• Antigen presentation;
  • Cytotoxic T cells;
  • Peptide epitopes;
  • Virus


  1. Top of page
  2. Abstract
  3. Introduction
  4. Results
  5. Discussion
  6. Materials and methods
  7. Acknowledgements
  8. Note added in proof

Infection with West Nile virus (WNV) causes fatal encephalitis in immunocompromised animals. Previous studies in mice have established that T cell protection is required for clearance of WNV infection from tissues and preventing viral persistence. The current study assessed whether specific WNV peptide epitopes could elicit a cytotoxic T lymphocyte (CTL) response capable of protecting against virus infection. Hidden Markov model analysis was used to identify WNV-encoded peptides that bound the MHC class I proteins Kb or Db. Of the 35 peptides predicted to bind MHC class I molecules, one immunodominant CTL recognition peptide was identified in each of the envelope and non-structural protein 4B genes. Addition of these but not control peptides to CD8+ T cells from WNV-infected mice induced IFN-γ production. CTL clones that were generated ex vivo lysed peptide-pulsed or WNV-infected target cells in an antigen-specific manner. Finally, adoptive transfer of a mixture of envelope- and non-structural protein 4B-specific CTL to recipient mice protected against lethal WNV challenge. Based on this, we conclude that CTL responses against immundominant WNV epitopes confer protective immunity and thus should be targets for inclusion in new vaccines.


envelope protein


Kunjin virus


non-structural protein


transporter associated with antigen processing


West Nile virus


  1. Top of page
  2. Abstract
  3. Introduction
  4. Results
  5. Discussion
  6. Materials and methods
  7. Acknowledgements
  8. Note added in proof

West Nile virus (WNV) is a single-stranded positive-polarity RNA flavivirus and the cause of West Nile encephalitis. WNV is maintained in an enzootic cycle between Culex mosquitoes and several bird species but also infects humans, horses, and other vertebrates. Humans develop a febrile illness with a subset of cases progressing to a meningitis or encephalitis syndrome 1. Currently, no specific therapy or vaccine has been approved for human use.

Host factors influence the expression of WNV disease in humans. The elderly or those with impaired immune systems are at greatest risk for severe neurological disease 13. Similarly, in animals, the integrity of the immune system correlates with resistance to WNV infection 47. Experiments in small animal models demonstrate that T lymphocytes are an essential component of protection against WNV 814. Consistent with this, individuals with hematologic malignancies and impaired T cell function have an increased risk of neuroinvasive WNV infection 15, 16. Upon recognition of a WNV-infected cell, cytotoxic T lymphocytes (CTL) proliferate, release proinflamatory cytokines 11, 17, 18, and lyse cells directly. Mice deficient in CD8+ T cells or class I MHC molecules have higher and sustained WNV burdens in the spleen and CNS and increased mortality 10, 14. CD8+ T cells appear to require perforin to control North American WNV strains, as mice deficient in perforin molecules had increased central nervous system viral burdens and lethality 8.

Prior studies have identified or predicted immunodominant CTL epitopes against flaviviruses in both structural and non-structural genes in mice and humans 1924. CTL lines have been generated that lyse flavivirus-infected or peptide-pulsed autologous mouse or human cells 22, 23. Many of the experiments that characterize the significance of CTL epitopes in flavivirus biology have been performed with the distantly related dengue virus. Somewhat surprisingly, to date, specific CTL epitopes against epidemic strains of WNV have not been defined in mice, and only have been predicted to bind HLA-B-07 proteins in humans by computational analysis 19. Importantly, no study has demonstrated that CTL against specific flavivirus epitopes protect animals against disease.

In this study, we identified two WNV peptides that bound the MHC class I proteins, Kb or Db. CTL generated against these peptides lysed WNV-infected target cells in an antigen-specific manner, demonstrating that they can detect the naturally processed WNV antigens. Our identification of immunodominant protective WNV epitopes enables mechanistic studies that probe the specific function of CD8+ T cells in primary and memory responses against WNV, information that is critical for the design of maximally effective WNV vaccines.


  1. Top of page
  2. Abstract
  3. Introduction
  4. Results
  5. Discussion
  6. Materials and methods
  7. Acknowledgements
  8. Note added in proof

Computational prediction of WNV class I peptides

To evaluate whether specific WNV peptide epitopes could elicit CTL responses that protected against virus challenge, we needed to identify dominant MHC class I-restricted viral epitopes to generate cytolytic T cell clones. We used computational methods to identify possible antigenic peptides from the WNV New York 1999 genome that are presented in H-2b C57BL/6 mice. Prediction of potential antigens was done employing hidden Markov models as implemented in the HMMER software package 25, 26. We generated a total of four models (two for each MHC allele) using aligned peptide sequences found in MHC peptide databases 27, 28. For H-2Kb, one model was used to predict eight-residue (8-mer) antigens and another predicted 9-mer antigens. Analogously, for H-2Db, two models predicted 9-mer and 10-mer antigens, respectively (Table 1).

Table 1. WNV Peptide binding to MHC class I moleculesa)
A. H-2Kb Peptide predictions
  1. a) WNV peptides that were predicted to bind MHC class I by a hidden Markov algorithm. Cell surface induction assays confirmed which peptides bound H-2Kb and H-2Db when pulsed on RMA.S cells. WNV proteins from which the peptides are derived are listed. Peptides 1–16 are 8-mers predicted to bind H-2Kb, peptides 17–25 are 9-mers predicted to bind H-2Kb, and peptides 26–35 are 9- and 10-mers predicted to bind H-2Db.

  2. b) Peptides in bold were identified as immunodominant by restimulation assays. Fold increase of H-2b molecules was determined by flow cytometic analysis of RMA.S cells compared to cells that received no peptide addition.

Peptide No.SequenceWNV proteinFold increase in H-2Kb surface induction
B. H-2Db Peptide predictions
Peptide No.SequenceWNV proteinFold increase in H-2Db surface induction

Surface induction of MHC by predicted peptides

Our computational models predicted 35 WNV peptides that could potentially bind H-2Kb and H-2Db. To assess whether these peptides bound MHC class I molecules, we utilized an MHC class I surface induction assay with RMA.S cells 29. RMA.S cells are deficient in transporter associated with antigen processing (TAP) molecules and fail to transport peptide into the endoplasmic reticulum. As a result, MHC class I heavy chain and β2-microglobulin accumulate in the lumen of the endoplasmic reticulum and the cells basally express low levels of MHC class I molecules on their surface. However, if exogenous peptide is added, MHC class I molecules assemble with peptides, and are stabilized and displayed on the cell surface.

To determine the relative binding strength of peptides to Kb and Db, RMA.S cells were incubated with increasing concentrations of candidate peptides and analyzed for MHC class I expression by flow cytometry. Of the 35 WNV peptides tested, 17 peptides efficiently stabilized Kb whereas seven peptides induced expression of Db. Two WNV peptides (30 and 35) stabilized expression of either MHC class I molecule and 11 peptides failed to induce surface expression of either Kb or Db (Table 1, Fig. 1).

thumbnail image

Figure 1. MHC class I cell surface induction by WNV peptides 3 and 33. Levels of H-2Kb or H-2Db expression on RMA.S cells correlate with peptide binding strength for MHC class I. RMA.S cells were cultured overnight in the presence or absence of peptide 3 or 33. Cells were stained with anti-Kb or anti-Db antibodies and analyzed by flow cytometry. One representative experiment of several is shown.

Download figure to PowerPoint

Identification of WNV-specific immunodominant Kb- and Db-restricted peptides

To test whether the peptides identified by our in vitro analysis were presented on MHC class I molecules during WNV infection, we restimulated splenocytes from day 7 WNV-infected mice ex vivo with candidate peptides, and performed intracellular IFN-γ and CD8 co-staining: we tested the peptides that exhibited high levels of surface stabilization of MHC class I molecules (peptides 3, 4, 19, 20, 25, 27, 28, 31, 32, 33, 33, and 36). Notably, many of these peptides failed to stimulate IFN-γ production, suggesting that they are not processed in cells during WNV infection or that C57BL/6 mice lack the T cells capable of detecting these determinants (e.g. possibly due to negative selection).

However, peptides 3 and 33 induced IFN-γ expression on CD8+ T cells from WNV but not mock-infected amimals. Peptide 3 is an 8-mer derived from the WNV envelope protein (E) and strongly induced expression of Kb, whereas peptide 33 is a 10-mer derived from non-structural protein (NS) 4B that strongly induced Db expression (see Table 1). In response to restimulation with peptides 3 or 33, on average, 1.1 ± 0.4% and 3.5 ± 1.1% of CD8+ splenocytes produced IFN-γ, respectively, whereas these peptides did not induce IFN-γ expression in CD8+ cells from uninfected mice (Fig. 2A and data not shown). This level of IFN-γ production was significant as restimulation with the agonists phorbol ester and ionomycin activated ∼10% of CD8+ T cells over baseline (Fig. 2D). Based on the data from the MHC class I surface induction and intracellular IFN-γ staining after restimulation, we conclude that peptides 3 and 33 encompass important WNV-specific H-2b MHC class I epitopes during primary infection.

thumbnail image

Figure 2. Production of IFN-γ after ex vivo stimulation of CD8+ T cells with WNV peptides. (A) Representative flow cytometry profiles of IFN-γ production by CD8+ splenocytes (harvested from mice at day 7 after infection) after restimulation with WNV peptide 3 or 33, or a negative control (peptide 17). In the top right corner, the percentage of IFN-γ+ CD8+ T cells is indicated. (B) Comparison of the relative binding affinities of the length variants of the WNV peptide 33 using the MHC class I surface induction assay. RMA.S cells were incubated overnight with titrated amounts of the WNV peptide 33 10-mer (SSVWNATTAI) or 9-mer (SSVWNATTA). For comparison the well-characterized and naturally processed Db ligand ASNENMETM (FluNp 366–374) was included. (C) Percentage of IFN-γ+ CD8+ splenocytes after peptide 33 restimulation from mice (day 7) infected with 102 PFU, 104 PFU, or 105 PFU of WNV. (D) Percentage of IFN-γ+ CD8+ splenocytes after peptide 33 or PMA and ionomycin restimulation from mice at day 4, 7, or 9 post-WNV (102 PFU) infection. Mock-infected mice were used as negative controls.

Download figure to PowerPoint

During the course of our experiments, we became aware that another group (Brien et al., published in this issue of EJI) had identified a Db-restricted immunodominant peptide in NS4B that was similar to peptide 33. However, this group used an alternative strategy, fusion protein sensitization and overlapping peptide epitope mapping, to identify a corresponding NS4B 9-mer (SSVWNATTA) as the dominant peptide. Because Db can efficiently present 9- or 10-mers to CD8+ T cells, we directly compared the ability of the peptide 33 9-mer (SSVWNATTA) and 10-mer (SSVWNATTAI) to stabilize Db expression on RMA.S cells (Fig. 2B). Notably, the peptide 33 10-mer showed an about tenfold enhanced ability to stabilize Db expression and comparable activity compared to a well-characterized naturally processed Db ligand, the influenza nucleoprotein 9-mer peptide (NP 366–374; ASNENMETM) 30. As a result, for all subsequent studies, we used the 10-mer form of peptide 33.

To extend our initial results, a more comprehensive analysis of CD8+ T cell reactivity with peptide 33 was conducted. An in vivo time course and viral dose response were performed followed by ex vivo stimulation with peptide 33. Splenocytes were collected at day 4, 7, 9 or 45 after infection, restimulated with negative control or peptide 33, and CD8+ T cells were analyzed for intracellular IFN-γ production by flow cytometry.

At 4 days after infection, only 0.2% CD8+ T cells produced IFN-γ in response to peptide 33. By day 7, this number increased to 3.5% of CD8+ T cells, and by day 9, the percentage decreased to 1.9% (Fig. 2D), consistent with the likely emigration of antigen-specific CD8+ T cells to areas of significant tissue infection, including the central nervous system 31. The effect was relatively dose-insensitive as mice infected at 102, 103, or 104 PFU of WNV all had similar levels of restimulation (Fig. 2C). In all cases, as a control, splenocytes from WNV-infected mice showed no IFN-γ production after restimulation with peptide 17 (data not shown). Thus, maximal IFN-γ expression by CD8+ T cells in the spleen in response to the peptide 33 occurred around 1 wk after infection. Notably, this time point corresponds to complete clearance of WNV infection in the spleen in wild-type mice 14. At 30–45 days after WNV infection, ∼2.2% of splenic CD8+ T cells produced IFN-γ in response to restimulation with peptide 33, suggesting that a stable memory T cell population was generated against this epitope (data not shown).

CTL activity of CD8+ T cells recognizing peptides 3 and 33

CD8+ T cells have important effector functions during WNV infection including perforin-dependent killing of target cells in the CNS 8, 11, 32. To confirm that CD8+ T cells that recognize WNV-specific peptides 3 and 33 had cytolytic activity, we performed classical CTL killing assays. We initially tested whether the precursor frequency of CD8+ T cells to peptides 3 and 33 was sufficient to generate a primary in vitro cytolytic response. Although such peptide-specific primary responses are rare, they have been reported and are thought to reflect T cell immunodominance 33, 34. Naive C57BL/6 CD8+ T cells stimulated in vitro with peptide 3 generated a low level of peptide-specific cytotoxicity, whereas the same cells stimulated in vitro with peptide 33 generated a significant CTL response (Fig. 3). These results strongly suggest that the CD8+ T cell responses to peptide 33 are immunodominant.

thumbnail image

Figure 3. Primary CTL cell response to WNV peptides. As shown in each panel, four different effector populations were tested. (Upper left) Naive C57BL/6 splenocytes were cultured for 5 days in the absence of irradiated C57BL/6 stimulators (anti-0). (Upper right) Naive C57BL/6 splenocytes were cultured for 5 days with irradiated C57BL/6 stimulators (anti-B6). (Lower left) Naive C57BL/6 splenocytes were cultured for 5 days with 100 μM of WNV peptide 3 (anti-B6 + p3). (Lower right) Naive C57BL/6 splenocytes were cultured for 5 days with 100 μM of WNV peptide 33 (anti-B6 + p33). All four effector populations were tested on RMA.S target cells cultured with no peptide, or 10 μM of the peptides 3 or 33 peptides as indicated.

Download figure to PowerPoint

Because peptide-generated CTL can be of low affinity, CD8+ T cell lines specific for peptides 3 and 33 were generated from memory CD8+ T cells by serial restimulation with peptide-pulsed congenic TAP–/– splenocytes. The bulk lines were assayed for specific killing of the appropriate peptide-expressing target cell using a standard 51Cr-release assay. CD8+ T cell lines amplified with peptide 3 specifically lysed more than 50% of RMA.S targets pulsed with peptide 3 at a 10:1 effector-to-target (E:T) ratio, whereas background killing of targets lacking peptide or with an irrelevant WNV peptide was less than 1% (Fig. 4A). Analogously, CD8+ T cell lines amplified with peptide 33 lysed ∼50% of peptide-loaded RMA.S target cells at a 10:1 E:T ratio (Fig. 4B). CTL killing was robust as even at a 2.5:1 E:T ratio peptide 3 and peptide 33 lysed 30% of target cells, respectively.

thumbnail image

Figure 4. Cytolytic activity of WNV peptide-specific CTL. (A–B) Lysis of RMA.S target cells by peptide 3 (A) or peptide 33 (B) specific CTL in the presence or absence of peptide. (C–D) Lysis of MC57GL, MC57GL-WNV replicon, or MC57GL KUNV-infected target cells by peptide 3 (C) or peptide 33 (D) specific CTL. CTL activity was assayed by a standard 51Cr-release assay. The data are from one representative experiment of three performed in triplicate. (E) Expression of WNV E in MC57GL cells infected with KUNV. Uninfected cells were incubated with E16 anti-E antibody (filled histogram), or infected cells were incubated sequentially with either an isotype control (bold line) or the E16 anti-E antibody (light line) and Alexa 647-conjugated anti-mouse IgG and analyzed by flow cytometry. (F) Expression of WNV NS1 protein in MC57GL-WNV replicon cells. Cells lacking the WNV subgenomic replicon were incubated with 16NS1 anti-NS1 antibody (filled histogram). Cells expressing the replicon were incubated with either an isotype control (bold line) or the 16NS1 anti-NS1 antibody (light line). All cells were subsequently incubated with Alexa 647-conjugated goat anti-mouse IgG and analyzed by flow cytometry.

Download figure to PowerPoint

To determine whether bulk CTL were capable of killing targets expressing endogenous viral peptides, we used MC57GL syngeneic fibroblasts that propagate a subgenomic WNV replicon (MC57GL-WNV replicon 35). This replicon encodes only non-structural genes (NS1–NS5) and thus, the cells should be lysed by NS4B-specific (peptide 33) CTL but not by E-specific (peptide 3) CTL. Expression of the non-structural proteins in these cells was confirmed by immunostaining with an anti-NS1 specific mAb (Fig. 4F). As expected, CTL generated by peptide 33 restimulation lysed MC57GL-WNV replicon cells with much greater efficiency than cells lacking the replicon (25% compared to 1%, E:T ratio of 10:1), whereas CTL against peptide 3 did not lyse either target significantly (Fig. 4C, D).

To address the killing activity of CTL against peptide 3, we infected MC57GL cells with Kunjin virus (KUNV), a related lineage I WNV that can be used at biosafety level 2. KUNV has 97.6% sequence identity with WNV 3000.0359, including 100% identity of peptides 3 and 33. MC57GL target cells were incubated with KUNV and 72 h later, infection was confirmed by immunostaining with an anti-E specific mAb (Fig. 4E). Subsequently, KUNV-infected target cells were incubated with peptide 3- or 33-specific CTL lines. Although both lines showed specific killing (Fig. 4C, D), peptide 3-specific CTL lysed KUNV-infected targets slightly more efficiently (57% with a 20:1 E:T ratio) than peptide 33-specific CTL (40% with a 20:1 E:T ratio). Both CTL lines lysed KUNV-infected targets at a low E:T ratio (2.5:1) but, importantly, did not lyse uninfected targets. As an additional control, CTL lines specific for an irrelevant Kb-restricted peptide (OVA) did not lyse infected targets (data not shown).

Taken together, our experiments show that WNV-specific CTL against two peptides recognize and kill targets that express endogenous viral proteins. This data strongly suggests that peptides 3 and 33 are naturally processed WNV epitopes. Formal identification will require elution of peptide from MHC class I molecules and sequencing. However, even if peptides 3 and 33 are not identical to the naturally processed peptides, they clearly stimulate CTL that are highly proficient at specific recognition of the naturally processed peptide bound to self class I MHC molecules.

Although our in vitro experiments suggested that peptide-specific CTL could kill WNV targets, we wanted to confirm their function in vivo in controlling infection. To determine whether peptide-specific CTL could provide similar protection, we transferred a 1:1 mixture of 2×106 peptide 3- and peptide 33-specific CTL into 4-wk-old wild-type C57BL/6 mice that were infected 24 h prior with 102 PFU of WNV (Table 2). Mice that received naive cells or 2×106 OVA-specific CTL had low survival rates. Mice that received a mixture of peptide 3- and peptide 33-specific CTL demonstrated an increased survival rate (70%, p<0.05). Thus, CTL specific for viral epitopes can protect against lethal WNV infection.

Table 2. In vivo protection of WNV infection by peptide-specific CTLa)
Survived/Total% Survivalp value
  1. a) No cells, 2×106 OVA-specific CTL, or 1×106 of each peptide 3- and 33-specific CTL were adoptively transferred into 4-wk-old C57BL/6 wild-type mice 1 day before infection with 102 PFU of WNV. Survival was monitored over 28 days and statistical significance was evaluated by the log-rank test. The results are the composite of two independent experiments.

No cells transferred2/1020
OVA-specific CD8+ T cells0/100not significant
Peptide 3- and 33-specific CD8+ T cells7/1070<0.05


  1. Top of page
  2. Abstract
  3. Introduction
  4. Results
  5. Discussion
  6. Materials and methods
  7. Acknowledgements
  8. Note added in proof

In this report, we used computational prediction algorithms to identify 35 candidate H-2b-restricted CTL peptide epitopes that were present in the WNV genome. After testing these peptides for stabilization of MHC class I molecules on the surface of TAP–/– cells and for restimulation of in vivo primed CD8+ T cells, two immunodominant epitopes were identified, corresponding to peptide sequences in the E and NS4B genes. CTL primed against these peptides in vitro efficiently lysed peptide-pulsed or virally infected target cells. Moreover, adoptive transfer of WNV-specific CTL against these peptides diminished lethal infection in mice.

Algorithms that predict CTL peptide epitopes broadly fall into three categories – statistical, structural, and neural approaches. Some statistical approaches classify data systematically based on mathematical descriptions of alignment profiles. One such method of pattern recognition is the use of hidden Markov models 25, 26. These models are probabilistic mathematical descriptions of the consensus of multiple sequence alignments. We employed this computational approach to identify possible WNV-specific antigenic peptides. Since MHC proteins impose allele-specific sequence constraints on the presented peptides, we hypothesized that a statistical description of these constraints could be used to scan the WNV genome to identify binding peptides for a given MHC allele. With this approach, we predicted that 35 peptides could bind H-2Kb or H-2Db MHC class I molecules. Of these, 25 peptides efficiently stabilized MHC class I expression on the surface of TAP–/– cells, and four peptides restimulated CTL that were primed in vivo during WNV infection, two of which were strongly immunodominant.

Although CTL epitopes for other flaviviruses in mice and humans have been identified 20, 21, 24, 3639, there are no published studies on dominant epitopes for virulent North American WNV isolates in either animal system. One recent report used a bioinformatics approach to search the WNV (New York 1999 strain) genome for human HLA-B-07-restricted CTL epitopes 19. This study identified 95 of 3433 WNV peptides as likely to bind HLA-B-07. Of these, four peptides strongly stabilized the expression of HLA-B-07 molecules on TAP–/– cell lines. However, these candidate CTL peptides were never tested against HLA-B-07-restricted CTL from human patients infected with WNV.

During the course of our studies, we became aware of another group performing investigations in parallel to identify immunodominant CTL epitopes against WNV in C57BL/6 mice (Brien et al., published in this issue of EJI). In contrast to using a prediction algorithm, this group used an approach of protein overload and overlapping peptide pools for restimulation of CTL ex vivo. As distinct strategies were used, results were exchanged. Notably, both approaches identified a conserved peptide in NS4B as dominant for its ability to restimulate WNV-specific CTL ex vivo. Interestingly, whereas the computational model suggested that this peptide would function best as a 10-mer for its interaction with Db, the empiric testing protocol identified the peptide as a 9-mer. Subsequent comparison studies have shown that although both peptides stabilized Db expression and facilitated CTL recognition, the 10-mer appears to preferentially bind. The results from the independent approaches did vary, however. Our model identified a Kb-restricted immunodominant epitope (E771–778; IALTFLAV), which was not initially detected by screening of peptide pools, and the empiric approach identified a different Kb-restricted immunodominant epitope (E347–354; RSYCYLAT), which was not predicted by the hidden Markov model. Overall, our combined data suggest that both empiric and predictive analysis may be complementary for identifying immunodominant peptides for CTL recognition against WNV.

The WNV genome encodes ∼3400 amino acids, which can be separated into an almost equivalent number of overlapping 10-mer peptides 19. Despite the large number of possible peptides, we identified only a small number of epitopes that were efficiently recognized by WNV-specific CTL. Similar results have been reported for dengue virus-specific human and murine CD8+ CTL with as few as one to three dominant CTL epitopes identified 24. Perhaps more relevant, CD8+ CTL that were generated against KUNV or a lineage II WNV (Sarafend) strain, recognized non-structural protein determinants (NS3, NS4A, and NS4B) in the context of vaccinia virus expression of individual and overlapping gene constructs 36, 38. This group localized the dominant CTL response to KUNV in H-2k, H-2d, and H-2q mice to a 98 amino acid segment at the NS3-NS4A junction, and an immunodominant epitope was proposed in an adjacent NS4A-NS4B region for H-2b mice although the precise peptide determinants were never identified 36. Our experiments that distinguish the NS4B peptide SSVWNATTAI as a critical H-2b-restricted CTL epitope against WNV and KUNV confirm and extend work from the prior studies.

The kinetics of recognition of WNV peptides by CTL correlated with time course of activation of CD8+ T cells in vivo and clearance of WNV from tissues. For example, in the spleen of C57BL/6 mice, WNV titers peak on day 4 and are cleared by day 8 40, a process that requires the presence and activation of CD8+ T cells 14. Analogously, we observed very low levels of E or NS4B peptide restimulation of CD8+ T cells that were derived from the splenocytes from mice infected with WNV for 4 days. In comparison, large numbers (∼1–4%) of CD8+ T cells derived from splenocytes at day 7 after infection were activated and produced IFN-γ in response to WNV-specific peptides. Consistent with this, a high proportion (∼25%) of CD8+ T cells in the brain were NS4B-specific at day 9 after WNV infection 31.

Previous studies in mice have established that T cells serve a critical function in the control of primary WNV infection for clearance of infected cells and preventing persistent infection in visceral and CNS tissues 8, 10, 14. Our present studies demonstrate that peptide-specific CTL protect against WNV infection in vitro or in vivo. Protection after adoptive transfer of antigen-specific CTL lines agrees with studies that suggest a limited role of bystander, non-antigen restricted CD8+ T cells, that could contribute to protection through antigen-independent production of inhibitory or immunomodulatory cytokines 41. Our results, showing an important contribution of antigen-specific CD8+ T cells to protection, support a role for inclusion of a broad range CTL epitopes in vaccines against WNV, and potentially other flaviviruses.

Given that non-structural proteins comprise a significant percentage of immunodominant CTL epitopes against flaviviruses, vaccines that include non-structural proteins could enhance cell immunity. Correspondingly, the absence of homologous non-structural proteins could limit the durability of chimeric vaccines that insert WNV, JEV, or DENV structural proteins onto a heterologous yellow fever virus non-structural protein backbone 42. Indeed, in studies that compared cellular immunity derived from subunit-based (prM-E) and live-attenuated homologous vaccines, we observed enhanced CD8+ T cell-mediated protection with vaccines that contained non-structural proteins (B. Shrestha and M. Diamond, unpublished results). As proof of principle, immunization studies with NS4B peptide-loaded single-chain MHC trimers 43, 44 are planned to establish the limitations of CD8+ T cell immunity against WNV in the absence of immune antibody. If successful, this strategy could be used to optimize cellular immune responses in vaccine development against flaviviruses, and perhaps other viruses.

Materials and methods

  1. Top of page
  2. Abstract
  3. Introduction
  4. Results
  5. Discussion
  6. Materials and methods
  7. Acknowledgements
  8. Note added in proof

Viruses and propagation

The WNV strain 3000.0259 was isolated in New York in 2000 from an infected mosquito and was passaged once in C6/36 Aedes albopictus cells. For inoculation of mice, virus was diluted in Hanks’ balanced salt solution (HBSS) and 1% heat-inactivated fetal bovine serum (FBS). KUNV (strain 16-532) was obtained (gift of J. Anderson, New Haven, CT) and propagated once in C6/36 cells. The generation and culturing of MC57GL and MC57GL-WNV replicon cells were previously described 14, 35.

Surface expression of WNV peptide-MHC class I complexes

The WNV peptides IALTFLAV (peptide 3, Kb-restricted) and SSVWNATTAI (peptide 33, Db-restricted) were synthesized commercially (Sigma-Aldrich, St. Louis, MO). Peptides were solubilized in 1% DMSO (1 mg/mL) and then diluted and in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% FBS. RMA.S cells are TAP–/– and at baseline do not express classical MHC class I molecules at high levels. These cells were cultured overnight at 37°C in DMEM medium in the presence or absence of 200 μM of peptides. To analyze RMA.S cells for surface expression of Kb or Db molecules after peptide incubation, 4×105 cells were incubated 20 μL of culture supernatants of anti-Kb (B8-24-3) or anti-Db (B22/249 45) at 4°C for 30 min in PBS supplemented with 1% BSA and 0.1% NaN3. After washing, cells were incubated with phycoerythrin-conjugated goat anti-mouse IgG at 4°C for 30 min (BD Biosciences, San Jose, CA). After additional washing, cells were fixed in 1% paraformaldehyde and processed on a FACSCalibur flow cytometer using CellQuest Pro Software (BD Biosciences).

CTL assays

For generation of bulk CTL clones, wild-type mice were infected subcutanenously by footpad injection with 102 PFU of WNV (strain 3000.0259). At 4 wk after infection, splenocytes were harvested washed with medium (RPMI 1640 containing 10% FBS), and resuspended at 7.5×106 cells/mL. In parallel, splenocytes from uninfected congenic TAP–/– C57BL/6 mice were harvested, depleted of erythrocytes with ACK buffer (0.15 M NH4Cl, 10 mM KHCO3, 0.1 mM Na2EDTA), washed with RPMI medium, irradiated (2000 rad), and counted. TAP–/– splenocytes were resuspended at 3.5×106 cells/mL and 2 µM of either peptide 3 or peptide 33 was added. CTL were generated by incubating immune splenocytes with peptide-pulsed TAP–/– splenocytes at a final peptide concentration of 1 µM in a 24-well tissue culture plate. At day 5, an aliquot of cells were assayed for their ability to lyse RMA.S target cells incubated with specific or control peptide by a 51Cr-release assay (see below). Remaining CTL were resuspended at 3×106 cells/mL in a 24-well tissue culture plate. Seven days later, CTL were restimulated with 3.5×106 irradiated TAP–/– splenocytes pulsed with 1 μM peptide 3 or 33. At day 14 and thereafter, bulk CTL against peptide 3 or 33 were stimulated at weekly intervals with irradiated wild-type splenocytes, 1 μM peptide, and 10 U/mL rIL-2.

Standard 51Cr-release assays were conducted to measure CTL activity. RMA.S or MC57GL target cells (3×106) were labeled with 150 μCi of 51Cr (PerkinElmer Life Sciences, Wellesley, MA) in 200 µL of RPMI 1640 medium supplemented with 10% FBS at 37°C in 5% CO2 for 1 h with mixing every 15 min. Effector CTL (1.6×105, 2.0×104, 4.0×104, and 2.0×104 cells/well) were added in triplicate to a 96-well plate containing washed target cells (2.0×103). The plates were incubated for 4 h at 37°C in 5% CO2. To determine maximum lysis, 2.5% Triton X-100 (Sigma-Aldrich) was added to a set of control wells. To determine spontaneous lysis, target cells were incubated without effector cells. Supernatants were collected and read by an Isomedic γ-counter (ICN Biomedicals, Huntsville, AL). The percentage of 51Cr release was calculated by [(experimental 51Cr release – control 51Cr release) / (maximum 51Cr release – control 51Cr release)] × 100.

Intracellular IFN-γ staining

Spleens were harvested from wild-type mice at days 4, 7, 9, or 60 after WNV infection. After lysis of erythrocytes, 106 splenocytes were added to a 96-well round-bottom immunoassay plate (TPP, Trasadingen, Switzerland) in the presence of 1.0 μg/mL peptide 33 and 1.0 μg/mL Golgi plug (BD Biosciences). Controls were incubated without peptide or artificially stimulated with 50 ng/mL PMA and 1.0 μM ionomyocin. Plates were incubated at 37°C in 5% CO2 for 4 h. Cells were washed with PBS containing 5% normal goat serum and incubated with fluorescein isothiocyanate (FITC)-conjugated anti-CD8 or an isotype FITC-conjugated antibody (Becton Dickinson Biosciences) for 30 min at 4°C. After three washes, cells were fixed with 4% paraformaldehyde in PBS, and permeabilized with saponin buffer (HBSS, 10 mM HEPES, 0.1% saponin, 0.025% sodium azide). Cells were then stained with allophycocyanin-conjugated anti IFN-γ antibody or an allophycocyanin-conjugated isotype control mAb in HBSS buffer containing 0.1% saponin for 30 min at 4°C. Cells were washed three times, and analyzed by flow cytometry. The percentage of CD8+ T cells that produced IFN-γ was calculated using CellQuest software (Becton Dickinson).

Mice and adoptive transfer experiments

C57BL/6J (H-2b) inbred wild-type and TAP–/– congenic mice were obtained commercially from Jackson Laboratories (Bar Harbor, ME). Mice were bred in the animal facility at Washington University and experiments were performed in accordance with Washington University's animal studies guidelines. WNV peptide-specific CTL (2×106) were transferred adoptively into 4 wk-old wild-type C57BL/6J mice. Mice were challenged 24 h later with 102 PFU WNV subcutaneously by footpad injection and monitored for lethality.


  1. Top of page
  2. Abstract
  3. Introduction
  4. Results
  5. Discussion
  6. Materials and methods
  7. Acknowledgements
  8. Note added in proof

The authors thank members of our laboratories for critical review of this manuscript and Dr. J. Nikolich-Žugich for sharing of data prior to submission of this manuscript. We also thank Ms. Tina Primeau for assistance with the CTL assays. The work was supported by U54 AI057160 to the Midwest Regional Center of Excellence for Biodefense and Emerging Infectious Diseases Research.

  • 1


  • 2


  • 3


  • 4


  • 1
    Hubalek, Z. and Halouzka, J., West Nile fever – a reemerging mosquito-borne viral disease in Europe. Emerg. Inf. Dis. 1999. 5: 643650.
  • 2
    Asnis, D. S., Conetta, R., Teixeira, A. A., Waldman, G. and Sampson, B. A., The West Nile virus outbreak of 1999 in New York: The Flushing Hospital experience. Clin. Infect. Dis. 2000. 30: 413418.
  • 3
    Tsai, T. F., Popovici, F., Cernescu, C., Campbell, G. L. and Nedelcu, N. I., West Nile encephalitis epidemic in southeastern Romania. Lancet 1998. 352: 767771.
  • 4
    Eldadah, A. H., Nathanson, N. and Sarsitis, R., Pathogenesis of West Nile virus encephalitis in mice and rats. 1. Influence of age and species on mortality and infection. Am. J. Epidemiol. 1967. 86: 765775.
  • 5
    Eldadah, A. H. and Nathanson, N., Pathogenesis of West Nile virus encepahlitis in mice and rats. II. Virus multiplication, evolution of immunofluorescence, and development of histological lesions in the brain. Am. J. Epidemiol. 1967. 86: 776790.
  • 6
    Weiner, L. P., Cole, G. A. and Nathanson, N., Experimental encephalitis following peripheral inoculation of West Nile virus in mice of different ages. J. Hyg. (Lond.) 1970. 68: 435446.
  • 7
    Samuel, M. A. and Diamond, M. S., Pathogenesis of West Nile virus infection: A balance between virulence, innate and adaptive immunity, and viral evasion. J. Virol. 2006. 80: 93499360.
  • 8
    Shrestha, B., Samuel, M. A. and Diamond, M. S., CD8+ T cells require perforin to clear West Nile virus from infected neurons. J. Virol. 2006. 80: 119129.
  • 9
    Camenga, D. L., Nathanson, N. and Cole, G. A., Cyclophosphamide-potentiated West Nile viral encephalitis: Relative influence of cellular and humoral factors. J. Infect. Dis. 1974. 130: 634641.
  • 10
    Wang, Y., Lobigs, M., Lee, E. and Mullbacher, A., CD8+ T cells mediate recovery and immunopathology in West Nile virus encephalitis. J. Virol. 2003. 77: 1332313334.
  • 11
    Kesson, A. M., Blanden, R. V. and Mullbacher, A., The primary in vivo murine cytotoxic T cell response to the flavivirus, West Nile. J. Gen. Virol. 1987. 68: 20012006.
  • 12
    Liu, Y., Blanden, R. V. and Mullbacher, A., Identification of cytolytic lymphocytes in West Nile virus-infected murine central nervous system. J. Gen. Virol. 1989. 70: 565573.
  • 13
    Nathanson, N. and Cole, G. A., Immunosuppression and experimental virus infection of the nervous system. Adv. Virus Res. 1970. 16: 397448.
  • 14
    Shrestha, B. and Diamond, M. S., The role of CD8+ T cells in the control of West Nile virus infection. J. Virol. 2004. 78: 83128321.
  • 15
    Murray, K., Baraniuk, S., Resnick, M., Arafat, R., Kilborn, C., Cain, K., Shallenberger, R. et al., Risk factors for encephalitis and death from West Nile virus infection. Epidemiol. Infect. 2006. 134: 13251332.
  • 16
    Pruitt, A. A., Central nervous system infections in cancer patients. Semin. Neurol. 2004. 24: 435452.
  • 17
    Douglas, M. W., Kesson, A. M. and King, N. J., CTL recognition of West Nile virus-infected fibroblasts is cell cycle dependent and is associated with virus-induced increases in class I MHC antigen expression. Immunology 1994. 82: 561570.
  • 18
    Kulkarni, A. B., Mullbacher, A. and Blanden, R. V., In vitro T-cell proliferative response to the flavivirus, West Nile. Viral Immunol. 1991. 4: 7382.
  • 19
    De Groot, A. S., Saint-Aubin, C., Bosma, A., Sbai, H., Rayner, J. and Martin, W., Rapid determination of HLA B*07 ligands from the West Nile virus NY99 genome. Emerg. Infect. Dis. 2001. 7: 706713.
  • 20
    Spaulding, A. C., Kurane, I., Ennis, F. A. and Rothman, A. L., Analysis of murine CD8(+) T-cell clones specific for the dengue virus NS3 protein: Flavivirus cross-reactivity and influence of infecting serotype. J. Virol. 1999. 73: 398403.
  • 21
    Mathew, A., Kurane, I., Rothman, A. L., Zeng, L. L., Brinton, M. A. and Ennis, F. A., Dominant recognition by human CD8+ cytotoxic T lymphocytes of dengue virus nonstructural proteins NS3 and NS1.2a. J. Clin. Invest. 1996. 98: 16841691.
  • 22
    Zivny, J., Kurane, I., Leporati, A. M., Ibe, M., Takiguchi, M., Zeng, L. L., Brinton, M. A. and Ennis, F. A., A single nine-amino acid peptide induces virus-specific, CD8+ human cytotoxic T lymphocyte clones of heterogeneous serotype specificities. J. Exp. Med. 1995. 182: 853863.
  • 23
    Murali-Krishna, K., Ramireddy, B., Ravi, V. and Manjunath, R., Recognition of nonstructural protein peptides by cytotoxic T lymphocytes raised against Japanese encephalitis virus. Microbiol. Immunol. 1995. 39: 10211024.
  • 24
    Rothman, A. L., Kurane, I., Lai, C. J., Bray, M., Falgout, B., Men, R. and Ennis, F. A., Dengue virus protein recognition by virus-specific murine CD8+ cytotoxic T lymphocytes. J. Virol. 1993. 67: 801806.
  • 25
    Eddy, S. R., Profile hidden Markov models. Bioinformatics 1998. 14: 755763.
  • 26
    Eddy, S. R., What is a hidden Markov model? Nat. Biotechnol. 2004. 22: 13151316.
  • 27
    Rammensee, H., Bachmann, J., Emmerich, N. P., Bachor, O. A. and Stevanovic, S., SYFPEITHI: Database for MHC ligands and peptide motifs. Immunogenetics 1999. 50: 213219.
  • 28
    Bhasin, M., Singh, H. and Raghava, G. P., MHCBN: A comprehensive database of MHC binding and non-binding peptides. Bioinformatics 2003. 19: 665666.
  • 29
    Townsend, A., Ohlen, C., Bastin, J., Ljunggren, H. G., Foster, L. and Karre, K., Association of class I major histocompatibility heavy and light chains induced by viral peptides. Nature 1989. 340: 443448.
  • 30
    Falk, K., Rotzschke, O., Deres, K., Metzger, J., Jung, G. and Rammensee, H. G., Identification of naturally processed viral nonapeptides allows their quantification in infected cells and suggests an allele-specific T cell epitope forecast. J. Exp. Med. 1991. 174: 425434.
  • 31
    Sitati, E. and Diamond, M. S., CD4+ T cell responses are required for clearance of West Nile virus from the central nervous system. J. Virol. 2006. 80: 1206012069.
  • 32
    Kesson, A. M., Blanden, R. V. and Mullbacher, A., The secondary in vitro murine cytotoxic T cell response to the flavivirus, West Nile. Immunol. Cell Biol. 1988. 66: 2332.
  • 33
    Carbone, F. R., Moore, M. W., Sheil, J. M. and Bevan, M. J., Induction of cytotoxic T lymphocytes by primary in vitro stimulation with peptides. J. Exp. Med. 1988. 167: 17671779.
  • 34
    Alexander, M. A., Damico, C. A., Wieties, K. M., Hansen, T. H. and Connolly, J. M., Correlation between CD8 dependency and determinant density using peptide-induced, Ld-restricted cytotoxic T lymphocytes. J. Exp. Med. 1991. 173: 849858.
  • 35
    Geiss, B. J., Pierson, T. C. and Diamond, M. S., Actively replicating West Nile virus is resistant to cytoplasmic delivery of siRNA. Virology J. 2005. 2: 53.
  • 36
    Hill, A. B., Mullbacher, A., Parrish, C., Coia, G., Westaway, E. G. and Blanden, R. V., Broad cross-reactivity with marked fine specificity in the cytotoxic T cell response to flaviviruses. J. Gen. Virol. 1992. 73: 11151123.
  • 37
    Zivna, I., Green, S., Vaughn, D. W., Kalayanarooj, S., Stephens, H. A., Chandanayingyong, D., Nisalak, A. et al., T cell responses to an HLA-B*07-restricted epitope on the dengue NS3 protein correlate with disease severity. J. Immunol. 2002. 168: 59595965.
  • 38
    Parrish, C. R., Coia, G., Hill, A., Mullbacher, A., Westaway, E. G. and Blanden, R. V., Preliminary analysis of murine cytotoxic T cell responses to the proteins of the flavivirus Kunjin using vaccinia virus expression. J. Gen. Virol. 1991. 72: 16451653.
  • 39
    Lobigs, M., Arthur, C. E., Mullbacher, A. and Blanden, R. V., The flavivirus nonstructural protein NS3 is a dominant source of cytotoxic T cell peptide determinants. Virology 1994. 202: 195201.
  • 40
    Diamond, M. S., Shrestha, B., Marri, A., Mahan, D. and Engle, M., B cells and antibody play critical roles in the immediate defense of disseminated infection by West Nile encephalitis virus. J. Virol. 2003. 77: 25782586.
  • 41
    Welsh, R. M., McNally, J. M., Brehm, M. A. and Selin, L. K., Consequences of cross-reactive and bystander CTL responses during viral infections. Virology 2000. 270: 48.
  • 42
    Lai, C. J. and Monath, T. P., Chimeric flaviviruses: Novel vaccines against dengue fever, tick-borne encephalitis, and Japanese encephalitis. Adv. Virus Res. 2003. 61: 469509.
  • 43
    Huang, C. H., Peng, S., He, L., Tsai, Y. C., Boyd, D. A., Hansen, T. H., Wu, T. C. and Hung, C. F., Cancer immunotherapy using a DNA vaccine encoding a single-chain trimer of MHC class I linked to an HPV-16 E6 immunodominant CTL epitope. Gene Ther. 2005. 12: 11801186.
  • 44
    Yu, Y. Y., Netuschil, N., Lybarger, L., Connolly, J. M. and Hansen, T. H., Cutting edge: Single-chain trimers of MHC class I molecules form stable structures that potently stimulate antigen-specific T cells and B cells. J. Immunol. 2002. 168: 31453149.
  • 45
    Klein, J., Huang, H. J. S., Lemke, H., Hammerling, G. J. and Hammerling, U., Serological analysis of H-2 and Ia molecules with monoclonal antibodies. Immunogenetics 1979. 8: 419432.

Note added in proof

  1. Top of page
  2. Abstract
  3. Introduction
  4. Results
  5. Discussion
  6. Materials and methods
  7. Acknowledgements
  8. Note added in proof

The article cited in this text by Brien et al., as “published in this issue of EJI” shows that the transfer of highly enriched antigen-primed CD8+ T cells protects the majority of mice against lethal WNV infection has the following [ ]reference