Purpose: We previously reported that autophagy in tumor cells plays a critical role in cross-presentation of tumor antigens and that autophagosomes are efficient antigen carriers for cross-priming of tumor-reactive CD8+ T cells. Here, we sought to characterize further the autophagosome-enriched vaccine named DRibble (DRiPs-containing blebs), which is derived from tumor cells after inhibition of protein degradation, and to provide insights into the mechanisms responsible for their efficacy as a novel cancer immunotherapy.
Experimental Design: DRibbles were characterized by Western blot and light or transmission electron microscopy. The efficiency of cross-presentation mediated by DRibbles was first compared with that of whole-tumor cells and pure proteins. The mechanisms of antigen cross-presentation by DRibbles were analyzed, and the antitumor efficacy of the DRibble vaccine was tested in 3LL Lewis lung tumors and B16F10 melanoma.
Results: The DRibbles sequester both long-lived and short-lived proteins, including defective ribosomal products (DRiP), and damage-associated molecular pattern molecules exemplified by HSP90, HSP94, calreticulin, and HMGB1. DRibbles express ligands for CLEC9A, a newly described C-type lectin receptor expressed by a subset of conventional and plasmacytoid dendritic cells (DC), and cross-presentation was partially CLEC9A dependent. Furthermore, this autophagy-assisted antigen cross-presentation pathway involved both caveolae- and clathrin-mediated endocytosis and endoplasmic reticulum–associated degradation machinery. It depends on proteasome and TAP1, but not lysosome functions of antigen-presenting cells. Importantly, DCs loaded with autophagosome-enriched DRibbles can eradicate 3LL Lewis lung tumors and significantly delay the growth of B16F10 melanoma.
Conclusions: These data documented the unique characteristics and potent antitumor efficacy of the autophagosome-based DRibble vaccine. The efficacy of DRibble cancer vaccine will be further tested in clinical trials. Clin Cancer Res; 17(22); 7047–57. ©2011 AACR.
A perfectly designed cancer vaccine should target antigens that are abundantly present on the surface of tumor cells and have not been exposed to the immune system. These targets would be packaged with the appropriate chaperones, include Toll-like receptor signals, and be directed to the most efficient antigen-presenting cells, such as dendritic cells, that can migrate into or reside in lymph nodes, so that vaccines would stimulate potent therapeutic activity. This report identifies these properties in “DRibble” autophagosome-enriched vaccines. These findings provide insights into why this vaccine strategy, already being investigated in a clinical trial, efficiently cross-presents tumor/tumor-associated antigens. This is an important step, but effective immunotherapy will likely require a multimodality combination therapy approach.
Autophagy is a fundamental cellular process in which damaged cytosolic proteins and superfluous organelles are sequestered in autophagosomes and delivered to lysosomes for their clearance. Autophagy is seen as a double-edged sword in the development of tumor-targeted therapies; it can be either a tumor suppression or survival mechanism in response to metastatic stress and antitumor therapy (1, 2). Emerging evidence suggests an important role for autophagy in both innate and adaptive immunity (3–5). It has been reported that autophagy enhances MHC I presentation of HSV-1 viral antigens via formation of autophagosomes in antigen-presenting cells (APC; ref. 6), and constitutively delivers antigens for MHC II presentation through autophagosomes (7). We recently showed that autophagy in tumor cells played a critical role in cross-presentation of tumor antigens and identified autophagosomes as the novel, efficient carriers for cross-presentation of tumor-associated antigens (TAA; ref. 8).
On the basis of these findings and the well-established role of cross-presentation in priming antitumor T-cell responses, we developed a novel tumor vaccine comprising enriched autophagosomes that favors cross-presentation of multiple TAAs. TAAs are degraded by 2 major proteolysis pathways in the tumor cells. It is generally believed that the long-lived proteins are degraded by the lysosomes through the autophagy pathway (9), whereas short-lived proteins (SLiP), including defective ribosomal products (DRiP), are ubiquitinated and degraded by proteasomes (10, 11). With induction of autophagy and inhibition of lysosomal/proteosomal activity, a broad spectrum of cellular antigens, including long-lived proteins, SLiPs, and DRiPs, is sequestered in autophagosomes (12). We refer to these autophagosome-enriched, DRiPs-containing blebs as DRibbles.
Efficient cross-presentation of TAA is pivotal for the success of cancer vaccines. The process of cross-presentation involves antigen internalization, processing, and presentation of peptides on MHC I molecules by APCs, such as dendritic cells (DC; ref. 13). What determines the efficiency of antigen cross-presentation is not well understood. Two mechanisms are known so far. First, specific cell surface receptors on DCs mediate distinct antigen endocytosis and intracellular routing for MHC I or MHC II presentation (14, 15). Mannose receptor (CD206)–mediated uptake of soluble ovalbumin (OVA) favors MHC I cross-presentation because it specifically targets OVA to early endosomes, but not lysosomes, whereas DC-SIGN (CD209A) routes antigens into lysosomes for MHC II-restricted presentation (15, 16). Recently, CLEC9A (DNGR-1), a newly identified C-type lectin receptor on conventional DCs, was found to be required for efficient cross-presentation of dead cell-associated materials (17). Like the mannose receptor, CLEC9A may also divert cargo away from lysosomes, because it localizes only in nonlysosomal compartments. Second, the size of the exogenous materials may dictate the route of endocytosis and direct cargo either to nonacidic compartments or to lysosomes (18, 19). Extracellular particulate materials were documented to enter cells either by clathrin-dependent endocytosis or caveolae-mediated endocytosis based on their size. Clathrin-mediated entry of particulate materials with a size of less than 200 nm was targeted for lysosomal degradation, whereas caveolae-mediated internalization of particles larger than 200 nm retained cargo in nonacidic compartments (19) and thus may lead to efficient cross-presentation (16).
In this study, we showed that tumor cell-derived DRibbles sequestered a broad range of antigens and were highly efficient at cross-priming of antigen-specific CD8+ and CD4+ T cells in vitro and in vivo. Interestingly, we found CELC9A was required for efficient cross-presentation of DRibble antigens. We also delineated mechanisms by which DRibbles were internalized and intracellular proteolysis pathways and machinery involved in cross-presentation of antigens delivered by DRibbles. Furthermore, we isolated DRibbles from a variety of tumor cell lines by augmenting autophagy and inhibiting protein degradation and tested their therapeutic efficacy in mice bearing established B16F10 melanomas and 3LL Lewis lung carcinomas.
Materials and Methods
Mice, cell lines, and cell culture
C57BL/6 mice were purchased from the Charles River Laboratories. OT-I T-cell receptor (TCR) transgenic mice (recognize the H-2Kb-restricted OVA 257–264 peptide) were obtained from The Jackson Laboratory. Pmel-1 transgenic mice (recognize mouse and human H-2Db restricted gp10025–33 peptide) were kindly provided by Dr. Nicholas P. Restifo (National Cancer Institute) and bred in our facility. All mice were used in accordance with the protocol approved by the Animal Care and Use Committee of the Earle A. Chiles Research Institute.
HEK 293T cells were maintained in Dulbecco's modified Eagle's medium supplemented with 2 mmol/L L-glutamine, 1 mmol/L sodium pyruvate, and 10% FBS. Murine B16-F10 melanoma was obtained from American Type Culture Collection (ATCC) and clone 3 was derived from an in vivo passage. The F10 cell line was characterized for expression of the melanoma-associated gp100 and envelope protein of endogeneous retrovirus melREV as determined by flow cytometry analysis. The F10 cell line was also shown to stimulate the proliferation of pmel-1 naïve T cells in vivo. The Lewis lung 3LL tumor cells were obtained from ATCC. These cells have not undergone further testing. These cell lines were cultured in RPMI1640 with 2 mmol/L L-glutamine, 1 mmol/L sodium pyruvate, and 10% FBS.
DNA construction and generation of cell lines
HEK 293T cells that expressed the M-GFP-OVA or gp100 model antigens, GFP-ubiquitin (Ub), or tdTamato-LC3, were generated as described before (8). To characterize the antigens sequestered in DRibbles, we used the following different forms of OVA: Ub-M-GFP-OVA (long-lived cytosolic protein), Ub-R-GFP-OVA (short-lived cytosolic protein), Ub-M-GFP-TfR-OVA (long-lived, membrane-bound protein), and Ub-R-GFP-TfR-OVA (short-lived, membrane-bound protein). These formulations consisted of M (methionine), R (arginine), V (valine), and TfR (transmembrane domain of the transferrin receptor).
Carboxy-fluorescein diacetate succinimidyl ester-labeling and in vitro and in vivo CFSE dilution assay
Cross-presentation of antigens to naïve OT-I or pmel-1 T cells was measured by carboxy-fluorescein diacetate succinimidyl ester (CFSE) dilution of labeled T cells by flow cytometry analysis. Flt3L DCs were isolated from spleens of C57BL6 mice at day 15 after sequential i.v. injection of plasmid DNA encoding murine Flt3 ligand (2 μg DNA in 2 mL PBS, day 1) and granulocyte macrophage colony-stimulating factor [(GM-CSF) day 10; ref. 8]. For the in vivo CFSE assay, DRibbles were injected s.c. into both flanks either directly or after loading onto DCs in vitro for 6 hours, or injected directly into both inguinal lymph nodes of C57BL6 mice. On the same day, 3 × 106 CFSE-labeled Thy1.1+ OT-I T cells were adoptively transferred into these mice. Lymph nodes were collected 4 or 5 days later, and single cell suspensions were prepared and analyzed by flow cytometry after staining with antibodies against thy1.1 and CD8.
Preparation of DRibbles
Autophagosome-enriched DRibbles were prepared as described previously (8). Briefly, tumor cells were treated with Bortezomib (velcade, 100 nmol/L) and NH4Cl (10 mmol/L) for 24 to 48 hours, cells were disrupted by mild sonication at 115 V, 56-60HZ using the G112SP1G Special Ultrasonic Cleaner (Laboratory Supplies Co., Inc.). The resulting suspension was precleared by centrifugation at 300×g for 10 minutes, and was then separated into the crude autophagosome-containing large vesicles (DRibbles) and the supernatant consisting of cytosolic components by a 15-minute centrifugation at 10,000×g. DRibbles were stored in PBS at 4°C for short term (less than 1 month) or −80°C for long term.
Western blot was conducted as previously described (8). The primary antibodies included mouse anti-GFP (1:1,000; StressGen), rabbit anti-ubiquitin (1:1,000; Upstate), rabbit anti-LC3 (1:1,000; Novus), rabbit anti-HSP90α (1:1,000; Chemicon), rabbit anti-HSP94, rabbit anti-calreticulin (1:500, Upstate), and rabbit anti-HMGB 1 (1:1,000; Abcam). The secondary antibodies were goat anti-rabbit horseradish peroxidase (HRP; 1:10,000, Jackson ImmunoResearch), and goat anti-mouse HRP (1:10,000, Jackson ImmunoResearch).
Fluorescent light microscopy
Images of live cells were taken using a Zeiss inverted microscope capable of digital epifluorescence imaging. A GFP filter (excited at 470/40, dichromatic mirror at 495, and long-pass emission filter 500LP) and an Orange filter (excited at 525/50, dichromatic mirror at 555, and band pass emission filter 590/50) were used to capture fluorescent images. Cell images were processed with Photoshop. Pseudo-green and red colors were applied to GFP- and tomato-positive cells respectively, and resulting images were overlaid to show colocalization of GFP and tomato colors.
Transmission electron microscopy
DRibbles containing autophagosomes were prepared from the culture media of murine Lewis lung 3LL tumor cells treated with 100 nmol/L bortezomib and 10 mmol/L ammonium chloride for 48 hours. DRibble samples were fixed in 100 mmol/L sodium cacodylate (pH 7.2), 2.5% glutaraldehyde, 1.6% paraformaldehyde, 0.064% picric acid, 0.1% ruthenium red, gently washed, and postfixed for 1 hour in 1% osmium tetroxide plus 0.8% potassium ferricyanide in 100 mmol/L sodium cacodylate, pH 7.2. After thorough rinsing in water, samples were dehydrated, infiltrated overnight in 1:1 acetone:Epon 812, infiltrated for 1 hour with 100% Epon 812 resin, and embedded in the resin. After polymerization, 60- to 80-nm thin sections were cut on a Reichert ultramicrotome, stained 5 minutes in lead citrate, rinsed, poststained for 30 minutes in uranyl acetate, rinsed, and dried. Electron microscopy was done at 60 kV on a Philips Morgagni transmission electron microscope, equipped with a CCD camera, and images were collected at original magnifications of ×1,000 to ×37,000.
Detection of CLEC9A ligand and CLEC9A expression
To detect the expression of the CLEC9A ligand, a recombinant soluble FcmILZ-CTLD9A receptor that contains a c-type lectin domain fused to C-terminal of human IgG and trimerization isoleucine zipper (ILZ) was made. The human Fc fragment of the soluble fusion receptor was mutated at the Fc receptor-binding site to avoid soluble CLEC9A-binding to DC membrane through the Fc and Fc receptor interaction. DRibbles harvested from HEK 293T cells were incubated with the soluble fusion receptor FcmILZ-CTLD9A for 1 hour. The soluble FcmILZ OX40 ligand fusion protein (constructed in our laboratory) was used as negative control. DRibbles were washed extensively to remove excess free soluble receptor. PE-conjugated anti-human Fc Fab fragment was used to detect ligand expression on DRibbles by flow cytometry. A sensitive photo multiplier tube was added on the Forward Side Scatter parameter of a FACS Aria II cell sorter, which allows detection of small-size particles to 200 nm or greater. To detect CLEC9A on DCs, a single cell suspension of splenic DCs was stained with APC-conjugated anti-CLEC9A antibody (7H11 monoclonal antibody specific for mCLEC9A, kindly provided by Dr. Caetano Reis e Sousa at London Research Institute) and PE-conjugated anti-CD11c antibody. Flow cytometry was done to detect CD11c and CLEC9A. All events were gated on CD11c+ population.
Investigation of the pathways involved in cross-presentation of DRibble antigens
To examine which endocytic pathways are involved in DRibble cross-presentation, DCs were pretreated with chlorpromazine (10 μg/mL, Sigma) to inhibit clathrin-mediated endocytosis, or filipin III (5 μg/mL, Sigma) to block caveolae-mediated endocytosis, or both, for 30 minutes before loading with DRibbles isolated from OVA-expressing cells (OVA-DRibble). DCs loaded with SIINFEKL peptide (250 ng/mL) were used as a control. Antigen-pulsed DCs were then incubated with CFSE-labeled OT-I T cells for 4 days. Division of OT-I T cells was monitored by flow cytometry analysis. To assess the requirement of the endoplasmic reticulum–associated degradation (ERAD) pathway, proteasomes, and lysosomal activity, DCs were pretreated with exotoxin A (10 μg/mL, Sigma), bortezomib (100 nmol/L), or NH4Cl (10 mmol/L, Sigma), respectively, for 30 minutes before OVA-DRibble or SIINFEKL peptide was loaded. To test the role of TAP1 in peptide loading, DCs from TAP1 knockout mice were used as APCs in parallel with wild-type DCs.
For treatment of 3LL primary tumor, C57BL/6 mice were s.c. injected with 2 × 105 3LL tumor cells. Subsequently, DRibbles-pulsed DCs with or without adjuvant were injected s.c. into both flanks of C57BL/6 mice on days 6, 8, and 10. Tumor growth was measured 3 times a week. Five or 6 mice were included in each group. In the 3LL lung metastasis model, 2 × 105 3LL tumor cells were injected i.v. into C57BL/6 mice. Mice were then treated with s.c. DCs loaded with 3LL-derived DRibbles, dead cells, and tumor cell lysate in the presence of adjuvant at day 3. All mice were sacrificed, and the metastases in the lungs were counted on day 28.
Log-rank nonparametric analysis was used to analyze the tumor-free survival data. Each group consisted of at least 5 mice, and no animal was excluded from the statistical evaluation. Student t test was used to analyze the number of T cells and tumor metastases. A 2-tailed P < 0.05 was considered significant.
Generation and characterization of DRibbles from HEK 293T cells
We previously reported that tumor cell autophagy was required for efficient cross-presentation of TAA (8). We also showed that treatment of tumor cells with bortezomib, a proteasome inhibitor, induced autophagy in tumor cells and enhanced cross-presentation of TAA. Moreover, treatment of tumor cells with NH4Cl, which blocks lysosomal degradation, also augmented cross-presentation (8). Norbury and colleagues recently reported that cross-presentation favored long-lived proteins, whereas SLiPs including DRiPs were not cross-presented because they were quickly degraded by proteasomes (20). To show whether SLiPs and DRiPs were shunted into autophagosomes and cross-presented, we generated stable HEK 293T cells that expressed the short-lived Ub-R-GFP-OVA or long-lived Ub-M-GFP-OVA fusion proteins (9). Consistent with the observations of Norbury and colleagues, proteosome inhibition resulted in accumulation of Ub-R-GFP-OVA protein and increased cross-presentation to the level induced by long-lived M-GFP-OVA (Supplementary Fig. S1). When HEK 293T cells were treated with bortezomib and NH4Cl concurrently, a large number of tomato-LC3+ blebs, indicating autophagosomes, were accumulated in the cells. Interestingly, we also observed that many tomato-LC3+ blebs were also released into the culture media after prolonged treatment (Fig. 1). To determine whether protein antigens or ubiquitinated proteins were sequestered into autophagosomes, we transfected HEK 293T cells that expressed tomato-LC3 with plasmid DNA encoding M-GFP-OVA, R-GFP-OVA, or GFP-Ub fusion proteins. Both the long-lived Ub-M-GFP-OVA (Fig. 1A) and the short-lived UB-R-GFP-OVA (Fig. 1B) were sequestered in punctuates and colocalized with tomato-LC3 after treatment with bortezomib and NH4Cl. Interestingly, GFP-Ub also colocalized with tomato-LC3 in the punctuates, suggesting that ubiquitinated proteins, the substrates of proteosome degradation, were also packaged in the autophagosomes (Fig. 1C).
We isolated DRibbles from HEK 293T cells that expressed the different forms of OVA fusion proteins by differential centrifugation as described before (8). Western blot analysis of these DRibbles showed the presence not only of long-lived proteins, but also short-lived proteins, short-protein fragments, possibly DRiPs, and large amounts of ubiquitinated proteins (Supplementary Fig. S2A). The model antigens, including M (methionine)-GFP-OVA (long-lived, cytosolic), R (arginine)-GFP-OVA (short-lived, cytosolic), M-GFP-TfR-OVA (long-lived, membrane-bound), and R-GFP-TfR-OVA (short-lived, membrane-bound), were all found in DRibbles.
As expected, the typical autophagosome marker, LC3-II, was present in DRibbles. Compared with vesicles isolated from untreated cells (Natural DRibbles), DRibbles from bortizomib-treated cells were enriched in antigens. These data suggest that DRibbles packaged a broad range of cellular antigens capable of priming a broad repertoire of T cells that would be specific for both long- and short-lived TAAs. Furthermore, HSPs, such as HSP90 and HSP94, were also present in the DRibbles (Supplementary Fig. S2B). HSP has been shown to chaperone protein fragments and to augment antigen cross-presentation by binding to HSP receptors on DCs (21, 22). Interestingly, calreticulin, a calcium-binding chaperone located in the endoplasmic reticulum (ER), and the high mobility group box 1 protein (HMGB1), were also found in DRibbles. The presence of damage-associated molecular pattern (DAMP) signals, such as HSP, HMGB1, and calreticulin, in the DRibbles suggests that DRibbles may be able to activate the innate pattern-recognition receptors and bridge innate and adaptive immune responses (23).
DRibbles from HEK293T cells were efficient in activating naive CD8+ T cells when loaded onto DCs
To determine whether DRibbles could efficiently stimulate naive CD8+ T cells, we loaded DRibbles from HEK 293T cells that expressed the M-GFP-OVA or gp100 onto induced splenic DCs generated by sequential i.v. injection of Flt3-ligand and GM-CSF encoding plasmid DNA (8). Subsequently, we examined the activation and proliferation of naïve OT-I T cells or pmel-1 T cells by flow cytometry analysis. DCs loaded with DRibbles containing the M-GFP-OVA were highly efficient at stimulating OT-I T-cell proliferation (63% T cells divided; Fig. 2A). In contrast, DCs pulsed with an equivalent number of irradiated whole cells barely stimulated OT-I T cells. Similarly, DCs pulsed with gp100-DRibbles showed robust capacity at activating naïve pmel-1 T cells, whereas DCs loaded with irradiated whole cells failed to do so. Surprisingly, when loaded onto DCs, DRibbles at a concentration of 1 μg/mL total protein could drive 98% to 99% OT-I T cells to proliferate, whereas a much higher concentration of the pure OVA protein (10 μg/mL) was needed to drive a similar level of T-cell proliferation (Fig. 2B). These results showed that, when loaded onto DCs, DRibbles, as a source of tumor antigens, were superior in activating antigen-specific CD8+ T cells as compared with irradiated whole cells or even pure proteins.
Cross-presentation of DRibble antigens involved CD8+ DC-specific C-type lectin receptor, CLEC9A
Recently, a novel C-type lectin receptor, CLEC9A, was shown to be critical for efficient cross-presentation of dead cell materials by binding to a preformed ligand that is exposed only when cells undergo necrosis (17). Because DRibbles were released from bortezomib- and NH4Cl-treated dead or dying tumor cells, we hypothesized that DRibbles could express CLEC9A ligand and that CLEC9A might be a receptor for efficient cross-presentation of DRibble antigens. To test this hypothesis, we prepared a soluble receptor fusion protein by fusing the mutated human IgG1 Fc domain with isolucine zipper and the C-type lectin domain of CLEC9A (FcmILZ-CTLD9A). The size of DRibbles was evaluated by using microbeads with standard sizes on a modified flow cytometer (Supplementary Fig. S3). We found abundant expression of the CLEC9A ligand on our DRibble preparation (Fig. 3A). Meanwhile, CLEC9A receptor is expressed by the majority of the Flt3L-induced splenic DCs that were used in the cross-presentation assay (Fig. 3B). Sancho and colleagues (17) and Caminschi and colleagues (24) have reported CLEC9A expression on the CD24high subset DCs that were generated by culturing bone marrow cells in media supplemented with recombinant Flt3L. CD24+ DCs were the precursors of CD8α+ DCs. Consistent with these findings, our unpublished data showed that the majority of our Flt3L-induced splenic DCs expressed CD24, and these CD24+ DCs were CLEC9A positive. Most importantly, we also found that preincubation of DRibbles with the soluble FcmILZ-CTLD9A but not the FcmILZ-OX40L (control fusion protein) significantly inhibited cross-presentation (Fig. 3C and D). In contrast, the same FcmILZ-CTLD9A fusion protein did not affect cross-presentation of the soluble OVA protein, which is known to be endocytosed through the mannose receptor (25).
Cross-presentation of OVA antigen in DRibbles by DCs involved both caveolae and clathrin-mediated endocytosis, ERAD, and proteasomes/TAP1 pathways
The diameter of DRibbles, as measured by electron microscopy, varied from 300 nm to 900 nm. The sizes are in the range of typical autophagosmes of mammalian cells (26). Thus, it is most likely that the phagocytosis pathway is involved in the cross-presentation of DRibbles. First, we examined the contribution of clathrin- and caveolae-dependent pathways in the internalization of DRibbles with selective inhibitors of these 2 pathways. We pretreated DCs with either chlorpromazine, which inhibits clathrin-mediated endocytosis by perturbing clathrin processing, or filipin III, which complexes with membrane cholesterol and blocks caveolae-mediated endocytosis before the DCs were pulsed with OVA-DRibbles or the SIINFEKL peptide (18). Cross-presentation of OVA-DRibbles was significantly reduced when DCs were pretreated with either chlorpromazine or filipin III (Fig. 4A). In contrast, these inhibitors did not affect the ability of peptide-pulsed DCs to stimulate OT-I T cells. The data suggested involvement of both pathways in DRibble cross-presentation. Moreover, inhibition by filipin III was more effective than chlorpromazine, suggesting that cross-presentation of DRibbles by DCs predominantly uses the caveolae-mediated endocytosis pathway. Because caveolae-mediated endocytosis routes antigens away from lysosomes, its involvement could contribute to the high efficiency of antigen cross-presentation mediated by DRibbles.
The intracellular routes of cross-presentation after antigen uptake include the proteosome/TAP-independent peptide loading on MHC I in ER-phagosomes (16, 27, 28), and the most common proteosome/TAP-dependent peptide loading on MHC I in ER, endosomes, or phagosomes (20, 29, 30). Next, we explored whether cross-presentation of DRibble antigens requires proteosome, TAP1, or lysosomal activity in DCs. As shown in Fig. 4B, the OVA antigen in DRibbles was cross-presented via a TAP-dependent intracellular pathway. TAP1-deficient DCs loaded with OVA-DRibble barely activated OT-I T cells. Inhibition of proteosome activity in DCs dramatically diminished cross-presentation of the OVA-DRibbles. The data showed that like most exogenous antigens, proteosome degradation, and TAP1-dependent peptide loading onto MHC I in DCs were critical for cross-presentation of antigens in DRibbles. These results indicate that antigens inside DRibbles either managed to exit the double membrane of autophagosomes or the autophagosomes involved in cross-presentation were not yet fully closed. Although the mechanisms by which the exogenous antigens exit the endosomes and enter the cytosol are still controversial, some studies have shown that ERAD machinery, present in the endosomes/phagosomes, is involved in transporting proteins to the cytosol (31). To assess whether the ERAD pathway is also involved in the cross-presentation of antigens in DRibble, we pretreated DCs with exotoxin A (ExoA) before pulsing them with DRibbles. ExoA is a toxin derived from Pseudomonas aerugenosa. Other investigators have used ExoA to successfully block translocation of ERAD substrates into the cytosol through the Sec61 channel and showed that ERAD was critical for cross-presentation of soluble OVA protein or OVA-IgG immune complexes (32, 33). Using this approach, we found that pretreatment of DC with ExoA partially inhibited T-cell activation by DRibble-loaded DCs (Fig. 4B), suggesting that ERAD-mediated translocation was involved in the cross-presentation of OVA-DRibbles. Treatment of DCs with NH4Cl, which inhibits lysosomal-mediated degradation, did not decrease cross-presentation of OVA-DRibbles. The ability of peptide-loaded DC to activate OT-I T cells was not affected by treatment with ExoA, bortezomib, and NH4Cl (Fig. 4B). In short, cross-presentation of DRibble antigens involved the ERAD pathway and required proteosome activity and TAP1, but not lysosomal degradation.
Tumor cell–derived DRibbles contained large amount of autophagosomes
Next, we investigated the feasibility of making DRibbles from a variety of tumor cell lines. DRibbles were visualized by light microscopy after tumor cells were treated with bortezomib and NH4Cl. These included human [non–small cell lung cancer (NSCLC) and melanoma] and mouse (3LL and B16F10) cell lines (Fig. 5A). DRibbles derived from 1 × 106 cells typically contained 10 to 100 μg proteins depending on the cell line used. To visualize the fine structure of DRibbles, we examined DRibbles from the 3LL mouse Lewis lung cancer cells with the transmission electron microscope (Fig. 5B). Many vesicles were released from the treated cells. DRibbles were harvested from the culture supernatant of the treated cells using differential centrifugation. A large number of vesicles with a unique double-membrane structure were found in the DRibble preparation. The majority of them were 300 nm to 1 μm in size, which falls in the range of the typical size of autophagosomes from mammalian cells (26). The light density of these vesicles varied too, indicating that, whereas some of the vesicles were autophagosomes, others may be amphisomes (Fig. 5B-c, d; ref. 33). Some of the vesicles included multiple smaller double-membrane structures (Fig. 5B-e). A phagophore that was still packaging cellular materials was also found in DRibble preparations (Fig. 5B-f).
Antitumor efficacy of DCs loaded with 3LL-dervied DRibbles in 3LL Lewis lung cancer model
First, we assessed whether 3LL-derived DRibbles could cross-prime T cells. DRibbles were prepared from 3LL tumor cells that expressed the Ub-M-GFP-OVA model antigen. When DCs were loaded with 20 μg 3LL-OVA DRibbles before they were used to stimulate naïve OT-I T cells, 96% of the OT-I T cells proliferated in vitro (Fig. 6A). IFN-γ induces DC to produce interleukin (IL)-12 p70 immunoproteasomes and polarizes the immune response to Th1 type (34, 35). Toll-like receptor (TLR) agonists engage TLR on DCs to induce vigorous inflammatory responses and set the stage for later adaptive immune responses (36, 37). Hence, in this tumor model, we used TLR3 agonist poly [I:C] in combination with IFN-γ to treat DCs during antigen loading to improve the efficacy of DRibble vaccine. Mice bearing 6-day established 3LL tumors (s.c.) were vaccinated with DCs pulsed with DRibbles collected from parental 3LL tumor cells, or DCs loaded with DRibble in the presence of IFN-γ and poly [I:C]. Untreated mice or mice treated with unloaded DCs served as controls. Vaccination with DCs loaded with DRibbles alone inhibited tumor growth (Fig. 6B) and significantly improved survival (Fig. 6C; P < 0.05) compared with untreated control. The treatment of DCs with IFN-γ and poly [I:C] remarkably enhanced the therapeutic efficacy of the DRibble-pulsed DC vaccine; 5 out of 6 treated mice were rendered tumor free and survived more than 80 days.
To compare the antitumor effect of DRibbles with other forms of tumor antigens, we injected 3LL tumor cells i.v. into C57BL/6 mice, and vaccinated these mice with DCs loaded with 3LL DRibbles, dead cells, and cell lysate in the presence of IFN-γ and poly [I:C] 3 days later. Vaccination with DRibble-loaded DCs showed significantly better ability in reducing the number of lung metastases as compared with DCs loaded with dead tumor cells and lysates (Fig. 6D).
We also evaluated the antitumor efficacy of DRibble vaccination in the B16F10 tumor model. C57BL/6 mice bearing 6-day B16F10 tumors were irradiated at 500 cGy, and then were adoptively transferred with 5 × 106 naïve spleen cells from pmel-1 TCR transgenic mice. These mice received 4 s.c. vaccinations at days 7, 10, 13, and 20 with 3 × 106 DCs loaded with DRibbles made from HEK 293T cells that expressed gp100. IFN-γ was used to treat DCs to improve DC function. We observed that IFN-γ increased production of IL-12 p70 (Supplementary Fig. S4A) by DCs. Vaccination with DCs loaded with gp100-DRibbles without IFN-γ slightly increased the number of pmel-1 T cells in the peripheral blood. DCs loaded with DRibbles and IFN-γ induced a dramatic expansion of pmel-1 T cells (Supplementary Fig. S4B). Accordingly, DCs pulsed with DRibble alone delayed tumor growth (P < 0.05; Supplementary Fig. S4C) and prolonged the median survival time of mice (38 days for vaccinated group vs. 26 days for control; Supplementary Fig. S4D), whereas DCs pulsed DRibbles and IFN-γ markedly enhanced median survival time of mice (from 38 to 59 days). DCs alone did not have any antitumor effect in this model (data not shown).
In the past decade, a number of cancer vaccines have been used in specific active immunotherapy trials for patients with melanoma, breast, and prostate cancers (38). These vaccines are based on peptides, partial or full-length proteins, whole-tumor cells, and TAA-loaded DCs (39, 40). The overall clinical response rate in cancer patients following active immunotherapy with vaccine was reported to be only 3.3%, using conventional response criteria (41). These unsatisfactory clinical outcomes may be a consequence of one or more of the following factors: (i) poor immunogenicity, likely due to poor cross-presentation of TAAs in vaccines, (ii) tumor immune escape due to antigen loss or defects in the antigen processing machinery, and (iii) immune suppression induced by myeloid-derived suppressor cells and T regulatory (Treg) cells. Therefore, efficient cross-presentation, inclusion of multiple tumor antigens, and reversal of immune suppression are important elements for the success of cancer vaccines.
We developed a novel autophagosome-based DRibble vaccine that facilitates efficient cross-presentation of multiple antigens to CD8+ T cells. Four unique characteristics of this novel vaccine could explain the high efficiency of cross-presentation of antigens in DRibbles and their superior antitumor activity. First, the DRibble vaccine incorporated SLiPs and DRiPs in autophagosomes. These SLiPs and DRiPs are not efficiently cross-presented by DCs under normal conditions. After inhibition of proteosome- and lysosome-mediated degradation, SLiPs and DRiPs became available for cross-presentation because they were shunted into autophagosomes when their degradation was blocked. Cross-presentation of SLiPs or DRiPs may be particularly beneficial for antitumor response, because they may give rise to most of the peptides presented on MHC class I molecules expressed on tumor cells (42). Cross-priming of CD8+ T cells specific for tumor SLiPs could increase the efficiency of tumor recognition and destruction. Second, an immune response to a diverse repertoire of TAAs in the vaccine may avoid immune escape via differential or altered expression of a particular antigen or MHC I molecule. Vaccines consisting of multiple epitopes or undefined antigens derived from tumor cells have given better clinical responses than singe peptide or protein (43). Third, DAMP molecules, such as heat-shock protein (HSP)-90, HSP94, calreticulin, and HMGB1 were also present in the DRibbles. HSPs, best known as chaperones, have been shown to play a critical role in preserving the antigen repertoire and augmenting cross-presentation of antigens (44). Calreticulin, an ER calcium-binding chaperone, was shown to be essential for triggering immunogenic responses by dying tumor cells (23). HMGB1, a DAMP molecule released by necrotic cells, augments cross-presentation through interacting with receptors for advanced glycosylation (RAGE), TLR2, and TLR4 on DCs (45, 46). Thus, these DAMP molecules in the DRibble preparation could serve as natural adjuvant during the vaccination for cancer therapy. DCs also express a variety of C-type lectin proteins, which function as TLR independent pathogen pattern recognition receptors and recognize viruses, bacteria, fungi, and even necrotic cell corpses. These receptors include DC-SIGN (CD209), langerin (CLEC4K, CD207), Dectin 1 (CLEC7A), Dectin 2 (CLEC6A), MICL (CLEC12A), DNGR 1 (CLEC9A), DCIR (CLEC4A), Dec 205 (CD205), and others. C-type lectin receptor-mediated signaling after pattern recognition induces differential cytokine production from DCs and thereby helps shape the immune response (47). CLEC9A, a member of C-type lectin family, is expressed on CD8α+ DCs and their precursors (CD8 α− CD24+ DCs), which are best known for cross-presenting cell-associated antigens (24, 48). Recently, CLEC9A was shown to recognize dead cell remnants and play a critical role in cross-presentation of antigens from necrotic cells (17). These features suggest that targeting CLEC9A could be a novel strategy in future DC-based vaccine design (17). Interestingly, DRibbles could directly bind to CLEC9A, and we show that CLEC9A was involved in cross-presentation of DRibble antigens. These results suggest involvement of CLEC9A in recognition or subsequent processing of DRibbles by DCs. Consistent with the findings from Caminschi and colleagues (24), we observed that CLEC9A did not affect the uptake of DRibbles. CLEC9A may facilitate cross-presentation at the intracellular routing or processing level. Because of these unique characteristics and to distinguish it from conventional pathways, we refer to this novel autophagosome-mediated cross-presentation pathway as the autophagy-assisted antigen cross-presentation pathway.
Imaging by light microscopy and transmission electron microscopy revealed that autophagosomes are the major component of the DRibbles. Autophagosomes are characterized by their double-membrane structure. Our data show that successful cross-presentation of antigens in DRibbles requires intact caveolae and clathrin-mediated endocytosis, as well as proteasomes and TAP. The ERAD pathway was also involved. Nevertheless, lysosomal activity seemed not to be necessary. On the basis of these findings, we envision the following pathway for cross-presentation of autophagosome antigens: Initially, the autophagosomes enter cells via caveolae-dependent and clathrin-dependent endocytosis. During entry, the CLEC9A receptor binds to autophagosomes and directs them into nonacidic compartments, for example, early phagosomes/endosomes or other intracellular compartments suitable for efficient cross-presentation. Subsequently, the antigens are first released from autophagosomes into phagosomes/endosomes, translocated into the cytosol through the ERAD machinery, and degraded by the proteosome. The peptide products are imported into either the ER or endosomes/phasosomes, loaded onto MHC I complex, and presented on the cell surface (Supplementary Fig. S5).
Finally, we showed that DRibbles, or DRibbles loaded onto DCs, in combination with IFN-γ and TLR agonist, induced robust antitumor responses against established 3LL lung carcinoma and B16F10 melanoma. Additionally, we recently reported that vaccination with autophagosomes, unlike vaccination with whole tumor cells, primes T cells which recognize a spectrum of related cancer cells and provides cross-protection against a panel of methylcholanthrene-induced tumors (49). Combining DRibble vaccine with agents that reverse the immune suppression in tumor-bearing hosts could further enhance the antitumor effect. On the basis of these findings, we have begun a phase I/II clinical trial of the DRibble vaccine in patients with NSCLC. We are planning a similar trial in breast cancer patients. Future studies will be focused on analysis of potential DAMP molecules, the nature of tumor antigens, CLEC9A ligands that are involved in cross-presentation, and the detailed analysis of immune responses induced by DRibble vaccines.
Disclosure of Potential Conflicts of Interest
H-M. Hu and B.A. Fox are cofounders of UbiVac, which has licensed the autophagosome intellectual property.
We thank Dr. Reis e Sousa for the generous gift of anti-CLEC9A antibody. We thank Eric Barklis, Mike Webb, and the OHSU EM Core Facility for electron microscopy sample preparation and imaging. We also thank Drs. E. Akporiaye, S. Shu, and C. Twitty for their input.
This work was supported by NIH grants CA107243 (H.-M. Hu), CA141278 (H.-M. Hu), and CA 123864 (W.J. Urba), the Susan G. Komen Breast Cancer Foundation (KG091164, H.-M. Hu and Y. Li), the Providence Portland Medical Foundation, and the Safeway Foundation.
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.