Altered Cytokine Responses of Dengue-Specific CD4⁺ T Cells to Heterologous Serotypes ¹ Free

The CD4⁺ T cell plays a crucial role in protection against viral infection and in the development of memory B and CD8⁺ T cells. One of the primary functions of CD4⁺ T cells during a viral infection is the production of Th1 cytokines IFN-γ and TNF-α. These cytokines, among others, induce an antiviral state in the host, activate professional APCs for presentation of viral epitopes to CD8⁺ T cells, and play a role in the modulation of humoral and cellular immune response during the course of a viral infection (1, 2, 3, 4).

The dengue (DEN) ³ group of viruses is composed of four closely related but serologically distinct viruses, serotypes 1–4. DEN viruses usually cause a flu-like self-limited infection that imparts long-term immunity to the infecting serotype. However, subsequent DEN viral infection involving a heterologous serotype leads to an increased risk for the more severe DEN hemorrhagic fever (DHF) (5, 6, 7, 8). The humoral and cellular arms of the immune system are hypothesized to play a role in the immunopathogenesis of DHF during a secondary DEN virus infection. Cross-reactive nonneutralizing Abs from infections involving heterologous serotypes have been shown to increase viral infection of monocytes in vitro (9). DEN-immune T cells stimulated with DEN Ags and live virus produce IFN-γ and TNF-α (10, 11, 12, 13). In this study, we looked at the ex vivo cytokine responses of T cells from DEN-immune individuals to Ag and peptide stimulation at the single-cell level. We hypothesized that the patterns of cytokine production by DEN-specific memory T cells elicited after a primary infection plays a significant role in the development of the severe manifestations of secondary DEN viral infections.

We measured the percentage of DEN-specific CD4⁺ T cells in the peripheral blood of six donors immunized with live attenuated DEN vaccine candidates by staining for intracellular IFN-γ and TNF-α production after stimulation with cognate Ags. Cross-reactive responses to heterologous DEN Ags were observed. IFN-γ responses were highest against Ags derived from the homologous serotype, whereas the ratios of TNF-α- to IFN-γ-producing cells were higher in stimulations with heterologous DEN serotypes. Of seven HLA class II-restricted DEN epitopes examined, two elicited IFN-γ responses highest against the sequence from the homologous serotype. One peptide (NS3 584–598) showed only a TNF-α response. We compared peptide-specific CD4 T cell frequencies measured directly using HLA class II tetramer and by intracellular cytokine staining after peptide stimulation. Altered CD4⁺ T cell cytokine responses to heterologous Ags and HLA class II-restricted peptide-specific cytokine response patterns may play a role in DHF immunopathogenesis.

We studied six volunteers who had previously received candidate live attenuated DEN vaccines (14, 15, 16, 17, 18, 19) (Table I) as well as five flavivirus naive Massachusetts residents. PBMC were isolated from whole blood by density gradient centrifugation using Ficoll-Hypaque. The cells were cryopreserved in RPMI 1640 medium containing 20% FCS, 2% l-glutamine, 1% penicillin/streptomycin, and 10% DMSO. The cells were stored in liquid nitrogen and thawed immediately before Ag stimulation. Informed consent was obtained from volunteers, and the investigators have adhered to the policies for protection of human subjects as prescribed in Office of the US Army Surgeon General. These studies were approved by the institutional review board at the University of Massachusetts Medical School.

Virus Ags for each of the four DEN serotypes were prepared as lysates of DEN-infected Vero cells as previously described (20). Control Ags were prepared in a similar fashion using uninfected Vero cells.

Peptides from the Core and NS3 proteins of the DEN 3 and 4 viruses (Table II) were synthesized with a Symphony automated peptide synthesizer (Apex 396; Advanced Chemtech) at the Protein Core Facility of the University of Massachusetts Medical School. The peptides were solubilized in pure, sterile DMSO. Five peptides were previously defined as HLA class II-restricted epitopes recognized by CD4⁺ T cell clones (18, 21, 22, 23, 24, 25). Two were defined in this study. Serial 10-fold dilutions were used to stimulate PBMC from each donor to determine the optimum peptide concentration for stimulation (data not shown). A final concentration of 10 μg/ml was used for all peptides because this concentration induced maximal IFN-γ responses. Stimulation with the same concentrations of an irrelevant peptide did not induce any peptide-specific cytokine responses. The ratio of TNF-α to IFN-γ was not dose dependent and ranged 6–9 ratio points over a peptide dose range of 0.01–100 μg. This range of peptide concentrations did not induce peptide-specific IFN-γ or TNF-α in five flavivirus naive donors for all peptides (data not shown).

Cytokine production and measurement of cytokine production after Ag stimulation was done as previously described (26) with the following modifications. Cell phenotype was determined by cell surface staining using allophycocyanin-Cy7-conjugated anti-CD4, PE-Cy7-conjugated anti-CD3 (Caltag Laboratories), and PE-conjugated anti-CD56 (BD Immunocytometry Systems). Intracellular cytokines and T cell activation markers were detected using PE-Cy5-conjugated anti-CD69, FITC-conjugated anti IFN-γ, and allophycocyanin-conjugated anti-TNF-α mAbs (BD Pharmingen). Data acquisition was performed on a BD LSR II flow cytometer. A total of 300,000 events from the lymphocyte gate were collected for each analysis. The data were analyzed using FlowJo version 6.1.1 software (Tree Star). CD4⁺CD3⁺CD56⁻ cells were gated and analyzed for cytokine expression. A panel of HLA class II-restricted peptides and their corresponding peptide sequences from all four DEN virus serotypes were used to stimulate PBMC as described above at a final concentration of 10 μg/ml with a total stimulation time of 6 h. Brefeldin A and peptide were added to the PBMC at the same time. Cell surface phenotype staining was done using PE-conjugated anti-CD56, PerCP-Cy5.5-conjugated anti-CD3 (BD Immunocytometry Systems) and PECy7-conjugated anti-CD4 mAbs (Caltag Laboratories). Intracellular cytokines were stained with FITC-conjugated anti-IFN-γ and allophycocyanin-conjugated anti-TNF-α mAbs. Data acquisition and analysis was performed as described above.

Peptide-specific cells were quantified using PE-conjugated DRB*1501 HLA class II tetramers (generously provided by Immunomics, Beckman Coulter) loaded with either peptide DEN3 NS3 141 or DEN3 NS3 348. PBMC were incubated with tetramers for 2 h at 37°C, and then stained for cell surface phenotype using allophycocyanin-Cy7-conjugated anti-CD19 (as a dump channel), PECy7-conjugated anti-CD4, and PerCP-Cy5.5-conjugated anti-CD3 mAbs.

PBMC were stimulated with inactivated DEN viral Ags for 18 h to measure cytokine-producing CD4⁺ T cells. In keeping with our earlier results, maximal frequencies of IFN-γ⁺CD4⁺ T cells were seen against Ag preparations from the DEN serotype used to vaccinate each subject (Fig. 1, a and b). Varying degrees of cross-reactivity to Ags from heterologous serotypes were also observed. The frequency of IFN-γ⁺CD4⁺ T cells did not exceed 0.02% in PBMC from any of the flavivirus naive donors in response to any DEN Ag above the response to control Ag. Maximal TNF-α⁺CD4⁺ frequencies were also observed in response to Ag homologous to the candidate vaccine serotype in most cases (Fig. 2, a and b). Donor 4 was the exception with the highest response to heterologous DEN 2 Ag stimulation. Cross-reactive TNF-α responses were also observed against all heterologous serotypes. In general, the frequencies of TNF-α⁺-producing CD4⁺ T cells were higher compared with IFN-γ⁺CD4⁺ T cells. To account for the difference in magnitudes between the two cytokines measured, we looked at IFN-γ vs TNF-α staining in CD4⁺CD69⁺ T cells (Fig. 3 a). The majority of cytokine-positive cells produced either TNF-α alone or TNF-α⁺IFN-γ⁺; fewer cells produced IFN-γ alone, and this population was most apparent in samples stimulated with the homologous DEN Ag.

We measured the ratio of all cells positive for TNF-α (cells that were single-positive for TNF-α and double-positive for TNF-α and IFN-γ) to all cells positive for IFN-γ (cells that were single-positive for IFN-γ and double-positive for TNF-α and IFN-γ) against each DEN serotype Ag for each donor and compared responses against homologous and heterologous serotype Ags (Fig. 3,b). The ratio was measured for each experiment per donor, and the means of the ratios by serotype for each donor were calculated. We compared the ratio of cytokine responses between the homologous serotype Ag and each heterologous serotype Ag for each donor (Fig. 3 b). Significant differences of the ratio of responses at p < 0.05 was observed for donor 1 (homologous vs DEN 2 and DEN 4 Ag), donor 4 (homologous vs all heterologous Ags), donor 5 (homologous vs DEN 1 Ag), and donor 6 (homologous vs DEN 3 Ag).

For the group as a whole, mean TNF-α:IFN-γ ratios in the three samples per donor stimulated with heterologous Ags were significantly higher than the TNF-α:IFN-γ ratios in samples stimulated with heterologous Ags by Wilcoxon signed ranks test (p = 0.046). Highest variation was seen in donors 4 and 6, where fewer than five independent experiments were performed. The highest TNF-α:IFN-γ ratio was always found to be against Ags from a heterologous DEN serotype. The TNF-α:IFN-γ ratio was lowest in samples stimulated with the homologous serotype in most donors.

Two peptides (DEN3 NS3 141 and DEN3 NS3 348) that elicited dominant IFN-γ responses in an earlier study (26) were loaded onto DRB*1501 tetramers and used to stain peptide-specific CD4⁺ T cells directly (Fig. 4,a). DEN3 NS3 141 and DEN3 NS3 348-specific CD4⁺ T cells were measured to be 0.042 and 0.089% of total CD4⁺ T cells, respectively. The frequency of DRB*1501 DEN3 NS3 141 tetramer binding CD4⁺ T cells was the same as the frequency of DEN3 NS3 141-specific IFN-γ⁺CD4⁺ T cells (Fig. 4,b) but less than TNF-α⁺CD4⁺ T cells (Fig. 4 c). The frequency of DRB*1501 DEN3 NS3 348 tetramer binding cells was greater than both DEN3 NS3 348-specific IFN-γ- and TNF-α-positive cells.

HLA class II-restricted epitopes in the C and NS3 region were previously identified using cytotoxicity assays in conjunction with limiting dilution of virus-stimulated PBMC bulk cultures (Table II). We identified two additional HLA class II-restricted epitopes (Table II) using intracellular cytokine staining of PBMC from DEN 4-immune donor after stimulating with pools of overlapping 15-mer peptides that span the NS3 protein of the DEN 4 virus (data not shown).

We looked at peptide-specific responses by stimulating PBMC from donor 4 (DEN 3 vaccinee) and donor 5 (DEN 4 vaccinee) with HLA class II-restricted epitopes (Table II). Peptides NS3 141, 223, 251, and 348 were tested on PBMC from donor 4 (Fig. 5, a and c) and peptides C 81, NS3 186, and 584 were tested on PBMC from donor 5 (Fig. 5, b and d). Broad cross-reactive responses against all sequences from the four serotypes were observed with peptides NS3 141 and NS3 186 for both IFN-γ and TNF-α. Peptide NS3 223 induced IFN-γ responses to the DEN 2, 3, and 4 sequences and TNF-α responses to all four serotype sequences. Peptide 251 induced a DEN 1 serotype sequence-specific IFN-γ response but TNF-α responses to all four sequences. Peptide NS3 348 induced a DEN 3 serotype-specific response for both IFN-γ and TNF-α. Peptide C81 induced IFN-γ responses to DEN 2 and 4 and high TNF-α responses to all serotypes. Peptide NS3 584 did not induce any IFN-γ response but had TNF-α responses to all four serotypes. Although the ratios of TNF-α to IFN-γ responses for each of the seven peptides did not follow a single pattern, the lowest ratio of TNF-α:IFN-γ responses based on total response to all peptides of a given serotype was against the homologous serotype (Fig. 5 e).

TNF-α vs IFN-γ double cytokine plots of peptide-stimulated donor PBMC were analyzed to determine the functional phenotypes of the different DEN peptide-specific populations (Fig. 6). The gates shown on the figures represent IFN-γ single-positive (IFN-γ⁺), TNF-α single-positive (TNF-α⁺), and IFN-γ and TNF-α double-positives (IFN-γ⁺TNF-α⁺) Results from donor 4 PBMC are shown in Fig. 6,a. NS3 141, 223, and 251 induced a predominantly TNF-α⁺ response while NS3 348 had a predominantly IFN-γ⁺ response with all peptides inducing IFN-γ⁺TNF-α⁺ responses. Results from donor 5 PBMC are shown in Fig. 6 b. C81 and NS3 186 showed predominantly TNF-α⁺ responses to all four serotypes with moderate double cytokine responses. NS3 584 showed only TNF-α⁺ responses to all four sequences. C81 was the only peptide to induce only IFN-γ⁺TNF-α⁻ responses in this donor.

We report differential cytokine production from CD4 T cells in response to Ags from heterologous DEN virus serotypes in donors that received monovalent candidate live attenuated DEN vaccines. The pattern of DEN-specific CD4⁺ T cell IFN-γ responses in the vaccine recipients are consistent with our earlier report (26), in which the highest CD4 T cell frequencies were seen when homologous Ag was used to stimulate donor PBMC. Given the antiviral activity of IFN-γ against DEN, this may contribute to long-term immunity to reinfection with the same serotype (27). Cross-reactive IFN-γ responses were observed that may impart partial heterologous immunity to other DEN serotypes. However, we observed different patterns and levels of cross-reactive IFN-γ production particular to each donor.

In contrast to IFN-γ CD4 T cell responses in these donors, TNF-α CD4 T cell responses had broader cross-reactivity for Ags from heterologous serotypes and, in general, higher frequencies. This observation implies that memory T cells could produce relatively higher amounts of TNF-α upon encountering Ag from heterologous DEN serotypes. In particular, stimulation of PBMC with Ag from heterologous serotypes was shown to significantly induce a higher ratio of TNF-α-producing cells to IFN-γ-producing T cells when we analyze all donors as a group (Fig. 3 b, p = 0.046). When we exclude data from donor 6 from whom we observe low frequencies of DEN virus-specific CD4⁺ T cells, the data is no longer significant at p < 0.05 but there still is a trend toward significance (p = 0.08). When we exclude data from the donors with the highest variance and had four or less independent experiments (donor 3 and 6), the calculated p = 0.001. We also observe that the high variance calculated for donors 3 and 6 (DEN 2 and 4 vaccinee, respectively) is seen only against DEN serotypes 1 and 3. In general, DEN serotypes 2 and 4 genomic sequences have higher homology to each other as compared with serotypes 1 and 3. The higher variance observed may be attributed to this difference in sequence homology. For four of six subjects, stimulation of PBMC with homologous Ag gave rise to the lowest ratio of TNF-α- to IFN-γ-producing T cells. Furthermore, our results suggest that T cell cytokine responses to a secondary infection with a heterologous DEN serotype are qualitatively altered from the responses that are seen with the homologous serotype.

Four donors showed significant differences in the ratio of responses between homologous and heterologous Ags with varying patterns of differences (Fig. 3 b). It is difficult to interpret the results from this particular analysis with respect to possible relevance to DHF immunopathogenesis because these donors have not been observed to experience a secondary DEN infection.

We looked in more detail at the basis for altered cytokine responses by stimulating PBMC from two donors with specific HLA class II-restricted peptides from all four serotypes. The results obtained imply that the different viral epitopes contribute in different ways to overall cytokine responses.

Each of the peptides studied in donor 4 (DEN 3 vaccinee) elicited IFN-γ⁺, TNF-α⁺, and IFN-γ⁺TNF-α⁺ CD4 T cells. Peptide NS3 141, 223, and 251 showed varying degrees of serotype cross-reactivity while NS3 348 showed specific responses to the DEN 3 serotype peptide sequence as well as a higher frequency of IFN-γ⁺ CD4 T cells. The PBMC sample from donor 4 was tested 14 years postinfection and the DEN3 NS3 348 tetramer-specific cells were observed at a relatively high frequency. We speculate that DEN3 NS3 348 may be an epitope that contributes to long-term protection against the homologous serotype and is less likely to contribute to enhanced immunopathogenesis.

The peptides used to stimulate PBMC from the DEN 4 vaccinee induced more varied cytokine responses. IFN-γ⁺ responses were seen for only one epitope, C81; this peptide was isolated from the structural protein C. Although all three peptides elicited predominantly TNF-α responses, the two peptides from nonstructural proteins elicited IFN-γ⁺TNF-α⁺ and TNF-α⁺ responses only. Peptide epitopes from structural proteins can be presented on HLA class II molecules to CD4 T cells through the exogenous pathway of virus uptake, whereas epitopes from the nonstructural proteins go through the endogenous pathway in the cytoplasm and must be cross-presented on HLA class II molecules. Future studies involving additional DEN epitopes from the structural and nonstructural proteins are needed to determine whether the pathway of epitope presentation affects the cytokine profiles of epitope-specific CD4 T cells.

The ratio of TNF-α- to IFN-γ-positive T cell frequencies for the responses to individual peptides did not show the same pattern as stimulation with viral Ag. However, taking the sum of the responses of all the peptides by serotype we found a pattern of ratios of TNF-α:IFN-γ similar to Ag stimulation. Isolation of more peptide epitopes from other regions of the DEN genome would be needed to further strengthen the relationship between the results from stimulation with viral Ag and peptide epitopes.

We also compared results from intracellular cytokine staining and HLA class II tetramer staining to determine whether there are any differences between direct and function-based assays in measuring DEN-specific CD4⁺ T cells (Fig. 4). For DEN3 NS3 141, we observed a higher frequency of TNF-α secreting CD4⁺ T cells than tetramer binding cells, implying that there are tetramer-negative CD4⁺ T cells that can respond to the peptide by TNF-α production. In contrast, DEN3 NS3 348 tetramer binding CD4 T cells had a higher frequency than cytokine-producing cells (Fig. 6 a). This implies that there is a population of peptide-specific CD4⁺ T cells that did not produce either IFN-γ or TNF-α; whether these cells fail to respond or have other specific functions is yet to be determined. Studies using murine immunodominant epitopes from the influenza virus showed a hierarchy for cytokine production by peptide-specific CD8⁺ T cells (28). In contrast to our results with inactivated Ag stimulation and another study using HBV peptide pools (29), La Gruta et al. (28) found that TNF-α- and IL-2-producing CD8 T cells formed subsets of the larger IFN-γ⁺ population. Similar cytokine profiles were observed from DEN3 NS3 348-specific CD4⁺ T cells. This peptide may be an immunodominant DEN CD4 T cell epitope when we consider its higher IFN-γ production, long-term stability, and optimal binding with the tetramer.

To our knowledge, this is the first direct measurement of CD4⁺ T cells from an acute, nonrecurring, nonpersistent virus infection. Previous attempts to measure peptide-specific memory CD4 T cells from an acute, nonpersistent infection may have used epitopes that did not induce high enough frequencies of peptide-specific memory CD4 T cells detectable by tetramer staining. Alternately, previously defined CD4 T cell epitopes may have had less-than-optimal binding characteristics when used in HLA class II tetramers. The high mean fluorescence intensity (MFI) of DEN3 NS3 348 tetramer in Fig. 4 a (MFI = 758) illustrates its optimal binding properties compared with DEN3 NS3 141 (MFI = 121).

Differences in T cell responses following sequential heterologous viral infections have been demonstrated in mouse CD8⁺ T cells with unrelated viruses such as vaccinia virus, Pichinde virus, murine CMV, and lymphocytic choriomeningitis virus (30, 31). These viruses gave different but not necessarily reciprocal levels of cross-protection upon challenge with a heterologous virus. Within the DEN group of viruses, the differences in DEN-specific T cell TNF-α responses to heterologous virus serotypes suggest that a primary DEN infection can alter T cell responses to a secondary DEN infection. Because the donors studied did not experience secondary DEN infection, we are unable to determine which responses are associated with cross-protection or immune enhancement. The pattern of cytokine responses (i.e., higher TNF-α⁺ CD4 T cell ratios against Ags from heterologous vs homologous serotypes) would suggest that the balance of cytokines secreted by memory T cells during a secondary infection is important in conferring protection or enhancement of symptoms.

High systemic levels of TNF-α have been shown to cause capillary leakage (32). Serum levels of TNF-α and its receptors have been shown to be elevated in DEN infection and to correlate with DEN disease severity (33, 34). Although activated monocytes and macrophages produce TNF-α, it has been shown that DEN-specific T cell clones are also capable of producing TNF-α (12) and the present study shows the potential breadth of such responses in DEN-immune individuals. TNF-α production of presecondary infection PBMC after stimulation with inactivated Ag appears to be a risk factor for DHF during secondary infection (13). DEN-specific CD4⁺ T cells elicited from a primary infection may be activated to produce higher-than-normal amounts of TNF-α by Ags from a heterologous DEN serotype and could lead to capillary leakage.

We thank Jurand Janus, Mina Seedhom, and the staffs of the University of Massachusetts Worcester campus Core Flow Cytometry laboratory and Peptide Core Facility for technical assistance. We also thank Drs. Michelle Catalina and Rachel Gerstein for their expert technical advice on intracellular staining and flow cytometry. We thank Drs. Francis A. Ennis, Daniel Libraty, and Sharone Green for critical reading of this manuscript. We are especially grateful for all the subjects that volunteered for the experiments in this report.

The authors have no financial conflict of interest.