 interact directly with the 3′UTR sequence of TNFα (Fig. 3D). Other miRNAs downregulate TNFα secretion indirectly. Because the TNFα target sequence of hsa-miR-498 was concordant according to miRWalk, RNAhybrid, and TargetScan, we mutated it in our reporter vector (Fig. 4A). We found that the introduction of 2 mismatches abolished the interaction between hsa-miR-498 and the 3′UTR of TNFα as shown by the absence of luminescence loss in our reporter cells transfected with mutated plasmids and hsa-miR-498 (Fig. 4B).
hsa-miR-498 targets the TNFα 3′UTR. A, Schematic representation of wild-type and mutated constructs in the miR-498–binding site. B, Percentage of relative luminescence unit of HEK-293 cells transfected with miR-498 (10 μmol/L) and our different constructs (20 ng). All assays were conducted in triplicate. WT, wild type.
To conclude, hsa-miR-498 targets the 3′UTR of TNFα to downregulate its expression and consequently its secretion by CD8+ T cells.
hsa-miR-3187-3p in melanoma-derived exosomes reduces T-cell receptor signaling
By flow cytometry, we checked for expression of proteins implicated in the T-cell receptor (TCR) signaling pathway in CD8+ T cells after exposure to melanoma-derived exosome during 24 hours. One of the first molecules to be activated is CD45, a phosphatase that regulates signaling in T cells (29–31). We found that CD45 expression was downregulated on CD8+ T cells after exposure to melanoma-derived exosomes (Fig. 5A). In CTL03.1 cells, CD45 expression decreased by more than 20% with melanoma-derived exosomes (Fig. 5B). This reduction was dependent of the dose of melanoma-derived exosomes added to T cells (Supplementary Fig. S7A) and was also specific to the melanoma-derived exosomes (Supplementary Fig. S7B). We also observed a downregulation of PTPRC (protein tyrosine phosphatase, receptor type, C) transcripts, the coding gene of CD45, in CTL03.1 cells exposed to melanoma-derived exosomes. This reduction was evident after 5 hours of coculture and persisted until 24 hours (Fig. 5C).
Melanoma-derived exosomes decrease CD45 expression on CD8+ T cells. A, Representative flow cytometry plots of CD45 expression in CD8+ T cells. In blue, CD8+ T cells alone; in red, CD8+ T cells exposed to exoM113; and in gray, the isotype control. MFI, median fluorescent intensity (n = 6). B, CD45 expression measured by flow cytometry on CTL03.1 cells exposed to M113 or M117 exosomes expressed as percentage of reduction (percentage of reduction relative to nonexposed CD8+ T cells). Results are expressed as mean ± SEM (n = 12). Wilcoxon tests are applied. C, PTPRC expression in CTL03.1 cells exposed to M113 or M117 exosomes determined by qPCR after 5 hours of coculture. The 2−ΔΔCt method was used to calculate relative changes in expression. Results were normalized to the level of RPLP0 and PPIA gene expressions (n = 3). D, CD45 expression in CTL03.1 transfected with 50 μmol/L of miR-3187-3p or miR-498 expressed as percentage of reduction (percentage of reduction relative to nontransfected CD8+ T cells). Results are expressed as mean ± SEM (n = 9). E, Relative luminescence unit of HEK-293 cells transfected with miR-3187-3p. All assays were conducted in triplicate and were normalized to a control containing only transfection reagents. F, Relative luminescence unit of HEK-293 cells transfected with an increasing dose of miR-3187-3p. All assays were conducted in triplicate.
Next, candidate's miRNAs were transfected in CTL03.1 cells, and CD45 expression was measured by flow cytometry 24 hours after transfection. We found that hsa-3187-3p decreased CD45 expression by 18% (P = 0.0035) and hsa-miR-498 by 14% (P = 0.0038; Fig. 5D) whereas other miRNA mimics had no effect on CD45 expression. To test the direct interaction between PTPRC 3′UTR and hsa-miR-3187-3p or hsa-miR-498, PTPRC 3′UTR sequence was inserted downstream of the RenSP luciferase reporter gene and transiently cotransfected with miRNA mimics into HEK-293 cells. We found that only hsa-miR-3187-3p directly targeted PTPRC 3′UTR (Fig. 5E).
Using prediction software, we failed to find any consensus sequence on PTPRC 3′UTR targeted by hsa-miR-3187-3p, so we decided to validate this interaction by different ways. First, we transfected an increasing dose of hsa-miR-3187-3p with our reporter plasmid in HEK-293 cells, and we observed a dose-response validating its interaction (Fig. 5F). Next, we used an hsa-antimiR-3187 transfected alone or with hsa-miR-3187-3p directly in T cells. We showed that miR-3187-3p silencing by its antimir rescued CD45 expression on CTL03.1 cells, whereas its antimir alone did not modify CD45 expression (Fig. 6A). Finally, to validate the functional ability of hsa-miR-3187-3p included in exosome, we transfected the hsa-antimiR-3187 directly in melanoma-derived exosomes and then exposed CTL03.1 cells to these purified transfected exosomes. Again, the silencing of the native hsa-miR-3187-3p carried in the melanoma-derived exosomes by its antimiR rescued the CD45 expression on T cells (Fig. 6B).
hsa-miR-3187-3p from melanoma-derived exosomes targets PTPRC transcripts. A, One representative FACS plot of CD45 expression in CTL03.1 transfected with hsa-miR-3187-3p alone or with antimiR-3187-3p. In blue, CD8+ T cells alone; in pink, CTL03.1 cells transfected with hsa-miR-3187-3p mimic; in yellow, CTL03.1 cells transfected with hsa-antimiR-3187-3p; in green, CTL03.1 cells transfected with hsa-miR-3187-3p mimic and its antimir; and in gray, the isotype control. MFI, median fluorescent intensity. P values were calculated from three independent experiments in triplicate. B, One representative FACS plot of CD45 expression in CTL03.1 exposed to melanoma-derived exosomes transfected with hsa-miR-3187-3p. In blue, CD8+ T cells alone; in red, CD8+ T cells exposed to exoM117; in green, CTL03.1 cells exposed to exoM117 transfected with hsa-antimiR-3187-3p; and in gray, the isotype control. MFI, median fluorescent intensity. P values were calculated from three independent experiments in triplicate. C, Cytotoxicity assay against M113 cells by CTL03.1 cells (n = 3). Cytotoxicity was measured at different target (T)/effector (E) ratios with no exosome (white-filled triangle) and with CTL03.1 preincubated overnight with M113 (black circle) or M117 exosomes (gray square).
To conclude, hsa-miR-3187-3p regulated CD45 expression through its transfer from melanoma cells to T cells by exosome.
These results show that melanoma-delivered miRNAs regulating gene expression in CD8+ T cells lead to functional changes and abnormal immune response against tumor. We investigated the ability of CD8+ T cells exposed to melanoma-derived exosomes to kill tumor cells in vitro by a luminescence-based cytotoxicity assay. Exposure of CTL03.1 cells to melanoma-derived exosomes resulted in a significant decline in their ability to induce the death of melanoma cell lines (P = 0.0109; Fig. 6C).
Discussion
miRNAs participate in the intercellular communication (32). Accordingly, extracellular miRNAs secreted by tumor cells may be delivered into recipient cells in the microenvironment where they will function as endogenous miRNAs leading to altered gene expression. Our study describes this mechanism in human melanoma. We showed that two miRNAs, hsa-miR-498 and hsa-miR-3187-3p, when transferred from tumor cells to CD8+ T cells through exosomes, can diminish the immune response against tumor.
By comparing the miRNA content of exosomes and their parent cells, we identified 206 miRNAs differentially expressed between melanoma-derived exosomes and melanoma cell lines. This signature may participate at the tumor immune escape because some miRNAs are able to downregulate the CD8+ T-cell response. Indeed, hsa-miR-3187-3p, carried by melanoma exosomes, decreases CD45 expression. The interaction between CD45 and Src kinases is vital for successful antigen receptor signaling in T cells (31, 33). CD45 dephophorylates the Src family protein tyrosine kinases to initiate and potentiate intracellular signaling in T cells (34). Loss of CD45 phosphatase activity reduces the abundance of active kinases and decreases TCR signaling (35), but also blocks thymocyte development (30, 31). Despite the requirement for CD45 in T-cell function, mechanisms of action remain unclear. Although hsa-miR-3187-3p does not target directly TNFα transcripts as do hsa-miR-498 or hsa-miR-181a and b, the observed loss of TNFα secretion driven by hsa-miR-3187-3p may be a consequence of the downregulation of the TCR signaling. Likewise, other miRNAs included in melanoma-derived exosomes have been implicated in the indirect regulation of TNFα secretion. This complexity makes it difficult to drive recovery of TNFα secretion by transfection of only one antimiR. Tumor exosomes from nasopharyngeal carcinoma decreased also TNFα released from CD8+ T cells, although the miRNAs driving this outcome were not identified (36).
Although molecular mechanisms of miRNAs sorting in exosomes remain unknown, we identified a common motif that could direct loading of specific miRNAs in melanoma-derived exosomes. The ability of short nucleotide sequences to guide transport of RNAs to different subcellular compartments, such as mitochondria or nucleus, has been previously reported (37, 38). Further analysis and mutagenesis need to be performed to confirm this hypothesis.
Exosomes are complex vesicles that contain an assortment of proteins, RNAs and DNA. Thus, we cannot claim that miRNAs encompassed in melanoma-derived exosomes are the only driver of reduced T-cell response. Indeed, we observed for TNFα as well as for CD45 a lower loss in transcript amount than in protein level. This could be explained by (i) miRNA-inhibiting translation without changing amounts of mRNA or (ii) other mechanisms that regulate protein amounts.
We have showed that CD8+ T cells can internalize tumor exosomes even if these cells are not as likely as other immune cells to internalize vesicles. Few studies have investigated the uptake of tumor exosomes by T-cell populations because these cells are less equipped to internalize vesicles than are macrophages or dendritic cells. In mouse, Wu and colleagues have shown that melanoma exosomes could be taken up by CD8+ T cells to deliver a biological payload (39), but in human, the uptake of tumor exosomes by T cells is less well understood. Maybruck and colleagues suggested that exosomes from human colon carcinoma and head and neck squamous cell carcinoma cell lines interact with T cells but may not necessarily be internalized (40). Exosomes from human prostate cancer cells could cause CD8+ T-cell apoptosis through an interaction with Fas/FasL (41), although resting or activated CD8+ T cells did not internalize exosomes from head and neck cancer cell lines, even after 72 hours of coincubation (42). Here, using different technologies (electronic and confocal microscopy, cytometry) we showed that exosomes from melanoma, mesothelioma, or colon colorectal carcinoma cells could be internalized by active or resting CD8+ T cells. However, our study does not indicate whether phenotypic and functional changes in T cells induced by tumor exosomes are common to all cancers. Additional studies of the miRNA content of these different kinds of cancer cells are required to address this question. Our study showed that certain miRNAs in tumor exosomes induce a favorable environment for the tumor by reducing immune response.
Disclosure of Potential Conflicts of Interest
No potential conflicts of interest were disclosed.
Authors' Contributions
Conception and design: V. Vignard, N. Labarrière, D. Fradin
Development of methodology: V. Vignard, G. André-Grégoire, D. Fradin
Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): V. Vignard, M. Labbé, N. Marec, G. André-Grégoire, N. Jouand, J.-F. Fonteneau, D. Fradin
Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): M. Labbé, N. Marec, G. André-Grégoire, N. Jouand, D. Fradin
Writing, review, and/or revision of the manuscript: M. Labbé, N. Jouand, N. Labarrière, D. Fradin
Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): D. Fradin
Study supervision: D. Fradin
Acknowledgments
This work was supported by grants from “Ligue contre le cancer - Comités 22, 29, 35, 44 et 56,” from “Cancéropole Grand Ouest – AO Structurant - ExomiR,” and from “Vaincre le Mélanome.” This work was realized in the context of the LabEX IGO program supported by the National Research Agency via the investment of the future program ANR-11-LABX-0016-01. We thank the cytometry facility “CytoCell” (SFR Santé Nantes), the microscopy facility “MicroPICell” (SFR Santé Nantes), the electron microscopy facility “SC3M” (SFR Santé Nantes), and the genomic facility “GENOM'IC” (Institut Cochin Paris) for their expert technical assistance. We thank Joëlle Veziers (SC3M, SFR Santé Nantes) for her work, help, and expertise in electron microscopy. We thank Dr. C. Blanquart for providing the mesothelioma cell line and T cells, and Dr. A. Jarry for providing the colon colorectal cancer cell line. We thank Dr. E. Mortier and Dr. B. Ory for critical reading of the manuscript.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Footnotes
Note: Supplementary data for this article are available at Cancer Immunology Research Online (http://cancerimmunolres.aacrjournals.org/).
Cancer Immunol Res 2020;8:255–67
- Received July 9, 2019.
- Revision received October 14, 2019.
- Accepted December 10, 2019.
- Published first December 19, 2019.
- ©2019 American Association for Cancer Research.