Human T cell responses to Japanese encephalitis virus in health and disease

Abstract

Japanese encephalitis (JE) virus (JEV) is an important cause of encephalitis in children of South and Southeast Asia. However, the majority of individuals exposed to JEV only develop mild symptoms associated with long-lasting adaptive immunity. The related flavivirus dengue virus (DENV) cocirculates in many JEV-endemic areas, and clinical data suggest cross-protection between DENV and JEV. To address the role of T cell responses in protection against JEV, we conducted the first full-breadth analysis of the human memory T cell response using a synthetic peptide library. Ex vivo interferon-γ (IFN-γ) responses to JEV in healthy JEV-exposed donors were mostly CD8(+) and targeted nonstructural (NS) proteins, whereas IFN-γ responses in recovered JE patients were mostly CD4(+) and targeted structural proteins and the secreted protein NS1. Among patients, a high quality, polyfunctional CD4(+) T cell response was associated with complete recovery from JE. T cell responses from healthy donors showed a high degree of cross-reactivity to DENV that was less apparent in recovered JE patients despite equal exposure. These data reveal divergent functional CD4(+) and CD8(+) T cell responses linked to different clinical outcomes of JEV infection, associated with distinct targeting and broad flavivirus cross-reactivity including epitopes from DENV, West Nile, and Zika virus.

Figure 1.

Breadth of the T cell response to peptide pools of JEV in JEV-exposed healthy donors. (A) Relative frequency of ex vivo IFN-γ responses were measured by ELISPOT or ICS in JEV-exposed healthy donors (n = 35, 29 for ELISPOT, and 6 for ICS). Peptide pools are shown grouped by viral proteins. For a subset of five subjects, ICS and ELISPOT were performed at least three times with consistent results. C, core. E, envelope. (B) Spot-forming cells (SFCs) per million PBMCs were measured by ELISPOT in 13 healthy JEV-exposed donors (18 responses, black circles) and three DENV-exposed subjects (four responses, red triangles). (C) Proliferative responses were measured by CFSE dilution and flow cytometry in healthy JEV-exposed donors once per subject. Data are relative frequency (n = 24) for CD4⁺ and CD8⁺ T cells. (D) Based on data from ICS assays, the proportion of the total IFN-γ response produced by CD8⁺ T cells in each healthy JEV-exposed donor was calculated. The bar depicts the median. n = 11.

Figure 2.

CD8⁺ T cell responses are highly flavivirus cross-reactive in healthy JEV-exposed donors. (A) ICS assays were used to detect IFN-γ⁺/TNF-α⁺ cells from healthy JEV-exposed donor H008/4. Example flow cytometry data from an ex vivo assay (top) and a short-term T cell line (bottom) show responses to variant peptides of JEV NS5 MTTEDMLQVW, gated on live, CD3⁺, and CD8⁺ cells, representative of three experiments. Similar results were obtained with DENV4 and WNV peptides (not depicted). Axes are log₁₀ fluorescence units. (B) IFN-γ responses to peptide titrations of the same NS5 peptides as in A and WNV peptide MTTEDMLEVW were measured by ex vivo ELISPOT. The results are representative of two independent experiments. SFC, spot-forming cell. (C) Cytokine (IFN-γ⁺, TNF-α⁺, or MIP-1β⁺ in any combination) responses to NS3 peptide titrations of JEV, DENV1–4, and yellow fever virus (YFV) presented on a B cell line matched for HLA B*08:01 were measured by ICS. Responding cells were CD8⁺ T cell lines (TCL) from a subject reporting dengue illness and yellow fever vaccination but not JEV exposure (H001/4), expanded with JEV (left) or DENV (right) peptides, each assayed against all three peptides. The black diamonds indicate peptides with no B cell line. Open squares indicate a peptide-pulsed HLA-mismatched B cell line. Peptide titrations by ex vivo ELISPOT or ICS gave similar results, and expansion of T cell lines from two further independent samples showed equal cross-reactivity. (D) IFN-γ⁺/TNF-α⁺ cells of a CD8⁺ T cell line from healthy JEV-exposed donor H007/3 were measured by ICS to a JEV NS5 peptide and the DENV variant peptides indicated, showing epitope conservation in variant peptides. The same experiment was performed twice ex vivo with similar results. (E) Cross-reactivity of ex vivo responses and short-term T cell lines were measured by ICS. Cross-reactivity index (variant response/JEV response from the same assay) between JEV and DENV1–4/WNV in five subjects of ex vivo measurements correlated with T cell lines (Spearman’s R = 0.62; P = 0.002). Ex vivo assays were performed at least twice in all donors except one with similar results. (F) Cross-reactivity of short-term T cell lines expanded with JEV peptides with WNV, and DENV1–4 measured by ICS. Data are cross-reactivity indices as in E of all T cell responses identified in healthy JEV-exposed donors. Wilcoxon signed rank tests showed no significant differences from a hypothetical value of 1 (indicating an equal response to JEV and variant peptides). In two subjects, these assays were repeated three times with good agreement. (G) CD8⁺ T cells from donor H008/4 were stained with JEV peptide–HLA class I tetramers (x axis) and variant peptide tetramers (y axis) and analyzed by flow cytometry. Data are representative of two independent experiments. (H) CD8⁺ T cells were tetramer stained and analyzed as in G. Titrations of three different tetramers are shown: JEV (MTTEDMLQVW), WNV (MTTEDMLEVW), and DENV2+3 (MTTEDMLTVW; common epitope). Data are median fluorescence intensity (left) and percentage of CD8⁺ T cells stained (right). Repeating this experiment using a short-term T cell line expanded with JEV peptides gave similar results. (I) Neutralizing antibodies against JEV and all four DENV serotypes were measured by PRNT₅₀. The bars depict the median. Four subjects had some assays repeated with similar results.

Figure 3.

Function of T cell responses to peptide pools of JEV in healthy JEV-exposed donors. (A) WB ICS assays were used to measure cytokine⁺ cells from 11 healthy JEV-exposed donors. Example data from a WB ICS experiment (donor H014/3) are shown. Axes are log₁₀ fluorescence units. (Top) IFN-γ versus TNF-α. (Bottom) IFN-γ versus MIP-1β. (B) WB ICS data were analyzed using SPICE software for 10 donors; 1 donor did not have the full flow cytometry panel performed. The bars indicate median and interquartile range (IQR). Pie slices indicate the fraction of response of the group that expresses cytokines four, three, two, and one in any combination. (C) Magnitude of CD4⁺ and CD8⁺ WB ICS IFN-γ⁺/TNF-α⁺ responses in healthy JEV-exposed donors measured by ICS (n = 11, 15 CD8⁺ responses and 3 CD4⁺ responses). (D) Secreted cytokine concentrations were measured by bead array from 12 healthy JEV-exposed donors after 5 d of stimulation with JEV peptide pools. The heat map depicts log₁₀-transformed picograms/milliliter. Some subjects had more than one peptide pool assayed. Data are depicted after subtraction of values obtained from unstimulated cells. This experiment was performed once. (E) Relative frequency of secreted cytokine responses from the same data as in D after 2 and 5 d of stimulation with JEV peptide pools. Subjects were considered positive if any pool was positive at either time point.

Figure 4.

CD8⁺ T cell responses from JEV-exposed donors show a cytotoxic phenotype. (A) Degranulation of CD8⁺ T cells in response to JEV and DENV variant peptides was measured by CD107a staining/flow cytometry. Representative data are from one of two experiments in a healthy JEV-exposed donor (H008/4). (B) Degranulation responses to variant flavivirus peptides were measured as in A in naturally JEV-exposed subjects (n = 7; six healthy and one recovered JE patient) and subjects reporting dengue illness (n = 2). Dengue illness and recovered JE are indicated by open, red triangles and purple circles, respectively. (C) Degranulation responses were measured as in A to titrations of JEV and DENV variant peptides in two healthy JEV-exposed donors (left, H001/1; middle, H008/4) and a subject reporting dengue illness (right, H001/4). (D and E) JEV-specific CD8⁺ T cells were identified by IFN-γ ICS and costained for perforin (D) and granzyme B (E). Representative data from one (perforin, JE054/2; granzyme B, H008/4) of four subjects are shown. (F) IFN-γ/granzyme B⁺ and CD107a/granzyme B⁺ double-positive CD8⁺ T cells were detected in the same experiments as in D and E in one healthy JEV-exposed and two dengue-exposed donors in response to all peptides tested: four JEV and two DENV peptides. (G) Degranulation of a short-term T cell line from a healthy JEV-exposed donor (H007/2) in response to JEV and DENV variant peptides measured as in A and representative of five donors tested. (H) Data from the same experiment as in G showing cross-reactivity for DENV peptides. Cross-reactivity was observed in two independent experiments, and one included CD107a staining. (I) CD8⁺ T cell IFN-γ, TNF-α, MIP-1β, and CD107a responses were measured by ICS and analyzed using SPICE. Six healthy JEV-exposed donors (seven JEV responses), four of whom had variant DENV peptides tested (15 DENV peptides), were included. In two subjects, repeat experiments showed similar results. The bars indicate median and IQR, and pie slices indicate the fraction of response showing four, three, two, and one function in any combination. (J) Peptide-pulsed, CFSE-labeled, HLA-matched targets were incubated with CD8⁺ T cell line effector cells, and the percent specific killing was measured by flow cytometry in response to JEV and DENV peptides. The peptides used were: donor H001/4 (DVMCHATL [JEV] and DLMCHATF [DENV]) and donor H008/4 (MTTEDMLQVW [JEV], MTTEDMLSVW [DENV1], and MTTEDMLTVW [DENV2/3]). Diamonds indicate lines expanded with JEV peptide, and squares indicate DENV peptide. Assays were performed in duplicate for each T cell line/peptide pair. Error bars represent standard error of the mean.

Figure 5.

Breadth of the T cell response to JEV peptide pools in recovered JE patients. (A) WB ICS assays were used to measure cytokine⁺ cells in recovered JE patients. Example flow cytometry data from four patients are shown. Axes are log₁₀ fluorescence units. (Top) CD8⁺ responses, IFN-γ versus TNF-α (left) and IFN-γ versus MIP-1β (right). (Bottom) CD4⁺ responses, IFN-γ versus TNF-α (left) and IL2 versus TNF-α (right). (B) Ex vivo IFN-γ responses from 39 recovered JE patients were measured by ELISPOT (n = 17) or ICS (n = 22), and relative frequency of responses to each peptide pool was calculated. (C) The relative proportion of the cytokine response produced by CD4⁺ T cells in individual patients was measured by ICS in the same experiments as in A and B. The bar depicts the median. (D) Proliferation responses were measured by CFSE dilution and flow cytometry in 18 recovered JE patients of whom 15 showed responses. The relative frequency for CD4⁺ T cells and CD8⁺ T cells is shown. (E) Proliferative responses from the same experiments as in D were equal in size between good outcome and poor outcome for CD4⁺ cells (25 responses in eight patients with good outcome and 13 responses in four patients with poor outcome; Mann Whitney U test); CD8⁺ responses were larger in poor outcome (23 responses in eight patients with good outcome and 17 responses in six patients with poor outcome). The bars depict the median and IQR, whiskers are 1.5 × IQR, and outliers are shown as points. (F) Cross-reactivity of short-term T cell lines expanded with JEV peptides with DENV1–4 and WNV were measured by ICS. Data are cross-reactivity indices (as in Fig. 2 E) of 14 T cell responses identified in seven recovered JE patients. Wilcoxon signed rank tests showed significant differences from a hypothetical value of 1 (i.e., the response to JEV and variant peptides was not equivalent) for all viruses except DENV1. Number of peptides tested: WNV = 13; DENV1 = 9; DENV2 = 14; DENV3 = 11; DENV4 = 11. (G) NAbs against JEV and all four DENV serotypes were measured by PRNT₅₀. The bars depict the median. Assays were generally performed once per person, and nine subjects had some assays repeated with similar results. (H) JEV PRNT₅₀ were the same in JE patients with good (n = 7) and poor (n = 9) outcome (Mann Whitney U test). The bars depict the median and IQR. C, core protein. E, envelope protein.

Figure 6.

T cell cytokine responses to JEV peptide pools in recovered JE patients. (A–C) Recovered JE patients showing CD4⁺ T cell responses were categorized into good outcome (complete recovery; n = 16) and poor outcome (recovery with residual neurological deficit/disability >6 mo after disease; n = 11). Cytokine responses were measured by WB ICS once per subject. 30 responses were detected in the good outcome group and 22 responses in the poor outcome group. (A) IFN-γ responses were smaller in those with poor outcome (Mann Whitey U test). (B) TNF-α responses showed no difference with outcome (Mann Whitey U test). (C) CD4⁺ T cell responses were analyzed using SPICE according to the outcome from JE. Patients with poor outcome had fewer polyfunctional cells (SPICE pie comparison, 10,000 replicates), with the difference largely accounted for by fewer IFN-γ⁺/TNF-α⁺/IL2⁺ cells (P = 0.001 for a Mann Whitney U test) and more TNF-α⁺ cells (P < 0.0001). The data were analyzed for individual responses; summing by subjects retained the strong significance of the result. (D) Secreted cytokine concentrations were measured by bead array from 11 recovered JE patients after 5 d of stimulation with JEV peptide pools. The heat map depicts log₁₀-transformed picograms/milliliter. Some subjects had more than one peptide pool assayed. Data are depicted after subtraction of values from unstimulated cells and are displayed according to outcome from JE. (Top) Good outcome, n = 7. (Bottom) Poor outcome, n = 4. This experiment was performed once. C, core protein. prM, pre-membrane protein. E, envelope protein. NS, non-structural protein. (E) Relative frequency of secreted cytokine responses from the same experiments as in D after 2 and 5 d of stimulation with JEV peptide pools. Subjects were considered positive if any pool was positive at either time point. (F) CD8⁺ T cell responses from recovered JE patients were assayed once per subject by WB ICS (n = 14; 18 responses) and analyzed using SPICE. (G) There was no difference in years since admission with clinical JE and sampling for this study between the good and poor outcome groups (Mann Whitney U test). Good outcome, n = 22. Poor outcome, n = 16.

Figure 7.

Peptide epitopes of JEV identified in healthy JEV-exposed donors, recovered JE patients, and DENV-exposed donors. (A–C) Peptides were identified in ICS assays of short-term T cell lines expanded with peptide pools showing ex vivo responses or by ex vivo ELISPOT. Both peptides that have been mapped down to the minimal epitope (eight peptides) and library peptides of 15–18 amino acids are shown. CD4⁺ responses are green, and CD8⁺ responses are blue. Location of peptide responses in healthy JEV-exposed donors (A), recovered JE patients (B), and two subjects reporting dengue illness (C) are shown. Three peptides in common between JE patients and either healthy group are marked by dotted lines. (D) Shannon entropy (H index) was calculated using all the available DENV and JEV complete polyprotein sequences from India (13 DENV1 sequences, 10 DENV2 sequences, 8 DENV3 sequences, 5 DENV4 sequences, and 8 JEV sequences) and the sequences of the DENV1–4 and JEV peptide libraries used in this study. The H index varies from 0, corresponding to a single conserved amino acid residue at that position, to 4.322, where all 20 amino acids are represented equally. (E) Schematic representation of JEV proteins corresponding to their size. C, core protein. prM, pre-membrane protein. ENV, envelope protein. NS, non-structural protein. (F) The mean Shannon entropy of the flavivirus regions corresponding to 15 unique peptides identified in the healthy JEV-exposed group and 25 unique peptides identified in the recovered JE group (using the same virus polyprotein sequences as in D) was significantly lower in the epitopes identified from recovered JE patients (Mann Whitney U test). The bars depict the median and IQR, whiskers are 1.5 × IQR, and outliers are shown as points.