Cell-Cell Fusion Induced by Measles Virus Amplifies the Type I Interferon Response

Abstract

Measles virus (MeV) infection is characterized by the formation of multinuclear giant cells (MGC). We report that beta interferon (IFN-β) production is amplified in vitro by the formation of virus-induced MGC derived from human epithelial cells or mature conventional dendritic cells. Both fusion and IFN-β response amplification were inhibited in a dose-dependent way by a fusion-inhibitory peptide after MeV infection of epithelial cells. This effect was observed at both low and high multiplicities of infection. While in the absence of virus replication, the cell-cell fusion mediated by MeV H/F glycoproteins did not activate any IFN-α/β production, an amplified IFN-β response was observed when H/F-induced MGC were infected with a nonfusogenic recombinant chimerical virus. Time lapse microscopy studies revealed that MeV-infected MGC from epithelial cells have a highly dynamic behavior and an unexpected long life span. Following cell-cell fusion, both of the RIG-I and IFN-β gene deficiencies were trans complemented to induce IFN-β production. Production of IFN-β and IFN-α was also observed in MeV-infected immature dendritic cells (iDC) and mature dendritic cells (mDC). In contrast to iDC, MeV infection of mDC induced MGC, which produced enhanced amounts of IFN-α/β. The amplification of IFN-β production was associated with a sustained nuclear localization of IFN regulatory factor 3 (IRF-3) in MeV-induced MGC derived from both epithelial cells and mDC, while the IRF-7 up-regulation was poorly sensitive to the fusion process. Therefore, MeV-induced cell-cell fusion amplifies IFN-α/β production in infected cells, and this indicates that MGC contribute to the antiviral immune response.

Measles virus (MeV) is an important human pathogen responsible for ∼345,000 deaths in 2005 (http://www.who.int/mediacenter/factsheets/fs286/en/). However, most humans clear this viral infection provided that they have functional cellular and adaptive immunity (19). MeV infection begins in the respiratory tract and then spreads to local lymphoid tissues, where virus replication can occur in macrophages and possibly conventional dendritic cells (DC) (cDC) (14, 17, 22, 49, 70). This allows MeV to spread to other lymphoid organs and to the whole body. MeV induces a cytopathic effect characterized by the fusion of neighboring cells into multinucleated giant cells (MGC). In vivo, 7 days after MeV infection of rhesus monkeys, MGC are found in the respiratory and genitourinary tracts as well as in the esophagus and skin epithelia (47). In addition, a specific subset of infected MGC, called Warthin-Finkeldey cells (WFC), initially described in infants dying of acute measles, is found in primary and secondary lymphoid organs (19). WFC are usually observed in germinal centers and interfollicular areas of secondary lymphoid organs and in the thymus. They are heterogeneous and display either B- or T-cell markers (54), although macrophage and DC markers have not yet been investigated. In vitro, MeV infection of human cells, including primary epithelial cells and cDC, induces the formation of MGC, also referred to as syncytia (17, 78, 82). MeV-induced cell-cell fusion is governed by the interaction of the viral envelope H and F glycoproteins with the cellular receptors CD46 and CD150, which are expressed ubiquitously and solely on immune-activated cells, respectively (19).

The hallmark of the immune response to a viral infection is the rapid production of a range of cytokines, most prominently type I interferon (IFN) (alpha/beta IFN [IFN-α/β]). IFN-α/β enables cells to be protected against viral infection via pleiotropic activities such as the inhibition of protein synthesis and cell proliferation and the enhancement of apoptosis of infected cells (reviewed in references 5 and 21). They also activate natural killer (NK) cells and cytotoxic T lymphocytes (CTL) that are capable of eliminating the viral pathogen by killing infected cells. IFN-α/β can act directly on CTL or indirectly by inducing the maturation of cDC, which facilitates the cross-presentation of viral antigens to CTL (42, 43). The single IFN-β gene and most of the IFN-α genes differ in their promoter regions; the former is activated by IFN regulatory factor 3 (IRF-3) and IRF-7 heterodimers (56), whereas only IRF-7 homodimers (25) and/or IRF-7/IRF-8 heterodimers, as recently reported for mouse DC (75), can activate the latter. The recognition of peculiar danger molecular motifs of viruses is mediated by host pattern recognition receptors (PRR) (36), which can recognize virus nucleic acids. Two groups of PRR are involved in IFN production in DC, Toll-like receptors (TLR) and RIG-like receptors (RLR). To date, TLR are mainly responsible for IFN-α production by plasmacytoid DC (pDC) via TLR7 and TLR9 (1), which respond to viral nucleic acids within the endosomal compartment. Nucleic acid recognition results in the activation of the IFN-α genes through the phosphorylation and nuclear translocation of IRF-7 (26, 37). Indeed, in comparison with other cells, pDC express high levels of IRF-7 (8, 30), and this allows pDC to produce 100- to 1,000-fold more IFN-β and IFN-α than other cell types (26). TLR are also involved in IFN production by cDC via TLR3 after phagocytosis of infected cells (64). However, the IFN response in cDC relies principally on RLR (RIG-I and MDA-5) (34, 35), which are cytosolic PRR expressed in almost all nucleated cells and dedicated to respond to viral nucleic acids produced during replication (28, 60, 87). MDA-5 and RIG-I recognized double-stranded RNA (20) and 5′-triphosphate-ended RNA (28, 60, 62), respectively. Accordingly, they recognize different types of RNA viruses, with MDA-5 being activated by the Picornaviridae and RIG-I being activated by the Flaviviridae, Orthomyxoviridae, and Mononegavirales (20, 35). MeV activates RIG-I, which recognizes the 5′-triphosphate end of the small RNA leader transcript (62). The activation of RLR induces the interaction with an adaptor protein called IPS-1 (also known as MAVS, VISA, and CARDIF) that leads to NF-κB, IRF-3, and IRF-7 phosphorylation and their transient translocation into the nucleus, resulting in early IFN-β gene induction (34, 57). In contrast to the short-living IRF-7, the more stable IRF-3 is highly expressed in all cells (25). IFN-β is then secreted and binds to the IFN-α receptor (IFNAR) to produce late IFN-β and all IFN-α subtypes. Indeed, early IFN-β induces the transcription of numerous genes known as IFN-stimulated genes (25), including IRF-7. The activation of this supplemental source of IRF-7 allows the delayed production of a boost of both IFN-β and IFN-α, according to a robust positive-feedback loop, which amplifies the antiviral response (25). The crucial role of the RLR in the IFN-α/β response and control of viral replication has recently been highlighted using IPS-1 (or MAVS)-deficient mice, which show normal IFN-α secretion by pDC (41, 74).

In the case of MeV, IFN-α/β production has been found after in vitro infection of various human cell types including epithelial cells (from various tissues), endothelial cells, glial cells, and peripheral blood mononuclear cells (52, 84, 85). MeV propagates more efficiently in mature cDC (mDC) than in immature cDC (iDC) and induces higher level of MGC formation in mDC (17, 66). Given the ability of MeV to induce MGC formation in the epithelium and secondary lymphoid organs, we aimed at investigating, in both human epithelial cells and cDC, the role of MeV-induced cell-cell fusion in the regulation of IFN-α/β production.

MATERIALS AND METHODS

Reagents.

Antibodies used were monoclonal antibodies (MAbs) clone 55 (anti-H), clone 25, and clone 120 (anti-N); G28.5 (anti-CD40); and rabbit polyclonal anti-human IRF-3. The following reagents were used: Hoechst 33342 (Sigma), fusion-inhibiting peptide (FIP) Z-D-Phe-L-Phe-Gly-OH (Neosystem), recombinant human granulocyte-macrophage colony-stimulating factor and recombinant human interleukin-4 (generously provided by Schering-Plough), recombinant human IFN-β (Calbiochem, San Diego, CA), and Draq5 (Alexis Biolabs).

Cells and phenotypic analysis.

Human kidney epithelial 293T/17 cells (ATCC) expressing CD46 (293T/CD46⁺), HeLa cells, African green monkey Vero fibroblasts (ATCC), 293T/CD150⁺ cells (77), Huh7.5 cells, a subline of Huh7 cells defective in RIG-I (73), and human cortical thymic epithelial cells (TEC) (clone P1.4D6) from postnatal thymus (16) were maintained in Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum or as described previously (16, 73). Monocyte-derived DC were generated in vitro from human blood (Etablissement de Transfusion Sanguine, Lyon, France) as previously described (17). After 6 days of culture in 200 ng/ml recombinant human granulocyte-macrophage colony-stimulating factor and 10 ng/ml recombinant human interleukin-4, >95% of the cells were iDC. The mDC were derived by treating iDC for 48 h with 10 μg/ml of anti-CD40 mAb. cDC phenotyping was determined as previously described (17, 82).

MeV infection.

MeV (Hallé strain) and recombinant chimeric MGV (72) were maintained in Vero cells. For MeV infection experiments, cells were seeded overnight at the appropriate density in a 24-well plate according to the duration of the observation to avoid cell crowding. 293T cells and TEC were infected at a multiplicity of infection (MOI) of 1, unless otherwise indicated, and then treated or not treated with FIP. DC were infected with MeV at an MOI of 0.1 as previously detailed (17, 83). As controls, all cells were pulsed with a mock preparation corresponding to uninfected Vero cell supernatant. In some experiments, FIP (100 μg/ml) was added to the cDC cultures infected or not infected with MeV.

Transient cell transfections.

A total of 4 × 10⁵ 293T/CD46⁺ cells were plated onto six-well plates (BD, Falcon). The next day, cells were transfected with either 1 μg pCXN2-F (expressing Edm-F) plus 1 μg pCXN2-EdH (expressing the CD46-binding H^Ed), 1 μg pCXN2-KAH (expressing the H^KA from wild-type MeV isolate KA, which does not bind to CD46) (79), or 2 μg of pBSK(+) carrier plasmid (Stratagene). All transfections were performed using Lipofectamine (Invitrogen) or Dreamfect (OZ Biosciences) according to the manufacturer's instructions.

Time lapse.

A total of 2 × 10⁶ 293T/CD46⁺ cells were plated onto a six-well plate and infected with MeV at an MOI of 1 as described above. After 24 h of culture in the presence of 10 μg/ml FIP, FIP was removed, and the plate was placed at 37°C and heated in a 5% CO₂ atmosphere (Carl Zeiss, Jena, Germany). Cells were imaged by Metamorph software v6 with a Coolsnap HQ monochrome camera associated with a time lapse microscope (Axiovert 100 M) and a 10× (numerical aperture, 0.25) Plan-Apochromat objective (Zeiss). Meta Imaging Series 4.5 (Universal Imaging, West Chester, PA) was used to make Quick-Time movies from image stacks from metamorph software. One picture was made every 10 min for 60 h, and every second of movie represents 235.4 min (3.92 h) of culture (see Fig. S3 in the supplemental material). Images extracted from stacks were processed with Adobe Photoshop 6.0 (Adobe Systems, San Jose, CA).

Coculture experiments.

A total of 8 × 10⁴ Huh7.5 and Vero cells were plated into 24-well plates (Costar) and were infected, 24 h later, with MeV at an MOI of 1. At 8 h postinfection (p.i.), Vero cells were trypsinized and added to the Huh7.5 cell monolayers at a 1:1 ratio in the presence or absence of 10 μg/ml of FIP. Cell-free supernatants were harvested at different times and frozen before analysis for IFN-α/β using a biological assay.

IFN-α and IFN-β detection assay.

IFN-α/β contents in supernatants were determined using a bioassay as detailed elsewhere previously (84). IFN-α and IFN-β levels were determined by enzyme-linked immunosorbent assay (ELISA) using the IFN-α kit from Bender MedSystems (detection limit, 8 pg/ml) and the IFN-β kit from PBL Laboratories (detection limit, 250 pg/ml), respectively.

RNA extraction, cDNA reverse transcription, and real-time quantitative PCR analysis.

A detailed procedure for viral RNA quantification was reported previously (61). Primer sets for human IFN-β and IRF-7 mRNA quantitation were TGGGAGGATTCTGCATTACC (forward) and CAGCATCTGCTGGTTGAAGA (reverse), respectively. The primer sets were purchased from Search-LC (Heidelberg, Germany). Results were normalized according to amounts of 18S rRNA and expressed in mRNA copy number/25 ng of total RNA.

MTT colorimetric bioassay.

A total of 4 × 10³ 293T/CD46⁺ cells were plated into a 96-well plate (Costar). At 30 h after infection or transfection, cells were treated with 225 ng/well of 3-(4,5-dimethyl-2-thiazolyl)-2,5-diphenyl-2H-tetrazolium bromide (MTT; Sigma) to measure mitochondrial activity of metabolically active cells. Four hours later, the supernatant was removed, and the cells were lysed with 100 μl/well dimethyl sulfoxide containing 0.04 N HCl. Absorbance was then measured at 490 and 650 nm (18).

Quantitative fusion assay.

The quantitative fusion assay is based on the conditional expression of β-galactosidase (β-Gal) under the control of the T7 polymerase promoter and was performed as previously described (7).

Subcellular localization of GFP-IRF-3 proteins.

A total of 5 × 10⁴ 293T/CD46⁺ cells, seeded into 24-well plate, were infected or not infected with MeV at an MOI of 1 for 2 h and then transfected by the green fluorescent protein (GFP)-IRF-3 expression plasmid (46) using Lipofectamine. Twenty-four to 48 h later, GFP fluorescence in living cells was analyzed with a Leica DM IRB microscope at a magnification of ×400. The percentage of IRF-3-labeled nuclei was calculated by counting, within a microscope field, the total number of nuclei (belonging to mononuclear cells or MGCs [labeled and unlabeled]) and the number of nuclei labeled with GFP-IRF-3. For each condition, >100 nuclei were counted.

Stainings.

For nuclear staining, cell monolayers were stained with Hoechst 33342 dye as detailed previously (82), and the stained nuclei were observed using a Leica DM IRB microscope.

MeV H-cell surface expression was detected using mAb cl55 plus goat anti-mouse immunoglobulin G phycoerythrin-conjugated secondary antibody (Jackson ImmunoResearch Laboratories) as previously described (82). MeV N intracellular detection was performed using a Cytofix/Cytoperm kit (Becton Dickinson, Pharmingen), and 293T/CD46⁺ cells were labeled using biotinylated mAb cl25 plus streptavidin-phycoerythrin (Caltag Laboratories) as previously described (82). After labeling, cells were analyzed using a FACSCalibur flow cytometer (Becton Dickinson Cellquest software). Integrated fluorescence was measured, and data were collected from at least 5,000 events.

Triple stainings were performed to visualize GFP-IRF-3 (green), MeV N protein (red), and nuclei (blue) by confocal microscopy assays. A total of 2.5 × 10⁴ 293T/CD46⁺ cells were seeded onto precoated poly-l-lysine (10 μg/ml overnight at 4°C; Sigma) glass coverslips in 24-well plates and incubated for 20 min at 37°C. Twelve hours later, cells were infected with MeV at MOIs of 0.1, 1, 2, and 4, followed by transfection of the GFP-IRF-3 expression plasmid as described above, and were then cultured in the absence or presence of FIP at 10 μg/ml. Forty-eight hours later, cells were fixed in 2% paraformaldehyde-phosphate-buffered saline (PBS) for 20 min at room temperature, treated with 0.1% glycine-PBS for 10 min at room temperature, and permeabilized with 0.5% Triton X-100-PBS for 5 min at room temperature. After washes with PBS, the fixed cells were blocked in a solution containing bovine serum albumin, human and goat sera, and Triton X-100-PBS overnight at +4°C. Cells were then incubated with anti-N mAb cl120 for 90 min at +4°C. Cells were washed three times for 5 min in PBS before incubation with goat anti-mouse immunoglobulin G-Alexa 568 for 30 min at +4°C. After three washes in PBS, cells were mounted onto glass slides with mounting medium (Dako) containing Draq5 as a nuclear marker. Labeled cells were imaged with a confocal microscope (1 μm; Zeiss LSM510) using a zoomed (×2) 63× (numerical aperture, 1.4) PlanFluor objective. To prevent cross-contamination between fluorochromes, each channel was imaged sequentially using a multitrack recording module before merging.

Double staining was performed to visualize endogenous IRF-3 (green) and nuclei (blue) by Axioplan2 Imaging microscopy assays. Briefly, 8 × 10⁴ mDC were seeded onto precoated poly-l-lysine glass coverslips in 24-well plates and incubated for 20 min at 37°C. Twenty-four hours later, cells were infected at an MOI of 0.1 and then cultured in the absence or presence of FIP at 100 μg/ml. At 3 days p.i., cells were then fixed as described above for triple staining. The fixed cells were then blocked in bovine serum albumin-serum-Triton X-100-PBS for 1 h at +4°C and incubated with rabbit anti-IRF-3 serum for 1 h at +4°C. Cells were washed three times for 5 min in PBS before incubation with a goat anti-rabbit immunoglobulin G Alexa 488-conjugated secondary antibody (Molecular Probes) for 30 min at +4°C. After three washes in PBS, cells were incubated with 1 μg/ml of Hoechst 33343 stain for 15 min at room temperature before being mounted onto glass slides with mounting medium (Dako). Labeled cells were analyzed with an Axioplan2 Imaging microscope (0.3 μm; Zeiss) and then imaged by Metamorph software V6. Images extracted from z stacks were processed with Adobe Photoshop 6.0 software.

RESULTS

MeV-induced cell-cell fusion amplifies IFN-β activation in epithelial cells.

While MeV induced a high number of syncytia containing numerous nuclei in cells expressing both MeV receptors (293T/CD46⁺ CD150⁺), syncytia were smallest and less numerous in 293T/CD46⁺ cells expressing only CD46 (supplemental data may be found at hal.archives-ouvertes.fr /hal-00169132). MeV infection of 293T/CD46⁺ cells led to the accumulation of F viral transcripts and to the induction of IFN-β gene expression (supplemental data may be found at hal.archives-ouvertes.fr/hal-00169132). The accumulation of viral F mRNA and the transcription of the IFN-β gene were both enhanced in 293T/CD46⁺ CD150⁺ cells compared to 293T/CD46⁺ cells (data not shown). Thus, the level of IFN-β activation was correlated to virus transcription and/or to cytopathic effects.

To decipher how each of these factor could influence the IFN response, we took advantage of the availability of an FIP. FIP efficiently blocks MeV-induced syncytium formation without inhibiting the cell-to-cell virus spreading (19). The addition of FIP repressed both syncytium formation and IFN-α/β production in an MeV-infected 293T/CD46⁺ cell culture without affecting the proportion of cells expressing MeV N protein (Fig. 1A, left). Whereas the fusion was totally inhibited by FIP, IFN-α/β production by day 3 (not shown) and day 7 p.i. (Fig. 1A, left) was inhibited by ∼85 to 96%. This residual IFN-α/β production was significant, since no IFN-β could be detected in uninfected cells (data not shown). However, 293T cells expressing only MeV Ed-H and F glycoproteins readily fused into MGC but did not secrete any detectable IFN-α/β (data not shown). This indicates that cell-cell fusion, per se, does not activate the IFN-α/β response.

FIG. 1.

Role of cell-cell fusion and MeV infection in the IFN-β response. (A) Syncytium formation (scored as a minus sign [−] for no syncytium and from + to 6+, graded by integrating the numbers and sizes of live adherent syncytia plus floating dead syncytia) percentage of cells expressing the MeV N protein determined by flow cytometry, and IFN-α/β production at 7 days p.i. of 293T/CD46⁺ cells with MeV (left) or recombinant chimerical MGV (right) in the absence (dotted columns) or presence (black columns) of 10 μg/ml of FIP. (B) Syncytium formation and accumulation of MeV N (top) and IFN-β (bottom) transcripts at 23 h p.i. in 293T/CD46⁺ cells transiently expressing H^EdF (grid columns) or H^KAF (checked columns), respectively. 293T/CD46⁺ cells were infected with MGV at different MOIs, and they were then transfected at 2 h p.i. (C) Dose-response relationship of MeV F transcription and IFN-β mRNA accumulation at 30 h p.i. and IFN-α/β secretion at 3 days p.i. with the MOIs of MeV used to infect 293T/CD46⁺ cells in the absence (dotted columns) or presence (black columns) of 10 μg/ml FIP. Data are mean values ± standard deviations (SD) from two to three independent experiments. ND, not detected. † indicates cell cytoxicity.

A recombinant chimeric virus, MGV, where the H and F genes have been substituted with that of the vesicular stomatitis virus G glycoprotein (72) infected and propagated more slowly in 293T/CD46⁺ cells without inducing syncytium formation. Accordingly, it induced a low level of IFN-β production, which was insensitive to FIP (Fig. 1A, right) and equivalent to that observed in FIP-treated MeV-infected 293T/CD46⁺ cells. Thus, MGV triggered only a basal level of IFN-α/β response, in agreement with the observation of the low IFN-α/β-inducing ability of a nonfusogenic MeV variant (48). To confirm the enhancing effect of cell-cell fusion on the MeV-induced IFN-β response, we compared the effect of expressing a fusing (H^EdF) and a nonfusing (H^KAF) (79) glycoprotein combination in trans on the IFN-β response induced by MGV infection. As expected, the expression of H^KAF affected neither virus nor IFN-β transcription (not shown). In contrast, compared to H^KAF, the expression of H^EdF induced large syncytia into 293T/CD46⁺ cells, a minor increase in viral transcription, and a significant increase in IFN-β gene activation (Fig. 1B). Furthermore, the enhancing effect on the IFN-β response was much more pronounced under conditions ensuring that every single got infected, i.e., at an MOI of 4, with a ∼25-fold enhancement of IFN-β mRNA accumulation compared to a limited 2.5-fold increase in virus transcript accumulation.

When 293T/CD46⁺ cells were infected with fusogenic MeV at an MOI ranging from 0.01 up to 4 and analyzed at 30 h p.i., the viral transcription of the F messenger exhibited a dose-response curve between MOIs of 0.01 and 1 and then reached a plateau (Fig. 1C, top). The identical levels of viral transcription between MOIs of 1 and 4 suggested that some viral interference occurred. In addition, at an MOI higher than 1, a cytotoxicity, increasing with the MOI used, was observed, which likely resulted in part from cell fusion from without (i.e., fusion between adjacent cells bridged by viral particles) (6), a reminder of virally induced hemolysis of CD46-expressing Vervet monkey red blood cells (59). Furthermore, syncytium formation was much reduced compared to that induced by a lower MOI, likely because of the strong down-regulation of CD46 upon contact with the large amount of hemagglutinin brought about by the high viral inoculum (39, 51). The level of IFN-β transcription followed the same dose response between MOIs of 0.01 and 1 to reach a plateau at MOIs of 1, 2, and 4 (Fig. 1C, middle). This correlation between viral transcription and IFN-β gene activation agrees with their parallel kinetics observed at an MOI of 1 (62). Surprisingly, the production of IFN-α/β in the supernatant measured at 3 days p.i. was almost identical between MOIs of 0.1 and 4 (Fig. 1C, bottom). At an MOI of 0.01, a small amount of IFN-α/β was detected only later, at 7 days p.i. (data not shown). The addition of FIP, which inhibited the formation of syncytia at all MOIs (Fig. 1C), had minimal effects on viral transcription but strongly inhibited IFN-β gene transcription and protein secretion (Fig. 1C, black columns), except at an MOI of 4. Interestingly, in the presence of FIP, the amount of IFN released into the supernatant and the MOI correlated for MOIs between 0.1 and 2 (Fig. 1C and data not shown). At an MOI of 0.01, the level of released IFN detected at 7 days p.i. was also inhibited by FIP (not shown). The lack of an FIP effect at an MOI of 4 was likely reflecting the side effects of the too-high viral load mentioned above.

Furthermore, after infection at an MOI of 1, FIP inhibition of the cell-cell fusion observed at 30 h p.i. and 3 days p.i. was dose dependent, as assessed by nucleus staining with Hoechst 33342 (Fig. 2A and data not shown) and the quantification assay using β-Gal as a reporter gene for intercellular fusion (Fig. 2B). A similar inhibition curve was also observed for IFN-β mRNA (Fig. 2C and not shown). Strikingly, the best mathematical equation describing these two dose-dependent responses had similar slopes (−11) and ordinates at the origin (+42 and +39).

FIG. 2.

Similar dose-dependent inhibition of cell-cell fusion and IFN-β gene transcription by FIP. 293T/CD46⁺ (A and C) or HeLa (B) cells were infected with MeV at an MOI of 1 prior to the addition of increasing amounts of FIP. (A) Micrographs of adherent cells stained with Hoechst 33342 (magnification, ×400) at 72 h p.i. Cells containing more than three nuclei were considered to be syncytia (white arrows). (B) Dose-dependent inhibition of cell-cell fusion by FIP quantified by colorimetric β-Gal reporter gene expression assay. OD, optical density. (C) Dose-dependent inhibition of IFN-β mRNA accumulation by FIP at 30 h p.i. Data are means ± SD of triplicate experiments.

In summary, MeV replication triggers a basal IFN-β response independently of H/F proteins, and MeV-induced cell-cell fusion robustly amplifies this response in a dose-dependent manner at both low and high MOIs.

High morphological plasticity of MeV-induced MGC.

Virally induced fusion is usually correlated with apoptosis (13). As we observed an increased synthesis of IFN-β in MGC, we further analyzed the morphological plasticity of these cells. Thirty hours p.i., syncytia were found to be metabolically active and able to convert MTT into intracellular formazan crystals, a hallmark of mitochondrial activity in viable epithelial cells (supplemental data may be found at hal.archives-ouvertes.fr/hal-00169132).

Studies by time lapse microscopy of MeV-infected 293T/CD46⁺ cells over 60 h showed the syncytia to be dynamic, exhibiting a morphology that varied with time (Fig. 3A and B) (see Fig. S3A in the supplemental material). In the first stage, the initial flat adherent syncytium increased in size and nucleus contents (Fig. 3A). Dynamic pseudopodia emerged from the syncytium to contact surrounding cells or syncytia. In a second stage, the adherent syncytium retracted into highly refringent smooth balls (third stage), where nuclei were no longer visible, except upon examination under confocal microscopy of z stacks after Hoechst staining (see below). Balls were highly mobile and rolled around. When they encountered surrounding healthy adherent mononuclear cells, they spread out into a secondary flat adherent syncytium with visible nuclei (fourth stage). The secondary adherent syncytium then retracted again (fifth stage) into an irregular ball with protruding blisters, giving it a cauliflower appearance (sixth stage), which rolled around. The duration of each stage was highly variable (see mean values in Fig. 3B), and most of the initial syncytia passed through stages 1, 2, and 3; half of them passed through stages 4 and 5 to reach stage 6; and another half directly passed from stage 3 to stage 6 (Fig. 3B). Of 17 initial syncytia recorded during three different experiments, none appeared to die before 60 h. Furthermore, when individual smooth or blistered balls were transferred onto a fresh uninfected 293T/CD46⁺ monolayer, they readhered. This suggests that syncytia may have an indefinite life span provided that they find a healthy cell monolayer within their vicinity (Fig. 3B and data not shown). In contrast, when transferred onto a plastic dish covered or not covered by collagen, both smooth and blistered balls became senescent and finally died, becoming floating and optically clear bubble-like structures (data not shown). Thus, MeV-induced syncytia are not prone to dying quickly; instead, they may remain a viable entity. As controls, cell-to-cell fusion was observed in neither uninfected nor FIP-treated MeV-infected 293T/CD46⁺ cells, indicating that none of the observed syncytia was due to merging senescent 293T/CD46⁺ cells (Fig. 3C and data not shown).

FIG. 3.

MeV-induced syncytia are dynamic entities with an extended life span. MeV-infected 293T/CD46⁺ cells were cultured overnight in the presence of FIP and then cultured in the absence (A and B) or the presence (C) of FIP (10 μg/ml), with recorded imaging for the next 60 h by time lapse microscopy. Images at a magnification of ×10 (numerical aperture, 0.25) were extracted from Fig. S3A in the supplemental material and another video not shown at 12.83 h, 15 h, 16.66 h, 25.33 h, 32.5 h, and 35.33 h. (B) The duration of each stage was evaluated and expressed as means ± SD of 17 microscopic areas from three to four separate experiments. The frequency was estimated and indicated as the proportion (percent) that underwent transition through a given stage.

Cell-cell fusion brings together both the danger activation signal and the IFN-β gene to trigger IFN-β induction.

We then studied whether MeV-induced fusion can trigger trans complementation using Huh7.5 cells and Vero cells that have disabled RIG-I (73) and lack the IFN-β gene (11, 12), respectively. IFN-α/β was secreted and accumulated over time from both of the MeV-Huh7.5/Vero and MeV-Vero/Huh7.5 cell coculture combinations, as syncytia were observed (Fig. 4). Moreover, the addition of FIP completely blocked both cellular fusion and IFN-α/β production to undetectable levels. As expected, MeV infection of isolated Huh7.5 or Vero cells induced syncytia but did not trigger any IFN-α/β response. Thus, the RIG-I defect in human Huh7.5 cells and the IFN-β gene defect in the simian Vero cells could be trans-complemented in fused cells, allowing the triggering of the human IFN-β gene.

FIG. 4.

Reciprocal trans-complementation of RIG-I- and IFN-β-deficient cells by MeV-induced fusion. RIG-I-deficient Huh7.5 or IFN-β-deficient Vero cells were infected with MeV at an MOI of 1 and cocultured 8 h later with uninfected Vero and Huh7.5 cells (ratio, 1:1), respectively. The cocultures were treated or not treated with 10 μg/ml of FIP. Cell-free supernatants were collected at 30 and 60 h p.i. to measure IFN-α/β production. At the end of the coculture, the cell monolayers were stained for fluorescent nuclei (magnification, ×400) for counting within every syncytium indicated by arrows. Data are from one representative experiment out of two. ND, not detected.

The IRF-7 expression level is not influenced by MeV-induced fusion in epithelial cells.

We then investigated whether the enhancement of IFN-β gene expression mediated by MeV-induced cell-cell fusion correlates with the up-regulation of IRF-7. To this end, the level of IRF-7 transcripts was analyzed in 293T/CD46⁺ cells and in primary TEC, which are IFNAR competent, in the presence or absence of FIP (Fig. 5). In both of these cell types, MeV infection resulted in the activation of the IFN-β and IRF-7 genes. However, the level of IRF-7 transcripts remained unchanged when syncytium formation was blocked by FIP treatment and did not correlate with the IFN-β mRNA level. Thus, the amplification of IFN-β activation by MeV-induced cell-cell fusion is not directly or solely controlled by the level of IRF-7 expression.

FIG. 5.

Unlike the IFN-β gene, IRF-7 gene expression does not correlate with cell-cell fusion. 293T/CD46⁺ cells (left) and TEC (right) were either treated with 1,000 IU/ml of recombinant human IFN-β or infected with MeV at an MOI of 1 and cultured in the absence or presence of FIP (10 μg/ml). Expression of IFN-β (top dotted histograms) and IRF-7 (bottom black histograms) mRNA was quantified at 30 h p.i. Data from one representative experiment out of two or three are shown. ND, not detected.

MeV-induced cell-cell fusion amplifies IFN-α and IFN-β responses in human mDC but not iDC.

The phenomenon of the amplification of the IFN-α/β response by MeV-induced cell-cell fusion was then examined in human monocyte-derived DC. More than 95% of iDC and mDC were CD1a⁺ and CD14⁻ (data not shown). While the immature phenotype was confirmed by the low or negative expression of major histocompatibility complex class II, CD83, CD40, CD80, and CD86, mDC expressed high levels of these molecules (data not shown). In agreement with previous reports (17, 50), both iDC and mDC, which have CD46⁺ CD150^Low and CD46⁺ CD150^High phenotypes, respectively (Fig. 6A) (50), were sensitive to MeV infection, as shown by MeV F transcription (Fig. 6B). However, the sensitivity to infection differed between iDC and mDC. MeV replication was faster in mDC than in iDC, with ∼230-fold-higher transcription at 3 days p.i. (Fig. 6B). While the iDC only poorly fused, the MeV-infected mDC contained numerous giant MGC (Fig. 6), which expressed both viral proteins and mDC markers (17) (data not shown). The addition of FIP to the cDC did not significantly affect MeV F transcription in the iDC and only partially reduced MeV F transcription in mDC (Fig. 6B), while FIP efficiently inhibited the formation of MGC (Fig. 6). We then investigated IFN-α/β production in cDC following MeV infection. MeV-infected iDC secreted significant levels of bioactive IFN-α/β (Fig. 6C) and IFN-α (Fig. 6D). No IFN-β production was detected (Fig. 6E), even though IFN-β transcripts were observed (Fig. 6F), because of either the limited sensitivity of the ELISA or the consumption of IFN-β by MeV-infected iDC (25). According to the very low level of cell-cell fusion observed within MeV-infected iDC, the addition of FIP did not affect IFN-α/β production (Fig. 6C and D). MeV-infected mDC produced significant levels of bioactive IFN-α/β, IFN-α, and IFN-β (Fig. 6C to E) and IFN-β mRNA (Fig. 6F). Both MGC formation and IFN-α/β production by MeV-infected mDC were strongly inhibited in the presence of FIP (Fig. 6C to F). Thus, MeV infection induces IFN-α/β responses in both iDC and mDC. However, virus-induced cell-cell fusion amplifies both IFN-α and IFN-β production only in mDC. Interestingly, MeV infection induced an IRF-7 up-regulation in iDC but not in mDC (Fig. 6G). Furthermore, FIP did not significantly affect the expression of IRF-7 in both DC types (Fig. 6G).

FIG. 6.

Mature, but not immature, DC exhibit fusion-dependent amplification of IFN-α and IFN-β responses. iDC and mDC were mock infected or infected with MeV at an MOI of 0.1 in the absence (dotted columns) or the presence (black columns) of FIP (100 μg/ml). Syncytium formation was scored for each condition as described in Fig. 1. (A) CD150 expression in iDC and mDC cultures was analyzed by flow cytometry. (B) MeV F transcript accumulation in iDC and mDC cultures was measured at 3 days p.i. (C) Secreted bioactive IFN-α/β in cell-free supernatants collected at 3 days p.i. (C and D) IFN-α (D) and IFN-β (E) were measured by ELISA at 3 days p.i. (F and G) Accumulation of IFN-β (F) and IRF-7 (G) transcripts at 3 days p.i. in cDC cultures in the absence or the presence of FIP (100 μg/ml). Data are mean values from two to five separate experiments. ND, not detected; MFI, mean fluorescence intensity.

As a control, the nonfusogenic chimeric MGV was also used to infect iDC and mDC. MGV induced, in both iDC and mDC, only a basal IFN response, which was not sensitive to FIP (supplemental data may be found at hal.archives-ouvertes.fr/hal-00169132). We noticed that, contrary to MeV, MGV replicated better in iDC than in mDC (supplemental data may be found at hal.archives-ouvertes.fr/hal-00169132) and, accordingly, induced stronger IFN-β gene transcription in iDCs. We speculate that the stronger endocytosis ability of iDC over mDC can favor MGV entry, which relies on the acidic endosomal pathway mediated by the vesicular stomatitis virus G glycoprotein.

Thus, MeV infection induces IFN-α/β responses in both iDC and mDC. However, the virus-induced cell-cell fusion amplifies both IFN-α and IFN-β production only in mDC.

Sustained nuclear translocation of IRF-3 in MeV-induced MGC derived from epithelial cells and mDC.

MeV infection induces the transactivation of the IFN-β gene through the phosphorylation and nuclear translocation of IRF-3 (65). Therefore, we analyzed the changes in the subcellular IRF-3 localization using GFP-tagged IRF-3 (GFP-IRF-3) transfected into 293T/CD46⁺ cells. As expected (46), GFP-IRF-3 was localized exclusively within the cytoplasm of uninfected cells (Fig. 7A). After MeV infection, nuclei from a single syncytium exhibited a diverse level of GFP-IRF-3 staining, thus looking asynchronous, and a large proportion of syncytia contained nuclear IRF-3 whatever their stage. In addition, few mononuclear MeV-infected cells surrounding outside MGC displayed nuclear localization of GFP-IRF-3 (not shown). In the absence of FIP, 50% ± 24% of the nuclei belonging to MGC were labeled with GFP-IRF-3, whereas only 6% ± 7% of the small amount of single cells, which remained outside MGC, had their nuclei labeled (Fig. 7A). In the presence of FIP, although most cells were infected (Fig. 1A and 7B) (see supplemental data at hal.archives-ouvertes.fr/hal-00169132), nuclear translocation of GFP-IRF-3 was observed in only 4% ± 2% of them (2α = 0.01) (Fig. 7A), compared to 50% ± 24% of nucleus labeling in the absence of FIP. In all cases, the nuclei were intact, including those in syncytia, as shown by Hoechst staining (Fig. 7A).

FIG. 7.

Nuclear translocation of IRF-3 can be triggered within MeV-induced syncytia. (A) Nuclear translocation of GFP-IRF-3 within syncytia of 293T/CD46⁺ cells infected by MeV at an MOI of 1. Microphotographs (magnification, ×400) show morphology (top panels), Hoechst-labeled nuclei (middle panels), and GFP-IRF-3-labeled nuclei (bottom panels) at 30 h p.i. Micrographs of uninfected cells (mock) and cells infected with MeV in the absence (MeV) or the presence (10 μg/ml) (MeV + FIP) of FIP are shown. Data are from one representative experiment out of four. (B) Three-color overlays of confocal images showing the distribution of GFP-IRF-3 (green), N (red), and nuclei (Draq5) (blue) in 293T/CD46⁺ cells infected or not infected with MeV at an MOI of 0.1 in the presence or absence of FIP and transfected with GFP-IRF-3. Syncytium images were taken at three morphological stages, flat adherent, retracting, and smooth ball, respectively. The whole set of one-color images used to build the overlays is shown in supplemental data at hal.archives-ouvertes.fr/hal-00169132. (C) Nuclear localization of endogenous IRF-3 in MGC derived from MeV-infected mDC. mDC were mock treated (left) (magnification, ×630) or infected with MeV at an MOI of 0.1 in the absence (middle) (magnification, ×400) or the presence (right) (magnification, ×630) of FIP (100 μg/ml). At 3 days p.i., mDC cultures were stained with anti-IRF-3 (green) (left) and with nucleus stain (Hoechst 33343) (blue). Cells were analyzed with an Axioplan microscope.

Confocal analysis of GFP-IRF-3 distribution in 293T/CD46⁺ cells infected with MeV at MOIs of 1 and 2 confirmed the high rate of nuclear IRF-3 labeling in MGC compared to mononuclear cells and the presence of cytoplasmic N protein in all but rare cells (data not shown). At a lower MOI of 0.1, similar results were obtained. Upon confocal analysis, many nuclei within MGC were labeled with GFP-IRF-3, whereas GFP-IRF-3 remained excluded from the nuclei of noninfected cells (Fig. 7B and data not shown). When the fusion was prevented by the addition of FIP, only a few mononuclear cell nuclei displayed GFP-IRF-3 labeling, although most of them were clearly infected, as shown by cytoplasmic N labeling (Fig. 7B, red dots and patches, and data not shown). Again, GFP-IRF-3 was translocated into the nuclei of MGC at all stages of the adherent-ball cycle shown in Fig. 3 (Fig. 7B and data not shown). Interestingly, the distribution of both N and GFP-IRF-3 tended to change with the morphological stage of the MGC. When the syncytium was flat and adherent, GFP-IRF-3 showed a dotted and/or reticulated distribution in the nuclei, and the N protein had a dotted cytoplasmic distribution. When the syncytium underwent retraction into a smooth ball, the intensity of the nuclear GFP-IRF-3 labeling became stronger and more diffuse, and the N protein tended to aggregate further into larger patches at the periphery of the cytoplasm (Fig. 7B and data not shown). Thus, the amplification of IFN-β production induced by MeV-mediated fusion correlates with a sustained and strong nuclear translocation of IRF-3 within MGC at all their morphological stages.

Localization of endogenous IRF-3 during infection of mDC was also studied. As expected, endogenous IRF-3 distributed exclusively within the cytoplasm of uninfected mDC (Fig. 7C, left). At 3 days p.i., IRF-3 (green) staining of many intact nuclei (blue stain) in syncytia of MeV-infected mDC was observed (Fig. 7C, right) (see Fig. S7C in the supplemental material). Interestingly, the distribution of endogenous IRF-3 in the nuclei of mDC-derived syncytia looked very similar to that of GFP-IRF-3 in nuclei of 293T-derived syncytia (compare Fig. 7C with A and B and data not shown). The addition of FIP strongly inhibited both MeV-infected MGC from mDC and nuclear localization of endogenous IRF-3 (Fig. 7C, middle). Thus, endogenous IRF-3 tended to remain translocated in nuclei within MGC from MeV-infected mDC, in agreement with the finding of sustained nuclear translocation of exogenous GFP-IRF-3 within MGC derived from epithelial cells.

DISCUSSION

We report here that MeV-induced MGC or syncytia derived from epithelial cells and mDC are metabolically active and long-living and display both sustained nuclear translocation of IRF-3 and enhanced activation of the IFN-β gene. First, there is a correlation between the fusogenic activity of the MeV strain and IFN-β production. Second, the fusion-enhancing effect is observed at both low and high MOIs. Third, the FIP inhibits, in a dose-dependent manner, both cell-cell fusion and IFN-β production. Fourth, cell-cell fusion mediated by MeV H and F glycoproteins does not activate any IFN-β response in the absence of viral infection but does it after infection with MGV, a nonfusogenic chimeric virus. Fifth, the cytosolic PRR RIG-I and the IFN-β gene are trans complemented during the fusion process. Sixth, although both iDC and mDC are infected by MeV, only mDC undergo massive cell-cell fusion. In the mDC, robust IFN-α/β production is mediated by MeV-induced cell-cell fusion. Seventh, in both epithelial cells and mDC, the fusion-enhancing effect on the IFN-β response appears to be mediated by a sustained nuclear IRF-3 localization but does not directly correlate with the up-regulation of IRF-7 expression. Finally, in response to MeV, iDC also produce IFN-α/β production, but it is independent of the fusion process and likely amplified via IRF-7. Altogether, our results indicate that MeV-induced MGC in epithelial cells and mDC are important sources of IFN-α/β and that the fusion can mediate an enhancement of IFN-α/β production without modulating the expression of IRF-7.

Cell-cell fusion is a hallmark of many viral infections, and the resulting MGC were thought to be short-lived. Indeed, syncytia induced by the human immunodeficiency virus type 1 glycoprotein died by apoptosis by at least three different mechanisms: transient lipid exchange, activation of several kinases and transcription factors, and contagious apoptosis (58). Surprisingly, MeV-induced MGC from epithelial cells were found to be viable and dynamic entities capable of transducing intracellular signals throughout their morphological stage changes. Thus, from our in vitro observations, we predict that the physiopathological MGC (WFC) observed in lymph nodes and thymus from MeV-infected children and primates should have a rather long life span in vivo (54, 76). However, as described by us and others (13, 58, 63, 82), syncytium apoptosis finally occurs, probably depending on cellular environment deprivation.

In nonpathological situations, the contents of a cell nucleus should be tightly regulated to ensure that every cell harbors a single nucleus. Notable exceptions are the fusion of cellular precursors undergoing a specific maturation process, such as myotubes, osteoclasts, and syncytiotrophoblasts. In the two latter cases, a survival program is turned on (15, 31). Whether such a mechanism occurs for the survival of MeV-induced MGC remains to be determined. Strikingly, both of the MeV-induced MGC and syncytiotrophoblasts need to recruit fresh mononuclear cells in order to survive (29). Furthermore, given that IFN-β is used as retrocontrol feedback to limit the size of the osteoclasts by preventing further recruitment of new mononuclear cells (9), it could also regulate the dynamics of MGC formation induced by MeV, as observed for other viruses (53, 80, 81, 86).

MeV-induced cell-cell fusion results in MGC harboring an important function in innate immunity, and it could be questioned if the fusion per se acts as an activation signal. Indeed, the artificial fusion of a human cell line with chicken erythrocytes results in the activation of both human and chicken IFN-β (23), with the latter being indicative of a reactivation of the dormant chicken erythrocyte nucleus. MeV-induced cell-cell fusion is mediated by H binding to the CD46 or CD150 cellular receptor, which results in the activation of the fusion F protein (19). H binding to CD46 has been reported to activate the IFN-β response and NO· production in murine macrophages expressing human CD46 (33). In human epithelial cells, the H/F- and CD46-mediated fusion, per se, was unable to trigger the IFN-β response, which required virus transcription (62). Likewise, we can exclude that the interaction of H with TLR2 is involved in IFN-β activation since (i) the signaling downstream of TLR2 occurred independently of the F glycoprotein, (ii) the use of a wild-type MeV strain with the H protein unable to bind to CD46 and TLR2 (4) gave similar results (not shown), and (iii) the TLR2 signaling pathway is not linked to IFN-α/β production. Altogether, MeV-induced cell-cell fusion in human epithelial cells is not directly sensed as a danger signal by the innate cellular machinery.

MeV infection of human epithelial cells triggers the production of IFN-β, which is differentially regulated in mononuclear cells and MGC. At the beginning of viral infection, the activation of IFN-β occurs in MeV-infected mononuclear cells, where cytosolic RIG-I is activated upon recognition of the 5′-triphosphate end of MeV leader RNA (62), and this results in the activation of IRF-3. IRF-3 then undergoes phosphorylation, homo- or heterodimerization, nuclear translocation, fixation on IFN-sensitive responsive elements, and degradation by the ubiquitin-proteasome pathway (3, 65). In a later phase, the amplification of IFN-β production in MeV-infected mononuclear epithelial cells can be mediated by classical IFNAR/IRF-7-dependent positive feedback, as described previously for other viruses (27), since IRF-7 is up-regulated after MeV infection. In contrast, the robust IFN-β production mediated by MGC from epithelial cells could not be explained solely by IRF-7 up-regulation, because the latter was poorly sensitive to the fusion process. As described previously for infection with respiratory syncytial virus, IRF-3 nuclear translocation occurs early, within a few hours after infection, and it then drops rapidly within 15 h because of the anti-IFN activity of nonstructural proteins (71). Therefore, the presence of a high level of IRF-3 nuclear translocation within MGC at a late time (30 h p.i.) of MeV infection is unexpected and supports an essential role for IRF-3 in the MGC-mediated amplification of IFN-β production. It is possible that MeV proteins with IFN antagonist activity are diluted out upon fusion of MGC with uninfected cells, thus allowing stronger and more sustained IFN production. It remains to be determined if the sustained IRF-3 nuclear localization within MGC occurs as phosphorylated IRF-3 homodimers or IRF-3/IRF-7 heterodimers.

What could the mechanism that enables cell fusion to boost IFN-β activation be? At low MOIs (virus-to-cell ratio of <1), we propose the following model. Since the RIG-I and IFN-β gene loci are trans complemented during cell-cell fusion, syncytium formation can boost IFN-β transcription by bringing uninfected cells into contact with viral pathogen-associated molecular patterns (PAMPs). At a given time, the level of IFN-β activation results from the balance between available trigger viral RNA (PAMPs), RIG-I (PRR), pathway components (i.e., IRF-3), and viral IFN antagonists. The sustained nuclear localization of IRF-3 in the nuclei within syncytia at low MOIs is compatible with the continuous recruitment of noninfected cells, which can result in a weakening concentration of viral antagonists and/or the recruitment of “naive” RIG-I/pathway component molecules by MeV leader RNA, which would be produced in excess over the amount of RIG-I/pathway components available in a single cell. At a high MOI, an alternative model should be made, since all individual cells get infected prior to the fusion event, and the fusion-mediated amplification of the IFN-β response is even higher (Fig. 1B). IRF-3 nuclear translocation could be sustained within MeV-induced MGC because of a synergistic activity such as the stabilization of phosphorylated IRF-3 by the activation of the DNA-dependent protein kinase (32). This will require further investigations. Presently, our data thus argue for two nonexclusive mechanisms involved in the fusion-enhancing effect on IFN-β activation: one, evidenced at a low MOI, is the recruitment of noninfected cells to MeV-infected MGC, and the other, at a high MOI, is a synergistic effect of cell fusion and virus infection. In both cases, there is a sustained nuclear translocation of IRF-3 within MGC, the underlying mechanism of which remains to be more deeply examined. In every case, the amplification of the IFN-β response by MGC derived from MeV-infected epithelial cells upgrades the alert level of the innate immune response against viral infection in peripheral tissues.

During natural infection, MeV infects lung epithelial cells and/or resident iDC in epithelia and mucosa and likely induces local IFN-α/β production, which could limit MeV replication (45). Infected iDC can then migrate and disseminate the virus to the draining lymph nodes. There, they can be stimulated via CD40L by encountering naive T lymphocytes and become activated and more permissive to MeV replication, as shown experimentally (66). Because mDC are prone to fuse with surrounding cells, they form MGC, which could correspond to the WFC found in lymphoid organs. This results in a high level of virus progeny, which can propagate throughout the body. There are several examples of IFN-α/β production by human or mouse iDC (2, 34, 35) infected in vitro by a few viruses, including MeV (38). Here, we demonstrate that upon MeV infection, both iDC and mDC produce IFN-β and IFN-α in vitro. iDC display low permissiveness to MeV infection and rapidly produce high levels of IFN-α and IFN-β independently of cell-cell fusion, probably through IFNAR/IRF-7 signaling, as judged by the up-regulation of IRF-7 expression. Since IFN-α/β is quickly produced and secreted by iDC after infection with MeV (not shown), IFN-α/β can protect cells against the propagation of MeV and strongly limit the formation of MGC. As iDC are present in peripheral tissues and secrete IFN-α/β, they can contribute to the establishment of the innate antiviral state by enhancing the cytotoxicity of NK cells and activating macrophages (10, 40). In addition, iDC constitute a critical link between innate and adaptive immunity (44, 67). Indeed, IFN-α/β induces the up-regulation of costimulatory molecules CD80, CD86, and CD40 on cDC (24) and the expression of TRAIL on iDC, which become cytotoxic (83). Thus, the IFN-α/β produced by MeV-infected iDC is a signal that upgrades the alarm level of cellular innate immunity for detecting the invasion of a possible pathogen.

In contrast to iDC, mDC are highly susceptible to MeV infection, form large MGC, and produce high levels of IFN-α/β. IFN-α/β is therefore less efficient in limiting MeV growth and MGC formation within mDC than within iDC. The opposite phenotypes of iDC and mDC could originate from different relative kinetics of infection and the innate antiviral response. Indeed, when the strength of the initial activation of IFN-β is too low compared to the virus growth kinetics, the rapid accumulation of MeV-encoded anti-IFN V and possibly C proteins can block the intracellular IFNAR signaling pathway (55, 69), and this paves the way for unlimited virus growth. We therefore favor that MGC can be promoted or repressed according to the respective speed and strength of virus growth and IFN-α/β production. Upon infection, mDC produce IFN-α/β mostly from MGC, without any IRF-7 up-regulation. This suggests that the IFNAR/IRF-7 feedback loop is not directly involved. These data are in agreement with the down-regulation of IFNAR in cDC upon their maturation (68). As for epithelial cells, the robust production of IFN-α/β by MeV-mediated MGC from mDC also correlated with the sustained activation of IRF-3. However, the induction of IFN-α independently of IRF-7 up-regulation in MGC is questionable, and the mechanism remains to be determined. The IFN-α/β produced by MGC from MeV-infected mDC could rather be involved in the establishment of MeV-specific adaptive immune responses in the secondary lymph nodes. By providing a high viral antigen load and IFN-α/β-dependent enhancement of the cross-priming to T cells (43), the paradoxical accumulation of virus and IFN-α/β within MeV-mediated MGC probably contributes to the stimulation of the MeV-specific adaptive immune response, which will finally clear the virus from the organism. Finally, because of the different abilities of various laboratory, vaccine, and wild-type MeV strains to counteract cellular innate immunity, the virus strain dependency of cell fusion-mediated amplification of the IFN-α/β response is also currently under investigation.