TECHNICAL REPORTS Fusion of tumor cells with antigen-presenting cells (APCs) has been proposed for the preparation of cancer vaccines. However, generation of these hybrids, using physical or chemical methods such as electrofusion or polyethylene glycol (PEG), has been difficult to standardize. Characterization of cell fusion has also been problematic because of difficulties in differentiating fusion from cell aggregation, leakage of cellular dyes and dendritic-cell (DC) phagocytosis of tumor material. In this report, we describe a new method to generate hybrid cell vaccines, based on gene transfer of a viral fusogenic membrane glycoprotein (FMG) into tumor cells, and incorporate a genetic method by which true hybrid formation can be unambiguously detected. We describe a new class of tumor cell–DC hybrid that can be rapidly isolated after cell fusion. These hybrids are highly potent in in vitro antigen presentation assays, target lymph nodes in vivo and are powerful immunogens against established metastatic disease. The fusion of tumor cells with APCs has been proposed as an efficient method to render antigenic tumor cells highly immunogenic, as the hybrids will express a full repertoire of tumor antigens along with the antigen-presenting machinery of the APC 1 . Hybrid cell vaccines have shown considerable promise in a variety of preclinical model systems 1 and clinical trials 2,3 . Hybrid cell formation has thus far relied on the use of chemical reagents such as PEG 4,5 or physical techniques such as electrofusion 6 . In both of these methodologies, however, the nature of the cell fusion event is difficult to characterize and standardize. Fluorescence-activated cell sorting (FACS) analysis can be used to determine the percentage of fusion within the cultures using coalescence of fluorescent dyes or costaining with tumor cell–derived and APC-derived markers 7 . However, this can lead to false positivity because of leakage of dyes between cells and phagocytosis by unfused DCs of debris from dying tumor cells. In addition, accurately distinguishing between aggregated, as opposed to fused, cells by FACS and microscopy can be problematic. 1 Molecular Medicine Program, Mayo Clinic, 200 First Street SW, Rochester, Minnesota 55905, USA. 2 CRUK Oncology Unit, St. James’ Hospital, Leeds LS9 7TF, UK. 3 Université Libre de Bruxelles, 6041 Gosselies, Belgium. 4 AstraZeneca, Alderley Park, Cheshire, SK10, UK. 5 Department of Otorhinolaryngology, Mayo Clinic, 200 First Street SW, Rochester, Minnesota 55905, USA. 6 INSERM U517, Faculty of Medicine, 21079 Dijon Cedex, France. 7 Department of Immunology, Mayo Clinic, 200 First Street SW, Rochester, Minnesota 55905, USA. Correspondence should be addressed to R.G.V. (vile.richard@mayo.edu). Published online 17 August 2003; doi:10.1038/nm923 A new genetic method to generate and isolate small, short-lived but highly potent dendritic cell–tumor cell hybrid vaccines Vy Phan 1 , Fiona Errington 2 , S Chiat Cheong 3 , Tim Kottke 1 , Michael Gough 1 , Sharon Altmann 1 , Annick Brandenburger 3 , Steve Emery 4 , Scott Strome 5 , Andrew Bateman 1 , Bernard Bonnotte 6 , Alan Melcher 2 & Richard Vile 1,7 NATURE MEDICINE VOLUME 9 | NUMBER 9 | SEPTEMBER 2003 1215 Our previous work has shown that viral FMGs can be used as effec- tive cytotoxic genes for treatment of local 8,9 and systemic 8,10,11 disease. These viral proteins normally mediate the fusion of the viral particle with the target cell’s membrane by binding to specific receptors on the target cell. We hypothesized that we might be able to exploit the hyper- fusogenicity of FMGs to mediate fusion between tumor cells and DCs and thus generate hybrid cell vaccines. We describe here a new genetic method by which tumor cell–DC hybrids can be prepared for use in vaccination, as well as a new class of hybrids that present tumor anti- gens to T cells, migrate from subcutaneous injection sites to draining lymph nodes and are potent immunogens in vivo. RESULTS A genetic assay for fusion To develop an assay that genuinely detects cell fusion between DCs and tumor cells, we constructed a retroviral vector in which the melanoma- specific mouse tyrosinase promoter was inserted into the 3′ long terminal repeat (LTR) 12,13 (Fig. 1a). Upon infection of mouse dendritic cells, the 3′ LTR is reverse-transcribed into the 5′ LTR of the provirus, which drives expression of a downstream gene 12,13 (Fig. 1a). The melanoma-specific promoter, which is integrated in the genome of the DC, is normally silent, but upon genuine fusion between the DC and a melanoma cell, transcription factors from the melanoma cell should become available to transactivate the promoter. To investigate the in vivo properties of tumor cell–DC fusion interactions, we used a mouse model system in which the vesicular stomatitis virus (VSV)-G FMG is used to fuse B16 melanoma cells 10 to DCs. Populations of DCs infected with the Tyr–green fluores- cent protein (GFP) reporter virus (Tyr-GFP DCs; Fig. 1a) do not express detectable GFP, nor do DCs cocultured with B16 melanoma cells in the absence of FMG expression (Fig. 1b).When Tyr-GFP DCs are cocultured with tumor cells engineered to express VSV-G, however, GFP-expressing cells can be detected by FACS analysis (Fig. 1b). Only true hybrids, in which there is genuine cell membrane fusion and mixing of cytoplasm, would be able to induce expression of the GFP marker gene from the © 2003 Nature Publishing Group http://www.nature.com/naturemedicine