Background
Epithelial ovarian cancer (EOC) is the most lethal gynecological cancer in the United States, with an estimated 22,530 new cases diagnosed and 13,980 deaths in 2019 [
1,
2]. Despite the standard first-line chemotherapy following cytoreductive surgery, the mortality rate has not improved in the era of targeted-therapy; the 5-year survival rate is around 39% for all EOC and 30% for in patients at stage III and IV [
1]. More than half of cases are diagnosed at advanced stages with extensive intraperitoneal disseminated metastasis and chemotherapeutic resistance [
3,
4]. Growing evidence suggests that the peritoneal tumor microenvironment (TME) plays an essential role in ovarian cancer progression, metastasis, and the development of drug resistance [
5]. During metastasis of ovarian cancer, the intraperitoneal tumor microenvironment forms an immunosuppressive milieu with accumulated ascites that contain a large number of tumor spheroids and various stromal and immune cells, such as macrophages and lymphocytes, as well as soluble pro-tumor mediators [
6]. The macrophages that infiltrate into the tumor microenvironment are defined as tumor-associated macrophage (TAMs), which generally display an anti-inflammatory and pro-tumor M2-like phenotype, and thus play an important role in facilitating the peritoneal dissemination of ovarian cancer cells [
7]. Increased recruitment of TAMs into tumor spheroids floating in the ascites has been positively correlated with poor outcomes in patients with advanced ovarian cancer [
6]. It remains unclear how ovarian tumor cells evade immune surveillance, a key obstacle to develop effective immunotherapy for ovarian cancer treatment.
Exosomes are nanometer-sized membrane encapsulated vesicles released by all cell types to convey information to neighboring or distant cells by transporting cytosolic biomolecules, such as proteins, DNA, mRNAs, and miRNAs, and thus influence recipient cells [
8,
9]. Among all of the cell types, tumor cells secrete at least 10-fold more exosomes than other cells [
10]. Exosomes generated by tumors carry cargos that partially mimic parent cell content, and thereby they reprogram recipient cells into active contributors to angiogenesis, metastasis, and immunosuppression [
11]. Tumor cells generate distinct exosomes that are regulated by external signals, particularly from oxidative stress [
12]. We previously showed that ovarian cancer cells produce excessive reactive oxygen species (ROS) with NOX4 being the major contributor [
13]. ROS can regulate the biogenesis and expression of miRNAs, which may in turn affect redox signaling pathways and thus promote tumorigenesis and progression [
14,
15]. However, it has yet to be determined how these miRNAs convey the ROS signal to downstream effectors leading to alterations in the extracellular milieu that creates a redox microenvironment. The current study aimed to investigate whether and how ROS influence the TME to promote ovarian cancer development through tumor exosomal miRNAs. We found that ROS decreased the amount of tumor exo-miR-155 that was taken up by macrophages, resulting in enhanced macrophage infiltration and T cell inactivation characterized by upregulation of programmed death ligand 1 (PD-L1). Targeting ROS/miR-155-5p is a promising strategy to prevent the formation of the suppressive tumor microenvironment in ovarian cancer.
Methods
Cell lines, reagents, and antibodies
Human ovarian cancer cell line A2780, Ovcar-3, and SKOV-3, murine ovarian cancer cell line ID8, and human peripheral blood monocytes THP-1 were purchased from the American Type Culture Collection (ATCC). Peripheral blood mononuclear cells (PBMC) were purchased from Stemcell Technologies. All cell lines were cultured in RPMI 1640 or in Dulbecco’s modified Eagle’s (DMEM) media supplemented with 10% FBS. All cell lines did not have mycoplasma contamination determined by RT-PCR. N-acetyl-L-cysteine (NAC) was purchased from Selleckchem, and phorbol 12-myristate 13-acetate (PMA), catalase-PEG, and rotenone were from Sigma Aldrich. The sources of antibodies used in immunoblotting, flow cytometry, and IHC were shown in Suppl. Table
2.
Exosomes purification from cell culture supernatant, ascites, and serum
Exosomes from cell culture medium were isolated by ultracentrifugation. Ovarian cancer cells were grown in T75 flasks until they reached to 70% confluency, and then were rinsed with PBS and then replaced with serum-free medium for 48 h. The supernatant was collected and centrifuged at 2000×g for 30 min to discard cellular debris. Next, the collected medium were centrifuged at 12000×g for 30 min at 4 °C followed by filtration through 0.22-μm pore filters (Steriflip, Millipore) to remove larger extracellular vesicles. The supernatant were then and ultracentrifuged at 110,000×g for 2 h at 4 °C (Beckman Coulter, Ti45 rotor). The exosomes pellets were re-suspended in 10 ml PBS, and loaded over 10 ml of 40% sucrose solution, and was ultracentrifuged at 110,000×g for 90 min. The lower sucrose layer, which contained the exosomes was diluted with PBS and ultracentrifuged at 110,000×g for 90 min. The pellet was resuspended in 100 μl PBS and stored in − 80 °C. Exosomes from ascites in mice were isolated using the same method except for erythrocytes lysis with ACK lysis buffer (Thermo Fisher Scientific) after the initial centrifuge at 1500×g for 15 min to remove debris. Exosomes from the serum in mice were isolated using Exosome Isolation Reagent (Invitrogen, Thermo Fisher Scientific). The serum exosomes were characterized by NTA.
Nanoparticle tracking analysis (NTA)
NTA was performed using a NanoSight NS300 instrument according to the manufacturer’s instructions. Isolated exosome samples were diluted with PBS at 1:100 before analysis. The settings after NTA collection were optimized and remained constant between samples. Three videos of 60 s length were recorded for each sample and analyzed to give estimates of mean size and number of particles.
Nanostring nCounter miRNA expression assay
All samples were prepared and processed according to NanoString nCounter Expression CodeSet Design Manual. 100 ng of exosomal RNAs were used as input for Nanostring nCounter miRNA sample preparation. Raw data were normalized to the top 100 miRs using nSolver analysis 2.5 software (Nanostring). Expression heatmaps were generated with the R package pheatmap.
Fluorescence in situ hybridization
In situ hybridization was performed using double digoxigenin (DIG)-labeled miRCURY LNA miRNA detection probe hsa-miR-155-5p (Advanced Cell Diagnostics, CA, USA) in human ovarian carcinoma tissue microarray (US Biomax, MD, USA). The TMA sections contain 160 cases of ovarian cancer tissue sections and 40 cases of normal human ovary tissue sections. The experiments were performed according to the miRNAscope LS Automated Assay (ACD, CA, USA) protocol. The signal was detected with miRNAscope™ LS Reagent Kit - RED (ACD, CA, USA). Slow Fade Gold anti-fade reagent with DAPI (Life Technologies) was employed as the mounting medium. The positive and negative control of the method was verified employing miRNAscope LS Positive Control Probe (ACD, CA, USA) and miRNAscope LS Negative Control Probe (ACD, CA, USA), respectively. Bright field pictures were scanned by Pannoramic slide scanner (3DHISTECH) and converted to fluorescence signal with Caseviewer 2.2 software. Pictures were taken by Image-Pro Plus 6.0 software. Slides were analyzed with Quant Center 2.1 (Thermo Fisher Scientific) and Graphpad Prism software (La Jolla, CA, USA).
RT-qPCR
Total RNAs were extracted using Trizol (Thermo Fisher Scientific). The cDNA synthesis was performed using oligo (dT)18 primers and M-MLV reverse transcriptase. The 100 ng of RT product was used for PCR reaction using Power SYBR Green PCR Master Mix (Thermo Fisher Sci). The primer sequences are listed in Supplementary Table
1.
Two-step Taqman-qPCR analysis was performed to assess miRNA levels using Taqman miRNA reverse transcription kit and Taqman universal PCR master mix (Thermo Fisher Sci) in accordance with manufacturer’s instructions.
Fluorescent labeling of exosomes with DiR
Exosomes were fluorescently labeled using lipophilic carbocyanine DiR dye (Thermo Fisher Scientific). In brief, 5 μl of DiR (200 μg/ml) in ethanol was mixed with 200 μg exosomes in 100 μl PBS for 1 h. The unincorportated DiR was removed by a Sepharose CL-4B column (Sigma-Aldrich).
Dual luciferase reporter assay
THP-1 cells were seeded in 12-well plates at a number of 5 × 105 and were co-transfected with PD-L1-WT 3′-UTR reporter plasmids (Addgene) and miR-cont, miR-155-5p mimics, miR-152-3p mimics, or miR-137 mimics, using jetPRIME® reagent. The transfected cells were harvested 48 h post-transfection using Passive Lysis Buffer (Promega, US), and firefly and Renilla luciferase activities were measured in cell lysates using Dual-Luciferase Reporter Assay System (Promega).
3D-coculture tumor model and 3D live cell imaging
A2780-miR-con and A2780-miR-155 cells were seeded onto 1% agarose (Sigma-Aldrich) substrate in the cover glass bottom of 48-well plates (MatTek, MA, USA) at 1 × 104 cells/well. After 4 days of formation of tumor spheroids, macrophages were added and allowed to co-culture with tumor spheroids for an additional 4 days. The diameters of spheroids were measured at 3 d, 7 d, 10 d and 12 d, respectively. For the 3D live cell imaging, the cells were stained with Cell tracker green (Thermo Fisher Scientific) before seeding onto 1% agarose concave surface for spheroid formation. The infiltrated macrophages were stained by eFluor670 (Thermo Fisher Scientific) before adding into tumor spheroids. The pictures of the 3D co-culture were taken by an A1R+ Nikon confocal microscope for 3D live cell imaging. The number of A2780 and infiltrated macrophages were presented as accumulative average intensity of fluorescence. Z stack-overlay video was taken by A1R+ Nikon confocal microscope.
Flow cytometry and fluorescence-activated cell sorting (FACS)
For cultured cells or 3D co-culture cells, single-cell suspensions were incubated with 2% human Fc Receptor block (eBioscience, Frankfurt, Germany) in PBS for 20 min on ice. After washing with PBS, cells were stained with conjugated antibodies for 30 min at room temperature in the dark. For tumors, spleen, and ascites from mouse models, single-cell suspensions were prepared by tissue extraction, mashing, digestion with Liberase DL, TL (Roche, USA) and DNase І (Sigma-Aldrich, Germany). The erythrocytes were lysed with ACK lysis buffer (Gibco, Thermo Fisher Scientific), filtered by 40 μm pore size filter (Corning, NY, USA), incubated with 2% human Fc receptor blocking (eBioscience, Frankfurt, Germany), and stained with conjugated antibodies. Isotype IgG control was used in each experiment as reference. Flow cytometry was performed using BD Celesta, BD LSRII, or BD Fortessa (BD Bioscience, US). FACS was performed using BD Aria II or BD Melody (BD Bioscience, US). Data were analyzed using FlowJo software (Tree Star Inc., CA, USA).
Human CD3+ T cell preparation
The human PBMCs were purchased from Stemcell Technologies (Cambridge, MA). CD3+ T cells were separated using CD3 microbeads (Miltenyi Biotec, USA) together with MACS columns and separators according to the manufacturer’s instructions. For stimulation of T cells, 24-well plates were coated with 10 μg/ml CD3 purified antibody. After washing with PBS twice, isolated CD3+ T cell were plated in CD3-coated plate in culture medium, and stimulated with CD28 purified antibody in the presence of 30 U/ml recombinant human IL-2 (Peprotech, USA).
In vivo tumor models
Animal experimental protocols were in consistent with the Care and Use of Laboratory Animals Guide and approved by the Institutional Animal Care & Use Committee of Thomas Jefferson University (No. 01159). For establishing human ovarian cancer xenograft model in nude mice, 1 × 106 A2780 cells suspended in 100 μl PBS with 50% Phenol Red-free Matrigel (Corning, NY, USA) were injected into flanks of 6-week-old female NCr nude mice (Taconic, US). Tail vein injections of Exo-con or Exo-NAC (100 μg in 100 μl PBS) were performed every 3 days. Tumors were measured with calipers and tumor volume was calculated with formula (width)2 × length/2. Mice were sacrificed 21 days after cell inoculation or the maximum tumor dimension reached 2.0 cm before 21 days.
For establishing immunocompetent mouse ovarian cancer model in C57BL/6 mice (Taconic), 8 × 106 ID8-miR-con or ID8–155 mouse ovarian cancer tumor cells were inoculated intraperitoneally (i.p.) into C57/BL6 mice. Exosomes derived from ID8-miR-con (Exo-miR-con) or ID8-miR-155 (Exo-miR-155) tumor cells were prepared and injected via tail vein every 3 days. Starting at 60 days after tumor inoculation, mice were given IgG isotype control or anti-PDL1 antibodies at 200 μg/mouse every 3 days via intraperitoneal injection (IP). Mice were sacrificed when weight gain more than 30% and/or more than 40% waist circumference. Ascites, tumor nodules, spleens, and blood samples were collected and processed for further analysis.
Statistical analysis
Data are shown as means ± SD. All experiments were repeated at least in triplicate. Statistical analysis were performed by Graph-Pad Prism software (Graph-Pad Software Inc., La Jolla, CA, USA). Differences between two groups were calculated using the unpaired two-tailed Student’s t-test. Statistical analyses of three or more groups were compared using one-way analysis of variance (ANOVA). Cumulative probabilities of overall survival were computed with the Kaplan−Meier analysis and comparisons between groups were analyzed using the log-rank test. P < 0.05 was considered statistically significant.
Discussion
Here we provide evidence that the redox status of tumor cells serves as a critical regulator of the microenvironment in the control of immunologic mechanisms involved in their response to therapy. Compared with normal cells, cancer cells exhibit an increased level of intrinsic reactive oxygen stress because of mitochondrial dysfunction and metabolic alteration, reflecting a disorder of redox homeostasis [
15,
27,
28]. Recent studies have shown that ROS promote tumor immune escape by creating an immunosuppressive tumor microenvironment [
29,
30]. For example, excessive ROS production can reduce the infiltration of lymphocytes and facilitate recruitment and accumulation of regulatory T-cells and M2-like tumor-associated macrophages [
30,
31]. A recent study showed that the inhibition of NADPH oxidase subunit NOX4 potentiates immunotherapy by preventing the formation of immunosuppressive phenotypes of cancer-associated fibroblast [
32]. Consistent with this finding, we found inhibition of ROS by NAC was able to reprogram macrophages with tumor exosomes to suppress ovarian cancer development. We and other groups have shown that ROS-sensitive microRNAs contribute pivotally with respect to how cancer cells respond to ROS [
14,
15,
33]. In this study, we identified exosomal miR-155-5p as a major downstream effector of ROS that mediates tumor immune responses. Given the dual roles of ROS in cancer development, the use of antioxidants in cancer treatment has had limited or even unexpected effects. A study showed that NAC and vitamin E accelerated lung cancer progression in a mouse model [
34]. It is plausible that targeting downstream ROS-sensitive miRNAs, such as miR-155-5p, may provide a more specific approach to manipulation of ROS effects.
A key question in miRNA research is how their expression is regulated under specific contexts. Studies from different groups show that ROS are able to influence microRNAs through altering miRNA biogenesis, transcription factors, and epigenetic modulation [
15]. Our prior study showed that endogenous ROS inhibit miR-199a and miR-125b genes expression through DNA hypermethylation in ovarian cancer cells [
35]. However, the mechanisms by which ROS inhibit cellular and exosomal miR-155-5p levels in ovarian cancer cells are not clear. It is known that the transcription factor nuclear factor-κB (NF-κB), a redox-sensitive factor, can transactivate
miR-155-5p gene expression [
15]. Some studies have suggested that TNF-α-induced ROS accumulation decreases NF-κB expression and the level of its target miR-155-5p [
36]. However, other groups argued that the rise of intracellular ROS levels, induced by TNF or IL-1, can up-regulate JNK-mediated NF-κB activation [
37]. ROS production exerts opposing effects on NF-κB, inducing activation in the cytoplasm and inactivation in the nucleus [
38]. To add complexity, reciprocal regulations exist between ROS and NF-κB signaling [
39,
40]. Whether and how ROS regulate miR-155-5p through NF-κB remains unclear and is a topic worthy of further investigation.
TAMs represent the most abundant infiltrating immune cells in the peritoneal tumor microenvironment to influence ovarian cancer initiation, growth, and metastasis [
41]. A meta-analysis showed that a high density of CD163
+ TAMs infiltration was associated with poor prognosis in ovarian cancer [
42]. It has been reported that exosomes derived from ovarian cancer cells are capable of educating macrophages to obtain the tumor-promoting M2 phenotype [
43,
44]. We found tumor exosomal miR-155-5p inhibited tumor growth through macrophage infiltration inhibition and T cell activation. PD-L1 is an inhibitory checkpoint molecule known for its role in negative regulation of cytokine production and T cell function. PD-L1 is widely expressed in tumor cells, tumor infiltrating lymphocytes, and tumor stromal cells, especially tumor-associated CD68
+ macrophages in ovarian cancer [
45]. Expression of PD-L1 on dendritic cells and macrophages in ovarian cancer and melanoma patients correlated with the efficacy of treatment with either anti-PD-1 alone or in combination with anti-CTLA-4 [
46]. However, it remains elusive how PD-L1 expression is upregulated in macrophages. Studies have shown that miR-155-5p inhibits PD-L1 expression by binding the 3′ UTR of PD-L1 mRNA [
21,
22]. In this study, we found PD-L1 expression was lower in tumor and spleen macrophages in mice treated with Exo-miR-155-5p. Moreover, overexpression of PD-L1 reversed tumor exosomal miR-155-5p-induced CD8
+ T cell proliferation, suggesting PD-L1 contributes to miR-155-5p-mediated anti-tumor immune response.
To date, the blockade of immune checkpoints in ovarian cancer has produced mixed results in preclinical setting and clinical trials [
47,
48]. Krempski et al. reported that PD-1 antibody alone facilitated tumor regression and T cell function and activation while many others found a lack of response [
49,
50]. In this study, we found PD-L1 antibody alone delayed tumor progression and ascites formation. Notably, exo-miR-155-5p inhibited tumor growth more potently than PD-L1 antibody, and the combination of the two agents failed to achieve better efficacy than exo-miR-155-5p alone. This indicates that PD-L1 inhibition by miR-155-5p may not entirely account for the immunosuppressive effects observed, suggesting that other pathways also contribute to miR-155-5p-mediated anti-tumor responses.
miR-155-5p has been recognized as a pro-inflammatory factor that can enhance the production of IL-1β, IL-6, IL-8, and TNF [
51,
52]. It has been implicated in innate and adaptive anti-tumor immune response. Deletion of miR-155-5p reduces the capability of CD8
+ cytotoxic T cells to respond to viral infection or tumor development [
53], and aberrant expression of miR155-5p correlates with inflammation through targeting and degrading SHIP1 and WEE1 genes involved in inflammation [
54,
55]. In addition to PD-L1, overexpression of miR-155-5p can reprogram tumor-associated macrophages to pro-inflammatory, antitumor macrophages, possibly by targeting C/EBP-ß and NF-ĸB and thus suppressing their signaling cascades [
33,
56]. Our results showed that tumor exo-NAC or exo-miR-155-5p decreases CXCL10 and CCL2 expression levels in macrophages. Co-culture with macrophages treated with tumor exo-NAC or exo-miR-155-5p increased TNF-α, IFN-ɣ, and IL-2 levels in T cells. It remains to be investigated whether miR-155-5p regulate the chemokines directly or indirectly through PD-L1 or other pathways. Notably, the direct role of miR-155-5p in ovarian cancer cells is not clear yet. There is no significant change of cell proliferation in A2780-miRcont cells vs. A2780-miR-155-5p cells by MTT assay (72 h, data not shown). However, when co-cultured with macrophages in a 3D model, A2780-miR-155 cells in the spheroids grew slower than control cells starting at day 7 (Fig.
S3B). In line with the mice study, we consider that miR-155-5p inhibits tumor growth mainly through the tumor microenvironment and adaptive immune response. However, this warrants further investigation.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.