Cell-free Tumor Microparticle Vaccines Stimulate Dendritic Cells via cGAS/STING Signaling

Success of a tumor vaccine lies in its ability to provide both tumor antigens and immunostimulatory signals to antigen-presenting cells (APC), leading to tumor-specific T-cell immune responses. Despite the continual discovery of new tumor antigens, using tumor cells as anticancer vaccines remains an attractive strategy, because it offers a broader repertoire of tumor-associated antigens to reduce the possibility of immune escape and development of resistance. Furthermore, the immunogenicity of whole tumor-cell vaccines can be augmented by modifying tumor cells with expression of GM-CSF or other immunostimulatory genes to generate improved antitumor T-cell immunity. To date, autologous and allogeneic whole tumor-cell–based vaccines have been intensely studied in clinical trials in patients with melanoma, renal cell and hepatocellular carcinomas, lung, prostate, breast, colorectal, cervical, pancreatic, or ovarian cancer (1–3). However, there is evidence that tumor cell–derived cytokines (e.g., VEGF, IL10, and TGFβ) and biologic factors (e.g., galectin-1, indoleamine 2,3-dioxygenase, lipid droplets) suppress dendritic cell (DC) maturation and T-cell activation, thus limiting the efficacy and efficiency of whole cell–based vaccination (4, 5). Therefore, a vaccine platform that can simultaneously deliver multiple tumor antigens and immunostimulatory signals to DCs remains to be developed.

Exosomes are small vesicles (60–100 nm) that are released from the cell upon fusion of multivesicular bodies with the plasma membrane (6). Exosomes are secreted by multiple cell types including tumor cells, and it is likely that tumor-released exosomes contain tumor-specific antigens expressed in the parental tumor cells, representing a new cell-free source of shared tumor antigens for in vivo immune priming or tumor vaccine design (7). Indeed, vaccination with tumor cell–derived exosomes or exosome-pulsed DCs was able to induce CD8⁺ T-cell–dependent antitumor effects in both mice and humans (7, 8). However, many reports also suggest that tumor cell–derived exosomes can confer immune suppression through different mechanisms including downregulation of DC maturation, raising concerns about the use of exosomes as tumor vaccine (9).

In addition to exosomes, tumor cells also secrete other types of vesicles of different sizes (10, 11). In response to stress, tumor cells change their cytoskeleton, leading to encapsulation of cytosolic contents by cellular membrane to form vesicles that are subsequently released into the extracellular space. These specialized subcellular vesicles with a diameter of 100 to 1,000 nm are called microparticles, which are different from exosomes in size, membrane structure, and contents (12). Recently, we have demonstrated that tumor cell–derived microparticles (T-MP) can serve as a safe and efficient carrier to deliver chemotherapeutic drugs to tumor cells, whereas their capacity as a novel vaccine platform remains to be determined (13). In another study, we demonstrated that microparticles released by Listeria monocytogenes–infected macrophages could transfer L. monocytogenes antigens to DCs, leading to the induction of protective adaptive immune responses against the bacteria (14), implicating the potential of microparticles as a vehicle for antigen delivery and DC targeting. In the present study, we provide evidence that T-MPs may represent a novel tumor vaccine due to their possession of both tumor antigens and innate DNA signals.

Female BALB/c, BALB/c-nude mice, and C57BL/6 mice (6–8 weeks old) were purchased from Centre of Medical Experimental Animals of Hubei Province (Wuhan, China). All animal experiments were approved by the Animal Care and Use Committee of Tongji Medical College. Mouse H22 hepatocarcinoma, B16 melanoma, and CT26 colon carcinoma tumor cell lines were purchased from the China Centre for Type Culture Collection (CCTCC), and cultured according to the manufacturer's guidelines. All cell lines were tested and determined to be free of Mycoplasma and other rodent pathogens by the CCTCC; no other authentication assay was performed.

Tumor cells were exposed to ultraviolet irradiation (300 J/m²) for 1 hour, and 12 hours later, supernatants were used for microparticle isolation as described previously (13). Briefly, supernatants were centrifuged at 1,000 g for 10 minutes to remove whole cells and then centrifuged for 2 minutes at 14,000 g to remove debris. The supernatant was further centrifuged for 60 minutes at 14,000 g to pellet microparticles. The pellets were washed three times and resuspended in culture medium for the subsequent experiments.

Exosomes were isolated as described (8). Briefly, H22 tumor cells were cultured for 48 hours, and the supernatants were collected and centrifuged (4°C) at 500 g for 10 minutes, 14,000 g for 30 minutes, and exosomes were then harvested by centrifugation at 100,000 g for 70 minutes. The exosome pellet was suspended in 20 mL of PBS and collected by ultracentrifugation as described above. Tumor-cell lysates were prepared by subjecting 1 × 10⁷/mL H22 tumor cells in PBS to three cycles of rapid freezing in liquid nitrogen and thawing at 37°C. The lysates were centrifuged at 2,000 g for 10 minutes to remove cellular debris.

For the H22 or CT26 tumor models, BALB/c mice were immunized three times in the lower left flank with T-MPs or PBS by s.c. injection on days −14, −13, and −7. On day 0, mice were challenged with H22 or CT26 tumor cells (3 × 10⁵ cells/mouse) in the contralateral right flank. Palpable tumors were measured by calipers, and tumor volume was calculated according to the formula: volume = (length × width²)/2, where length and width are measured in millimeters. For the B16 melanoma tumor model, C57BL/6 mice received s.c. injection of 3 × 10⁵ B16 melanoma cells after immunization with B16-MPs as described above. In the metastasis model, C57BL/6 mice were i.v injected with 1 × 10⁵ B16 melanoma cells after immunization with B16-MPs, and 21 days later, mice were sacrificed and the black melanoma nodules on the lungs were assayed.

For the immune-deficient model, nude mice were immunized with H22-MPs or PBS and then challenged with H22 tumor cells. In lymphocyte-depletion experiments, mice were injected i.p. with 0.1-mg anti-CD4 (clone GK1.5), anti-CD8 (clone 53-6.7), or anti-asialo GM1 (eBioscience) antibody at days −1, 2, 5, and 8 before/after tumor inoculation. Tumor size was measured every other day with calipers once tumors became palpable. Flow cytometric analysis confirmed greater than 95% depletion of specific lymphocyte populations.

Tumors were established in mice by s.c. injection of 3 × 10⁵ H22 or CT26 tumor cells in the right flank. Five days after tumor inoculation, mice were treated s.c. with T-MPs or microparticle-pulsed DCs three times at 5-day intervals. Control mice received PBS or unpulsed DCs. C57BL/6 mice carrying subcutaneous B16 tumors or metastases were treated similarly.

Bone marrow cells were harvested from femurs of mice and cultured in RPMI 1640 supplemented with 10% FBS, 2 mmol/L l-glutamine, 1 mmol/L sodium pyruvate, 1 mmol/L HEPES, 50 mmol/L 2-ME, 100 U/mL penicillin, and 100 mg/mL streptomycin. The cells were cultured in six-well plates with 20 ng/mL GM-CSF (PeproTech) and 20 ng/mL IL4 (PeproTech), and cytokines were replenished on days 3 and 5; nonadherent cells were harvested for experiments.

A tumor-specific CTL killing assay was performed as described previously (15). Briefly, splenocytes were harvested from mice preimmunized with H22-MPs, H22 exosomes, or H22 lysates 20 days after tumor implantation, and single-cell suspensions were prepared by homogenization using frosted glass slides. Splenocytes (4 × 10⁷) were cultured with 2 × 10⁶ irradiated (100 Gy) H22 tumor cells. Five days later, the cells were harvested and used as CTL effector cells in a standard lactate dehydrogenase cytotoxicity assay, in which H22 tumor cell targets were seeded at 10,000 cells per well. The percentage of specific killing was defined as (experimental value − effector cells spontaneous control − target cells spontaneous control)/(target cell maximum control − target cells spontaneous control) × 100%.

For phenotypic analysis of DCs, cells were stained with surface antibodies: anti-CD11c (clone N418), anti-CD80 (clone 16-10A1), anti-CD86 (clone GL1), anti-MHC II (clone M5/114.15.2), and anti-CCR7 (clone 4B12). In some cases, anti-IFNAR (clone MAR1-5A3) was added to the culture system to test DC maturation. To examine tumor-infiltrating T cells, tumor-infiltrating leukocytes (TIL) were isolated and stained with antibodies against anti-mouse CD3 (clone 17A2), CD4 (clone GK1.5), and CD8 (clone 53-6.7).

For intracellular cytokine staining, lymphocytes prepared from draining lymph nodes were restimulated with phorbol 12-myristate 13-acetate (PMA; 50 ng/mL) and ionomycin (1 μg/mL; Sigma-Aldrich) in the presence of Brefeldin A (1 μg/mL; eBioscience) for 4 hours, and then stained with CD11c antibody. After surface staining, cells were treated with Fix/Perm solution and restained with anti-IFNγ (clone XMG1.2) or anti-IL12 (clone C17.8) antibody. In some cases, TILs were stained with CD3 and CD8 antibodies and then restained with IFNγ antibody. All antibodies were purchased from eBioscience, and Flow cytometric analysis was performed with Accuri C6.

TILs were isolated and cultured in the presence of PMA and ionomycin for 6 hours. IFNγ in the supernatants was assessed by the Mouse mini ELISA Kit (PeproTech) according to the manufacturer's protocol. DCs were cocultured with or without H22-MPs, and supernatants were collected for 36 hours. IFNα and IFNβ were measured with the mouse IFNα or IFNβ ELISA Kit (R&D Systems).

Isolated H22-MPs were labeled with a green fluorescent cell linker (PKH67; Sigma-Aldrich), according to the manufacturer's protocol. Labeled H22-MPs were coincubated with DCs at 37°C for 2 to 12 hours before staining with PE-CD11c antibody (eBioscience), according to the manufacturer's protocol, and visualized by two-photon fluorescent microscopy (Confocal Laser Scanning Microscope Leica TCS SP8; Leica). In some cases, carboxyfluorescein diacetate succinimidyl ester (CFSE)-labeled apoptotic H22 tumor cells or H22-MPs were coincubated with DCs at 37°C for 12 hours before staining with PE-CD11c antibody (eBioscience) and then visualized by two-photon fluorescent microscopy.

T-MPs or tumor-cell lysate or DC lysate proteins were extracted and analyzed by Western blot using anti-mouse HSP70, anti-mouse HSP90, anti-mouse HMGB1, anti-mouse p44/42 MAPK (Erk1/2), anti-mouse Phospho-p44/42 MAPK (Erk1/2), anti-mouse p38 MAPK, anti-mouse Phospho-p38 MAPK, anti-mouse Phospho-IKKα/β, anti-mouse IRF-3, anti-mouse Phospho-IRF-3, and anti-mouse β-actin at suppliers' recommended dilutions, followed by secondary horseradish peroxidase–coupled antibodies and visualized by enhanced chemiluminescence according to the manufacturer's protocol (ECL kit; Pierce). All antibodies were purchased from Cell Signaling Technology.

Splenic T cells were purified from H22 tumor–bearing mice by CD3-positive selection, and then fluorescently labeled with 5 μmol/L CFSE (Sigma-Aldrich). DCs were incubated with H22-MPs for 12 hours to obtain H22-MP–loaded DCs. CFSE-labeled T cells were incubated with H22-MPs alone, H22-MP–loaded DCs, or empty DCs for 4 days before flow cytometric analysis. In some cases, anti-IFNAR was added to the culture system to test T-cell proliferation.

Total RNA was extracted from cells with TRIzol reagent (Invitrogen). Real-time PCR was performed using Fast SYBR Green PCR master mix (TOYOBO) on a CFX96 Touch real-time PCR detection system (Bio-rad). mRNA levels were normalized to GAPDH. The primer sequences were as follows: mouse GAPDH, sense 5′-AGGTCGGTGTGAACGGATTTG-3′, antisense 5′-TGTAGACCATGTAGTTGA GGTCA-3′; mouse IFNα, sense 5′-TGATGGTCTTGGTGGTGAT-3′, antisense 5′-TT GTGCCAGGAGTGTCAA-3′; mouse IFNβ, sense 5′-GGTGGAATGAGACTATTG TTG-3′, antisense 5′-CTTCAAGTGGAGAGCAGTT-3′; mouse RIG-I, sense 5′-AAT CAGACAGATCCGAGACAC-3′, antisense 5′-ACTTGCTGTTGTCTTTCTCC-3′; mouse cGAS, sense 5′-ACCGGACAAGCTAAAGAAGGTGCT-3′, antisense 5′-GCA GCAGGCGTTCCACAACTTTAT-3′; mouse stimulator of interferon genes (STING), sense 5′-CACCTCTCTGAGC CTCAACC-3′, antisense 5′-CCATCCACACAGGTCAACAG-3′. DNA was extracted from H22 tumor cells or H22-MPs using a Qiagen DNA isolation kit. Semiquantitative PCR was performed using primers: mouse GAPDH, sense 5′-GTGGAGATTGTTGC CATCAACG-3′, antisense 5′-CAGTGGATGCAGGGATGATGTTCTG-3′; mouse mitochondrial fragment, sense 5′-ACTATCCCCTTCCCCATTTG-3′, antisense 5′-GT TGGTCATGGGCTGATTA-3′.

siRNAs against murine RIG-I 5′-GCAGGUUACUGUGGACUUUdTdT-3′; murine cGAS 5′-GCAGCUACUAUGAACAUGUdTdT-3′; murine STING 5′-GGAG CCGAAGACUGUACAUdTdT-3′; and control siRNA were purchased from RiBoBio and transfected into bone marrow–derived DCs (BMDC) using Lipofectamine RNAIMax (Invitrogen) at a final concentration of 50 nmol/L. Forty-eight hours after transfection, cells were treated with H22-MPs for an additional 12 hours, and mRNA of IFNα or IFNβ was detected by real-time PCR. Secreted IFNα or IFNβ in the supernatants was measured by ELISA 36 hours after H22-MPs treatment. Knockdown of the indicated genes was verified by real-time PCR.

Results were expressed as mean values ± SEM and interpreted by repeated-measure analysis of variance or Kaplan–Meier analysis. Differences were considered to be statistically significant when the P value was <0.05. Other detailed experimental procedures are described in Supplementary Methods.

One advantage of using T-MPs as a vaccination platform is that they harbor a discrete set of tumor-specific antigens expressed in the parental tumor cells that can potentially stimulate broad antitumor immunity. To determine the immunogenicity of T-MPs, we immunized BALB/c mice with H22 hepatocarcinoma cell–derived microparticles (H22-MPs) followed by a challenge with 3 × 10⁵ H22 cells. As a comparison, different groups of mice received tumor-cell lysates or tumor cell–released exosomes, which were collected from the same number of H22 cells. As shown in Fig. 1A, 100% tumor formation was seen in mice immunized with PBS or tumor-cell lysates, whereas 50% of microparticle-immunized mice and 12.5% of exosome-immunized mice remained tumor free, suggesting that T-MPs are more immunogenic than the other tumor cell–derived materials. Consistent with the observed antitumor activity in mice, in vitro cytolytic analysis indicated that splenic T cells from mice immunized with H22-MPs were more potent in killing H22 targets than those from exosome- or cell lysate–immunized mice (Fig. 1B). In T-cell–deficient nude mice, H22-MPs failed to induce the antitumor effect (Fig. 1C). Consistently, in vivo depletion of CD4⁺ or CD8⁺ T cells, but not natural killer cells in wild-type mice, also diminished tumor protection, confirming that microparticle-induced antitumor immunity is T-cell dependent (Fig. 1D). To determine that immunogenicity of microparticles is not limited to one tumor type or a specific mouse strain, we immunized C57BL/6 mice with B16 melanoma cell–derived microparticles. Immunized animals were protected against subsequent challenge with B16 tumor cells via s.c. or i.v. delivery (Supplementary Fig. S1A and S1B). Similarly, BALB/c mice immunized with CT26 colon carcinoma cell–derived microparticles prevented CT26 tumor growth (Supplementary Fig. S1C), confirming that T-MPs are a highly effective and generally useful vaccination platform. Interestingly, mice immunized with H22-MPs did not affect CT26 colon carcinoma growth (Supplementary Fig. S1C), but resulted in protection against Hepa1-6, another murine hepatocarcinoma tumor cell line (Supplementary Fig. S1D), suggesting that common tumor antigens exist in microparticles from the same tumor types, which can be cross-presented through host APCs. This latter observation offers a possibility to simplify clinical application of T-MP–based vaccination therapy.

To investigate the therapeutic potency of T-MPs, BALB/c mice were inoculated s.c. with 3 × 10⁵ H22 tumor cells 5 days before vaccination with H22-MPs. Surprisingly, T-MP vaccine alone failed to affect the growth of preexisting tumor despite its immunogenicity in the prophylactic setting. In addition, in vitro studies showed that T-MPs had no direct effect on the proliferation and apoptosis of tumor cells (Supplementary Fig. S2A), suggesting that direct delivery of T-MPs may not be optimal to fully activate T cells, likely due to insufficient loading of endogenous DCs. This speculation led us to test whether ex vivo loading of autologous DCs could enhance therapeutic efficacy of T-MP vaccines. To this end, 1 × 10⁶ BMDCs pulsed with H22-MPs were injected s.c. into BALB/c mice carrying 5-day-old H22 tumors. As shown in Fig. 2A, vaccination with H22-MP–loaded DCs significantly inhibited H22 tumor growth, compared with either component alone or PBS control, leading to long-term tumor regression in 50% of treated mice. Consistently, intratumoral cellular analysis revealed that 7 days after vaccination, the number of infiltrating CD8⁺ T cells was three times higher in the H22-MP–loaded DC group than in the other groups (Fig. 2B), and they were more capable of producing IFNγ upon restimulation with PMA and ionomysin (Fig. 2C). We extended this investigation to other tumor types and demonstrated that microparticles derived from CT26 tumor cells were capable of inhibiting the growth of preexisting CT26 colon cancer after they were loaded onto DCs (Fig. 2D). Similarly, a therapeutic effect was observed after treatment with B16-MP–loaded DCs in mice that carried subcutaneous (Fig. 2E) or lung metastatic B16 melanoma (Fig. 2F). These results suggest that DCs may function as an adjuvant that improves therapeutic efficacy of T-MP vaccines.

To understand the mechanisms by which DCs enhance the immunogenicity of T-MPs, we investigated interactions between microparticles and DCs and their impact on DC maturation and antigen presentation. We first examined the uptake of microparticles by DCs in vitro. PKH67-labeled T-MPs were incubated with DCs for 2, 6, or 12 hours, and the percentage of DCs that were PKH67-positive was analyzed by flow cytometry. As shown in Fig. 3A, the uptake of T-MPs by DCs was evident as early as 2 hours (24.3%) and peaked at 12 hours (62.5%). The kinetics of microparticle uptake and their cellular localization were confirmed by confocal fluorescent microscopy (Fig. 3A). However, pulsing DCs with apoptotic H22 tumor cells (Apo-H22 cells) resulted in very little tumor materials in DCs within 12 hours, confirming that T-MPs were more efficiently taken up by DCs (Supplementary Fig. S2B and S2C). An in vitro T-cell proliferation assay indicated that the presence of DCs was required because T-MPs alone were not sufficient to stimulate T-cell proliferation (Fig. 3B), suggesting that the uptake of T-MPs is critical for optimal antigen presentation. We next turned our attention to the impact of T-MPs on the phenotypic and functional properties of DCs. Bone marrow–derived immature DCs were exposed to H22-MPs for 36 hours and characterized by flow cytometric analysis for expression of surface markers that are required for antigen presentation and DC migration. Upregulation of CD80, CD86, MHC class II, and CCR7 demonstrates that the uptake of T-MPs triggered DC maturation (Fig. 3C). Furthermore, consistent with upregulation of CCR7, a chemokine receptor important for DC homing to lymph nodes, we found that more DCs that were exposed to T-MPs accumulated in the draining lymph nodes 24 to 48 hours after s.c. injection, compared with control DCs (Fig. 3D). Locally, we noticed that popliteal draining lymph nodes were much larger 7 days after footpad injection with H22-MPs (Supplementary Fig. S2D), compared with PBS control, and cellular analysis indicated that both T lymphocytes and DCs were significantly increased (Supplementary Fig. S2E). Consistent with in vitro data (Fig. 3C), DCs recovered from T-MPs–inoculated lymph nodes expressed higher CD80, CD86, and MHC class II than those in the PBS group (Fig. 3E), and exhibited higher IL12 and IFNγ production (Fig. 3F). Local concentrations of IL12 and IFNγ were induced by T-MPs in the draining lymph nodes (Supplementary Fig. S2F). These results together suggest that T-MPs may stimulate DC maturation and migration, leading to subsequent T-lymphocyte recruitment or proliferation.

DC maturation is differentially regulated by MAPK, NF-κB, and IRF-3 pathways (16, 17). These pathways can be activated by certain endogenous proteins and nucleotide molecules released from tumor cells (18). To determine whether DC maturation by T-MPs that contain tumor cell–derived proteins and nucleotides activates these pathways, we first analyzed the phosphorylation of Erk, p38, and IKK in DCs pulsed with T-MPs. Data in Fig. 4A indicate that no further activation of NF-κB and MAP kinases was seen in DCs treated with T-MPs. Interestingly, however, phosphorylation of IRF-3, a key transcription factor responsible for type I IFN production, was evident 1 hour after exposure to T-MPs or 30 minutes after exposure to poly (I:C; Fig. 4B and Supplementary Fig. S3A), whereas DCs incubated with H22-exosomes showed no phosphorylation of IRF-3 and no induction of type I IFN (Supplementary Fig. S3A and S3B). Consistent with this finding, the increase of IFNα and IFNβ expression by DCs was detected in mRNA and protein levels after incubation with T-MPs (Fig. 4C). It has been shown that IFNα/β signaling is essential for DC maturation and immune rejection of tumors (19). To determine whether type I IFN was responsible for DC maturation induced by T-MPs, we used IFNAR-1 antibody to block IFN signaling during incubation with T-MPs. Our results indicate that upregulation of CD80, CD86, and MHC II during incubation with T-MPs was attenuated by IFNAR-1 blockade (Fig. 4D), leading to reduced capacity of microparticle-loaded DCs to stimulate T-cell proliferation (Fig. 4E). These data suggest that one of the mechanisms by which T-MPs trigger DC maturation is through activation of the type I IFN pathway.

Certain proteins and nuclear materials released from tumor cells, such as Hsp70, Hsp90, and HMGB1, can serve as endogenous danger signals that induce DC activation through TLR4, leading to activation of IRF-3 (20). However, although these proteins were indeed detectable by Western blot in tumor-cell lysates, T-MPs did not appear to contain many of these molecules (Fig. 5A). Furthermore, DCs deficient for MyD88, a downstream signal molecule of TLRs, underwent maturation during incubation with T-MPs (Fig. 5B). In addition to protein components, T-MPs also contain RNAs (21, 22) that are capable of stimulating DCs through the IFN pathway (23). To determine this possibility, we knocked down RIG-I, a key sensor for RNA-mediated type I IFN production, using specific siRNA during the incubation of DCs with T-MPs. Our results show that knockdown of RIG-I did not influence DC production of type I IFN by T-MPs stimulation (Fig. 5C and Supplementary Fig. S3C). Together, our data suggest that neither proteins nor RNAs in T-MPs are responsible for IFN-mediated DC maturation.

Another possible component that can engage IFN pathways is DNAs carried by T-MPs. Indeed, using PCR amplification we confirmed that abundant nuclear gene GAPDH fragments and mitochondrial DNA fragments were included in T-MPs (Supplementary Fig. S3D). Recent studies have highlighted Cyclic GMP-AMP synthase (cGAS) as a cytosolic DNA sensor that induces type I IFN production by catalyzing and synthesizing the second messenger Cyclic GMP-AMP (cGAMP) to activate STING (24). The minimum cGAS expression was seen in parental tumor cells and the corresponding T-MPs, but approximately 500 times more cGAS was detected in DCs (Supplementary Fig. S3E). To determine whether DNAs in T-MPs engaged the IFN signaling pathway, we used siRNA to knock down cGAS in DCs during incubation with T-MPs. siRNA oligo strongly inhibited type I IFN production. Alternatively, knocking down STING in DCs resulted in similar outcomes (Fig. 5D and Supplementary Fig. S3F), suggesting that DNAs in T-MPs are the primary component in stimulating DC maturation via the cGAS-STING pathway.

Microparticles are enriched in a selective repertoire of proteins and nucleic acids from parental cells and are thought to play an active role in conferring intercellular signals (12). We have recently demonstrated that microparticles can mediate antigen transfer from macrophages to DCs (14), which prompted us to examine whether T-MPs could function as a cell-free vaccine for cancer immunotherapy. In the present study, we show that T-MPs are not only capable of providing tumor antigens to DCs but also conditioning DCs toward an immunogenic phenotype via activation of type I IFN production, leading to potent antitumor immunity in both prophylactic and therapeutic settings. Here, we provide evidence that the cGAS-STING pathway is required for IFN-dependent DC maturation, suggesting that endogenous DNA in T-MPs may serve as an activator of innate immunity. To our knowledge, this represents the first report to reveal a novel property of T-MPs as a vaccination platform that can deliver both antigens and adjuvants to DCs.

Recent success of whole-exome sequencing provides a powerful means to identify mutated tumor antigens that can be recognized by cytotoxic T cells (25). However, how to develop therapeutic strategies to target multiple mutated antigens remains a challenge. Theoretically, whole tumor cells or tumor lysates harbor the tumor antigen repertoire that can generate epitopes for both CD8⁺ and CD4⁺ T cells, leading to enhanced antitumor immunity and greatly diminishing the chance of tumor escape. Furthermore, the immunogenicity of whole tumor-cell vaccines can be increased by modulating tumor cells with an inflammatory cytokine, such as GM-CSF or a replicating virus, which facilitates DC maturation and antigen presentation (26). However, it is still unclear in what form tumor antigens are transferred to DCs following exposure to tumor cells or their extracts. The difference between exosomes and microparticles in this regard is likely due to their unique biogenesis and their physical size, which determine their efficiency of interaction with and influence on DCs. In this study, when the efficiency of DC uptake of apoptotic tumor cells and microparticles was compared, microparticles were more effectively taken up by DCs. Although apoptotic cells are enriched in eat-me signal phosphatidylserine (27), they are bigger than microparticles and likely less efficient to be taken up and processed by DCs on a particle basis. Furthermore, in addition to being internalized and processed, microparticles can directly transfer MHC class I–peptide complexes onto DC surfaces, as we reported in a Listeria infection model (14), further enhancing the efficiency and potency of antigen presentation by microparticles.

An interesting and important finding of the present study is that T-MPs can activate DCs to upregulate costimulatory molecules and produce inflammatory cytokines simultaneously, suggesting that T-MPs contain endogenous activators of innate immunity. This intrinsic immunogenic property makes T-MPs an ideal vaccination platform. It was reported that tumor cell–derived exosomes could induce specific antitumor responses only when the parental tumor cells were genetically modified or subjected to stress conditions. For example, heat-shocked tumor cells release exosomes with increased levels of MHC or Hsp70, which elicit Th1/CTL antitumor immunity more efficiently than exosomes from untreated tumor cells (28–30). T-MPs in our study were generated from UV-treated tumor cells and appeared not to contain heat-shock proteins but instead harbored abundant genomic and mitochondrial DNA fragments. Indeed, further analysis indicates that microparticle-derived DNAs, but not proteins or RNAs, induced endogenous type I IFNs that stimulated DC maturation, confirming that exosomes and microparticles have different compositions of parental cell–derived molecular components that determine their biologic activities.

A role for microparticles in transferring cellular components has been well characterized (12), and our data suggest that microparticles may also deliver DNAs to amplify innate immune responses in certain circumstances. In eukaryotic cells, double-stranded DNA is normally sequestered in the nucleus or the mitochondria, preventing its direct contact with the cytosol. As such, cytosolic exposure of DNA resulting from microbial infections or damaged host cells turns on endogenous “danger” signals to trigger innate immune responses (31). It is now known that cytosolic DNA is sensed through the cGAS/STING pathway that leads to the activation of NF-κB and IRF3, respectively (24). Here, we provide evidence that knockdown of either cGAS or STING expression by siRNA resulted in the significant downregulation of type I IFN, whereas RIG-I knockdown had no effect on type I IFN production. The reason that RNA is not inducing type I IFN might be because the cell-derived RNAs in T-MPs lack the triphosphate structure at the 5 end of RNA for RNA sensor recognition (32). Type I IFN is capable of enhancing antigen cross-presentation and production of Th1 cytokines by DCs, particularly IL12 (33, 34). Depending on the amount and duration, DC-derived type I IFN may also act directly on CD8⁺ T cells for clonal expansion and memory formation (35). Our finding not only demonstrates an understanding of the innate immune signaling triggered by T-MPs but also provides a potential novel strategy for manipulating T-MP–based vaccines.

In summary, we demonstrate that T-MPs contain both tumor antigens and innate DNA signals that engage the cGAS/STING pathway for induction of type I IFN production. T-MPs can be readily taken up by DCs in vitro and in vivo, leading to multiple antigen presentation and DC maturation. Thus, T-MPs may represent a novel and effective vaccination platform to prime both CD4⁺ Th1 cells and CD8⁺ CTLs, limiting the risk of tumor escape through induction of antigen-loss variants. Moreover, we provide evidence that microparticles from allogeneic tumor cell lines appear to have shared tumor antigens that can be cross-presented by host DCs, offering a possibility to simplify clinical application of T-MP–based vaccination therapy.

No potential conflicts of interest were disclosed.

Conception and design: J. Ma, Y. Wan, B. Huang

Development of methodology: H. Zhang, J. Ma, S. Luo, Y. Wan, B. Huang

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): H. Zhang, K. Tang, Y. Zhang, R. Ma, Y. Li, S. Luo, X. Liang, T. Ji, Z. Gu

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): H. Zhang, S. Luo, W. He, Y. Wan, B. Huang

Writing, review, and/or revision of the manuscript: H. Zhang, Y. Wan, B. Huang

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): S. Luo, J. Lu, X. Cao

Study supervision: B. Huang

The authors thank Dr. Ning Wang (Huazhong University of Science and Technology) for technical support.

This work was supported by National Basic Research Program of China (2014CB542100 and 2012CB932500), National Science Fund for Distinguished Young Scholars of China (81225021), National Natural Science Foundation of China (81472653), and Special Fund of Health Public Welfare Profession of China (201302018). B. Huang was supported by Soundny (Sheng-Qi-An) Biotech.

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.