Expression of viral fusogenic membrane glycoproteins (FMGs) is a potent strategy for antitumor cytotoxic gene therapy in which tumor cells are fused into large multinucleated syncytia. To understand how local cell killing can potentiate activation of antitumor immune responses, we characterized the mechanism of FMG-mediated cell killing. Here, we show that syncytia are highly ordered structures over 24–48 h but then die through processes that, by multiple morphological and biochemical criteria, bear very little resemblance to classical apoptosis. Death of syncytia is associated with nuclear fusion and premature chromosome condensation as well as severe ATP depletion and autophagic degeneration, accompanied by release of vesicles reminiscent of exosomes (syncytiosomes). Dying syncytia produce significantly more syncytiosomes than normal cells or cells killed by irradiation, freeze thaw, or osmotic shock. These syncytiosomes also load dendritic cells (DCs) more effectively than exosomes from cells dying by other mechanisms. Finally, we demonstrate that syncytiosomes from either autologous or allogeneic fusing melanoma cells lead to cross-presentation of a defined tumor antigen, gp100, by DCs to a gp100-specific CTL clone. Cross-presentation was significantly more efficient than that with exosomes from normal, irradiated, or herpes simplex virus thymidine kinase/ganciclovir-killed tumor cells. Therefore, FMG-mediated cell killing combines very effective local tumor cell killing with the potential to be a highly immunogenic method of cytotoxic gene therapy. In addition, these data open the way for novel methods of loading DCs with relevant tumor-associated antigens for vaccine development.

Genetic transfer of viral FMGs6 is a novel approach to cytoreductive gene therapy for cancer that has shown potent antitumor efficacy both in vitro and in vivo(1, 2, 3, 4) and has potential for generating increased dispersion of viral vectors through tumors (5). Cytotoxicity results from the generation of large multinucleated syncytia when transduced cells expressing the FMG fuse with neighboring cells that express the appropriate viral receptor. This process of syncytial formation involves a significant bystander effect through recruitment of adjacent untransduced cells into the evolving syncytium (2, 3). Our own in vitro and in vivo analyses have shown that FMG-mediated cytoreductive gene therapy induces bystander cell killing in excess of 1 log relative to that produced by the HSVtk/GCV system under the same experimental circumstances (2).

Although the extent of direct and local bystander cell death per se has a significant bearing on the outcome of any cytoreductive gene therapy approach, no vector system can, at present, reliably deliver therapeutic transgenes systemically to metastatic cancer deposits disseminated throughout the body (6). Therefore, strategies that prime a systemic immune response against untransduced tumor cells and metastatic deposits offer the most immediate prospect of clinically beneficial gene therapy (7). In this respect, the biochemical mechanisms by which cancer cells are killed are of considerable importance to the subsequent immunogenicity of that death (8, 9, 10, 11). In general terms, cell death by the highly ordered process of apoptosis, with subsequent phagocytosis of neatly packaged cellular fragments by macrophages, escapes detection by the immune system and is actually immunosuppressive (12, 13). In contrast, cancer cell death through nonapoptotic (necrotic/oncotic) mechanisms, separate from nonphysiological processes such as freeze thaw or osmotic shock, is proinflammatory and has the potential to prime the generation of a systemic antitumor immune response (8, 9, 10, 11, 13). We have shown that delivery of a hyperfusogenic mutant of either the GALV or the measles virus F and H protein combination in the context of plasmid or viral vectors leads to therapeutic reductions in the growth of human tumor xenografts in nude mice (2, 3). However, because these FMGs do not fuse murine cells, it is not possible to assess the immunostimulatory value of local tumor treatments with these vectors directly in immunocompetent mice. Therefore, we have characterized the biochemical mechanisms by which two different FMGs induce cell killing. We show that syncytial formation is not associated with markers of classical apoptosis at early, intermediate, or late stages of syncytial development and disintegration. In contrast, syncytia undergo nuclear fusion, autophagic digestion, and nonapoptotic death that is associated with the release of immunologically active exosome-like vesicles (syncytiosomes) that can transfer a defined melanoma tumor antigen, gp100, into DCs for presentation to gp100-specific T cells. Taken together, our results show that fusion of cells through expression of viral FMG leads to a nonapoptotic mechanism of cell death that is highly immunostimulatory and may be valuable in the development of novel protocols for the loading of DCs ex vivo for production of cancer vaccines.

Vectors, Cells, and Transfections

cDNAs encoding two different FMGs were used in this study expressed from the cytomegalovirus promoter in the pCR3.1 expression plasmid (Invitrogen, Carlsbad, CA). F&H were expressed from the vectors pCR3.1-F and pCR3.1-H (2) by cotransfection into tumor cells. A hyperfusogenic mutant of the GALV envelope, as described in Refs. 1 and 2, was expressed by transfection with pCR3.1-GALV.

Cell lines were grown in DMEM (Life Technologies, Inc., Rockville, MD) supplemented with 10% (v/v) FCS (Life Technologies, Inc.) and l-glutamine (Life Technologies, Inc.). All cell lines were monitored routinely and found to be free of Mycoplasma infection. For HSVtk/GCV cell killing in vitro, medium was supplemented with GCV (Cytovene; Roche, Nutley, NJ) to a final concentration of 5 μg/ml.

All transfections were performed with 1 μg of plasmid DNA using Effectene lipid reagent (Qiagen, Valencia, CA) according to the manufacturer’s instructions. For cotransfections with F&H, 1 μg of each plasmid DNA was used.

Mel624 and Mel888 are two adherent human melanoma tumor cell lines that are HLA-A2+ and HLA-A2−, respectively, and were obtained from Dr. Esteban Celis (Mayo Clinic). Both cell lines express comparable levels of the melanoma-associated antigen gp100 by Western blot analysis (data not shown). Presentation of a gp100-derived, HLA-A2-restricted epitope to the human cytotoxic T-cell line αgp100-CTL (a kind gift from Dr. Esteban Celis) induces release of both IFN-γ and GM-CSF as well as direct target cell killing that can be assessed by chromium release assays.

The results presented in this paper are representative of the effects of either of the two FMGs used (GALV or F&H) in all five human histological types tested [fibrosarcoma (HT1080), osteosarcoma (Tel.CeB6), colorectal cancer (HCT-116), lung cancer (SCC9), and melanoma (Mel624 and Mel888)], unless specifically described otherwise. Experiments with exosome preparation and DC loading were performed only with the GALV FMG.

Western Analysis for Procaspase-3 Activation and Proteolytic Cleavage of PARP and for Analysis of Protein Composition of Syncytiosomes

HT1080 or Tel.CeB6 cells were transiently transfected with the F&H plasmids or with the H plasmid alone. Both adherent and floating cells were harvested for protein preparation at 18 and 40 h. Jurkat cells treated with etoposide (VP16) or FasL were used as positive controls. Western analysis was performed according to the previously reported protocol (14). Briefly, adherent cells in a 25-cm2 flask were washed and harvested by a 10-min incubation in a solution of 30 μl of phenylmethylsulfonyl fluoride (Sigma, St. Louis, MO) and 30 μl of β-mercaptoethanol in 3 ml of alkylation buffer [6 m guanidine HCl, 250 mm Tris-HCl (pH 8.5), and 10 mm EDTA]. Samples were then sonicated and allowed to reduce overnight. Thereafter, 300 μl of iodoacetamide (278 mg/ml) and phenylmethylsulfonyl fluoride (10 μl/ml) in alkylation buffer were added, mixed, and incubated for 1 h in the dark. At this time, 30 μl of β-mercaptoethanol were added, and the samples were dialyzed against 4 m urea and 50 mm Tris (pH 7.4) for 90 min and then dialyzed against 4 m urea for 90 min (×4) and, finally, dialyzed against 0.1% SDS for 90 min (×3). At this stage, an aliquot of 100 μl was removed for quantitation (Bio-Rad, Hercules, CA), and the remainder was frozen on dry ice and lyophilized to dryness. Subsequently, the protein sample was resuspended in SDS sample buffer [4 m deionized urea, 2% (w/v) SDS, 62.6 mm Tris-HCl (pH 6.8), and 1 mm EDTA] at a concentration of 5 μg/μl, heated to 65°C for 20 min, run on a 10% SDS-PAGE gradient gel, and transferred at 4°C to a nitrocellulose membrane at 60 V. Antibodies were obtained as a kind gift from Dr. Scott Kaufman (details available on request).

The protein composition of syncytiosomes, exosome, and whole cell lysates was analyzed by Western blotting. Nitrocellulose membranes were probed with antibodies against human caveolin-1 (PharMingen), human HSP70 (StressGen, Victoria, British Columbia, Canada), human HSP90, HLA-A2, gp96 and gp100 (Santa Cruz Biotechnology), and β-actin (Sigma).

ZVAD-fmk Assay

HT1080 human fibrosarcoma cells were plated in 6-well plates (2 × 105 cells/well) and transfected 24 h later. On each plate, transfections with FMG plasmids or control plasmids were performed. After overnight incubation, wells were washed, and the medium was replaced with normal medium or the same medium plus 50 mm/ml pan-caspase inhibitor ZVAD-fmk (Enzyme Systems Products, Livermore, CA). The medium from individual wells was collected and replaced with fresh medium (with or without ZVAD-fmk, as appropriate) for 4 consecutive days and stored at −70°C. The samples obtained were then analyzed using a lactate dehydrogenase cytotoxicity detection kit (Boehringer Mannheim). The supernatant was centrifuged at 3000 rpm for 5 min, triplicate 100-μl samples were plated on a microtiter plate, and 100 μl of reaction mixture (250 μl diaphorase solution/11.25 μl tetrazolium salt solution) were added to each well and incubated for 30 min at 25°C in the dark. The absorbance was determined at 492 nm with wavelength correction of 620 nm using the SPECTRAmax 190 plate reader (Molecular Devices, Sunnyvale, CA).

RNase Protection Assay for Apoptotic and Antiapoptotic Genes

RNA was prepared from control or FMG-transfected cells using a RNeasy Mini Kit (Qiagen) according to the manufacturer’s instructions. The concentrations of RNA were determined by spectrophotometry at 280 nm, and 10-μg samples were aliquoted before analysis. Two RNA protection assay kits (hAPO-2 and hAPO-3c) were obtained (BD PharMingen) containing cDNA templates to permit analysis of a wide range of RNA species (see “Results”). Assays were performed according to the kit instructions. Briefly, 32P-labeled RNA probes were synthesized from the provided cDNAs using T7 DNA-dependent RNA polymerase. Subsequently, the radiolabeled probes were treated with DNase I to degrade the templates, column-purified (Chroma Spin-30; Clontech, Palo Alto, CA), quantified, and diluted to contain 2.5 × 105 cpm/μl. The probes were hybridized to the 10-μg RNA test samples at 56°C overnight. Thereafter, the samples were treated with RNase A/T1, precipitated, and resuspended in running buffer before being run on a 6% polyacrylamide gel (Ambion Inc., Austin, TX) in Tris-borate EDTA buffer.

Digital Image Analysis

HT1080 and HCT-116 cells were grown on Labtek chamber slides (Nalge Nunc International, Nutting Lake, MA) and transfected with measles F&H or GALV. Untransfected cells served as a normal control. Syncytia development was studied at 24, 48, and 72 h for the HT1080 cell line and at 24, 48, 72, and 96 h for the HCT-116 cell line. Approximately 200 nuclei from each slide were captured and analyzed using a CAS 200 image analyzer (Bacus Laboratories, Lombard, IL). The nuclear morphometry features of area, DNA index, and average absorbance were obtained for each nucleus using Cell Sheet software. These features were summarized with means and SDs. Differences in the nuclear morphometry features among the slides were assessed using two-way ANOVA models with terms for transfection, time period, and the interaction between the two. These models were assessed separately for the two cell lines. The nuclear morphometry features were analyzed on a natural logarithmic scale to meet model-fitting assumptions required by ANOVA. All tests were two-sided, and Ps < 0.05 were considered statistically significant.

Studies of the Morphological Features of Syncytial Development

Nuclear Morphology.

Cells were plated at 1 × 105 cells/chamber in Labtek chamber slides and transfected with 1 μg of FMG plasmid. At time points up to 120 h after transfection, the slides were stained by the addition of 2 μl/ml DAPI or Hoechst 33342 and incubated at 37°C for 10 min. The cells were observed using a LSM 510 confocal microscope for the evolution of nuclear morphological changes. In addition, tumor cells transfected with pCR3.1-FMG were trypsinized, pelleted, and prepared for electron microscopic examination as follows. Cell pellets were fixed in Trump’s fixative [1% glutaraldehyde and 4% formaldehyde in 0.1 m phosphate buffer (pH 7.2)], rinsed for 30 min in 3 changes of 0.1 m phosphate buffer (pH 7.2), and incubated for 1 h in phosphate-buffered 1% OsO4. After rinsing in three changes of distilled water for 30 min, the cell pellet was stained en bloc with 2% uranyl acetate for 30 min at 60°C, rinsed in three changes of distilled water, dehydrated in progressive concentrations of ethanol and 100% propylene oxide, and embedded in Spurr’s resin. Thin (90 nm) sections were cut on a Reichert Ultracut E ultramicrotome, placed on 100–200 mesh copper grids, and stained with lead citrate. Micrographs were taken on a JEOL 1200 EXII operating at 60 kV.

Cytoskeletal Organization.

Tumor cells were transiently transfected with control plasmid or with pCR3.1-FMG. At time points up to 120 h after transfection, the cells were stained with anti-tubulin antibody (T5293; Sigma) and FITC secondary antibody (donkey antimouse IgG FITC-labeled; Jackson Immunoresearch Laboratories, West Grove, PA). Actin was stained with rhodamine phalloidin (P1951; Sigma). In all samples, nuclei were counterstained with DAPI.

Mitochondrial Distribution.

Tumor cells growing in Labtek chamber slides were transiently transfected with FMG or control plasmid. At different time points, samples were stained with murine monoclonal anti-HSP60 antibody (SPA-806; StressGen) and a donkey antimouse IgG tetramethylrhodamine isothiocyanate-labeled secondary antibody (Jackson Immunoresearch Laboratories). Control specimens were stained with phycoerythrin-conjugated secondary antibody alone. The nuclei were counterstained with DAPI.

Cytoplasmic Vacuoles.

Tumor cells growing in Labtek chamber slides were transfected with pCR3.1-FMG. Forty-eight h later, they were incubated with propidium iodide or LysoTracker Red DND-99 (Molecular Probes, Eugene, OR) diluted to 50 nm in medium. Cells were cultured for 30 min at 37°C, washed in PBS, and then viewed with an Olympus 1x70 tissue culture microscope. Images were captured with an Olympus SC35 Type 12 camera (Olympus Corp. Japan).

Preparation of Exosomes and Syncytiosomes from Cell Cultures

Exosomes were prepared as described in Wolfers et al.(15). Briefly, supernatants from cultures of previously transfected, irradiated, or otherwise treated tumor cells (2 × 106 to 2 × 107) were centrifuged sequentially at 300 × g (4°C) for 10 min, 2× at 800 × g for 15 min, and at 10,000 × g for 30 min. Exosomes were pelleted from the supernatants at 100,000 × g for 1 h and resuspended in 100 μl of PBS. Protein concentration was measured using a Bradford assay (Bio-Rad). The amount of protein recovered from this procedure from cells that had been freeze-thawed over three cycles or from osmotically shocked cells was very low and very variable. In DC loading experiments, all of the exosome preparation from these sources was used.

DC Loading and gp100 Antigen Transfer to DCs by Exosome Preparations

Tumor cells were labeled using the red dye Cell Tracker Orange [5- (and 6-) (((4-chloromethyl)benzoyl)amino) tetramethylrhodamine (Molecular Probes)] according to the manufacturer’s instructions. These cells were used as a source of exosome/syncytiosome production after transfection, irradiation, or other treatment. Equal amounts of protein from exosome preparations were loaded onto immature HLA-A2+ human DCs labeled with the green dye Cell Tracker Green (5′-chloro-methyl-fluorescein diacetate; Molecular Probes). Uptake of tumor exosomes by the DCs was measured by fluorescence-activated cell-sorting analysis for a double staining population of DCs. For confocal analysis, DCs loaded with exosomes were incubated on glass chamber slides, washed thoroughly in PBS, fixed in 4% (v/v) formaldehyde, and mounted with Vectashield (Vector Laboratories, Burlingame, CA) containing 2 μg/ml DAPI (Roche, Indianapolis, IN). Analysis was performed using a LSM510 confocal microscope (Zeiss, Oberkochen, Germany).

For antigen presentation assays, 107 Mel624 or Mel888 cells were irradiated (100 Gy), freeze-thawed, osmotically shocked, or transfected with either the GALV FMG or the HSVtk gene and grown in GCV. Forty-eight h later, cultures were used to prepare exosomes/syncytiosomes. A total of 0.2 μg protein/exosome preparation (or 100 μl of the exosome preparation from osmotically shocked or freeze-thawed cells) was incubated with 2 × 105 immature HLA-A2+ or HLA-A2− human DCs. Twenty-four h later, DC cultures were incubated with 2 × 106 human gp100/HLA-A2-specific T cells, and GM-CSF release from the T cells was measured as a surrogate of recognition of the gp100 antigen presented in the context of HLA-A2 by the DCs.

Syncytial Development Is a Highly Ordered Process at the Level of the Cytoskeleton.

Transfection of genes expressing FMG (either GALV or F&H) leads to the generation of large multinucleated syncytia (Fig. 1, A and B; Ref. 2). Typically, these syncytia remain metabolically active and alive (as seen by trypan blue exclusion) for anywhere from 12–120 h posttransfection. By 120 h posttransfection, >90% of cells in culture are dead (2). To grow, the syncytia recruit more cells by fusing nontransfected cells that express the relevant cell surface receptors. However, far from being chaotically organized, immunohistochemical staining for components of the cellular cytoskeleton revealed that the actin and tubulin networks are highly ordered and maintain the architecture of the large multinucleated structures that form the syncytium (Fig. 1, C and D). Newly fused cells contribute nuclei to the syncytium that subsequently cluster together, and the cytoskeletal network appears to facilitate tracking of these nuclei from the site of cell-cell fusion to the site of nuclear aggregation (Fig. 1,D). Immunohistochemical staining for anti-HSP60 antibody showed that mitochondria become widely distributed through the syncytia, largely correlating with the cytoskeletal organization (Fig. 1, E and F). At late stages of the syncytial life cycle, nuclear changes are seen, and disintegration of the cytoskeleton is accompanied by breakdown of the syncytial structure (see below). Finally, we confirmed that syncytia are not only seen in cultured cells in vitro. Thus, injection of adenoviral vectors into progressively growing human tumor xenografts produces areas of syncytial formation within solid tumors that closely resemble the structures seen in vitro (Fig. 1 G).

Development and Subsequent Death of Syncytia Are Associated with Nuclear Fusion.

DAPI-stained tumor cells transfected with pCR3.1-GALV exhibited prominent nuclear fusion at 48–72 h. Representative images of this process are presented in Fig. 2, A–C. At early stages of syncytial development (up to 24–48 h), nuclei remain clearly intact and distinct. At later time points, small areas in which adjacent nuclei within a single syncytium are fused together can be observed (Fig. 2,A). This fusion process progressively sweeps through the site of nuclear aggregation and may eventually involve all of the nuclei within a syncytium (Fig. 2,B). Occasional syncytia can sometimes be observed undergoing premature chromosome condensation (Fig. 2,C). At no time was nuclear condensation or fragmentation of a type that would be consistent with apoptotic cell death seen. Similar features of nuclear fusion were observed with DAPI and Hoechst 33342 staining in other cell types (data not shown). Electron microscopy also confirmed these observations, and an example of such giant nuclei is shown in Fig. 2 E.

FMG-mediated Syncytial Formation Is Not Associated with Procaspase-3 Activation or PARP Cleavage.

The morphological characteristics of syncytial development and death were inconsistent with the changes typically described for apoptotic cell death (16, 17, 18, 19). Therefore, we investigated whether biochemical characteristics of apoptosis could be detected at any stage of the killing process. Western blot analysis for procaspase-3 activation or PARP cleavage during FMG-mediated syncytial formation, development, and death only ever showed minimal evidence of procaspase-3 activation in tumor cells transfected with FMGs (Fig. 3). In contrast, treatment of cells with FasL generated a significantly reduced signal for procaspase-3 in the positive control population. Similarly, exposure of cells to either etoposide or FasL was associated with significant PARP cleavage, but this effect was absent from cultures of tumor cells transfected with FMG. Importantly, these data were similar both at early times of syncytial killing (0–24 h after DNA transfection) and at late times (72–96 h), by which time large amounts of cell death were observed in the transfected cell populations. These data indicate that procaspase-3- and PARP-mediated mechanisms of apoptosis play at most a minimal role in FMG-mediated cytotoxicity.

Cytochrome c Is Retained within Mitochondria during Syncytial Killing, and Other Markers of Apoptosis Are Absent.

Because the release of cytochrome c serves as an important signal for the initiation of many forms of apoptotic cell death (16), we investigated the subcellular localization of cytochrome c during the syncytial killing process. Cytochrome c was retained within mitochondria (Fig. 4) even until stages of the killing process where cell death was apparent through trypan blue exclusion. These data suggested that cytochrome c release is unlikely to serve as an initiator of apoptosis in this killing process. In addition, terminal deoxynucleotidyl transferase-mediated nick end labeling and DNA laddering assays were uniformly negative during syncytial killing (data not shown; Refs. 2 and 4), with the exception of at very late stages of the killing process, when the majority of tumor cells were already dead. The effect of the addition of the pan-caspase inhibitor ZVAD-fmk on the cytotoxicity of GALV- or F&H-mediated cell killing was also investigated intransiently transfected tumor cells. The profile of cytotoxicity for each FMG was the same in the absence or presence of ZVAD-fmk by lactate dehydrogenase assay (see “Materials and Methods;” Ref. 4; data not shown). Because apoptosis can occur through a variety of different pathways and the actions of a diverse set of pro- and antiapoptotic genes (16, 17, 19), we also used RNA protection assays to monitor a range of such genes during the killing of cells through syncytial development. No significant changes in the levels of such genes, including caspase 8, Fas, DR3, DR5, DR4, TRAIL, TNFRp55, RIP, bclx, bfl1, blk, bak, bax, bcl-2, and mel1, were observed at early time points or at later stages, by which time extensive cell death was evident (see “Materials and Methods;” data not shown).

Cells Recruited into Syncytia Accumulate in the G2-M Phase of the Cell Cycle.

The nuclear morphometry features of area, DNA index, and average absorbance for normal cells or cells recruited into syncytia at various time points are summarized in Table 1 for HCT-116 cells. Generally, the mean nuclear area was significantly greater in cells transfected with a FMG as compared with untransfected controls at each of the individual time points (P < 0.001). In addition, for each transfection condition (F&H, GALV, or control), there was a significant change in nuclear area across the 72-h study period (P < 0.001). The interaction between these two terms was highly significant (P < 0.001), indicating that the differences in area by time period were not the same for the two FMGs and the control group. There were also significant differences in mean DNA index between the F&H, GALV, and control groups at the various time points (P < 0.001) and across time (P = 0.001). Very similar results were obtained for other cell lines.

Measurements of DNA mass for HCT-116 cells showed this cell line to be tetraploid (mean DNA mass, 8 pg). By DNA mass criteria, cells transfected with control plasmid were distributed through the cell cycle with cells in G1, S, and G2-M at 24 h (Fig. 5,A). At 72 and 96 h, most of the cells were in G1 phase, with evidence of a progressive reduction of the proportion of cells in S and G2-M (Fig. 5, B and C). In contrast, cells transfected with a FMG showed a significant and progressive right shift in their distribution within the cell cycle (Fig. 5, D–F). As a consequence, by 96 h after transfection, the majority of the cells were in the G2-M phase of the cell cycle. Very few cells were seen in G1 or S, suggesting that the transfected cells were accumulating or blocking in G2-M. An identical pattern of changes was seen in HCT-116 cells transfected with GALV or F&H (data not shown). For HT1080 cells, the same pattern was observed (data not shown). Importantly, at none of the time points was a significant population of sub-G1 cells identified, confirming the morphological and biochemical observations that the cytotoxic process is nonapoptotic.

Syncytial Cell Killing Is Associated with Buildup of Intracellular Lysosomes, Extensive Vacuolation, and Cytoplasmic Blebbing.

At late stages of syncytial development, coincident with or following the nuclear changes described earlier, extensive cytoplasmic vacuolation was observed even though syncytia were still viable as determined by propidium iodide staining (Fig. 6). Staining with the dye LysoTracker Red, which concentrates in acidic organelles of viable cells, demonstrated enhanced staining in these vacuolated syncytia (Fig. 6). Similar features were seen in several other cell lines, showing that these effects are not cell line specific. These morphological and staining characteristics are consistent with an autophagic degeneration in which the death of a cell/syncytium occurs through proteolytic digestion from within (20, 21, 22). These data suggest that some signal (possibly an inability to control the expanding cell surface, volume of the syncytium, and/or a metabolic depletion within the cell) finally induces cell killing through autophagic destruction.

In addition, microscopy of late-stage but still viable syncytia formed after FMG transfection showed large numbers of blebs arising from the cell membranes (Fig. 7 A). These did not resemble apoptotic bodies. However, their existence at late stages of the life of the syncytia was suggestive of exosome-like structures that have been described as vehicles for the transfer of tumor antigens into DCs and are themselves clearly separate from apoptotic bodies (15, 23, 24, 25).

Biochemical and Structural Characterization of Syncytiosomes.

We compared the structure and composition of the FMG-mediated production of syncytiosomes with that of previously described tumor-derived or dendritic exosomes (15, 23, 24). Consistent with the structure of exosomes, electron microscopy of fusing cells indicated numerous multivesicular bodies. In some samples, apparent fusion of late endosomes with the plasma membrane was observed. In addition, the number of the exosomes/syncytiosome-like vesicles released from cultures undergoing FMG-mediated fusion was significantly increased compared with similar cultures either growing normally or undergoing cell killing by HSVtk-GCV (Fig. 7, B and C). After differential centrifugation for the preparation of exosomes, electron microscopy revealed a population of vesicles of a size consistent with that reported for DCs and tumor-derived exosomes (60–90 nm in diameter), often with typical cup-shaped morphology (15, 23, 24). In addition, syncytiosome preparations were also noticeably enriched with free membrane material in addition to the vesicles themselves (Fig. 7 D).

We also characterized the protein content of the syncytiosomes. Consistent with the proposed role of exosomes as rich sources of tumor antigens, the tumor antigen gp100 was readily detected in syncytiosomes from fusing melanoma Mel888 cells (Fig. 7,E). Western blot analysis for a variety of other molecules demonstrated a profile as summarized in Fig. 7,F. Notable in this analysis was that, unlike the reported composition of tumor-derived exosomes, levels of class I MHC were either very low or not detectable in most preparations of syncytiosomes. Also contrary to the reported composition of tumor-derived exosomes, we could consistently detect gp96 in syncytiosomes. Through a time course analysis, we also observed that inducible HSP70 is directed into syncytiosomes only at low levels at early time points of syncytial killing (up to 48 h after transfection). Thereafter, syncytiosomes rapidly become enriched for HSP70 (48–96 h after transfection; Fig. 7,G). These data suggest that HSP70 may even act as an intracellular indicator of the onset of death of the syncytia. Given some of these differences between the composition of tumor-derived exosomes and syncytiosomes, we also looked for the presence of some alternative proteins. Most striking in this analysis was the observation that during cell fusion, caveolin appears to be selectively excluded from the cytoplasm of syncytia and secreted within discrete particles that are not normally seen in normal cells (Fig. 7, H and I). Correspondingly, caveolin was detected at high levels in syncytiosomes, even from early times after the initiation of cell fusion (Fig. 7 J). Interestingly, we could not detect caveolin in preparations of standard tumor or DC-derived exosomes. These findings suggest that the pathways of syncytiosome generation, although clearly related to that of exosome production, may have significant differences.

FMG-mediated Cell Fusion Induces the Production of Exosomes in Higher Quantities and of Higher Quality than that from Normal, Freeze-thawed, or Osmotically Shocked Cells.

Therefore, exosomes were prepared from fusing tumor cells as described by Wolfers et al.(15) as well as from tumor cells mock-transfected or killed by irradiation (100 Gy), freeze thawing (three cycles), or osmotic shock. Measurement of the protein content of the resulting exosomes indicated that fusing tumor cells repeatedly generated more exosomes than the same number of irradiated cells and twice as many exosomes as normal growing cells (Table 2).

Exosomes from fusing cells were significantly more effective at labeling DCs than those from irradiated cells, and both sources were superior to preparations from normal cells growing in culture (Table 3). Confocal microscopy of the labeled DCs suggested that a major proportion of the double staining was due to uptake of tumor-derived exosomes rather than simply binding to the DCs (data not shown). Previously, we have shown that syncytia develop severe metabolic/ATP depletion (4) and therefore that cell death of the syncytia can be rescued in part by incubation with 20 mm fructose. Therefore, we also performed the experiment of Table 3 in the presence of fructose. Whereas exosome preparations from irradiated or normal cells were equally capable of labeling DCs in the presence or absence of fructose, exosomes from fusing cells grown in the presence of fructose were significantly less able to load the immature DCs (Table 3). This result is consistent with the hypothesis that the “quality” of the exosomes produced from fusing tumor cells is related to the amount of cell killing and possibly also to the degree of stress that the syncytia are undergoing at the time of syncytia release.

Exosomes from Fusing Cells Are a Superior Source of Tumor Antigen for Cross-Presentation to Cytotoxic T Cells.

Finally, we investigated whether the exosome-mediated uptake of tumor-derived material by DCs was immunologically relevant. Mel888 and Mel624 are HLA-A2− and HLA-A2+ human melanoma cell lines, respectively, that both express the gp100 tumor antigen. The cytotoxic T-cell line αgp100-CTL recognizes gp100 in the context of HLA-A2 class I molecules. Coculture of fusing melanoma cells in both human and murine systems leads to cross-presentation of tumor-associated antigens by DCs. Therefore, we assayed whether these observations could be explained by the release of the syncytiosomes described above. Exosomes derived from Mel888 or Mel624 cells were incubated with either human HLA-A2+ or HLA-A2− DCs, and the DCs were then incubated with αgp100-CTL. As expected, HLA-A2− human DCs were unable to stimulate GM-CSF production from αgp100-CTL, irrespective of the source of exosomes or even when incubated with gp100 peptide (data not shown). Fig. 8 shows that exosomes from cells growing in culture were poor sources of gp100 for presentation to the T cells when loaded into HLA-A2+ DCs. In one experiment, exosomes from irradiated Mel624 cells loaded HLA-A2+ DCs with gp100 (Fig. 8,A), but this was not repeated in two other experiments and was not seen at all when exosomes were used from Mel888 cells (Fig. 8,B). Exosomes from Mel624 or Mel888 cells undergoing cell death as a result of HSVtk transfection/GCV treatment were very poor sources of cross-presented antigen, but the low levels of GM-CSF production were more reproducible in both cases (Fig. 8, A and B; two of three experiments). However, exosomes from both Mel624 and Mel888 fusing cells were reproducibly able to serve as a very good source of gp100 for presentation to αgp100-CTL (Fig. 8, A and B). HLA-A2+ DCs loaded with exosomes from fusing cells but not incubated with the CTL did not produce GM-CSF (data not shown). Although fusing cells induce DC maturation (that is whole Mel624 cells dying by FMG-mediated killing), the isolated syncytiosomes do not. Mel888 exosomes were significantly more effective than Mel624 cells in this assay. We are currently investigating whether the explanation for this is trivial (relating, for example, to the levels of gp100 expressed in Mel888 compared with Mel624) or whether the presence of allo-MHC molecules may enhance the potency of the exosomes for loading into the DCs.

Coculture of syncytiosome-loaded DCs with αgp100-CTL in the presence of the pan-HLA antibody W6.32 inhibited GM-CSF release from the CTLs by over 85% or 95% using syncytiosomes from either Mel624 or Mel888 melanoma cells, respectively (Fig. 8,C). An isotype-matched irrelevant antibody at the same concentration was unable to block presentation of the gp100 epitope, confirming that cross-presentation of the gp100 antigen occurs through class I-mediated presentation pathways (Fig. 8,C). Unlike the dose response for loading DCs (Table 3), the levels of GM-CSF production did not increase when the DCs were loaded with an increasing amount (0.5 μg) of syncytiosome protein. We also confirmed by Western blot that purified preparations of syncytiosomes from the melanoma cells are enriched for the gp100 tumor antigen (Fig. 7 E). We have also demonstrated very similar syncytiosome-mediated cross-presentation of the model tumor antigen ova from murine B16-ova melanoma cells. Murine DCs were able to present the SIINFEKL epitope of ova to ova-specific OT1 T cells, and we have confirmed by Western blot analysis that these B16-ova-derived syncytiosomes contain ova antigen (data not shown). Taken together, these data demonstrate that syncytiosomes from FMG-mediated killing of melanoma cells contain tumor-associated antigens, can load DCs, and can lead to cross-presentation of these antigens to antigen-specific T cells.

Many different biochemical pathways leading to apoptosis have been described, and it is becoming increasingly unclear as to what exactly constitutes apoptotic as opposed to nonapoptotic death (17, 18, 22). However, by multiple different criteria, syncytia-associated cell death proceeds through pathways that lack the morphological, cytogenetic, or biochemical markers of classical apoptosis (16, 17, 18, 19). We could not detect significant changes in the level of expression of pro- or antiapoptotic genes at any point through syncytial formation (0–24 h after transfection), development (24–72 h), and disintegration (72–120 h). We also observed that syncytial development is structurally a highly ordered process. However, at late stages, disintegration of the syncytia occurs, usually marked by nuclear fusion, severe depletion of cellular ATP (which is known to inhibit many apoptotic pathways), and the appearance of multiple acidified vacuoles that resemble lysosomes. Subsequent death of the syncytia is rapid, probably through a process that is most akin to autophagy (20, 21, 22). The precise triggers that turn a viable, ordered syncytium into a self-digesting, dying aggregate may include the increasing volume of the syncytium, cell cycle-associated signals, and/or the severe metabolic depletion that occurs as a result of trying to maintain such a huge cellular structure.

We and others have previously shown that direct killing of tumor cells in vivo with the HSVtk suicide gene system can, in some cases, stimulate antitumor immunity (8, 26, 27). In particular, the mechanisms by which tumor cells are killed are critical to the attraction of professional antigen-presenting cells, such as DCs and macrophages, to the site of killing (9, 13, 28). In general, large amounts of apoptotic death, which overwhelms the local phagocytic capacity to clear it (9, 12, 29, 30), and/or nonapoptotic cell death, which is associated with activation of stress response programs (8, 10, 11, 13, 31), lead to activation of antitumor immune responses. Therefore, we hypothesized that the nonapoptotic killing of tumor cells that we observe with FMG may have significant immunostimulatory properties. Our data here show that fusing tumor cells release increased numbers of tumor-derived, exosome-like vesicles (15, 25) compared with normal cells growing in culture or tumor cells killed through irradiation, osmotic shock, or freeze thawing. Our characterization of syncytiosome structure suggests that they resemble previously described exosomes in general size, structure, and composition [including tumor antigen (15, 23, 24)]; however, our preliminary analysis suggests that syncytiosomes also differ in that they are produced from fusing cells at higher levels than normal cell-generated exosomes, and they contain gp96 but have only low levels of MHC class I antigens. In addition, their composition appears to evolve with time and the killing process, most notably in the sequestration of HSP70, suggesting that HSP70 may even act as an intracellular indicator of the onset of death of the syncytia. Finally, we observed enrichment for caveolin-1 specifically in syncytiosomes. Caveolin is a major component of caveolae, plasma membrane assemblies of glycosphingolipids and cholesterol that are associated with specific molecules including signaling proteins (32). Caveolin-1 has been associated with specific secretory pathways (33, 34) including viral budding (35), and it may be that syncytiosomes represent a variation on the exosome theme associated with secretion of cell vesicles during pathological/viral-type infection processes. Moreover, exosomes from FMG-mediated fusion are also of a significantly enhanced quality in terms of their ability to transfer tumor-derived material into immature DCs. Finally, our data also confirm that this loading of DCs is immunologically relevant because it allows the cross-presentation by DCs of a known tumor antigen gp100. This cross-presentation was significantly more reproducible and effective than using tumor-derived exosomes from other sources of cell killing, including irradiation and HSVtk/GCV-mediated killing. For these reasons, we believe that the exosomes produced from FMG-mediated fusing tumor cells, syncytiosomes, are qualitatively different from exosomes derived from other sources in ways that enhance their recognition, uptake, and loading abilities with regard to DCs. In particular, successful cross-presentation of gp100 epitopes from fusing Mel888 (HLA-A2−) cells by HLA-A2+ DCs clearly indicates that syncytial cell killing not only promotes antigen release but also allows entry of the antigen into the class I antigen processing pathways for (cross) presentation. Indeed, syncytiosomes from the allogeneic partner tumor cells were reproducibly (three of three experiments) significantly more effective than those from autologous cells in this assay. The explanation for this may be trivial (for example, relating to the levels of gp100 expressed in Mel888 compared with Mel624), or it may be significant, perhaps due to the presence of allo-MHC molecules in the syncytiosomes.

The mechanisms by which tumor antigens are transferred efficiently into the human DCs for subsequent MHC class I-mediated presentation are currently under investigation. It may be that stress proteins such as HSPs, which we have shown to be induced during the syncytial killing process (2) and are concentrated within exosomes (15, 23, 24), serve in some way as potentiators for uptake of the syncytiosomes (36, 37, 38). Other routes for uptake of antigen from dying cells and subsequent cross-presentation by DCs have also been reported that rely on levels and routes of antigen release (30, 39, 40, 41, 42). Although there is still debate about the mechanism by which radiation kills cells (43), and the mechanism of HSV-tk/GCV killing is also variable (8), our data suggest that syncytial killing across a wide range of solid tumor cell lines occurs through a common mechanism with highly immunostimulatory potential.

VSV-G induces low pH-induced syncytial formation in murine cells by mechanisms similar to those described here. Replicating VSV-G infection of B16 cells has recently been shown to be an effective antitumor therapy both locally and with the generation of anti-B16 CTLs (44); in addition, vaccination studies with murine melanoma cells fused by the VSV-G FMG ex vivo and used as in vivo vaccines have shown potent vaccinating capabilities in our own laboratory (45),7 suggesting that the immunostimulatory effects of GALV-mediated fusion described in the work reported here have in vivo relevance.

Our data here are significant in several ways. First, they demonstrate that syncytial killing occurs through a nonapoptotic pathway that is associated with an enhanced ability to deliver antigens into DCs for cross-presentation to T cells. This may represent a natural immunological adaptation to viral infection that allows the immune system to see viral-induced cell fusion as an immunogenic event (31). Second, these results highlight the possibility that transfer of cellular antigens (of which gp100 is an example) to DCs after infections with fusogenic viruses also may play a role in the etiology of some autoimmune diseases by helping to break tolerance to self-antigens (46). Third, our data indicate that specific mechanisms must also exist in the few situations where physiological cell fusion occurs to prevent both cell killing and the immunostimulatory nature of syncytiosome release from inducing potent autoimmune disease. Foremost among such examples is the formation of the syncytiotrophoblasts involved in placental formation, where endogenous retroviral FMGs have been mechanistically implicated (47, 48). Finally, and of direct relevance to our studies here, the functional efficiency of transfer of cellular tumor antigens into DCs using syncytiosomes may allow insights into mechanisms of antigen loading and presentation that will allow simpler, cell-free methods for the generation of antigen-loaded DC cancer vaccines.

Fig. 1.

Syncytia are highly ordered structures. HT1080 cells were transiently transfected with control plasmid (A, C, and E) or FMG-expressing plasmid (B, D, and F). After 48 h, cells were stained with either an anti-tubulin antibody (C and D) or an anti-HSP60 antibody (E and F) to detect mitochondria. In all samples, nuclei were stained with DAPI (blue). HT1080 tumors, grown in nude mice, injected with recombinant adenoviruses expressing FMG (G) generate areas of syncytial formation in vivo that are absent from tumors injected with an adenovirus expressing green fluorescent protein (H; ×500 magnification).

Fig. 1.

Syncytia are highly ordered structures. HT1080 cells were transiently transfected with control plasmid (A, C, and E) or FMG-expressing plasmid (B, D, and F). After 48 h, cells were stained with either an anti-tubulin antibody (C and D) or an anti-HSP60 antibody (E and F) to detect mitochondria. In all samples, nuclei were stained with DAPI (blue). HT1080 tumors, grown in nude mice, injected with recombinant adenoviruses expressing FMG (G) generate areas of syncytial formation in vivo that are absent from tumors injected with an adenovirus expressing green fluorescent protein (H; ×500 magnification).

Close modal
Fig. 2.

Nuclear fusion is seen in developing syncytia. Cells were transfected at day 0 and followed for nuclear changes by DAPI staining. A single multinucleated syncytium is shown. Nuclear fusion (F) can be detected in syncytia involving a portion of the nuclei within a syncytium (A) or the whole syncytium (B). In a proportion of syncytia, this process ultimately results in the breakdown of the nuclear aggregate altogether and the appearance of premature chromosome condensation with the deposition of chromosomes (C). Electron microscopy confirmed the presence of multiple nuclei (N) and nuclear fusion (NF) at 3 days after transfection of FMG (E) compared with the normal nuclei of untransfected cells (D) (magnification, ×2500). The white bar in each image represents 2 μm.

Fig. 2.

Nuclear fusion is seen in developing syncytia. Cells were transfected at day 0 and followed for nuclear changes by DAPI staining. A single multinucleated syncytium is shown. Nuclear fusion (F) can be detected in syncytia involving a portion of the nuclei within a syncytium (A) or the whole syncytium (B). In a proportion of syncytia, this process ultimately results in the breakdown of the nuclear aggregate altogether and the appearance of premature chromosome condensation with the deposition of chromosomes (C). Electron microscopy confirmed the presence of multiple nuclei (N) and nuclear fusion (NF) at 3 days after transfection of FMG (E) compared with the normal nuclei of untransfected cells (D) (magnification, ×2500). The white bar in each image represents 2 μm.

Close modal
Fig. 3.

Cells undergoing FMG-mediated syncytial formation do not show appreciable levels of activation of procaspase 3 or PARP cleavage. HT1080 or TELCeB.6 cells were transiently transfected with no DNA (N), irrelevant DNA (I), or FMG cDNA (FH). In this example, protein samples were collected at 18 and 40 h. Procaspase 3 activation typical of apoptosis is seen as a decreased signal intensity (Lane 3) compared with untreated control cells (Lane 1) undergoing apoptosis in response to FasL. Small reductions in the signal intensity, such as that seen in Lane 6 relative to the controls, were not reproducible over multiple experiments. Apoptosis-associated PARP cleavage is indicated by the appearance of a positive signal at Mr 85,000 as seen in cells undergoing apoptosis in response to treatment with VP16 or Fas (Lanes 2 and 3). Only minimal levels of the PARP cleavage product (see Lane 13) were ever observed in FMG-transfected cells at these or later time points of the cell killing process, which may be a residual effect of syncytial formation.

Fig. 3.

Cells undergoing FMG-mediated syncytial formation do not show appreciable levels of activation of procaspase 3 or PARP cleavage. HT1080 or TELCeB.6 cells were transiently transfected with no DNA (N), irrelevant DNA (I), or FMG cDNA (FH). In this example, protein samples were collected at 18 and 40 h. Procaspase 3 activation typical of apoptosis is seen as a decreased signal intensity (Lane 3) compared with untreated control cells (Lane 1) undergoing apoptosis in response to FasL. Small reductions in the signal intensity, such as that seen in Lane 6 relative to the controls, were not reproducible over multiple experiments. Apoptosis-associated PARP cleavage is indicated by the appearance of a positive signal at Mr 85,000 as seen in cells undergoing apoptosis in response to treatment with VP16 or Fas (Lanes 2 and 3). Only minimal levels of the PARP cleavage product (see Lane 13) were ever observed in FMG-transfected cells at these or later time points of the cell killing process, which may be a residual effect of syncytial formation.

Close modal
Fig. 4.

Cytochrome c remains within mitochondria in FMG-mediated syncytia formation. Tumor cells were transfected with control plasmid (left panels) or FMG plasmid (right panels). Cells shown were stained at 48 h for cytochrome c (FITC), mitochondria (phycoerythrin), and DAPI (blue). Integrity of the mitochondrial associated cytochrome c was also seen at later time points when syncytial death was observed by trypan blue.

Fig. 4.

Cytochrome c remains within mitochondria in FMG-mediated syncytia formation. Tumor cells were transfected with control plasmid (left panels) or FMG plasmid (right panels). Cells shown were stained at 48 h for cytochrome c (FITC), mitochondria (phycoerythrin), and DAPI (blue). Integrity of the mitochondrial associated cytochrome c was also seen at later time points when syncytial death was observed by trypan blue.

Close modal
Fig. 5.

Feulgen staining data indicate persistent DNA synthesis in nuclei contained within syncytia. HCT-116 cells were transfected with control plasmid (A−C) or FMG (D−F). Time points: 24 (A and D), 72 (B and E), and 96 h (C and F).

Fig. 5.

Feulgen staining data indicate persistent DNA synthesis in nuclei contained within syncytia. HCT-116 cells were transfected with control plasmid (A−C) or FMG (D−F). Time points: 24 (A and D), 72 (B and E), and 96 h (C and F).

Close modal
Fig. 6.

Tumor cells demonstrate marked vacuolation on FMG-mediated syncytia formation. A, syncytia in which nuclei are negative for propidium iodide staining also show enhanced staining for lysosomes. Cells were transfected with FMG plasmid. Forty-eight h later, they were incubated with propidium iodide (A) or LysoTracker Red (B). Propidium iodide can be seen to stain nonviable syncytia surrounding a viable syncytium (arrows) with extensive vacuoles (A). LysoTracker Red concentrates in acidic organelles of viable cells and shows enhanced staining in syncytia with vacuolation (arrows, B). These features are in keeping with autophagy occurring in these syncytia. C and D, electron micrographs of tumor cells identify large vacuolation (arrows) within syncytia produced by transient transfection of F&H genes. C, normal cells (×500). D, extensive syncytium formed after transfection with FMG and assayed at day 2 (×500). White bar, 10 μm.

Fig. 6.

Tumor cells demonstrate marked vacuolation on FMG-mediated syncytia formation. A, syncytia in which nuclei are negative for propidium iodide staining also show enhanced staining for lysosomes. Cells were transfected with FMG plasmid. Forty-eight h later, they were incubated with propidium iodide (A) or LysoTracker Red (B). Propidium iodide can be seen to stain nonviable syncytia surrounding a viable syncytium (arrows) with extensive vacuoles (A). LysoTracker Red concentrates in acidic organelles of viable cells and shows enhanced staining in syncytia with vacuolation (arrows, B). These features are in keeping with autophagy occurring in these syncytia. C and D, electron micrographs of tumor cells identify large vacuolation (arrows) within syncytia produced by transient transfection of F&H genes. C, normal cells (×500). D, extensive syncytium formed after transfection with FMG and assayed at day 2 (×500). White bar, 10 μm.

Close modal
Fig. 7.

Characterization of exosome-like vesicles produced from tumor cells undergoing FMG-mediated killing. A, light microscopy shows blebbing of tumor cells 72 h after FMG transfection and syncytia formation. B and C, electron microscopy observation of fusing (B) or control (C) tumor cells. Multiple areas of high concentrations of membrane particle release were observed in the fusing populations (arrows). D, electron microscopic observation of membrane particles purified from the centrifuged pellets of fusing tumor cells. Small vesicles of <100 nm in size were observed, some with a crescent- or cup-shaped morphology (arrow). E, Western blot analysis of purified syncytiosomes (Lane 1) or whole melanoma cell lysates (Lane 2) for gp100. F, time course for trafficking of HSP70 into purified syncytiosomes at 24 (Lane 1), 48 (Lane 2), or 72 (Lane 3) h after initiation of FMG-mediated cell killing. G, protein composition in syncytiosomes as assessed by Western blotting. Exosomes were prepared from human melanoma cells (Mel888) or human HLA-A2+ DCs and compared with syncytiosomes prepared 72 h after initiation of fusion of Mel888 cells and with whole Mel888 cell lysates. −, no signals in repeated experiments; −(?), only one weak signal relative to levels in the whole cell lysates detected from four experiments; +, consistently positive in Western blot analysis. ∗, see G above. H, normal tumor cells growing in culture express both caveolin-1 (red) and caveolin-2 (green). I, fusing tumor cells generate distinct punctate staining for caveolin-1 (red) over the surface of the syncytia (the limits of one syncytium are shown by the arrows), which is not observed other than during syncytial formation, perhaps representing the secretion of caveolin-1-enriched vesicles. J, Western blot analysis for caveolin-1 showing time course for trafficking of caveolin-1 into purified syncytiosomes at 24 (Lane 1), 48 (Lane 2), or 72 (Lane 3) h after initiation of FMG-mediated cell killing.

Fig. 7.

Characterization of exosome-like vesicles produced from tumor cells undergoing FMG-mediated killing. A, light microscopy shows blebbing of tumor cells 72 h after FMG transfection and syncytia formation. B and C, electron microscopy observation of fusing (B) or control (C) tumor cells. Multiple areas of high concentrations of membrane particle release were observed in the fusing populations (arrows). D, electron microscopic observation of membrane particles purified from the centrifuged pellets of fusing tumor cells. Small vesicles of <100 nm in size were observed, some with a crescent- or cup-shaped morphology (arrow). E, Western blot analysis of purified syncytiosomes (Lane 1) or whole melanoma cell lysates (Lane 2) for gp100. F, time course for trafficking of HSP70 into purified syncytiosomes at 24 (Lane 1), 48 (Lane 2), or 72 (Lane 3) h after initiation of FMG-mediated cell killing. G, protein composition in syncytiosomes as assessed by Western blotting. Exosomes were prepared from human melanoma cells (Mel888) or human HLA-A2+ DCs and compared with syncytiosomes prepared 72 h after initiation of fusion of Mel888 cells and with whole Mel888 cell lysates. −, no signals in repeated experiments; −(?), only one weak signal relative to levels in the whole cell lysates detected from four experiments; +, consistently positive in Western blot analysis. ∗, see G above. H, normal tumor cells growing in culture express both caveolin-1 (red) and caveolin-2 (green). I, fusing tumor cells generate distinct punctate staining for caveolin-1 (red) over the surface of the syncytia (the limits of one syncytium are shown by the arrows), which is not observed other than during syncytial formation, perhaps representing the secretion of caveolin-1-enriched vesicles. J, Western blot analysis for caveolin-1 showing time course for trafficking of caveolin-1 into purified syncytiosomes at 24 (Lane 1), 48 (Lane 2), or 72 (Lane 3) h after initiation of FMG-mediated cell killing.

Close modal
Fig. 8.

DCs loaded with exosomes derived from GALV-expressing melanoma cells are recognized by gp100-specific cytotoxic T cells. A, HLA-A2+ immature human DCs loaded with exosomes derived from Mel624 cells treated as shown were assayed for presentation of the gp100 antigen by coculture with αgp100-CTL and measurement of GM-CSF release (see “Materials and Methods”). DCs loaded with free gp100 peptide induced release of 1279 ± 54 ng/ml GM-CSF in the experiment shown here. ∗, this result is representative of three experiments, except that irradiated cells only generated GM-CSF at levels significantly higher than HSV-tk-transfected cells in one of the three experiments. B, the same experiment described in A using exosomes derived from Mel888 cells. Exosomes from either Mel624 or Mel888 were not able to induce GM-CSF release from CTLs incubated with HLA-A2−-loaded DCs (data not shown). C, the same experiment in B was carried out, but this time the HLA-A2+ human DCs were coincubated with mouse monoclonal antibody W6/32 (Neomarkers) for 24 h before the addition of the αgp100-CTL. W6/32 reacts with a monomorphic determinant of human MHC class I antigens. Incubation of the DCs with an isotype-matched (IgG2a) irrelevant antibody was carried out as a control.

Fig. 8.

DCs loaded with exosomes derived from GALV-expressing melanoma cells are recognized by gp100-specific cytotoxic T cells. A, HLA-A2+ immature human DCs loaded with exosomes derived from Mel624 cells treated as shown were assayed for presentation of the gp100 antigen by coculture with αgp100-CTL and measurement of GM-CSF release (see “Materials and Methods”). DCs loaded with free gp100 peptide induced release of 1279 ± 54 ng/ml GM-CSF in the experiment shown here. ∗, this result is representative of three experiments, except that irradiated cells only generated GM-CSF at levels significantly higher than HSV-tk-transfected cells in one of the three experiments. B, the same experiment described in A using exosomes derived from Mel888 cells. Exosomes from either Mel624 or Mel888 were not able to induce GM-CSF release from CTLs incubated with HLA-A2−-loaded DCs (data not shown). C, the same experiment in B was carried out, but this time the HLA-A2+ human DCs were coincubated with mouse monoclonal antibody W6/32 (Neomarkers) for 24 h before the addition of the αgp100-CTL. W6/32 reacts with a monomorphic determinant of human MHC class I antigens. Incubation of the DCs with an isotype-matched (IgG2a) irrelevant antibody was carried out as a control.

Close modal

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.

1

Supported by NIH Grants RO1 CA85931, RO1 CA094180, P5O CA91956 and by the Mayo Foundation.

6

The abbreviations used are: FMG, fusogenic membrane glycoprotein; DC, dendritic cell; HSVtk, herpes simplex virus thymidine kinase; GCV, ganciclovir; GALV, Gibbon ape leukemia virus; F&H, measles virus F and H proteins; GM-CSF, granulocyte macrophage colony-stimulating factor; PARP, poly(ADP-ribose) polymerase; DAPI, 4′,6- diamidino-2′-phenylindole dihydrochloride; HSP, heat shock protein; HLA, human leukocyte antigen; ZVAD-fmk, z-Val-Ala-Asp-fluoromethyl ketone; FasL, Fas ligand; VSV-G, vesicular stomatitis virus G protein.

7

E. Linardakis, A. Bateman, V. Phan, A. Ahmed, M. Gough, K. Olivier, R. Kennedy, F. Errington, K. Harrington, A. Melcher, R. Vile. Enhancing the efficacy of a weak allogeneic melanoma vaccine by viral fusogenic membrane glycoprotein-mediated tumor cell-tumor cell fusion. Submitted for publication.

Table 1

Nuclear morphometry features for HCT-116 cells after transfection with control or FMG plasmid DNA

Data are expressed as mean (±SD).

Time after transfection (h)
24487296
Nuclear area     
 Control 96.9 (23.5) 80.2 (17.0) 83.3 (32.4) 87.0 (18.6) 
 F&H 133.2 (40.9) 139.1 (39.9) 141.0 (40.2) 187.0 (58.9) 
 GALV 152.5 (32.1) 114.0 (31.3) 136.7 (54.0) 158.0 (42.2) 
DNA index     
 Control 1.47 (0.41) 1.53 (0.42) 1.35 (0.40) 1.31 (0.39) 
 F&H 1.73 (0.40) 1.85 (0.37) 2.07 (0.47) 2.18 (0.61) 
 GALV 1.81 (0.38) 1.88 (0.37) 1.92 (0.59) 1.90 (0.51) 
Average absorbance     
 Control 0.23 (0.05) 0.29 (0.06) 0.25 (0.07) 0.22 (0.04) 
 F&H 0.20 (0.05) 0.21 (0.05) 0.22 (0.03) 0.18 (0.04) 
 GALV 0.18 (0.02) 0.25 (0.06) 0.22 (0.05) 0.18 (0.03) 
Time after transfection (h)
24487296
Nuclear area     
 Control 96.9 (23.5) 80.2 (17.0) 83.3 (32.4) 87.0 (18.6) 
 F&H 133.2 (40.9) 139.1 (39.9) 141.0 (40.2) 187.0 (58.9) 
 GALV 152.5 (32.1) 114.0 (31.3) 136.7 (54.0) 158.0 (42.2) 
DNA index     
 Control 1.47 (0.41) 1.53 (0.42) 1.35 (0.40) 1.31 (0.39) 
 F&H 1.73 (0.40) 1.85 (0.37) 2.07 (0.47) 2.18 (0.61) 
 GALV 1.81 (0.38) 1.88 (0.37) 1.92 (0.59) 1.90 (0.51) 
Average absorbance     
 Control 0.23 (0.05) 0.29 (0.06) 0.25 (0.07) 0.22 (0.04) 
 F&H 0.20 (0.05) 0.21 (0.05) 0.22 (0.03) 0.18 (0.04) 
 GALV 0.18 (0.02) 0.25 (0.06) 0.22 (0.05) 0.18 (0.03) 
Table 2

Syncytia generate more exosome-associated protein than other treatments

TreatmentTotal exosome-associated protein/2 × 106 cells (μg)
GALV transfection 0.41 
Irradiation 0.17 
Mock transfection 0.09 
Osmotic shock <0.05 
Freeze thaw <0.05 
TreatmentTotal exosome-associated protein/2 × 106 cells (μg)
GALV transfection 0.41 
Irradiation 0.17 
Mock transfection 0.09 
Osmotic shock <0.05 
Freeze thaw <0.05 
Table 3

Exosomes from fusing tumor cells are effective at loading DCs with tumor-derived material

Source of exosomes% DCs labeled with tumor material
+ Fructose (20 mm)
Mock-transfected 16.3 17.8 
GALV-transfected 40.4 18.6 
Irradiated (100 Gy) 28.0 124.6 
Freeze-thawed ND ND 
   
Source of exosomes% DCs labeled with tumor material
+ Fructose (20 mm)
Mock-transfected 16.3 17.8 
GALV-transfected 40.4 18.6 
Irradiated (100 Gy) 28.0 124.6 
Freeze-thawed ND ND 
   
a

ND, Not determined.

We are grateful to Toni Higgins for excellent secretarial support and to Drs. Jeffrey Salisbury and Scott Kaufmann for advice and provision of reagents.

1
Fielding A. K., Chapel-Fernandes S., Chadwick M. P., Bullough F. J., Cosset F-L., Russell S. J. A hyperfusogenic Gibbon ape leukaemia envelope glycoprotein: targeting of a cytotoxic gene by ligand display.
Hum. Gene Ther.
,
11
:
817
-826,  
2000
.
2
Bateman A., Bullough F., Murphy S., Emiliusen L., Lavillette D., Cosset F. L., Cattaneo R., Russell S. J., Vile R. G. Fusogenic membrane glycoproteins as a novel class of genes for the local and immune-mediated control of tumor growth.
Cancer Res.
,
60
:
1492
-1497,  
2000
.
3
Diaz R. M., Bateman A., Emiliusen L., Fielding A., Trono D., Russell S. J., Vile R. G. A lentiviral vector expressing a fusogenic glycoprotein for cancer gene therapy.
Gene Ther.
,
7
:
1656
-1663,  
2000
.
4
Higuchi H., Bronk S., Bateman A., Harrington K. J., Vile R. G., Gores G. J. Viral fusogenic membrane glycoprotein expression causes syncytia formation with bioenergetic cell death: implications for gene therapy.
Cancer Res.
,
60
:
6396
-6402,  
2000
.
5
Li H., Haviv Y. S., Derdeyn C. A., Lam J., Coolidge C., Hunter E., Curiel D. T., Blackwell J. L. Human immunodeficiency virus type 1-mediated syncytium formation is compatible with adenovirus replication and facilitates efficient dispersion of viral gene products and de novo-synthesized virus particles.
Hum. Gene Ther.
,
12
:
2155
-2165,  
2001
.
6
Vile R. G., Russell S. J., Lemoine N. R. Cancer gene therapy: hard lessons and new courses.
Gene Ther.
,
7
:
2
-8,  
2000
.
7
Pardoll D. M. Paracrine cytokine adjuvants in cancer immunotherapy.
Annu. Rev. Immunol.
,
13
:
399
-415,  
1995
.
8
Melcher A. A., Todryk S., Hardwick N., Ford M., Jacobson M., Vile R. G. Tumor immunogenicity is determined by the mechanism of cell death via induction of heat shock protein expression.
Nat. Med.
,
4
:
581
-587,  
1998
.
9
Melcher A. A., Gough M. J., Todryk S., Vile R. G. Apoptosis or necrosis for tumour immunotherapy: what’s in a name?.
J. Mol. Med.
,
77
:
824
-833,  
1999
.
10
Basu S., Binder R. J., Suto R., Anderson K. M., Srivastava P. K. Necrotic but not apoptotic cell death releases heat shock proteins, which deliver a partial maturation signal to dendritic cells and activate the NF-κB pathway.
Int. Immunol.
,
12
:
1539
-1546,  
2000
.
11
Gallucci S., Lolkema M., Matzinger P. Natural adjuvants: endogenous activators of dendritic cells.
Nat. Med.
,
5
:
1249
-1255,  
1999
.
12
Savill J., Fadok V. Corpse clearance defines the meaning of cell death.
Nature (Lond.)
,
407
:
784
-788,  
2000
.
13
Gough M. J., Melcher A. A., Ahmed A., Crittenden M. R., Riddle D. S., Linardakis E., Ruchatz A. N., Vile R. G. Macrophages orchestrate the immune response to tumor cell death.
Cancer Res.
,
61
:
7240
-7247,  
2001
.
14
Kaufmann S. H., Wikstrom A. C., Gustafsson J. A., Shaper J. H. Binding of the glucocorticoid receptor to the rat liver nuclear matrix. The role of disulfide bond formation.
J. Biol. Chem.
,
261
:
11962
-11967,  
1986
.
15
Wolfers J., Lozier A., Raposo G., Regnault A., Thery C., Masurier C., Flament C., Pouzieux S., Faure F., Tursz T., Angevin E., Amigorena S., Zitvogel L. Tumor-derived exosomes are a source of shared tumor rejection antigens for CTL cross-priming.
Nat. Med.
,
7
:
297
-303,  
2001
.
16
Hengartner M. O. The biochemistry of apoptosis.
Nature (Lond.)
,
407
:
770
-776,  
2000
.
17
Leist M., Jaattela M. Four deaths and a funeral: from caspases to alternative mechanisms.
Nat. Rev. Mol. Cell. Biol.
,
2
:
589
-598,  
2001
.
18
Kitanaka C., Kuchino Y. Caspase-independent programmed cell death with necrotic morphology.
Cell Death Differ.
,
6
:
508
-515,  
1999
.
19
Denecker G., Vercammen D., Declercq W., Vandenabeele P. Apoptotic and necrotic cell death induced by death domain receptors.
Cell Mol. Life Sci.
,
58
:
356
-370,  
2001
.
20
Dunn W. A. J. Autophagy and related mechanisms of lysosome-mediated protein degradation.
Trends Cell Biol.
,
4
:
139
-143,  
1994
.
21
Liang X. H., Jackson S., Seaman M., Brown K., Kempkes B., Hibshoosh H., Levine B. Induction of autophagy and inhibition of tumorigenesis by beclin 1.
Nature (Lond.)
,
402
:
672
-676,  
1999
.
22
Clarke P. G. Developmental cell death: morphological diversity and multiple mechanisms.
Anat. Embryol. (Berl.)
,
181
:
195
-213,  
1990
.
23
Thery C., Boussac M., Veron P., Ricciardi-Castagnoli P., Raposo G., Garin J., Amigorena S. Proteomic analysis of dendritic cell-derived exosomes: a secreted subcellular compartment distinct from apoptotic vesicles.
J. Immunol.
,
166
:
7309
-7318,  
2001
.
24
Thery C., Regnault A., Garin J., Wolfers J., Zitvogel L., Ricciardi-Castagnoli P., Raposo G., Amigorena S. Molecular characterization of dendritic cell-derived exosomes. Selective accumulation of the heat shock protein hsc73.
J. Cell Biol.
,
147
:
599
-610,  
1999
.
25
Clayton A., Court J., Navabi H., Adams M., Mason M. D., Hobot J. A., Newman G. R., Jasani B. Analysis of antigen presenting cell derived exosomes, based on immuno-magnetic isolation and flow cytometry.
J. Immunol. Methods
,
247
:
163
-174,  
2001
.
26
Vile R. G., Nelson J. A., Castleden S., Chong H., Hart I. R. Systemic gene therapy of murine melanoma using tissue specific expression of the HSVtk gene involves an immune component.
Cancer Res.
,
54
:
6228
-6234,  
1994
.
27
Freeman S. M., Ramesh R., Marrogi A. J. Immune system in suicide gene therapy.
Lancet
,
349
:
2
-3,  
1997
.
28
Reiter I., Krammer B., Schwamberger G. Cutting edge: differential effect of apoptotic versus necrotic tumor cells on macrophage antitumor activities.
J. Immunol.
,
163
:
1730
-1732,  
1999
.
29
Albert M. L., Sauter B., Bhardwaj N. Dendritic cells acquire antigen from apoptotic cells and induce class I-restricted CTLs.
Nature (Lond.)
,
392
:
86
-89,  
1998
.
30
Bellone M., Iezzi G., Rovere P., Galati G., Ronchetti A., Protti M. P., Davoust J., Rugarli C., Manfredi A. A. Processing of engulfed apoptotic bodies yields T cell epitopes.
J. Immunol.
,
159
:
5391
-5399,  
1997
.
31
Matzinger P. Tolerance, danger and the extended family.
Annu. Rev. Immunol.
,
12
:
991
-1045,  
1994
.
32
Shin J. S., Abraham S. N. Cell biology. Caveolae–not just craters in the cellular landscape.
Science (Wash. DC)
,
293
:
1447
-1448,  
2001
.
33
Liu P., Li W. P., Machleidt T., Anderson R. G. Identification of caveolin-1 in lipoprotein particles secreted by exocrine cells.
Nat. Cell Biol.
,
1
:
369
-375,  
1999
.
34
Li W. P., Liu P., Pilcher B. K., Anderson R. G. Cell-specific targeting of caveolin-1 to caveolae, secretory vesicles, cytoplasm or mitochondria.
J. Cell Sci.
,
114
:
1397
-408,  
2001
.
35
Brown G., Aitken J., Rixon H. W., Sugrue R. J. Caveolin-1 is incorporated into mature respiratory syncytial virus particles during virus assembly on the surface of virus infected cells.
J. Gen. Virol.
,
83
:
611
-621,  
2002
.
36
Castellino F., Boucher P. E., Eichelberg K., Mayhew M., Rothman J. E., Houghton A. N., Germain R. N. Receptor-mediated uptake of antigen/heat shock protein complexes results in major histocompatibility complex class I antigen presentation via two distinct processing pathways.
J. Exp. Med.
,
191
:
1957
-1964,  
2000
.
37
Singh-Jasuja H., Toes R. E., Spee P., Munz C., Hilf N., Schoenberger S. P., Ricciardi-Castagnoli P., Neefjes J., Rammensee H. G., Arnold-Schild D., Schild H. Cross-presentation of glycoprotein 96-associated antigens on major histocompatibility complex class I molecules requires receptor-mediated endocytosis.
J. Exp. Med.
,
191
:
1965
-1974,  
2000
.
38
Basu S., Binder R. J., Ramalingam T., Srivastava P. K. CD91 is a common receptor for heat shock proteins gp96, hsp90, hsp70, and calreticulin.
Immunity
,
14
:
303
-313,  
2001
.
39
Buschle M., Schmidt W., Zauner W., Mechtler K., Trska B., Kirlappos H., Birnstiel M. L. Transloading of tumor antigen-derived peptides into antigen-presenting cells.
Proc. Natl. Acad. Sci. USA
,
94
:
3256
-3261,  
1997
.
40
Hirt U. A., Gantner F., Leist M. Phagocytosis of nonapoptotic cells dying by caspase-independent mechanisms.
J. Immunol.
,
164
:
6520
-6529,  
2000
.
41
Kurts C., Miller J. F. A. P., Subramaniam R. M., Carbone F. R., Heath W. R. Major histocompatibility complex class I-restricted cross-presentation is biased towards high dose antigens and those released during cellular destruction.
J. Exp. Med.
,
188
:
409
-414,  
1998
.
42
Mitchell D. A., Nair S. K., Gilboa E. Dendritic cell/macrophage precursors capture exogenous antigen for MHC class I presentation by dendritic cells.
Eur. J. Immunol.
,
28
:
1923
-1933,  
1998
.
43
Cohen-Jonathan E., Bernhard E. J., McKenna W. G. How does radiation kill cells?.
Curr. Opin. Chem. Biol.
,
3
:
77
-83,  
1999
.
44
Fernandez M., Porosnicu M., Markovic D., Barber G. N. Genetically engineered vesicular stomatitis virus in gene therapy: application for treatment of malignant disease.
J. Virol.
,
76
:
895
-904,  
2002
.
45
Eslahi N. K., Muller S., Nguyen L., Wilson E., Thull N., Rolland A., Pericle F. Fusogenic activity of vesicular stomatitis virus glycoprotein plasmid in tumors as an enhancer of IL-12 gene therapy.
Cancer Gene Ther.
,
8
:
55
-62,  
2001
.
46
Marrack P., Kappler J., Kotzin B. L. Autoimmune disease: why and where it occurs.
Nat. Med.
,
7
:
899
-905,  
2001
.
47
Blond J. L., Lavillette D., Cheynet V., Bouton O., Oriol G., Chapel-Fernandes S., Mandrand B., Mallet F., Cosset F. L. An envelope glycoprotein of the human endogenous retrovirus HERV-W is expressed in the human placenta and fuses cells expressing the type D mammalian retrovirus receptor.
J. Virol.
,
74
:
3321
-3329,  
2000
.
48
Mi S., Lee X., Li X., Veldman G. M., Finnerty H., Racie L., LaVallie E., Tang X. Y., Edouard P., Howes S., Keith J. C., Jr., McCoy J. M. Syncytin is a captive retroviral envelope protein involved in human placental morphogenesis[see comments].
Nature (Lond.)
,
403
:
785
-789,  
2000
.