In countries where HIV is highly prevalent, a recombinant MV-HIV vaccine might be administered to naïve infants as a standard measles immunization that would protect from measles whilst eliciting long-term memory to HIV that could be boosted later with another type of HIV vaccine. A MV-HIV recombinant vaccine might also be used to immunize the adolescent and adult populations who are already pre-immune to MV since their childhood vaccination. In that case, pre-existing immunity to measles might prevent or reduce the efficacy of the recombinant MV vaccine. However, numerous studies have shown that revaccinating already immunized individuals results in a boost of anti-MV immunity, indicating that the live vaccine replicates in spite of pre-existing immunity 
. Moreover, we previously demonstrated that a recombinant MV-HIV vector induced antibodies to HIV in mice and macaques in the presence of MV pre-existing immunity, provided that two injections with a higher dose are performed 
. Yet, this point needs to be further evaluated in human trials. Measles is still difficult to control, as evidenced by the large outbreaks occurring in Europe since 2010. Improving measles vaccination coverage is essential to containing and preventing further such outbreaks. An MV-HIV vector might be an effective and safe carrier for a HIV vaccine, whilst boosting pre-existing measles immunity.
This study was undertaken to evaluate the immunogenicity of a recombinant measles vector, MV1-F4, carrying an insert encoding HIV-1 clade B gag, RT and nef. CD46/IFNAR mice and cynomolgus macaques were chosen for pre-clinical evaluation of MV1-F4, as both species are susceptible to infection with MV vaccine strains and further macaques are susceptible to infection with wild-type MV 
. As a first and rapid assay, mice were inoculated with a single administration of escalating doses of recombinant vaccine and CMI were assessed as early as 7 days after immunization. MV1-F4 vaccine was immunogenic and induced strong CD4+ and CD8+ responses to HIV-1 F4 and to MV. CD4+ responses that released mainly IFNγ were observed, although the cytokine pattern in mice that lack type-I IFN receptor is likely unconventional. Dose-response effect was evident, the higher doses being more immunogenic. A relatively low dose (105
) was chosen to immunise macaques. To determine whether boosting with this vaccine improves immunogenicity, we compared responses in groups of macaques that received a single immunisation with those that received two. Superior immunogenicity was obtained with two immunisations, indicating that MV1-F4 humoral responses are boostable even in the presence of anti-vector antibodies. Titres of anti-F4 and anti-MV antibodies initially peaked at day 14 but were boosted after a second immunisation at day 28. By contrast, primary CD4+ T cell cytokine responses against MV1-F4 initially peaked between days 14 and 28 in vaccinees that responded, but boosting was only observed with 1 macaque out of 4, F52. From these data, it would appear that a boost interval of 28 days is too short for efficient re-stimulation of CD4+ T cell responses. For example, macaque F53 appears to have been boosted during its primary response suggesting that the peak reached on day 42 was due to the primary vaccination. Importantly the CD8 responses of F53 fell after boosting. Since peripheral CMI responses against MV1-F4 waned by day 84, in both groups of macaques, then, in hindsight, delaying the second vaccination to this later time may have resulted in superior boosting. This may explain why boosting of humoral responses contrasted with a lack of boosting of cellular responses. For example, macaque F54 was divergent in that it had had good anti-F4 and MV antibody responses that were boosted, but in comparison poor CMI responses. Alternatively, measurement of Th2 cytokines (IL-4, IL-5 and IL-13) may have better correlated with boosting of humoral responses. These results may guide further preclinical and clinical development of the MV1-F4 candidate vaccine.
In vitro responses to measles are dominated by CD4+ T cells that, depending on antigen dose, primarily produce a Th1-like pattern of cytokine release 
. Therefore it was not unexpected for CD4+ T cell cytokine responses to MV1-F4 to be greater than CD8+ T cell responses, following F4 peptide stimulation. Interestingly, responding CD4+ T cells were all CD154+, which acts as a potent maturation agent for dendritic cell priming of anti-viral CD8+ T cells, desirable for a candidate HIV vaccine 
. Although peripheral CMI responses were no longer readily detectable by day 84, CD8+ T cell responses could still be detected in lymphoid tissues taken at termination suggesting that MV1-F4 may stimulate long lasting immunity. This would be consistent with previous reports that MV-specific CD4+ and CD8+ T cells and MV-specific IgG can be detected up to 25 years after vaccination 
The cytokine profiles observed with MV1-F4 vaccinees were remarkably similar between responding animals. For CD4+ T cells, we observed a TNFα and IL-2 bias that contrasted with CD8+ T cell responses with a TNFα and IFNγ bias. Such responses are indicative of a T helper 1 (Th1) cell response characterized by the production of IFN-γ, IL-2 and TNF-α 
. However, IL-4 responses were not assessed here so a Th1/Th2 mixed pattern of cytokine release cannot be excluded. The bias of CD4+ T cell cytokine responses to TNFα and IL-2 suggests a central memory response rather than terminally differentiated CD4 effector cell response, which was characterised by higher levels of IFN-γ and TNF-α expression 
. CD8+ memory T cells can quickly produce a variety of cytokines including IFNγ, TNFα and to a lesser extent IL-2 
, matching the profile we observed against F4 peptides. The low frequency of multifunctional T cells, positive for all three cytokines, is likely due to the necessary use of peripheral blood lymphocytes deficient in effector memory T cells when compared with mucosal sites 
Despite group B receiving only a single MV1-F4 immunisation their CD8+ T cell responses were of a greater magnitude than group A that received two immunisations. This may have been the result of MHC haplotype bias of individuals within group B towards strong CD8+ T cell responses, which may have been more evident due to our small group size (n
4). Theoretically such variability could be minimised with larger group sizes or pre-selection of MHC matched animals to balance MHC-restricted CMI responses between groups. In a follow up study, using MV1-F4 encoding a C Clade HIV-1 insert, we plan to test this using larger groups (n
8) of MHC-characterised cynomolgus macaques assigned to each group evenly, rather than randomly assigning animals as previously done.
In SIV infection, Mauritian derived cynomolgus macaques with the rare M6 MHC haplotype (~4% of the population) are associated with a significant reduction in chronic phase viremia 
. If control of viremia is associated with superior MHC-restricted CMI responses then it might be expected that those individuals would also make superior vaccine responses. In this study, the only macaque with a M6 MHC haplotype was F57, a M3/M6 heterozygote, which coincidentally exhibited the best CD8+ T cell response and a strong CD4+ T cell response to both HIV-1 F4 and MV. However, the M2 MHC haplotype has also been associated with significant control of chronic phase SIV viremia 
, yet we failed to detect significant CD4+ or CD8+ T cell cytokine responses against HIV-1 F4 peptides in the M2/M3 heterozygote F54, despite two immunisations. At a simpler level, chronic SIV viremia in MHC homozygous macaques is reported to be 80 times worse than in MHC heterozygous Mauritian-derived cynomolgus macaques 
. MHC heterozygous advantage suggests recognition of a maximally diverse set of epitopes that provides a rationale for prophylactic vaccination to elicit broad CMI responses in such individuals 
. The only MHC homozygote in this study was F58, a M3/M3 homozygote, which had poor CMI responses to HIV-1 F4 peptides. However, there was no evidence of a heterozygous advantage for F51 and F54, M1/M3 and M2/M3 MHC haplotypes respectively, in terms of superior CMI responses to F4 peptides. More interestingly, all macaques with recombinant heterozygous MHC haplotypes, F52, F53 and F56, M3/M1+M2+M3, M2/M1+M2+M7 and M1/M1+M3 respectively, made significant CMI responses to F4 peptides, even though they are simple recombinants of the same alleles present in F51 and F54. Recombination between alleles and loci has been suggested as a mechanism responsible for generating diversity at MHC loci allowing recognition of new epitopes 
. Our data suggests that heterozygous recombinants have an advantage in terms of responses to vaccines, with the caveat that our group size and number of haplotypes is small. Nevertheless this is an issue that should be taken into account when assigning macaques between groups to avoid bias.
Immune escape is believed to be a significant force shaping viral evolution at the population level through a MHC “imprinting effect” in which escape mutations selected in the context of common MHC alleles may become predominant in the circulating viral population, unless they revert when transmitted to new hosts 
. If immunodeficiency viruses have evolved to escape commonly presented epitopes, then MHC homozygous and common heterozygous haplotypes may be at a disadvantage in terms of poor vaccine responses compared to recombinant or rare haplotypes. Mauritian derived cynomolgus macaques however, have not to our knowledge ever been naturally exposed to SIV and hence the virus won't have adapted to their MHC. In contrast, heterozygote advantage is probably due to twice as many alleles meaning twice as many potential T cell responses. As a result macaques with homozygous MHC haplotypes would be anticipated to be poor vaccine responders and so they should be equally divided between vaccine groups to avoid negative bias. By chance, F51 in group A and F55 in group B shared the same MHC heterozygous haplotype, M1/M3. However, although their CMI responses exhibited similarities, they were not identical, thus suggesting that other factors, possibly minor histocompatibility antigens also shape immune responses to MV1-F4 
. The main limitation of this study was the small group size complicated by a mix of responders and non-responders to the F4 insert. Retrospective analysis of MHC haplotypes greatly aided our interpretation of an otherwise potentially confusing pattern of vaccine responses. In future MHC typing of Mauritian derived cynomolgus macaque studies has the potential to greatly increase the power of new studies and reduce animal usage.
All vaccinees developed anti-MV neutralising antibodies after immunisation but this could have been induced by inert virus particles. Only humoral and CMI responses to MV1-F4 demonstrate replication of the vector because the vaccine preparation contained only MV particles and no F4 protein. Thus, F4 protein was expressed from vector replication in vivo
. By these criteria there may have been no take of MV1-F4 in F58. It is unclear why F55 failed to seroconvert to MV-ELISA antigen despite a detectable neutralising antibody titre, but it may be a technical or species issue associated with the use of recombinant proteins. Since most HIV vaccine candidates based on vectors are derived from naturally occurring human viruses, pre-existing immunity has the potential to blunt vaccine responses through neutralisation 
. However, boosting of humoral immunity with MV1-F4 was able to efficiently overcome pre-existing immunity in the presence of protective tires of measles neutralising antibody. It is possible that the intramuscular route of immunisation prevented rapid neutralisation of MV1-F4 allowing boosting to occur. This may have facilitated secretion of transgene proteins that could boost humoral immunity. Secretion of HIV proteins by infected muscle would require cross presentation by antigen presenting cells to boost CD8 responses, a process that is probably is probably less efficient in directing antigen to MHC class I than de novo synthesis. Alternatively, the aerosol route of administration is very effective in humans as a booster for the second MV immunization 
. In a recent study that also used a vector based upon an attenuated measles virus strain, expressing SIV Gag, transgene immunity was found to be weak, but immunisation was carried out in pre-immune rhesus macaques, albeit in the absence of protective titres of measles neutralising antibodies 
. That pre-existing immunity may have limited replication of attenuated vaccine virus resulting in low levels of transgene expression and immunogenicity. By contrast, immunisation of naive cynomolgus macaques with MV1-F4 in this study resulted in superior transgene and anti-vector responses.
A replicating vaccine vector capable of inducing potent cellular and humoral immune responses is likely to be required for the development of an effective HIV vaccine. That vector will also need to be safe and preferably boostable. The results of this study suggest that MV1-F4 is a promising component of a prime-boost vaccine strategy to limit the spread of HIV-1.