CD8 T cell cross-reactivity networks mediate heterologous immunity in human EBV and murine vaccinia virus infections

Abstract

In this study, we demonstrate complex networks of CD8 T cell cross-reactivities between influenza A virus and EBV in humans and between lymphocytic choriomeningitis virus and vaccinia virus in mice. We also show directly that cross-reactive T cells mediate protective heterologous immunity in mice. Subsets of T cell populations reactive with one epitope cross-reacted with either of several other epitopes encoded by the same or the heterologous virus. Human T cells specific to EBV-encoded BMLF1(280-288) could be cross-reactive with two influenza A virus or two other EBV epitopes. Mouse T cells specific to the vaccinia virus-encoded a11r(198-205) could be cross-reactive with three different lymphocytic choriomeningitis virus, one Pichinde virus, or one other vaccinia virus epitope. Patterns of cross-reactivity differed among individuals, reflecting the private specificities of the host's immune repertoire and divergence in the abilities of T cell populations to mediate protective immunity. Defining such cross-reactive networks between commonly encountered human pathogens may facilitate the design of vaccines.

Figure 1. Cross-reactive T-cell responses in humans

(A) Cross-reactive T-cells with specificity for EBV-BMLF1₂₈₀ and IAV-NP₈₅. CD8 T-cells derived from IM patient E1109 were cultured with BMLF1 peptide-pulsed T2 cells for 4-5 weeks. (i) The antigen specificity of the cell line was tested in a standard intracellular cytokine assay, where the cell line was restimulated for 5hr with a variety of different peptides. The percentage represents the proportion of the T-cell line producing IFNγ. (ii) A similar intracellular cytokine assay was combined with extracellular tetramer staining to demonstrate that a portion of T-cells bound to BMLF1-tetramer could also produce IFNγ following restimulation with IVA-NP₈₅. Tyrosinase peptide served as a non-specific stimulation control. The percentage represents the proportion of the T-cell line binding BMLF1₂₈₀-tetramer and producing IFNγ. Since the TCR was significantly downregulated upon BMLF1 restimulation, we have also included the total percentage of IFNγ-producing cells in parenthesis, as they would all presumably bind BMLF1-tetramer. (B) Cross-reactive T-cells with specificity for EBV-BMLF1₂₈₀ and EBV-LMP2₃₂₉. BMLF1-specific T-cell lines were derived from healthy donor D-002 and were cultured as described in (A). (i) Standard intracellular IFNγ assays are shown, where T-cell lines were restimulated with EBV-LMP2₃₂₉ or a non-specific peptide, tyrosinase. Positive and negative controls are also provided, BMLF1₂₈₀ or no restimulation respectively. The percentage of the T-cell line producing IFNγ is shown. (ii) An intracellular IFNγ stain was combined with an extracellular BMLF1₂₈₀-tetramer stain, where EBV-EBNA 3A peptide served as a non-specific stimulation control. The percentage of the T-cell line binding BMLF1₂₈₀-tetramer and producing IFNγ is shown. (C) Cross-reactive T-cells with specificity for EBV-BMLF1₂₈₀ and EBV-BRLF1₁₉₀ detected ex vivo. CD8 T-cells freshly isolated from (i) IM patient E1232 at day 13 post-presentation and (ii, iii) IM patient E1205 at days 0 and 19 post-presentation were co-stained with tetramers ex vivo, where (IVA)M1-loaded tetramers served as a negative control. (ii,iii)In the presence of EBV-BRLF1₁₉₀-loaded tetramer, EBV-BMLF1₂₈₀-specific cells co-stained dimly with BMLF1 tetramer or were blocked from staining with BMLF1 tetramer. . This competition between the tetramers appeared to be similar whether we added the tetramers together or sequentially.

Figure 2. T-cell cross-reactive network between unrelated human or murine viruses focusing on (a) EBV-BMLF1₂₈₀ and (b) VV-a11r₁₉₈

(A) Diagram of cross-reactive CD8 T-cell network in human EBV infection. The EBV-BMLF1₂₈₀ peptide can activate CD8 T-cell populations specific for 4 different peptides of rather dissimilar sequence derived from 2 different viruses, IAV and EBV. In the diagram of the cross-reactive T-cell responses focused on BMLF1_280, the fractions in the figure indicate the number of individuals of the total tested that demonstrated the indicated cross-reactive response. The epitopes in grey or white boxes are EBV and IAV specific, respectively. Cross-reactive responses were detected using both ex vivo and in vitro tetramer and intracellular cytokine staining assays. (B) Diagram of cross-reactive CD8 T-cell network in murine VV infection. The VV-a11r₁₉₈ peptide can activate CD8 T-cell populations specific for 5 different peptides of similar sequence derived from 3 different virus infections, LCMV, PV and VV. In the diagram of the cross-reactive responses focused on VVa11r_198, the fractions in the figure indicate the number of mice of the total tested that demonstrated the indicated cross-reactive response. There are three separate groups, where the light gray box indicates mice that are LCMV-immune (i.e. LCMV-contact), the white color box indicates mice that are VV-immune (i.e. VV-contact) and the dark gray box are mice that are PV-infected. Cross-reactive responses were detected using both ex vivo and in vitro tetramer and intracellular cytokine staining assays.

Figure 3. Cross-reactive T-cell responses in mice. Pattern of cross-reactivity varies between individual mice and prior infection history

VV-a11r-specific T-cell lines were generated from LCMV-immune (A-C), LCMV-immune+VV-immune (D), or VV-immune (E-e) C57BL/6 mice. VV-e7r-specific T-cell lines were generated from VV-immune mice (F-f). Indicated peptides were used for stimulation in ICS assays. Results are representative of 2 to 10 cell lines/group.

Figure 4

(A) Pattern of cross-reactivity is similar between 3 different cultures from one individual mouse. Three individual VV-a11r₁₉₈ cell lines (i, ii, iii) were generated from one LCMV-immune mouse and were stained with either one or two tetramers simultaneously. (B) Presence of cross-reactive CD8 T-cells ex vivo during VV infection (day 6) of LCMV-immune mice. Double-tetramer staining demonstrates the presence of cross-reactive CD8 T-cells that recognize both VV-a11r₁₉₈ and LCMV-GP₃₄ as well as VV-a11r₁₉₈ and LCMV-NP₂₀₅ (gated on CD8 cells). (C) Double tetramer staining of two a11r₁₉₈-specific T-cell lines generated from two different LCMV-immune mice (i) The majority of this 3 week culture of an a11r-specific T-cell line bound either LCMV-GP₃₄ or VV-a11r₁₉₈-tetramers when stained separately. Co-staining with both tetramers resulted in blocking of the LCMV-GP34 tetramer blocking binding of the VVa11r tetramer. (ii) In a second VV-a11r-specific T-cell line, after a short-term 10 day culture, co-staining experiments with LCMV-GP₃₄ tetramer blocked about 5% of the VV-a11r tetramer binding (gated on CD8).

Figure 5. Cross-reactive a11r-specific CD8 T-cells proliferate in vivo after LCMV or VV infection and reduce VV load

(A) This is a representative a11r-specific CD8 T cell line (L/a11r-4) derived from LCMV-immune splenocytes; it was predominantly cross-reactive with LCMV-GP34 but also recognized GP118 in an IFNγ ICS assay. (B) Cross-reactive a11r-specific line proliferates in response to LCMV and VV but not to PV-infected or PBS-treated controls, as assessed by loss of CFSE label by day 3 post the simultaneous adoptive transfer and infection of congenic LY5.1 mice. (C) Adoptive transfer of cross-reactive a11r-specific CD8 T-cell lines derived from an LCMV+VV-immune mouse (▲LV/a11r-3) or an LCMV-immune mouse (△L/a11r-2) led to a significant VV reduction compared to PBS controls (). VV titers in the testes were assayed on day 4 post-infection (LV/a11r-3 vs. PBS: p<0.01; L/a11r-2 vs. PBS: p<0.05). (D) Differential effect on VV titers upon the use of different cross-reactive a11r-specific CD8 T-cell lines, L/a11r-4 and L/a11r-5. There was a greater reduction of VV titers in testes day 4 post-infection in mice injected with L/a11r-4 () compared to L/a11r-5 (엯)(p<0.08, n=4), or PBS (●)(p<0.08, n=4). (E) Time course of weight loss after VV infection. Adoptive transfer of the T-cell line L/a11r-4 resulted in a 50% inhibition of weight loss (day 2: 2.4%±0.6% vs. 4.8%±0.7% weight loss, n=4, p=0.06; day 3: 2.3%±0.5% vs. 4.5%±0.7% weight loss, n=4, p=0.08) post VV-infection, while T-cell line L/a11r-5 did not affect weight loss when compared to the control, PBS injected, mice (p>0.3).

Figure 6. Two a11r-specific cross-reactive CD8 T-cell lines show different patterns of avidity to cross-reactive epitopes and different TCR repertoires

(A,B) Titration of peptide concentrations in ICS assays using a11r-specific lines L/a11r-4 and L/a11r-5. The indicated concentrations of VV-a11r₁₉₈, LCMV-GP₃₄, -GP₁₁₈, and -NP₂₀₅ peptides were used in a 4 hour ICS assay to stimulate the production of IFNγ and TNFα by the a11r-specific T-cell lines L/a11r-4 or L/a11r-5, which were generated from two different LCMV-immune mice. The percentage of IFNg+ or IFNg+ TNFa+ cells of total CD8+ is shown above each plot. The numbers above the upper quadrants represent the percent cytokine production (gated on CD8). The percentage of maximum cytokine response to each peptide concentration is graphed in B. (C) Private specificity in TCR Vβ repertoire: TCR Vβ mRNA expression of the L/a11r-4 and L/a11r-5 T-cell lines demonstrates different a11r-specific Vβ repertoires in lines derived from two different LCMV-immune mice.

Figure 7. Change of immune hierarchy in LCMV-immune mice infected with VV

(A). Lymphocytes were isolated from the peripheral blood of naïve mice (VV acute) or the same LCMV-immune mice before (LCMV-immune) or 6 days after VV infection (LCMV+VV acute). The lymphocytes were stimulated with the indicated peptide in a 6 hr intracellular IFNg assay. We show a representative example of how a typically subdominant VV-a11r₁₉₈ response during acute VV infection becomes immunodominant when the host has been previously exposed to LCMV. Percentage of CD44hi CD8 T-cells responding to the indicated peptide is shown. (B) A summary of the data derived from ICS assays shows that a shift in VV epitope immunodominance hierarchy among LCMV+VV mice occurred in 3 out of 9 mice analyzed (30%). The line drawn between the epitopes indicates the results from the same mouse.