INTRODUCTION

Top Abstract Introduction Materials & Methods Results Discussion References

Dengue hemorrhagic fever (DHF), the severe form of dengue illness, is hypothesized to be the result of the overactivation of the immune system (15). In its most serious form, dengue shock syndrome (DSS), the plasma leakage of DHF leads to shock and can be life threatening. Dengue virus infections are commonly seen in South Asia, Southeast Asia, the Caribbean, and South America. In the United States, the disease is seen in residents returning from travel to tropical countries. Antibodies to one serotype of dengue virus have been shown to enhance infection with another serotype of dengue virus in vitro by a process known as antibody-dependent enhancement (9). Epidemiological observations suggest that antibodies capable of mediating antibody-dependent enhancement in vitro are a risk factor for developing DHF or DSS and that complications of dengue illness are more commonly seen in secondary infections (20). These observations suggest that vaccines against dengue should induce protection against all four serotypes.

Our laboratory has been involved in the understanding of the human CD4⁺ and CD8⁺ cytotoxic T-lymphocyte (CTL) responses to dengue virus infection. We have identified CD4⁺ and CD8⁺ T-cell responses to several proteins of dengue virus in previously unexposed healthy Caucasians who received live candidate monovalent dengue virus vaccines (6, 19). These studies have shown a predominance of recognition of serotype-cross-reactive epitopes on the nonstructural proteins NS3 and NS1.2a (NS1 and NS2a collectively), with less prominent recognition of serotype-specific epitopes on structural proteins. We hypothesize that the cross-reactive T cells from the primary dengue virus infection are reactivated during a secondary infection and play a role in recovery from and/or immunopathology of infection.

We therefore examined dengue virus-specific CTL responses in peripheral blood mononuclear cells (PBMC) of four children 12 months after a secondary natural infection obtained as part of a prospective study of dengue virus infections in Thailand (10). In all four patients, CTLs were detected in bulk culture against at least one protein of dengue virus. In two of these patients, limiting-dilution cloning experiments generated numerous CD4⁺ and CD8⁺ CTL lines directed against serotype-cross-reactive epitopes on the nonstructural proteins NS1.2a and NS3. These results are consistent with our hypothesis that cross-reactive cells from the primary infection are reactivated after a secondary infection. This is the first report of dengue virus-specific CTL responses after natural secondary dengue virus infections.

	MATERIALS AND METHODS

Top Abstract Introduction Materials & Methods Results Discussion References

Viruses. Dengue 2 (D2) virus (New Guinea C strain) was provided by Walter E. Brandt (Walter Reed Army Institute of Research). Dengue 4 (D4) virus (Caribbean strain 814669) was provided by Jack McCown (Walter Reed Army Institute of Research). Viruses were propagated as previously described (13) and frozen at 70°C until use. Recombinant vaccinia viruses containing genes coding for dengue virus proteins produced as described previously (Vac.D2NS3 [vaccinia virus expressing D2 virus NS3], etc.) were kindly provided by C. J. Lai (National Institutes of Health, Bethesda, Md.), Enzo Paoletti (Virogenetics Corporation, Troy, N.Y.), and Margo A. Brinton (Georgia State University, Atlanta) (5, 22, 23).

Study protocol and PBMC. Blood samples were obtained from children enrolled in a prospective study of dengue virus infections at the Queen Sirikit National Institute for Child Health (formerly Bangkok Children's Hospital), Bangkok, Thailand, and the Kamphaeng Phet Provincial Hospital, Kamphaeng Phet, Thailand, in 1994 (10). Children with fever of 72 h or less in duration without an obvious cause and facial flushing were eligible to participate in this study. A diagnosis of acute dengue virus infection was based on serologic tests (antibody capture enzyme immunoassay and hemagglutination inhibition) and isolation of dengue virus (in Toxorhynchites splendens mosquitoes) (21). A diagnosis of secondary dengue virus infection was made on the basis of (i) a dengue virus immunoglobulin M (IgM)-to-IgG ratio of <1.8:1; (ii) with serial specimens, a twofold increase in IgG to dengue virus, with an absolute value of 100 U in the absence of IgM to dengue virus of 40 U; and (iii) a hemagglutination inhibition titer of >1:1,280 1 week after the onset of illness. Written consent was obtained from each subject's parents or guardians. The study protocol was approved by the Institutional Review Boards established by the Ministry of Health, Thailand, the Surgeon General's Office of the Department of the Army, and the University of Massachusetts Medical Center. PBMC obtained 12 months after dengue virus infection were separated and cryopreserved at 70°C until further use. Frozen PBMC samples were shipped to the University of Massachusetts Medical Center for testing. HLA-A, -B, and -C class I typing was performed by using a standard microlymphocytotoxicity assay, and HLA class II typing was performed by PCR-based amplification and hybridization with HLA allele DRB1, DRB3, DRB5, DQA1, DQB1, and DPB1 sequence-specific oligonucleotide probes as previously described (3). Four subjects with documented secondary D2 or D4 virus infections were selected for this study (Table 1).

Bulk culture of PBMC. PBMC were initially cultured with 0.25 ml containing approximately 10⁷ focus-forming units of the homologous infecting serotype of dengue virus in 0.75 ml of AIM-V medium (GIBCO BRL Life Technologies, Gaithersburg, Md.) containing 10% heat-inactivated human AB serum (Hu ABS; Advanced Biotechnologies, Inc., Columbia, Md.). PBMC were then restimulated 7 to 9 days later with 10⁶ gamma-irradiated (3,500 Rad) allogeneic PBMC and anti-CD3 monoclonal antibody 12F6 (0.1 µg/ml), kindly provided by Johnson Wong, in 0.5 ml of fresh medium containing 10% Hu ABS and 25 to 50 U of recombinant interleukin-2 (IL-2; Collaborative Biochemical Products, Bedford, Mass.) per ml as indicated previously (24). Bulk cultures were restimulated every 2 weeks. Cells were assayed 7 to 10 days after the last restimulation for cytolytic activity in CTL assays.

Cloning of PBMC. PBMC which had been stimulated in bulk culture for 7 days were collected and plated at a concentration of 10 and 30 cells per well in 96-well round-bottom microtiter plates (Costar, Cambridge, Mass.) in 200 µl of AIM-V medium containing 10% Hu ABS, 10⁵ allogeneic gamma-irradiated PBMC, anti-CD3 (0.1 µg/ml), and recombinant IL-2 (25 U/ml) (24). Every 3 to 4 days, cells were fed with fresh AIM-V medium containing 10% Hu ABS and IL-2 (25 U/ml). Cells were restimulated every 2 weeks. The T-cell lines were initially screened with vaccinia virus recombinants expressing dengue virus proteins. Growing cells that showed positive lytic activity against any dengue virus proteins were expanded into 48-well plates (Costar) and restimulated with 10⁶ allogeneic PBMC and anti-CD3 at a final volume of 1 ml.

Preparation of target cells. B-lymphoblastoid cell lines (BLCLs) were established by culturing PBMC with Epstein-Barr virus in 48-well plates (7). Allogeneic BLCLs were also obtained from the National Institutes of General Medical Sciences Human Genetic Mutant Cell Repository or the American Society for Histocompatibility and Immunogenetics Cell Bank and Repository. BLCLs (3 × 10⁵ to 5 × 10⁵) were infected with vaccinia viruses for 1.5 to 2 h at 37°C and then diluted in 2 ml of RPMI 1640-10% fetal bovine serum medium (Sigma Immunochemicals, St. Louis, Mo.) and further incubated for 12 to 16 h. Target cells were labeled with 0.25 mCi of ⁵¹Cr for 60 min at 37°C. After three washes, the target cells were counted and diluted to 10⁴ cells/ml for use in cytotoxicity assays.

⁵¹Cr release cytotoxicity assay. Cytotoxicity assays were performed in 96 round-bottom wells as previously reported (2, 19). Effector cells were added to 10^{3 51}Cr-labeled target cells at various effector-to-target (E/T) ratios. In CTL assays with synthetic peptides, peptides were added at 25 µg/ml to target cells and incubated at 37°C for 30 min, after which effector cells were added. Plates were centrifuged at 200 × g for 5 min and incubated 4 to 5 h at 37°C. Supernatant fluids were harvested by using the Skatron Instruments (Sterling, Va.) supernatant collection system, and ⁵¹Cr content was measured in a gamma counter. The percent specific ⁵¹Cr release was calculated as 100 × (cpm experimental release  cpm spontaneous release)/(cpm maximum release  cpm spontaneous release). All assays were performed in triplicate, and the results were calculated from the average of the triplicate wells. The standard error of the mean was <10% in all experiments.

	RESULTS

Top Abstract Introduction Materials & Methods Results Discussion References

Protein specificity of CTLs generated from convalescent PBMC of children after natural secondary dengue virus infection in Thailand. To detect dengue virus-specific CTL, we stimulated convalescent PBMC with the homologous dengue virus and tested for cytolytic activity in bulk culture against the homologous dengue virus proteins. We detected CTL activity in bulk culture PBMC from patient KPP94-037 against target cells infected with Vac.D2NS1.2a and to those infected with Vac.D2NS3 to a lesser degree (Table 2). For patients KPP94-024 and CHD94-020, the predominant CTL response was directed against the homologous NS3 protein (Table 2). For patient CHD94-134, low-level CTL activity was detected in bulk culture against all dengue virus-vaccinia virus recombinants tested. These results show that in bulk culture experiments, PBMC of all four patients had detectable cytolytic activity against target cells expressing nonstructural proteins NS1.2a and/or NS3.

Cross-reactivity of T-cell lines established from PBMC of patients KPP94-037 and KPP94-024. T-cell lines were established by limiting dilution from the bulk cultures obtained from patients KPP94-037 and KPP94-024. For patient KPP94-037, the lines were initially screened against Vac.control, Vac.D2NS3, and Vac.D2NS1.2a. All lines that had positive lytic activity were found to be directed against the nonstructural proteins NS1.2a (data not shown). Among 12 T-cell lines, four lines had the CD3⁺ CD4⁺ CD8 phenotype and eight lines had the CD3⁺ CD4 CD8⁺ phenotype (Table 3). The NS1.2a-specific CTL lines were tested for cross-reactivity against Vac.D1ENS1.2a and Vac.D4NS1.2a constructs. Recognition of NS1.2a by all the T-cell lines was cross-reactive to D1, D2, and D4 viruses (Table 3).

HLA restriction of the lysis of target cells by CD8⁺ and CD4⁺ CTL lines from PBMC of patients KPP94-037 and KPP94-024. For patient KPP94-037, allogeneic BLCLs which had HLA alleles in common with autologous cells were infected with Vac.D2NS1.2a and used as targets. Table 5, experiment 1, shows that three representative CD8⁺ lines were HLA B57 restricted because only targets having B57 in common with the autologous BLCLs were lysed efficiently by CD8⁺ CTLs. Experiments 2 and 3, using allogeneic BLCLs that had class II alleles in common with the autologous line, indicate that two of the CD4⁺ CTL lines from patient KPP94-037 were HLA DR7 restricted. For patient KPP94-024, allogeneic BLCLs sharing common class I alleles with autologous BLCLs were infected with D2NS3. Four representative CD8⁺ lines generated from patient KPP94-024 were B7 restricted (Table 5, experiments 4 and 5). All CD8⁺ lines established from patient KPP94-037 were B57 restricted, and all lines established from patient KPP94-024 were B7 restricted (data not shown).

Cross-reactivity, HLA restriction, and epitope analysis of bulk culture CTL generated from PBMC of patient CHD94-020. For patient CHD94-020, bulk culture CTLs lysed targets expressing the NS3 proteins of D2, D3, and D4 viruses (Tables 2 and 6). Using recombinant vaccinia viruses expressing truncated D3NS3, we were able to localize a CTL epitope to amino acids (aa) 1 to 176 of NS3 (Table 6, experiment 2). Using allogeneic targets having HLA class I alleles in common with CHD94-020, we found bulk culture CTL activity to be HLA A11.1 restricted, based on recognition of CHD94-134 BLCL target cells, which share only A11.1 in common with the autologous line (Table 6, experiment 3). These results indicate that the NS3-specific CTLs detected in bulk culture from this subject are A11.1 restricted and are also serotype cross-reactive. We tried to isolate CTL lines from this donor by the limiting-dilution method but were not successful.

Localization of the epitope within the NS3 protein recognized by T-cell lines from PBMC of patient KPP94-024. CTL lines from patient KPP94-024 lysed target cells expressing NS3; therefore, we infected target cells with vaccinia virus recombinants expressing truncations of the D3NS3 gene to localize the epitopes recognized by these CTL lines. We mapped the region on the NS3 protein recognized by all T-cell lines tested from patient KPP94-024 to aa 216 to 247 of NS3 (Table 7, experiment 1). We then checked for CTL activity against overlapping peptides which spanned that region and found lytic activity only against peptide 221, which is a 15-mer corresponding to aa 221 to 235 (LAPTRVVAAEMEEAL) (data not shown). We used synthetic peptides with truncations of peptide 221 to detect the minimum epitope recognized by the T-cell lines (Table 7, experiments 2 and 3). Peptide targets pulsed with truncations 221g and 221f were lysed by most CTL, whereas target cells pulsed with truncations 221a, 221c, and 221d were not lysed. T-cell lines 2F5 and 3C3 recognized aa 222 to 230 of NS3, and most of the other T-cell lines recognized targets that were pulsed with aa 221 to 232 of NS3.

	DISCUSSION

Top Abstract Introduction Materials & Methods Results Discussion References

Most of the research performed on human T-cell cytotoxic responses to dengue viruses has been in Caucasian volunteers who received experimental live monovalent dengue virus vaccines. These individuals had no known prior exposure to dengue viruses and therefore received the dengue virus vaccine as a primary infection. However, the complications seen in dengue virus infection (DHF and DSS) are more common in patients who have preexisting immunity to one serotype of virus during infection with another serotype (secondary infection) of virus (8). Virus-specific memory T cells have been implicated in contributing to the pathogenesis of severe dengue virus infection (16). Analysis of T-cell responses in patients after natural secondary infections is therefore important because it may provide insights into the mechanisms of T-cell-mediated immunopathology. In this study, memory CTL responses were detected against dengue virus proteins in the PBMC of all four patients examined. This is the first report of dengue virus-specific CTLs after natural secondary dengue virus infections.

We have previously shown that nonstructural proteins, in particular NS3, are predominant targets for both CD4⁺ and CD8⁺ CTLs in vaccine recipients (17, 19, 22). In the present study, we also found that three of four donors had CTL responses to NS3 and that the fourth donor had CTLs which recognized the nonstructural proteins NS1.2a. Proteins with a cytoplasmic intracellular localization have been noted to be the predominant source of cytotoxic T-cell antigenic peptides (18). The NS3 protein constitutes approximately 25% of the cytoplasmic region of the polyprotein, which may explain why it is an immunodominant target for CTL recognition. Numerous other factors (rate of proteolysis, efficiency of transport, peptide stability, etc.) are also likely to influence why peptides from nonstructural proteins are preferentially presented on major histocompatibility complex molecules (1). Since we have not tested vaccinia virus recombinants expressing NS5, it is possible that there are also epitopes on NS5.

Immune responsiveness is affected by the major histocompatibility complex, and the pathogenesis of some diseases has been associated with specific HLA alleles. In areas where DHF is endemic, only a small percentage of individuals exhibit severe disease, which suggests that host genetic factors may play a role in susceptibility to severe disease. One study found a positive association between HLA A2 and B blank and the development of DSS and a negative relationship for HLA B13 (4).

Since it is known that there is diversity in both HLA and non-HLA gene loci between Southeast Asians and Caucasians (3), it was important to analyze the recognition of dengue virus proteins by T cells in the context of HLA molecules in Thai patients. We previously isolated NS3-specific CD4⁺ and CD8⁺ CTL clones from Caucasian volunteers which were restricted by the HLA alleles B35 (17), DR15 (12), and DPw2 (14). The NS1.2a-specific CTLs from patient KPP94-037 in the present study were B57 restricted and DR7 restricted; the NS3-specific CTLs from patient KPP94-024 were B7 restricted, and the NS3-specific CTLs from patient CHD94-020 were A11.1 restricted. These results indicate that despite differences in HLA alleles between Thais and Caucasians, peptides from dengue virus nonstructural proteins are also predominantly recognized by T cells from Thai children. These studies confirm our previous results of the dominance of nonstructural proteins as CTL targets in Caucasian dengue virus vaccine recipients. All of the T-cell lines isolated from patient KPP94-024 recognized autologous targets pulsed with aa 221 to 235 of the NS3 protein. The 15-mer which is lysed by all the B7 restricted CD8⁺ CTLs from this patient contains a proline at position 223. These results are entirely consistent with reports of antigenic peptides bearing a proline at position 2 and aromatic or hydrophobic residues at the C terminus, preferentially binding to HLA B7 and related class I alleles sharing the B7-like supertype (21).

Our earlier work analyzing primary responses from PBMC of vaccine recipients, detected both serotype-specific and serotype-cross-reactive CTL responses against many proteins in individual donors (11, 19). We speculated that the serotype-cross-reactive memory T cells generated during the primary infection would be reactivated in a secondary infection and contribute to the immunopathology of DHF which is observed much more frequently in secondary infections. In this study, we have examined CTL responses after secondary infection in four Thai patients. All the NS1.2a-specific CTL lines generated from patient KPP94-037 and the NS3-specific CTLs from patient KPP94-024 in this study were cross-reactive with the other serotypes of dengue virus. The bulk culture CTLs from patient CHD94-020 were also cross-reactive with D2 and D3 virus NS3. We have not yet analyzed the primary CTL responses in Thai children; therefore, we cannot conclusively define the relationship between primary and secondary dengue virus infections. However, these results suggest that the CTL responses in secondary infections are predominantly due to reactivation of memory cross-reactive CTL from the primary infections, consistent with our hypothesis.

In conclusion, the CTL responses of Thai patients to natural secondary dengue virus infection are directed against nonstructural proteins and are mainly cross-reactive in nature.