The objective of this study was to evaluate the immunogenicity of a dengue vaccine candidate based on a pediatric measles vaccine expressing a minimal dengue antigen. This strategy provides a recombinant vaccine that might protect children simultaneously from measles and dengue and that might be affordable to populations through the EPI program in the regions affected both by dengue and measles infections. An efficient pediatric dengue vaccine is supposed to elicit durable protective humoral immune responses against all four dengue serotypes without risk of ADE 
. Regarding this objective, we assembled covalently the antigenic domain III from the DV1 envelope E glycoprotein and the pro-apoptotic ectodomain of DV-1 M protein to generate a dengue combined antigen, EDIII-ectoM. In the fusion construct, the N-terminal calreticulin peptide signal sequence directs EDIII-ectoM to the secretory pathway. The furin-dependent cleavage site of prM/M which links ectoM to EDIII allows the processing of the antigen by specific proteases throughout the Golgi apparatus. Expressed by recombinant MV vector, the EDIII-ectoM antigen induced in mice susceptible to MV specific antibodies to DV1 EDIII that did not cross-react with other DV serotypes and that neutralized DV1 infection in vitro. Immunization primed a long-term memory that was vigorously boosted when animals were inoculated with live DV.
Although DV disease pathogenesis and protection mechanisms are not fully clarified, disease severity is correlated with viremia levels and neutralizing antibody is generally used as a marker of vaccine effectiveness 
. Experimental mouse models of DV infection have been reported showing that adult AG129 mice, which are deficient for IFN α/β/γ receptors develop a dose-dependant transient viremia after peripheral injection of unadapted or mouse-adapted DV, whereas A129-IFNAR mice, which are deficient only for IFN-α/β receptor are less sensitive to DV infection 
. However, AG129 mice are not sensitive to MV infection and the prototype DV1 Hawaï strain did not replicate in these mice 
. Suckling mice develop lethal encephalitis after DV intracerebral inoculation, but in our study mice were 3–4 month-old after two MV immunizations and intracranial inoculation could not be performed. Moreover, this model is far from the human situation since DV does not infect the nervous system, nor lead to encephalitis in humans. The CD46-IFNAR mouse model sensitive to MV infection that we used did not allow documenting in vivo protection from DV replication. Therefore, to demonstrate the induction of anamnestic neutralizing antibody response upon live DV exposure, mice were peripherally inoculated with DV a long time after immunization. These experiments showed that neutralizing antibodies induced by immunization with MV-EDIII-ectoM were strongly boosted by live DV inoculation, thus suggesting a protective capacity. Indeed, the available vaccines against yellow fever, Japanese encephalitis and tick-borne encephalitis viruses have proven that anamnestic neutralizing antibodies play an essential role for protection against flaviviral infections 
The EDIII without ectoM was poorly immunogenic in the context of MV expression. It appeared, therefore, that the 40-residue long ectodomain of M plays a critical role in its immunogenicity. DV EDIII has been previously shown to be immunogenic in the form of recombinant chimeric proteins 
or expressed from a plasmid 
or from adenovirus vector 
. To determine whether the EDIII sequence inserted into MV vector was able to present neutralizing epitopes, we produced in E.coli recombinant EDIII proteins from DV1, 2, 3 and 4 corresponding to the same sequence and we coated plates with these proteins. Tested by ELISA on these plates, a neutralizing HMAF specific to DV1 recognized specifically the DV1-EDIII, but not the other serotypes (data not shown), indicating its specificity. This DV-1 HMAF recognized also specifically by immunofluorescence the DV1 EDIII expressed in cells infected by MV-EDIII, indicating the capacity of EDIII to expose serotype-specific epitopes. The neutralizing monoclonal antibody 4E11 recognized also the EDIII expressed by MV-EDIII infected cells, indicating that the epitope specific of this antibody is accessible within the EDIII expressed by MV. This epitope has been mapped and shown to be exposed on the native form of EDIII 
. Furthermore, we expressed the same EDIII sequence as a secreted protein by drosophila cells and showed that it was able to efficiently compete with the neutralizing activity of antibodies induced by the MV-EDIII-ectoM virus. Altogether, these observations suggest that EDIII expressed by MV was able to present a conformationally active neutralizing epitope. Recent studies evaluating the immunogenicity of West-Nile virus (WNV) EDIII showed that a high amount of EDIII was necessary to induce neutralizing antibodies, while EDIII fused to TLR ligands was immunogenic and conferred protection at lower doses 
. Therefore, the low immunogenicity of DV EDIII in our hands might be due to the lower amount of antigen expressed by recombinant MV as compared to the high protein or DNA doses administered by others. To increase the level of expression by MV, EDIII can be cloned upstream the N gene, as MV genes are expressed as a gradient from the 3′ to the 5′ end of the genome.
This small ectoM protein, which is highly conserved among the four serotypes of DV, has pro-apoptotic properties 
. High titers of MV-EDIII-ectoM induced apoptosis of infected U937 monocyte-like cells that was not observed at standard titers. This critical point in terms of safety needs to be evaluated further in the development of this vaccine candidate. Indeed, recombinant MV vector has to keep the high safety level of standard MV vaccine. However, this property might be determinant to the immunogenicity of EDIII because apoptotic infected cells express Toll-like receptor (TLR) ligands that increase the cross-presentation of viral epitopes by antigen presenting cells 
. Indeed, we observed that ectoM, in the context of MV replication, increased human DCs maturation and triggered the release of cytokines and chemokines determinant for the establishment of specific adaptive immunity. Therefore, its capacity to adjuvant EDIII might be still more efficient in humans than in CD46-IFNAR mice. Further studies are needed to address the mechanism of action at the molecular level.
In conclusion, we have produced a minimal antigen from DV1 able to induce long-term specific neutralizing antibodies to DV1 with no cross-reactivity with other serotypes. We have shown that the remarkable adjuvant capacity of ectoM to EDIII immunogenicity was correlated to its capacity to mature primary DCs and to activate the secretion of a panel of proinflammatory and antiviral cytokines, as well as numerous chemokines determinant for the establishment of specific adaptive immunity. The immunogenicity of this antigen was demonstrated through its expression by a recombinant MV vector, thus making the proof-of-concept of this strategy for dengue vaccine development. Using MV as a vaccination vector presents a number of advantages : vaccination against measles is mandatory, vaccine strains are genetically stable, MV does not recombine or integrate genetic material, and vaccine does not persist or diffuse. MV-specific CD8 T cells and IgG are detected in vaccinees up to 25–34 years after a single MV vaccination 
and boosting increases this memory 
. Using MV as a recombinant vaccine to immunize simultaneously against measles and dengue might be particularly attractive in areas where both diseases threaten children every year, such as Africa and South America. Taking advantage of the capacity of MV vector to express large amounts of heterologous genetic material very stably 
, we generated tetravalent dengue antigenic constructs inserted into single MV vectors that are currently characterized. Such a strategy should avoid the stability and interference problems encountered with tetravalent formulation of four attenuated viruses, as well as the reactogenicity problems 
. These new candidates will be evaluated in a much more appropriate non-human primate model.