In this study, we have demonstrated that memory CD8+ T cells of single specificity induced by immunization with DCs pulsed with viral peptides protect from an acute lethal viral disease. Furthermore, we showed that CD8+ T cells directed to the ID as well as to those SD determinants that were effective at inducing a significant CD8+ T cell response upon DC-peptide immunization were highly protective. Moreover, we showed that protection can be achieved even in the absence of NK cells, which are essential for resistance to primary ECTV infection.
Other laboratories have previously studied the differential protective abilities of memory CD8+
T cells specific for single ID or SD determinants during LCMV infection (20
). However, the pathogenesis of LCMV is very different from that of ECTV. Natural LCMV infection in the mouse occurs in utero
and results in a chronic infection rather than an acute disease (4
). Depending on the dose, route, and clone, experimental intraperitoneal (i.p.) or i.v. infection results in transient acute or chronic infection without major symptoms and causes fatal meningitis only after intracerebral inoculation.
Somewhat analogous studies have also been performed following infection with respiratory viruses. For example, Fu et al. generated a DNA construct encoding full-length NP with two mutations (NPmut) that eliminated the ID determinant NP147-155 from influenza virus A/PR/8/34. This allowed for the detection of the immunorecessive determinant NP218-226 (19
). NP218-226 behaves as a typical immunorecessive determinant in that specific CD8+
T cell response, which can be detected only when the ID determinant is absent during priming (39
). BALB/c mice were immunized intramuscularly with NPmut DNA and were protected against cross-strain challenge with A/HK/68 (H3N2). Also, Cole et al. demonstrated that the hierarchy of CD8+
T cell determinants recognized in Sendai virus can be selectively altered by immunization against an SD determinant, with the resulting CD8+
T cell response following virus challenge directed predominantly to the subdominant determinant (8
). In addition, Kast et al. showed that peptide immunization with the ID peptide of Sendai virus protected mice from a lethal challenge (27
). In these experiments, protection conferred by memory CD8+
T cells specific for an SD determinant was not assessed. These studies differed from ours because, different from ECTV infection, influenza and Sendai viruses produce disease by replicating at the primary site of infection rather than by spreading systemically. Moreover, we examined protection by CD8+
T cells against subdominant rather than immunorecessive determinants, and we found that the response to the ID determinant was not abrogated in the presence of memory cells to the SD determinant.
Regarding infection with the related OPV VACV, studies of DNA vaccines containing ID or SD determinants from simian or human immunodeficiency virus showed a reduction in virus titers in ovaries of mice infected i.p. with recombinant VACV expressing the relevant determinants (24
). Snyder et al. showed protection against lethal secondary intranasal (i.n.) VACV challenge in HLA-A2 transgenic mice by vaccination with an MHC-I-restricted T cell determinant. However, mice with a memory CD8+
T cell response to a single determinant did not have complete protection, as some mice lost weight and some mice died. This did not occur in mice previously immunized with the whole virus (46
). Cornberg et al. showed that VACV-E7R-specific memory CD8+
T cells reduced viral load in the fat pads of mice following a nonlethal dose of VACV inoculated i.p. Previously, we showed variable levels of protection against i.n. VACV challenge in mice immunized 12 days earlier with synthetic SD determinants (37
). In agreement, here we also have shown that immunization against the SD epitopes KSYNYMLL and SIFRFLNI resulted in high frequencies of peptide-specific CD8+
T cells and very strong protection. On the other hand, immunization with ITYRFYLI- or STLNFNNL-pulsed DCs was not effective at inducing memory CD8+
T cells, and protection was almost nil even though their affinity for MHC-I is very high, ~6 and ~12 nM, respectively (37
). Of interest, while in this study ITYRFYLI and STLNFNNL were very poor immunogens, they were immunogenic and protective against VACV in our previous report (37
). At this point we can only speculate about the reasons why DC immunization failed to induce protective responses to these peptides. A major difference with the previous report is that VACV is not a natural pathogen of the mouse, replicating poorly in this host. In addition, i.n. VACV infection is mainly a local infection that produces pneumonia (31
), while footpad infection with ECTV causes systemic disease following lympho-hematogenous spread (4
). As we have previously shown, a major mechanism whereby memory CD8+
T cells protect from mousepox is by curbing lympho-hematogenous spread. Another difference between the two studies is that here we used DC-peptide immunization and analyzed protection at 6 to 8 weeks after immunization, while in the previous report we examined protection by virus-specific CD8+
T cells at 12 days postimmunization with peptide in incomplete Freund adjuvant (IFA) and with an MHC-II helper peptide. The time of challenge and methods of immunization may have been responsible for the differences observed. For example, the challenge with ECTV was performed during the memory phase of the response, while the challenge with VACV was done when the CD8+
T cells were still effectors and when IFA inflammatory signals may have still been present at the time of challenge. It is also possible that DC immunization failed to selectively induce responses to some peptides, even though they have high affinity for MHC-I. In support, we have been unable to induce responses to the influenza virus A/PR8/34 NP immunodominant epitope ASNENMEM by DC immunization, even though it has an 8 nM affinity for Db
). DCs have an endopeptidase activity at their plasma membrane that has been shown to degrade the Kb
-restricted tyrosinase epitope YMDGTMSQV, precluding its recognition by CD8+
T cells (2
). Thus, it is possible that, similar to YMDGTMSQV, peptides such as ITYRFYLI, STLNFNNL, and ASNENMETM, but not the immunogenic peptides, are unsuitable for DC immunization because they are preferentially degraded at the surface of DCs. Another possibility is that, despite their high affinity for MHC-I, the half-lives of the peptide–MHC-I complexes at the surface of cells is shorter for the nonimmunogenic than the immunogenic peptides. As an example, the half-life of Db
-ASNENMETM at the cell surface is 6 h 15 min, quite shorter than that of TGICNQNII (9 h 30 min), another high-affinity NP peptide (45
While normally resistant to footpad infection, B6 mice infected with ECTV i.n. succumb with respiratory complications. Relevant to our studies, Tscharke et al. reported that B6 mice immunized with splenic DCs pulsed with TSYKFESV were partially protected from i.n. challenge with ECTV (48
). Those authors suggested that the lack of complete protection could have been due to insufficient numbers of TSYKFESV-specific CD8+
T cells induced by their method of immunization, and they indicated that there may have been at least ~50-fold fewer TSYKFESV-specific CD8+
T cells than with VACV infection. In support of this view, our prime-boost method of immunization resulted in strong responses to some but not all the peptides. We observed complete protection only when the frequency of memory CD8+
T cells was high. In agreement with our findings, West et al. demonstrated that a high frequency (105
) of virus-specific memory CD8+
T cells from P14 transgenic mice were able to rapidly reduce or clear LCMV clone 13 virus (55
). Thus, independent of the reason for the inability of ITYRFYLI- and STLNFNNL-pulsed DCs to induce a response, our data suggest that protection strongly correlates with productive immunization and that the immunogenicity of a peptide may vary with the method of immunization. Hence, when designing vaccines, it is important to determine the efficiency of CD8+
T cell induction by the different determinants with the specific immunization method.
Different from any of the other studies, we also analyzed the primary CD8+ T cell responses to the ID and SD determinants in mice with preexisting memory CD8+ T cells specific for the ID or an SD determinant. Interestingly, the SIFRFLNI SD response was undetectable in mice immune to the ID TSYKFESV. However, the presence of memory CD8+ T cells to SD SIFRFLNI did not override the ID response to TSYKFESV, suggesting that these primary effectors could contribute to the protection.
We previously showed that NK cells migrate to the D-LN of ECTV-infected mice and use perforin and IFN-γ-dependent mechanisms to reduce virus spread (10
). We have more recently shown that B6.D2-D6 mice are susceptible to mousepox because they lack CD94, resulting in deficient control of ECTV by NK cells (11
). Our finding that memory CD8+
T cells protect B6.D2-D6 from mousepox provided a first line of evidence that NK cells may no longer be required for resistance to mousepox when memory CD8+
T cells are present. However, a final conclusion could not be drawn because in B6.D2-D6, NK cells still migrate to the D-LN and produce IFN-γ following ECTV infection (11
). Thus, our results showing that B6 mice immunized with DC-TSYKFESV remain resistant to mousepox after NK cell depletion definitively demonstrate that NK cells are not required when protective memory CD8+
T cells are present.
In summary, our study provides us with a better understanding of the mechanisms of acquired protection to highly infectious OPV. In addition, because ECTV spreads through the lympho-hematogenous route, our findings may be relevant for the many unrelated viruses that spread via this route (17
). Moreover, our work contributes to the efforts of rational vaccine development by providing information about mechanisms of acquired protection that may be applicable to other pathogenic viruses that cause acute or chronic viral diseases.