Vaccine research efforts for filoviruses have been hampered by a lack of knowledge regarding the adaptive immune mechanisms needed for protection and by the difficulty of assessing cellular responses in animal models, such as the guinea pig and nonhuman primate. This study identified CD8+
CTLs as a protective immune mechanism induced by vaccination of mice against each of the VP proteins. It also extended our earlier findings and other reports of in vitro cytolytic activity (23
) by demonstrating protective CTL responses in adoptive transfer studies. This analysis of the CD8+
cellular responses to these six ZEBOV proteins offers insights on the specificity and breadth of responses that may be important to vaccine research efforts. BALB/c and C57BL/6 mice were protected after vaccination with VRPs expressing NP, VP30, VP35, and VP40; for each of these combinations, we identified or confirmed at least one protective CTL response. Twice we observed that an individual peptide induced a protective CTL response in both mouse strains (see VP30 in Table ), but not surprisingly, the peptide specificity of the majority of the responses was not the same, as the mouse strains express major histocompatibility complex (MHC) class I molecules with different peptide-binding motifs. The mice also differed in the number of CTL responses generated towards each protein. For example, C57BL/6 mice had CTL responses to one peptide derived from the Ebola virus VP40, whereas three protective VP40-specific CTL responses were identified in BALB/c mice. However, the number of VRP-induced responses to each protein was low (≤3) in both mouse strains.
The murine responses to VP24 were also interesting, in that the presence or absence of detectable CTL activity corresponded with protection. Vaccination to VP24 induced three CD8 T cell responses in BALB/c mice and protected them from challenge, while similarly vaccinated C57BL/6 mice failed to produce any detectable CD8+
T-cell response to VP24, and none survived challenge, even after modification of the vaccine regimen. The lack of detectable CTL responses to VP24 suggests that cells of the H-2b
haplotype do not present peptides from VP24 that induce lytic activity. Analysis of the VP24 sequence using predicted binding motifs for Kb
class I molecules indicates at least five peptide sequences with the potential to be presented by these molecules; however, studies comparing motif peptides and CTL responses for other viruses demonstrated that not all motif peptides actually induce CTLs (25
). The same observation may apply to C57BL/6 mice and VP24 predicted peptide sequences.
Alternatively, the failure to induce protection in VRP-VP24-vaccinated C57BL/6 mice could indicate a suboptimal vaccine regimen. Both mouse strains had equivalently low, but detectable, antibody responses to VP24 after vaccination. There may be differences in the levels of VP24 protein expressed in the mouse strains, or in the threshold needed to activate their CTL, such that expression of VP24 from another vector would be better able to induce protective CTLs to VP24 in these mice. However, this is not supported by our inability to detect VP24-specific T-cell responses in convalescent C57BL/6 mice after challenge.
We had some difficulty identifying protective CD8+
T-cell responses to GP, despite reports of in vitro lytic activity against target cells expressing GP with cells from mice vaccinated with plasmid vaccines (30
) or liposome-encapsulated virus (22
). The plasmid vaccine platforms induced solid lytic activity to cells expressing GP but did not identify the specific epitope(s) recognized. The study using liposome-encapsulated virus identified two CTL epitope sequences (GP 161 to 169 and GP 231 to 239). We did not observe any response to those peptides after vaccination with VRPs expressing GP; however, we did detect T-cell responses to the peptide at positions 231 to 239 after challenge (Table ). The reported lytic activity to those sequences was modest, even with the encapsulated virus, generally being less than 25% at E:T ratios as high as 100:1, and the different vaccine regimens used in the two studies likely contributed to the different observations. Alternatively, the protective antibody responses known to be induced in both mouse strains following vaccination with VRPs expressing GP (35
) may have contributed to clearance of antigen-expressing cells before these cytolytic T-cell responses were firmly established in these mice. The response to GP531
is interesting because it contains the putative fusion domain (7
) and because both CD4+
cells responded to it. Within this sequence there is some variation at amino acid position 544 (T to I) in both Ebola virus Zaire and Sudan isolates. The GP in the VRP has an I at amino acid 544, but the mouse-adapted challenge virus has a T. Preliminary data suggested that both sequences induced IFN expression by the CD4 and CD8 responses in C57BL/6 mice vaccinated with VRP (data not shown). Our expansion strategy used the peptide sequence containing the threonine residue since it is found in the mouse-adapted challenge virus.
Three assays were used to try to identify cellular immune responses and were examined for their ability to predict protection accurately. The ELISPOT assay was the least sensitive of the three, often detecting only the most dominant responses. The ICC assay was more sensitive, identified more epitopes, and was generally predictive, with the single exception of the GP531 response in C57BL/6 mice. The 51Cr assay was the most reliable indicator of protection, as it accurately predicted protection when cells were lysed and predicted the lack of protection by the CD8+ T cells specific for the GP531 epitope. The lack of cytolytic activity could be from competition between the CD4+ and CD8+ T cells for the peptide; however, we could not identify any cytolytic activity even when the CD8+ cells were separated from the CD4+ cells during restimulation. This may be another example of a peptide that is presented but does not activate a cytolytic response. The inability of these CD8+ Tcells to protect mice, despite inducing IFN-γ, lends further support to the role of CTL effectors in protection, although wecould not confirm that the cells continued to secrete IFN-γ in vivo.
Our demonstration of protection by CTLs specific for 20 different epitopes in six proteins provides evidence that cellular immunity has a role in resolving Ebola virus infection in mice. This study, together with prior reports (8
), indicates a growing consensus that, in the mouse model of Ebola virus infection, both humoral and cellular responses contribute to protection. This is supported by our prior studies identifying protective MAbs and CTLs, our observation that Ab responses to GP arise after challenge in mice provided CTLs, and the observation that CTLs arise after challenge of mice vaccinated with VRP-GP or pretreated with protective monoclonal antibodies. Prior lack of information regarding T-cell epitopes recognized by mice has limited the ability to examine de novo responses occurring in animals surviving challenge, including in our prior antibody study (37
). We conducted preliminary studies to further examine the contributions of both antibodies and CTL responses in protection. We found that mice vaccinated with one protein, or passively administered a mixture of antibodies, generated new CTL responses after challenge, suggesting that both antibody and cellular responses are working together to provide protection. Either response alone may be able to limit virus replication until both arms of the immune response are present to clear infection.
The induction of both antibody and cellular responses was also reported in the studies that successfully protected nonhuman primates against filovirus challenge with adenovirus vaccination against GP and NP (27
). T-cell responses, assessed according to the production of tumor necrosis factor alpha, were demonstrated in five of eight macaques successfully protected by vaccination (27
) or with chimeric viruses (14
). If CTL responses contribute to protection, it is likely that the epitope specificity of the responses will vary among species, and ultimately among humans, as it does among mouse strains. It is not feasible to test every recipient of an eventual human use vaccine to evaluate CTL responses; therefore, producing a vaccine that includes multiple proteins increases the chances of inducing CTL responses in the majority of recipients. This study indicates that there are at least six Ebola virus proteins capable of inducing protective CTL responses. The GP elicits protective humoral immunity in addition to cellular immunity, but inclusion of the other ZEBOV proteins may be required in an eventual human use vaccine formulation to induce sufficient CTL responses in humans.