The generation of memory T cells following primary exposure to a pathogen is a critical feature of the vertebrate immune system which has evolved as a protective mechanism in order to defend the host against repeated assaults by the patnogen. Memory T cells are long-lived, undergo rapid proliferation upon re-activation, mediate a robust secondary response and clear the pathogen much more efficiently. These aspects have made the generation of memory T cells an attractive goal for the production of both prophylactic and therapeutic vaccines. However, the degeneracy of the T cell receptor, whereby a given T cell recognizes more than one epitope, allows the T cell to be modulated by epitope variants which could be self-ligands, ligands related to the original epitope but altered in sequence, or completely unrelated epitopes. Experiments in both mice and humans show that such cross-reactive stimulation of memory T cells results in complete, partial, or no activation of T cells, and in some cases, even alters the functional identity of the T cell (for example, T helper 1 cells start secreting IL-4, IL-5 and become part of a T helper 2 response). In the context of secondary infection of immune organisms with pathogens containing mutated or related T cell epitopes, such alterations at the cellular level translate into drastic changes in the overall clinical outcome of the infection. Thus, the presence of cross-reactive T cells in the memory population implies that the protective or pathologic nature of the secondary immune response is a consequence of the host's infection history. Although several murine models of heterologous infection resulting in altered pathological outcome have been studied, the exact immune correlates of protection versus immunopathology are still unclear. This thesis addresses this issue in dengue virus infections in humans. Dengue fever (DF) and Dengue Hemorrhagic Fever (DHF) are two disease manifestations caused by infections of humans by the dengue viruses. These are a group of 4 serologically distinct flaviviruses (D1-4) which often co-circulate among endemic populations. While primary infection with any of the four serotypes can result in the more severe clinical disease characterized by DHF, epidemiological data from several outbreaks show that 80% - 90% of DHF cases occur among individuals with secondary infection. This implies that prior immunity to dengue is actually a risk factor for developing severe disease. In these DHF cases, there are increased numbers of CD69+ CD8+ T cells in circulation, with increases observed in the frequency of epitope-specific T cells, and the serum levels of several T cell produced cytokines, chemokines, and immune receptors are highly elevated. Since the four serotypes share 65% - 75% amino acid sequence homology, the possibility that unconserved T cell epitope sequences stimulated cross-reactive responses was borne out in in vitroexaminations. In these studies, peripheral blood mononuclear cells (PBMC) and cloned T cells from both vaccinated and infected donors contained large populations of memory T cells that were cross-reactive for heterologous viral serotypes in proliferation and CTL assays. These data suggest that the severity of disease seen in DHF patients can be attributed to an immunopathologic secondary response during heterologous infection, and highlight a role for serotype cross-reactive T cells in this process. This thesis addresses the hypothesis that the recognition of the natural variants of dengue virus T cell epitopes by serotype cross-reactive CD8+ T cells of a dengue-immune donor results in an altered secondary response profile, with the changes reflected in both the quantitative and qualitative nature of the response. In order to compare the functional profile of the secondary response of dengue-immune PBMC re-activated with heterologous serotypes, we focused on a panel of 4 donors who were vaccinated with live attenuated monovalent vaccines corresponding to D1, D2, or D4 serotypes. We screened a panel of peptides predicted to bind to HLA-A*0201 for cytokine responses and identified 4 novel epitopes that were highly immunogenic in all four donors. Direct ex vivo stimulation of donor PBMC with the heterologous sequences of these epitopes also showed sizeable serotype cross-reactive T cell populations. CFSE- and intracellular staining for cytokines and chemokines showed that these cross-reactive T cells not only expanded but also produced IFNγ, TNFα, and MIP-1β. Multi-parameter staining revealed functionally diverse populations comprised of single cytokine (IFNγ+, TNFα+, MIP-1β+, double cytokine (IFNγ+TNFα+, IFNγ+MIP-1β+, TNFα+MIP-1β+, and triple cytokine (IFNγ+TNFα+MIP-1β+ secreting sub-sets. Stimulation with the epitope variants altered the magnitude of the overall response as well as the relative sizes of these sub-sets. The patterns of responses revealed the effects of epitope immunogenicity, infection history and donor-specific variability. All 4 donors showed the highest cytokine response to a -single epitope (NS4b 2353). The same two peptide variants (D2 NS4a 2148 and D3 NS4b 2343) induced the highest response in all 4 donors regardless of the serotype of primary dengue infection. Interestingly, the epitope variants which showed the highest immunogenecity in our donors corresponded to the D2 and D3 serotypes which have been documented as being more virulent as well as a viral risk factor for DHF. In one donor, the response to all peptide variants was dominated by the same cytokine sub-sets. These data suggested that the dengue-immune memory T cell repertoire was functionally diverse and underwent alterations in size after secondary stimulation. Therefore, we also investigated the effect of epitope variants on dengue-specific CD8+T cell clones isolated from vaccinated and infected donors in order to determine if epitope variants induced altered functional outcomes at the clonal level. The epitope variants functioned either as strong agonists (particularly the D2 and D3 sequences), partial agonists, or null ligands. Some variants were able to induce cytolysis but not other effector functions at low concentrations. The variant ligands also influenced the hierarchy of cytokine responses within each clone. The third part of this thesis focused on the characterization of the frequency and phenotypic profile of epitope-specific CD8+ T cells in patients with DHF and DF at different times in the disease course in order to better understand the kinetics of the response and delineate any differences between the immune profile of severe vs. moderate disease. Tetramer staining for a previously identified HLA-B*07 restricted epitope was combined with staining for activation markers (CD69, CD38, HLA-DR), homing receptors (CCR7, CD62L), and programmed death receptor 1 (PD-1). The DHF subjects had early T cell activation with higher frequencies of tetramer+CD69+ cells as compared to DF subjects, in whom T cell frequencies peaked around the time of defervescence. While each subject had a unique phenotypic profile of tetramer+ cells, there was a difference between DF and DHF subjects in terms of CCR 7 expression; all subjects expressed low levels of CCR7 during acute illness but only the DHF subjects did not show upregulation of CCR7 on tetramer+ cells during convalescence. These data suggest that there is a sustained alteration in memory phenotype in those who recovered from severe dengue disease. A majority of the tetramer+cells also expressed PD-1 during acute illness but not during convalescence. Double-staining with variant tetramers allowed us to directly visualize serotype cross-reactivity of the epitope-specific population, and showed that secondary stimulation did induce the expansion of cells with low avidity for that secondary serotype and higher avidity to the variant. Furthermore, the ratios of these sub-sets changed during the course of the response. Taken together, these studies suggest that the immune response to heterologous secondary dengue infection is mediated by a heterogeneous population of serotype-cross reactive T cells that have different functional avidities to epitope variants and is influenced by the serotype of the secondary infection as well as the prior infection history of the individual. The preferential expansion of clones which secrete IFNγ but not inflammatory MIP-1β or TNFα or a repertoire characterized by a higher ratio of cytolytic to cytokine producing clones could limit immune mediated damage while efficiently clearing the virus. This information will be useful in the design of vaccine strategies aimed at inducing protective cross-reactive responses against all 4 dengue serotypes while preventing immunopathological outcomes following secondary infection.