Specificities of Human CD4⁺ T Cell Responses to an Inactivated Flavivirus Vaccine and Infection: Correlation with Structure and Epitope Prediction

ABSTRACT

Tick-borne encephalitis (TBE) virus is endemic in large parts of Europe and Central and Eastern Asia and causes more than 10,000 annual cases of neurological disease in humans. It is closely related to the mosquito-borne yellow fever, dengue, Japanese encephalitis, and West Nile viruses, and vaccination with an inactivated whole-virus vaccine can effectively prevent clinical disease. Neutralizing antibodies are directed to the viral envelope protein (E) and an accepted correlate of immunity. However, data on the specificities of CD4⁺ T cells that recognize epitopes in the viral structural proteins and thus can provide direct help to the B cells producing E-specific antibodies are lacking. We therefore conducted a study on the CD4⁺ T cell response against the virion proteins in vaccinated people in comparison to TBE patients. The data obtained with overlapping peptides in interleukin-2 (IL-2) enzyme-linked immunosorbent spot (ELISpot) assays were analyzed in relation to the three-dimensional structures of the capsid (C) and E proteins as well as to epitope predictions based on major histocompatibility complex (MHC) class II peptide affinities. In the C protein, peptides corresponding to two out of four alpha helices dominated the response in both vaccinees and patients, whereas in the E protein concordance of immunodominance was restricted to peptides of a single domain (domain III). Epitope predictions were much better for C than for E and were especially erroneous for the transmembrane regions. Our data provide evidence for a strong impact of protein structural features that influence peptide processing, contributing to the discrepancies observed between experimentally determined and computer-predicted CD4⁺ T cell epitopes.

IMPORTANCE Tick-borne encephalitis virus is endemic in large parts of Europe and Asia and causes more than 10,000 annual cases of neurological disease in humans. It is closely related to yellow fever, dengue, Japanese encephalitis, and West Nile viruses, and vaccination with an inactivated vaccine can effectively prevent disease. Both vaccination and natural infection induce the formation of antibodies to a viral surface protein that neutralize the infectivity of the virus and mediate protection. B lymphocytes synthesizing these antibodies require help from other lymphocytes (helper T cells) which recognize small peptides derived from proteins contained in the viral particle. Which of these peptides dominate immune responses to vaccination and infection, however, was unknown. In our study we demonstrate which parts of the proteins contribute most strongly to the helper T cell response, highlight specific weaknesses of currently available approaches for their prediction, and demonstrate similarities and differences between vaccination and infection.

INTRODUCTION

Flaviviruses are the most important causes of arthropod-transmitted viral infections in humans and include yellow fever (YF), dengue (DEN), Japanese encephalitis, West Nile (WN), and tick-borne encephalitis (TBE) viruses (1). With all of these viruses, the induction of neutralizing antibodies is believed to be responsible for long-term immunity after natural infection and vaccination (2, 3). Effective B cell responses, however, are strongly dependent on the induction of antigen-specific CD4⁺ T cells that contribute to affinity maturation, isotype switching, and immunological memory (4, 5). Their stimulation requires the uptake of antigen and its proteolytic processing into peptides by antigen-presenting cells (APCs) (6, 7), association of peptides with major histocompatibility complex (MHC) class II (MHC-II) molecules (HLA-DRB1, -DRB3/4/5, -DP, and -DQ) (8, 9), and transport of these complexes to the plasma membrane for specific interactions with T cell receptors (TCRs) (10). The process from antigen uptake to stimulation of CD4⁺ T cells is thus highly complex, and, at each of the steps, restrictions in the generation and selection of peptides can occur. These variables, together with individual-specific variations of the TCR repertoire that affect the selection of peptide–MHC-II complexes with appropriate affinities, limit the diversity of CD4⁺ T cell responses in individuals and are responsible for the phenomenon of immunodominance, i.e., the restriction of CD4⁺ T cell specificities to a limited number of epitopes from complex protein antigens. Furthermore, the mode of antigen delivery (i.e., infection and endogenous protein synthesis in APCs versus delivery of exogenous antigens after immunization with inactivated or subunit vaccines) as well as adjuvants can modulate the selection of peptides from proteins with the same primary structures (11–13).

Direct cell-cell interactions of B cells with CD4⁺ T cells, required for efficient antibody production, are mediated by the TCR specifically recognizing peptide–MHC-II complexes on the B cell. These peptides are generated from protein antigens that are bound and endocytosed by B cells through their cognate B cell receptors; i.e., they are derived from the same protein or protein complex which also carries the epitope recognized by the B cell. Data obtained with influenza virus and hepatitis B virus (HBV) (14–18) indicate that B cells can internalize whole virus particles, and peptides derived from internal proteins of the virion can therefore also function as helper T cell epitopes for the production of neutralizing antibodies targeting the envelope proteins.

In our study, we have used the TBE virus system to compare the specificity and immunodominance of CD4⁺ T cell responses induced by the viral structural proteins contained in the aluminum hydroxide-adjuvanted inactivated whole-virus TBE vaccine to those induced by the same proteins after natural infection. Such comparative analyses of CD4⁺ T cell responses to replicating and nonreplicating viruses have not been conducted so far. Based on the known atomic structures of viral proteins, we also investigated influences of protein structural factors on patterns of immunodominance and evaluated specific strengths and weaknesses of computer prediction of epitopes by comparison with experimentally determined data. The algorithms underlying these T cell epitope prediction programs consider only the interaction strength between peptides and MHC-II molecules and not the preceding steps of antigen processing and transport or interactions between peptide–MHC-II complexes and TCRs (19).

Like flaviviruses in general, TBE virus is a small enveloped virus (diameter, ca. 50 nm) that is composed of only three structural proteins, the capsid protein C and two membrane-associated proteins, prM/M and E (Fig. 1A). prM (precursor of M) is a component of immature particles and is proteolytically cleaved during virus maturation, leaving M in the membranes of fully infectious virions. The E (envelope) protein mediates viral entry functions (binding to cell surfaces and membrane fusion after receptor-mediated endocytosis) and is the principal target of neutralizing antibodies (20). Structurally, flaviviruses are among the best-studied enveloped viruses. The overall protein organizations of immature and mature virus particles have been determined by cryo-electron microscopy (21, 22), and X-ray crystallographic structures are available for soluble forms of E (sE) proteins (Fig. 1A) from several flaviviruses, including TBE virus (23). The structure of C has been determined for Kunjin (KUN) virus (a variant of WN virus) (1) and is assumed to be similar to that of TBE virus in terms of sequence homology and secondary structure predictions (24).

FIG 1.

CD4⁺ T cell response to the TBE virus structural proteins C, prM/M, and E. (A) Schematic representation of a flavivirus particle, showing an immature and mature virion. The virion has three structural proteins: C (capsid), prM/M (membrane), and E (envelope). The capsid contains the positive-stranded RNA and several copies of the capsid protein C. Immature virions are covered with prM-E heterodimers. The proteolytic cleavage of prM leads to the reorganization of the E proteins and the formation of particles covered with E dimers. sE, soluble form of E lacking the membrane anchor and stem; M, membrane-anchored cleavage product of prM. (Adapted from PLoS Pathogens [84].) (B) Magnitude of individual CD4⁺ T cell responses to TBE virus C, prM/M, and E from 40 booster-vaccinated, 45 infected, and 5 TBE-naive individuals determined by IL-2 ELISpot assay. Statistical comparisons between the data from vaccinated and infected individuals were performed using a Kruskal-Wallis test (P < 0.0001) and Dunn's multiple comparison tests. Significant differences were observed for the responses to C and E (indicated by stars). Medians are indicated by red lines. (C and D) Spearman correlation of individual C- and E-protein-specific CD4⁺ T cell responses of booster-vaccinated (Vacc) and infected (Inf) individuals. (E) E/C ratios of individual CD4⁺ T cell responses. Values below the cutoff of 21 spots/10⁶ cells were set at 10 for this analysis. (F) Percentage of spots contributed by C, prM/M, and E peptides in vaccinated (n = 40) and infected (n = 45) individuals.

Our data revealed immunodominant regions in both C and E that correlated with specific protein elements and domains, providing evidence for a strong influence of structural properties on antigen-processing and MHC-II-loading pathways. Some, but not all, of the immunodominant sequences were identical in vaccinated and infected individuals, suggesting that the generation and/or presentation of certain peptides differs between infection and immunization with an inactivated adjuvanted vaccine. An excellent match of experimentally determined and in silico predicted epitopes was found for the immunodominant regions in C, but much less concordance characterized the predictions for the envelope proteins prM/M and E, reflecting the protein-specific impact of steps in the antigen-processing and CD4⁺ T cell stimulation pathways that are not considered in the prediction programs.

MATERIALS AND METHODS

Study design and patients.

Analyses of CD4⁺ T cell responses were performed with peripheral blood samples. The first study group included 40 healthy Caucasians (age range, 18 to 77 years; median age, 63 years; 21 female and 19 male) who received a booster vaccination with an inactivated whole-virus TBE vaccine (FSME-Immun, 0.5 ml; Baxter). Peripheral blood samples were taken 13 to 38 days after booster vaccination. None of the participants had taken immunosuppressive drugs or had any acute infection, clinically significant disease, or any health condition known to influence immune responses.

The second study group consisted of 47 Caucasians (age range, 18 to 81 years; median age, 52 years; 25 female and 22 male) hospitalized with clinical symptoms of acute TBE virus infection at the Hospital České Budĕjovice, Czech Republic. TBE virus infection was confirmed by TBE virus-specific IgG and IgM analysis of acute plasma samples. Peripheral blood samples were taken 6 to 73 days (median, 22 days) after the first symptom. For three patients the exact day of symptom onset was not known.

As a negative control, samples of five TBE-seronegative Caucasians (age range, 21 to 75 years; median age, 24 years; 3 female and 2 male) who had no history of TBE vaccination or infection were analyzed. None of the negative controls had taken immunosuppressive drugs or had any acute infection, clinically significant disease, or any health condition known to influence immune responses.

Ethics statement.

The studies were approved by the ethics committees of the Medical University of Vienna, Austria (approval no. 590/2007), and the Hospital České Budĕjovice, Czech Republic (approval no. 8/2008). Written informed consent was obtained from all participants.

Preparation of blood samples.

For T cell assays, peripheral blood mononuclear cells (PBMCs) were isolated from whole-blood samples collected in sodium citrate tubes by density gradient centrifugation using Ficoll-Paque Plus (GE Healthcare) and cryopreserved for future use.

Serum or plasma samples were stored at −20°C until analysis.

CD8 depletion.

Frozen PBMCs were thawed and diluted 1:10 in RPMI 1640 medium (Sigma) containing CTL Wash Supplement (Cellular Technology Limited), 1% glutamine (Sigma), and 50 units/ml Benzonase (Novagen) according to the instructions of Cellular Technology Limited. PBMCs were depleted of CD8⁺ T cells using anti-CD8 antibody-coated immunomagnetic beads and LD columns (Miltenyi Biotec), according to the manufacturer's instructions. The CD8-depleted PBMCs were resuspended in serum-free medium (AIM-V; Gibco). After overnight incubation at 37°C in 5% CO₂, the cells were counted, centrifuged at 300 × g for 10 min, and resuspended at a final concentration of 2 × 10⁶ cells/ml in AIM-V medium for use in interleukin-2 (IL-2) enzyme-linked immunosorbent spot (ELISpot) assays.

The purity of CD8-depleted PBMC samples was assessed by flow cytometry using anti-CD3-phycoerythrin (PE), anti-CD8-allophycocyanin (APC), and 7-aminoactinomycin D (7-AAD) (all, BD Bioscience), which usually detected ≤1% CD8⁺ CD3⁺ T cells.

Peptides.

A total of 188 15-mer peptides overlapping by 11 amino acids (aa) to cover the entire amino acid sequences of the C, prM/M, and E proteins from TBE virus Neudörfl strain (NCBI GI 27596775, GI 27596776, and GI 27596778) were synthesized by and purchased from JPT (Berlin, Germany). The purity of all peptides was >70% as determined by high-performance liquid chromatography. The peptides were grouped into three maxipools, which contained all peptides that covered each of the TBE virus proteins C (n = 26), prM/M (n = 40), and E (n = 122). In addition, up to 12 peptides were arranged into minipools of C (n = 10), prM/M (n = 13), and E (n = 22), with each peptide present in two different minipools, as described previously (25). To confirm positive results obtained with these minipools, samples were then tested with single peptides. Peptide pools and single peptides were dissolved in dimethyl sulfoxide (DMSO) at a concentration of 1.25 mg of each peptide/ml and then diluted in AIM-V medium at a concentration of 8 μg of each peptide/ml. These stock solutions were kept frozen at −20°C until use.

IL-2 ELISpot assay.

The IL-2 ELISpot assay (Mabtech) was performed essentially according to the manufacturer's instructions. Briefly, polyvinylidene difluoride (PVDF)-ELISpot plates (Millipore) were treated with 70% ethanol for 30 min before being coated with 1 μg of anti-IL-2 antibody (IL-2-I/249). Plates were blocked with medium (RPMI 1640 medium; Sigma) containing 10% human serum (PAA), 1% penicillin/streptomycin/glutamine (Gibco), and 1% nonessential amino acids (Sigma) for 1 to 3 h at 37°C in 5% CO₂. After plates were washed with phosphate-buffered saline (PBS), 50 μl of AIM-V medium (Gibco) and 2 × 10⁵ CD8-depleted PBMCs in 100 μl of AIM-V medium (see “CD8 depletion” above) were added per well. The cells were then stimulated with 50 μl of either pooled peptide or single peptide at a final concentration of 2 μg of each peptide/ml or with 0.5 μg/ml phytohemagglutinin as a positive control. As a negative control, AIM-V medium was used. After incubation for about 45 h at 37°C in 5% CO₂, the plates were washed twice with PBS containing 0.05% Tween 20 and twice with PBS. Spots were developed with 0.05 μg of biotin-conjugated antibody (IL-2-II) for 2 h, streptavidin-alkaline phosphate (ALP; 1:1,000) for 1 h, and 5-bromo-4-chloro-3-indolylphosphate/nitroblue tetrazolium (BCIP/NBT; Sigma) for 30 min at room temperature. The dried plates were analyzed using a Bio-Sys Bioreader 5000 Pro-S/BR177 and Bioreader software, generation 10.

Data were calculated as spots per 1 × 10⁶ CD8-depleted PBMCs after subtraction of the negative control (mean number of spots from three to four unstimulated wells) as has been described for IL-2 ELISpot assays previously (26, 27). A positive test result was defined as >20 spots per 1 × 10⁶ CD8-depleted PBMCs.

The response to a single peptide was defined as positive if test results after stimulation with the maxi- and minipools containing the peptide and the confirmatory single peptide testing revealed >20 spots per 1 × 10⁶ CD8-depleted PBMCs.

ELISA.

TBE virus-specific IgG antibodies were analyzed by enzyme-linked immunosorbent assay (ELISA) using purified formalin-inactivated TBE virus strain Neudörfl as described previously (28). Serum or plasma samples were quantified in arbitrary units (AU) using a standard polyclonal human anti-TBE virus serum set at 1,000 AU. ELISA values above 220 AU were considered positive. At least two independent experiments were performed for each sample to calculate mean concentrations.

Neutralization assay.

Neutralization tests (NTs) were carried out in baby hamster kidney cells (ATCC BHK-21) as described previously (29). Serial dilutions of serum or plasma samples (in duplicates) were mixed with 25 PFU of TBE virus strain Neudörfl and incubated for 1 h at 37°C. Cells then were added, and incubation was continued for 3 days. The presence of virus in the supernatant was assessed by ELISA. The virus neutralization titer was defined as the reciprocal of the serum sample dilution that gave a 90% reduction in the absorbance readout in the assay compared to the control without antibody. At least two independent experiments were performed for each sample to calculate geometric mean titers.

Determination of molar ratios of structural proteins in virus particles.

To quantify the amounts of structural proteins in TBE virus particles, the membrane proteins E and prM/M were separated from the capsid by detergent solubilization and low-speed centrifugation as described previously (30). For this purpose, mature TBE virus (strain Neudörfl) was grown in primary chicken embryo cells, concentrated by ultracentrifugation, and purified by two cycles of sucrose density gradient centrifugation as described previously (31). The purified virus, at a final protein concentration of 100 μg/ml in TAN buffer (0.05 M triethanolamine, pH 8.0, 0.1 M NaCl), was mixed with β-d-dodecylmaltoside (DDM; final concentration, 1%) and incubated for 1 h at room temperature. The aggregated capsid was separated from solubilized membrane proteins by centrifugation at 14,600 rpm for 30 min at 4°C (Ti90 rotor; Beckman). The pellet was solubilized by incubation in TAN buffer containing 0.2% SDS for 30 min at 65°C. The purity of the capsid and membrane protein fractions was controlled by SDS-PAGE according to Laemmli (32), and the amount of protein in each fraction (E and prM/M in the supernatant; aggregated capsid in the pellet) was determined as described by Schaffner and Weissmann (33). In this procedure, proteins are precipitated by trichloroacetic acid, vacuum blotted onto a nitrocellulose membrane, stained with amido black, eluted with ethanol, and quantified colorimetrically. Protein concentrations were determined using bovine serum albumin as a standard.

Structural analysis.

For analysis of the C protein, the crystallographic structure of the flavivirus KUN C protein (24) was used as no crystallographic data exist for the TBE virus C protein. To label identified TBE virus C clusters in the KUN C structure, the amino acid sequences of TBE C (strain Neudörfl; NCBI GI 27596775) and KUN C (strain MRM61C; Swiss-Prot/GenBank accession number P14335) were aligned.

For structural analysis of the stem-anchor region of the E protein, the cryo-electron microscopy structure of the flavivirus DEN E protein (34) was used, as no crystallographic data exist for the TBE virus E stem-anchor region. To label the identified TBE virus E stem cluster in the DEN virus E structure, the amino acid sequences of TBE virus E (strain Neudörfl; NCBI GI 27596778) and DEN-2 virus E (strain Thailand/16681/1984; Swiss-Prot/GenBank accession number P29990) were aligned.

Alignment was done using Geneious Pro, version 5.0.4 (Geneious alignment, BLOSUM62; gap open penalty, 12; gap extension penalty, 3; global alignment [Needleman-Wunsch]).

HLA typing.

HLA typing was performed by nucleotide sequencing of exon 2 of HLA-DRB1, -DRB3/4/5, and -DQB1 genes and exons 2 and 3 of HLA-DPB1 genes (35, 36). Briefly, the amplification products were purified by polyethylene glycol (PEG) precipitation and directly sequenced by cycle sequencing with Big Dye Terminator chemistry on an ABI 3100 capillary sequencing device. The sequences were analyzed using GenDX SBT Engine software (GenDX, Utrecht, The Netherlands) and compared to the ImMunoGeneTics (IMGT)/HLA database, and the types were assigned accordingly.

Computer prediction.

The MHC-II binding predictions were made in November 2012 using the Immune Epitope Database (IEDB) analysis tool “IEDB recommended” (www.iedb.org) (37, 38). Query submission was automated using the framework provided by the PeptX project (39). Peptides with an IEDB percentile rank score of 5 or lower were used for further analysis.

The amino acid sequences of the TBE virus proteins C, prM/M, and E (strain Neudörfl; NCBI GI 27596775, GI 27596776, and GI 27596778) were entered separately, and predictions were done for all peptides used for experimental testing and the experimentally determined HLA class II alleles offered by IEDB from all individuals (see Tables S1 and S2 in the supplemental material).

Prediction of CD4⁺ T cell epitopes and labeling of the transmembrane (TM) domains was done for the following proteins: YF virus prM/M and E (GenBank U17066) and TM label (Swiss-Prot/GenBank P03314); DEN virus prM/M and E (GenBank P29990); severe acute respiratory syndrome (SARS)-coronavirus spike 2, E, and M (GenBank AAP41036.1) and TM label (Swiss-Prot/GenBank P59594, P59637, and P59596); influenza virus hemagglutinin 2 (HA2) (as published in reference 40), starting with amino acid position 328, and TM label (Swiss-Prot/GenBank P03437); herpes simplex virus 1 (HSV-1) B protein (Swiss-Prot P10211); vaccinia virus (VACV) extracellular enveloped virion (EV) A33R and B5R (Swiss-Prot/GenBank P68616 and P21115); HBV surface antigen (HBsAg) (Swiss-Prot/GenBank P31873); and HIV gp160 (Swiss-Prot/GenBank P04578). For these predictions, 14 human HLA class II alleles common in middle Europe (DRB1*01:01, 03:01, 07:01, 11:01, 13:01, and 15:01; DPB*02:01, 04:01, and 04:02; DQB*02:01, 03:01, 03:02, 05:01, and 06:02) were used.

Hydrophobicity of peptides.

The percentage of hydrophobic amino acids in selected TBE virus peptides was calculated using the web-based Peptide Property Calculator from GenScript (https://www.genscript.com/ssl-bin/site2/peptide_calculation.cgi) and the respective peptide sequence.

Statistical analysis.

Statistical tests were performed with GraphPad Prism, version 5.

A nonparametric Kruskal-Wallis test was used to perform comparisons of overall CD4⁺ T cell reactivity for the C, prM/M, and E proteins in groups of vaccinated and infected subjects.

A Wilcoxon signed-rank test was used to compare the numbers of single peptides that induced a CD4⁺ T cell response.

For identification of the peptides that most frequently induced CD4⁺ T cell responses, a Fisher's exact or chi-square test was used.

Two-by-two correlations were evaluated using a Spearman correlation coefficient.

RESULTS

Testing of peptide pools.

We first determined the overall extent of CD4⁺ T cell responses of 40 TBE-vaccinated, 45 TBE virus-infected, and 5 TBE-naive individuals in an IL-2 ELISpot assay using three peptide pools comprised of overlapping 15-mer peptides that covered the entire sequence of each of the TBE virus structural proteins C, prM/M, and E.

The rationale for using an IL-2 ELISpot assay was based on previous studies which showed that CD4⁺ T cells producing IL-2 but not those producing other cytokines (gamma interferon [IFN-γ] and tumor necrosis factor alpha [TNF-α]) correlate with antibody titers (41, 42) and on initial experiments which indicated that IL-2 and IFN-γ ELISpot assays detected similar magnitudes of TBE virus-specific CD4⁺ T cell responses in both vaccinated and infected individuals (data not shown).

The results displayed in Fig. 1B reveal significantly higher C and E peptide responses in vaccinated than in infected individuals (Kruskal-Wallis test, P < 0.0001), and an extensive degree of individual variation was observed in both groups (e.g., E-specific spots ranging from 39 to 633 per 10⁶ cells after vaccination and <20 to 430 per 10⁶ cells after natural infection). For prM/M peptides, the overall reactivities were very low and not significantly different between the two groups. TBE-naive individuals (n = 5) showed no response to any of the tested peptide pools, confirming the specificity of the analysis. Although the magnitudes of the responses to the E and C peptide pools correlated significantly in both vaccinated and infected humans (Fig. 1C and D), considerable individual variation of the ratios of E- and C-specific spots was observed, ranging from 0.6 to 30 in vaccinated and 0.1 to 10 in infected individuals (coefficients of variation, 140% for vaccinated and 71% for infected humans) (Fig. 1E). We also quantified TBE virus-specific antibodies in the blood samples used for cellular analyses and found a positive correlation between the IL-2 ELISpot data and ELISA concentrations as well as NT titers in both vaccinated and infected individuals (Fig. 2).

FIG 2.

Correlation of CD4⁺ T cell and antibody responses. The magnitude of individual CD4⁺ T cell responses to TBE virus C (A and B) and E (C and D) peptide maxipools was plotted against the corresponding ELISA units (solid dots) and NT titers (empty circles) in vaccinated (A and C) and infected (B and D) individuals. Correlations were calculated using a Spearman correlation coefficient. Linear regressions are indicated by solid (ELISA) or dashed (NT) lines.

An analysis of the contribution of peptides from the three structural proteins to the total response is displayed in Fig. 1F. Assuming that all three proteins have similar propensities to induce CD4⁺ T cell responses, peptides of the C protein are 2- to 3-fold overrepresented (∼32% of total response), considering its molecular mass of only 10.6 kDa (∼15% of total molecular mass in immature and ∼13% in mature virus particles). This suggested that peptides from the C protein are either specifically favored during CD4⁺ T cell stimulation or that they are overrepresented in the virus particle relative to the envelope proteins. Since published data on the contents of C in flavivirus virions do not yet exist, we determined the molar ratios of C relative to the envelope proteins (E and prM/M) in purified virus particles. For this purpose, we analyzed the protein contents in capsid and envelope protein fractions separated after detergent solubilization (see Materials and Methods) and compared the ratios obtained with the molecular weight ratios of these proteins (Table 1). Since the SDS-PAGE analysis revealed some uncleaved prM protein in the virus preparation (Fig. 3), we performed the calculation for both completely mature (containing only M) and completely immature (containing only prM) virions. As can be seen from Table 1, the experimentally determined C-to-envelope protein ratios were at least 3-fold higher than the ratios of the molecular weights, indicating at least a 3-fold molar excess of C relative to the envelope proteins in virions, which is in agreement with the ELISpot data.

TABLE 1.

Protein determinations of separated capsid and envelope protein fractions and calculation of molar ratios in the virion

Expt. no.	Ratio of protein content by fraction					Molar excess of C relative to:
	Based on exptl data			Based on MWa
	C (μg)	Env (μg)b	C to Env	C to E+M	C to E+prM	E+Mc	E+prMd
1	79.6	138.0	0.58	0.17	0.15	3.4	3.9
2	69.6	154.0	0.45	0.17	0.15	2.6	3.0
3	62.0	124.0	0.50	0.17	0.15	2.9	3.3

^aMolecular weight (MW) values are as follows: C, 10,600; E, 53,500; M, 8,300; prM, 18,500.

^bEnv, envelope proteins (E+prM/M).

^cAssuming that all particles are mature.

^dAssuming that all particles are immature.

FIG 3.

SDS-PAGE of TBE virus as well as its envelope proteins and capsid fractions after solubilization with DDM. The soluble fraction (containing the membrane-associated proteins E and prM/M) and aggregated fraction (containing protein C) were separated by low-speed centrifugation. Identical aliquots of the supernatant (SN) and the pellet (P) were analyzed by SDS-PAGE and stained with Coomassie blue. V, untreated virus control. The positions of E, prM, C, and M are indicated.

Single-peptide testing.

To obtain information on the epitope specificity and diversity of CD4⁺ T cell responses, we performed ELISpot analyses of TBE-vaccinated and -infected individuals with peptide minipools and single peptides covering the entire C, prM/M, and E polypeptides. As expected, the responses were restricted to a limited set of peptides, and a high degree of individual variation was observed with respect to the distribution of peptide responses in both study groups. Representative examples of two vaccinated and two infected individuals are shown in Fig. 4. Due to individual-specific variation of CD4⁺ T cell responses, patterns of immunodominance and possible differences between vaccinated and infected humans can become apparent only from cumulative data. We therefore calculated the frequency of obtaining a positive ELISpot assay result for each peptide using the results from all vaccinated and infected individuals obtained from the single-peptide testing. This frequency is displayed in Fig. 5A (vaccinated individuals) and Fig. 5B (patients) as the percentage of responders out of all individuals reacting with at least one peptide of a given protein. Consistent with the low response to the prM/M peptide pool (Fig. 1B), single peptides yielded only a few positive signals, and as such a meaningful analysis was not possible.

FIG 4.

Individual variation of CD4⁺ T cell responses. CD4⁺ T cell responses against TBE C, prM/M, and E single peptides were measured using an IL-2 ELISpot assay. Examples of two vaccinated (A and B) and two infected (C and D) individuals are shown.

FIG 5.

Mapping of immunodominant experimental and predicted CD4⁺ T cell responses. (A and B) Percentage of positively tested vaccinated (A) and infected (B) individuals recognizing a specific single peptide within the C (n = 31 vaccinated; n = 13 infected) and E (n = 34 vaccinated; n = 26 infected) proteins. prM/M-specific single-peptide responses were too low for evaluation. Peptides recognized significantly more often than the average were identified (Fisher's exact or chi-square test; significance level of P < 0.05, separately for each protein and each group) and are indicated by asterisks. Clusters of these peptides are numbered 1 to 9. (C) Crystallographic structure of the flavivirus Kunjin C protein (24) consisting of four helices (H1 to H4; left panel). For the N-terminal region (gray line), no crystallographic data exist. Crystallographic structure of the TBE virus soluble E (23) consisting of three domains (DI, DII, and DIII; right panel). Recent data suggest that the stem-anchor region consists of three alpha-helices in the stem and two alpha-helices in the transmembrane (TM) anchor (34) (boxes). Dominant clusters labeled in panels A and B are colored as follows: C, green; E DI, red; E DII, yellow; E DIII, blue; E stem, magenta. N- and C-terminal residues are indicated. (D and E) Cumulative computer prediction of CD4⁺ T cell epitopes for HLA class II alleles from vaccinated (n = 39 subjects; number of predicted alleles, 233) (D) and infected (n = 44 subjects; number of predicted alleles, = 264) (E) individuals. The percentage of alleles predicting a specific single peptide is shown for peptides of all three structural proteins (C, prM/M, and E). In panels A, B, D, and E, lines below the x axes indicate the TM anchor domain of prM/M (black) and the domains of the E protein (DI, red; DII, yellow; DIII, blue; stem, magenta; TM, black).

As can be seen in Fig. 5, the cumulative analysis showed specific regions of peptides that were positive in significantly more individuals (as assessed by chi-square or Fisher's exact test) than the average of all peptides (labeled 1 and 2 for protein C and 1 to 9 for protein E) and thus dominated the CD4⁺ T cell response. The same peptides also induced the highest numbers of spots in the ELISpot assays, and evaluations based on the magnitude of the responses yielded similar patterns (data not shown). Although some of these clusters were similar in vaccinated and infected individuals, specific differences also became apparent in some instances (see below). Since structural aspects of the proteins can potentially influence the patterns of CD4⁺ T cell responses (43–47), we determined the location of the dominant peptides in the three-dimensional (3D) structures of C and E (Fig. 5C). In the following paragraphs we describe this structure-related analysis of immunodominance and emphasize the similarities and differences observed between the two groups.

(i) C protein.

Two peptide clusters dominated the total CD4⁺ T cell response. C peptides of cluster 2 were recognized by up to 74% of vaccinated and 55% of infected responders. Structurally, these peptides corresponded to helix 2 and helix 4 of the protein (Fig. 5C), whereas peptides of helices 1 and 3 had much lower representation, with those for the N-terminal 20 amino acids almost completely absent.

(ii) E protein.

For E protein, the response was less focused than for C, and in the vaccinated group only a single peptide (in cluster 7) was positive in 50% of responders (Fig. 5A). Nevertheless, distinct patterns of dominance were discernible. In both vaccinated and infected individuals, peptides of domain III (DIII) were overrepresented and yielded three distinct peaks, labeled 6, 7, and 8 in Fig. 5. In vaccinated individuals, further clusters were found in the two other domains (Fig. 5A, peaks 1 to 5) although these were not significantly overrepresented in patients (Fig. 5B). On the other hand, patients displayed a dominant reactivity peak with peptides forming a helix in the so-called stem region of E (Fig. 5B, peak 9) that was absent in vaccinated individuals.

Since primary infection has been proposed to be associated with a more diverse CD4⁺ T cell response than immunization with exogenous nonreplicating antigens (12), we made comparative analyses of the breadth of C- and E-specific responses in our groups of vaccinated and infected individuals. For this purpose we selected pairs of individuals from both groups that had similar extents of reactivities with the respective peptide pools (Fig. 1) to avoid misinterpretations due to differences in response magnitude. The comparison of the numbers of peptides recognized by these individuals (Table 2) revealed no significant difference between natural infection and vaccination (Wilcoxon signed-rank test).

TABLE 2.

Number of C or E peptides recognized by matched pairs of vaccinated and infected individuals

Protein	No. of matched pairsa	No. of peptides by subject groupb						P
		Vaccinated			Infected
		Median	Min	Max	Median	Min	Max
C	11	4.0	0	11	3.0	0	9	0.18
E	12	5.0	0	14	3.5	0	31	0.75

^aThe magnitude of response to maxipools (≥50 spots/10⁶ cells) was comparable between the pairs of vaccinated and infected individuals included in this analysis.

^bMin, minimum; max, maximum.

Comparison with epitope predictions.

For epitope prediction based on peptide–MHC-II binding affinity, we used the IEDB database, which is the most comprehensive for human HLA class II alleles and which allowed predictions for all MHC-II DRB1 alleles, all but one of the DRB3/4/5 alleles, and about two-thirds of the DP and DQ alleles in our study participants. As a prerequisite for these analyses we first determined the HLA class II type of all individuals (see Tables S1 and S2 in the supplemental material). For each of the different alleles and peptides tested experimentally, the prediction program provided a percentile rank as a measure of their propensity to be a CD4⁺ T cell epitope. In order to compare these predictions with our experimental data, we selected the peptides with the best prediction values (IEDB percentile rank score of 5 or lower) and summed these data for all individuals in both groups. The results displayed in Fig. 5D and E show the distribution of these predicted CD4⁺ T cell epitopes in the sequences of C, prM/M, and E for the specific MHC-II alleles present in the two study groups.

The epitope predictions for the two groups were almost identical (compare Fig. 5D and E), indicating that HLA class II allele distributions were similar among the groups of vaccinated and infected individuals. For the C protein, an excellent match with the experimental data was obtained. The program most frequently predicted the same set of peptides derived from helices 2 and 4 of C. In E, a relatively good agreement/match was observed for the weakly dominant peaks 1 and 5 of domains I and II, as well as peptide peak 9 in the stem, whereas other experimentally dominant reactivities of peptides of DI and DII and those of DIII were not predicted to similar extents (Fig. 5D). Most importantly, the peptide yielding the highest percentage of responders in both vaccinated and infected individuals (in cluster 7, encompassing amino acids 329 to 343) was not within the top binders in the prediction. On the other hand, several frequently predicted epitopes did not have dominant counterparts in the experimental data.

Quantification of epitope predictability.

To obtain a quantitative measure of the congruence between epitope prediction and experimental data in our study system, we determined the following for C and E and every individual: (i) the percentage of experimentally identified peptides that were also positive in the prediction and (ii) the percentage of predicted peptides that were also identified experimentally. The prerequisite for inclusion in this analysis was the availability of prediction algorithms for all HLA class II alleles identified in these individuals (see Tables S1 and S2 in the supplemental material). The results of these comparisons are shown in Table 3. For C protein peptides, the epitope predictions yielded an excellent match with the experimental data (ca. 75%), which was substantially lower for E (ca. 28%). Since in many publications only DRB1 alleles were used as a basis for epitope predictions (48–53), we also made a comparative calculation using DRB1 alleles only. As shown in Table 3, the congruence of epitope prediction dropped from 75% to 42% for C peptides and from 28% to 16% for E peptides. Of all predicted epitopes, only 24% and 10% were also experimentally identified for C and E, respectively (Table 3).

TABLE 3.

Quantification of epitope predictability

Protein	No. of individualsa	% Predicted peptides by allele groupb		% Experimentally identified peptidesc (all HLA class II alleles)
		All HLA class II alleles	DRB1 alleles only
C	15	74.6	42.1	23.7
E	22	28.2	16.2	9.9

^aOnly individuals for whom predictions could be performed for all HLA class II alleles were included.

^bPercentage of experimentally identified peptides that were also positive in the prediction. Values are means.

^cPercentage of predicted peptides that were also identified experimentally. Values are means.

Transmembrane domains.

A striking feature of the predicted patterns was the dominance of peptides from the C terminus of the membrane proteins prM/M and E (Fig. 5), which did not have a counterpart in the experimental data. In both proteins, these sequences constitute a double membrane-spanning element (Fig. 5, labeled TM) that is derived from the specific mode of flavivirus polyprotein processing and anchors these proteins first in the endoplasmic reticulum membrane and then in the viral membrane (54). To analyze whether predicted immunodominance is a general feature of viral transmembrane (TM) protein domains, we performed CD4⁺ T cell epitope predictions for envelope proteins of viruses for which experimental data (including those for TM domain-derived peptides) were available. This included envelope proteins of HBV (50, 55–58), influenza virus (40, 59, 60), VACV-EV (61, 62), SARS-coronavirus (63, 64), HSV-1 (65), and HIV (66) as well as YF (67) and DEN (68) viruses. For each of these viruses the TM domain-derived peptides (both at the C terminus as in type I and at the N terminus as in type II membrane proteins) were among those with the highest prediction frequency (Fig. 6). However, in none of these cases was a dominant CD4⁺ T cell response to such peptides found experimentally, with a single exception: one of the four TM domains (proximal to the N terminus, amino acids 182 to 202) (Fig. 6) of hepatitis B virus surface antigen (HBsAg) contained CD4⁺ T cell epitopes that were described to be immunodominant in humans (50, 55–58). In conclusion, the high predictive values for CD4⁺ T cell epitopes in the TM domains of viral envelope proteins is not confirmed by experimental data in most instances.

FIG 6.

Computer prediction of potential CD4⁺ T cell epitopes from envelope proteins of different viruses. CD4⁺ T cell epitopes for surface proteins of hepatitis B virus (HBV), influenza virus, vaccinia virus (VACV) (extracellular envelope virion [EV]), SARS-coronavirus (SARS-CoV), herpes simplex virus 1 (HSV-1), HIV, and yellow fever (YF) virus were predicted using common human HLA class II alleles. Positions of transmembrane domains within the respective amino acid sequences are indicated by open rectangles. The prediction patterns were very similar for YF virus and dengue virus, and the latter was therefore not included in the figure.

DISCUSSION

In our study we exploited the opportunity to compare the specificities of CD4⁺ T cell responses after vaccination with an aluminum hydroxide-adjuvanted formalin-inactivated flavivirus vaccine to those after natural infection in two groups of individuals with very similar HLA class II allele distributions. Since direct or indirect information on the 3D structures of the viral structural proteins is available, we were able to correlate immunodominance patterns with structural characteristics of the antigens and to assess the congruence of in silico epitope predictions on the basis of specific features of individual proteins.

An important finding of our work is that the overall proportions of reactivities with peptide pools from the three proteins were similar in vaccinated and infected individuals (Fig. 1F) although virus replication in infected individuals could theoretically affect protein abundance and relative proportions of the three proteins in the host. In both groups, the response to the C protein was strongly overrepresented relative to its molecular weight, which was apparently due to a previously unrecognized 3-fold molar excess of C over prM/M and E in virus particles. The extents of experimentally determined reactivities to peptides from C, prM/M, and E were therefore completely concordant with the amounts and sizes of these proteins in virions, suggesting that the intrinsic propensities to induce a CD4⁺ T cell response were similar for all three proteins.

The strong CD4⁺ T cell response to C should also be seen in the light of its potential helper function in the production of neutralizing antibodies directed toward E. As shown for HBV and influenza virus, B cells expressing envelope protein-specific antibodies can internalize whole-virus particles and present peptides of internal proteins together with MHC-II at their plasma membranes. As a consequence, CD4⁺ T cells specific for internal proteins can provide help in the production of envelope protein-specific neutralizing antibodies (14–18). Consistent with this scenario, we found a significant correlation between the magnitudes of the C-specific CD4⁺ T cell and neutralizing antibody responses (Fig. 2).

In the course of flavivirus infections (in contrast to vaccination with inactivated vaccines) capsidless virus-like particles containing only prM/M and E can also be secreted (54). This could theoretically lead to a stronger envelope protein-specific CD4⁺ T cell response in TBE patients than in vaccinees. However, the overall reactivities with all three structural proteins were similar in vaccinated and infected individuals (Fig. 1F), suggesting that in both instances the whole virus particle is the principal source of peptides for MHC class II presentation. Surprisingly, the magnitude of the TBE virus-specific CD4⁺ T cell responses to C and E were significantly lower after infection than after vaccination (Fig. 1B). This can be due to viral antagonism of the host immune response during infection (69), an enhanced T cell response after booster vaccination, and/or aluminum hydroxide-related effects (70, 71).

Theoretically the reactivity pattern after infection could be less focused than that elicited in response to exogenous protein antigens because viral protein synthesis may override the DM editing and peptide selection machinery in infected APCs (12). However, we found no evidence for an increased breadth of CD4⁺ T cell responses after infection compared to results after vaccination although there is evidence that TBE virus replicates in dendritic cells (72). In a comparison of the number of peptides recognized by vaccinated and infected individuals that had about equal overall CD4⁺ T cell responses (Table 2), there was no statistically significant difference between the two groups, indicating that, in this specific case, infection does not lead to a broadening of the CD4⁺ T cell response.

Despite the strong variability of individual peptide specificities, characteristic patterns of immunodominance were observed at the population level, with both similarities and distinct differences between vaccinated and infected individuals (Fig. 5A and B). Evaluations based on either the percentage of positive responders or the magnitude of response to individual peptides yielded similar patterns. The response to single peptides of the membrane protein prM/M was very low (as was the case for DEN and YF viruses [67, 68]) and did not allow a meaningful interpretation. Therefore, our discussion of aspects of fine specificity is restricted to C and E. For both proteins, distinct epitope clusters were identified that, at least in some instances, were clearly linked to specific protein regions. This was most apparent for peptides derived from helices 2 and 4 of the C protein, which dominated the response in both vaccinated and infected individuals, whereas peptides from helices 1 and 3 were underrepresented (Fig. 5A, B, and C, left panel). Most strikingly, the N-terminal part of C (not resolved in the structure determined for KUN virus and not predicted to have a helical secondary structure) did not elicit any CD4⁺ T cell response.

The CD4⁺ T cell response to E was less focused than that to C, but distinct preferences for specific sequence elements were also observed in this case, especially with respect to DIII. Peptides of this domain were clearly overrepresented in the reactivity patterns of both vaccinated and infected individuals (Fig. 5A and B), and three hot spots (Fig. 5, peaks 6, 7, and 8) could be discerned. Structurally, these epitopes include β-sheets as well as loops extending from the lateral upper part of the domain (23) (Fig. 5C). These loops protrude from the surface of the virion and have been shown for different flaviviruses to be targets of strongly neutralizing antibodies in mice (1). Our findings are consistent with previous reports describing clusters of immunodominant epitopes in limited protein regions, often at exposed sites of the proteins (45, 46). A further structural link for immunodominance was provided by Landry in studies with influenza virus HA, in which dominant epitopes were found at the C-terminal flanks of conformationally stable protein segments (43).

Despite the concordance of certain immunodominance patterns between vaccinated and infected individuals, characteristic differences were found, especially with respect to certain regions in E. Specifically, none of the peptides in domains I and II (designated 1 to 5 in Fig. 5A and highlighted in Fig. 5C, right panels) that were classified as dominant in vaccinated humans reached this criterion in patients. Conversely, the two peptides designated 9 in Fig. 5B were dominant in infected but not vaccinated individuals. These peptides are located in the so-called stem region of E which, as resolved by cryo-electron microscopy of DEN virus particles, consists of three alpha helices that are located beneath the E protein shell and interact with the viral membrane (Fig. 1A) (34). Since the distributions of MHC-II alleles in the two groups were very similar (as was also reflected in the identical patterns of predicted epitopes), it is unlikely that the differences between vaccinated and infected individuals resulted from HLA class II heterogeneity. Instead, they more likely reflect an influence of the different routes of antigen presentation in the two groups and/or antigen modifications in the preparation of the vaccine. There is evidence from studies with influenza virus (73) that some peptides are generated only in the course of infection and protein synthesis in infected cells but not upon exogenous antigen uptake. On the other hand, virus treatment with formalin (as is used for preparing the TBE vaccine) results in protein cross-linking (74, 75) and can potentially affect antigen processing, the interaction of peptides with MHC-II, DM editing, and TCR recognition. Specifically, the cross-linking of E protein in the viral envelope is likely to prevent its conversion into the trimeric postfusion structure (76, 77) in the acidic endolysosomal compartment of the APCs, resulting in a different structural substrate for proteolytic processing. Differences in immunodominance patterns between vaccinated and infected individuals may also be due to the adjuvant (aluminum hydroxide) contained in the vaccine, which can modulate antigen processing and the TCR-based selection of CD4⁺ T cell clones (13). In this context it should also be considered that antigen processing and presentation can be influenced by antibodies bound to the antigen (78, 79) which, in the case of virion-antibody complexes, would be expected to modulate the CD4⁺ T cell response to peptides from the antibody-accessible E protein but not from the internally located C protein.

The comparison between experimentally determined and predicted epitopes yielded the major finding that the congruencies varied strongly between proteins and specific structures within these proteins. With the C protein, an excellent match was achieved (75% of experimentally determined epitopes were also predicted), and the patterns of immunodominance were virtually superimposable. In contrast, predictions for E were much less satisfactory, and about 70% of experimentally identified peptides did not have an IEDB percentile rank score of 5 or lower. Most importantly, the dominance of DIII-derived peptides observed after both vaccination and natural infection was not apparent in the prediction, suggesting the importance of protein-specific and protein domain-specific factors that can affect the match between epitope prediction and experimental data.

The comparative analysis of experimental and predicted data also revealed that their congruence was significantly lower when data were based on DRB1 alleles only. In this case, only 42% and 16% of experimentally determined C and E peptides, respectively, would have been predicted (Table 3). Considering these data, the common practice of using DRB1-based epitope predictions before the synthesis of peptides for experimental analyses may be problematic because a number of CD4⁺ T cell specificities that actually contribute to immune responses could be missed.

A very high risk of yielding strongly predicted epitopes that do not contribute to CD4⁺ T cell responses became apparent for the transmembrane domains of envelope proteins. The peptides derived from this region are relatively hydrophobic (containing up to 80% hydrophobic amino acids) (Table 4), which could bias their performance in the ELISpot assays. However, some of the most immunodominant peptides from the C protein are similarly hydrophobic (Table 4), and although this region was clearly underrepresented at the population level, some individuals did indeed have robust reactivities with peptides derived from the stem-anchor region of E (in both peptide pool and single-peptide analyses) (Table 5). Similarly, in a mouse immunization study with WN virus DNA and virus-like particle vaccines, a peptide of the WN virus E protein TM domain was found to elicit a strong CD4⁺ T cell response. The same peptide also enhanced the protective activity of a DEN virus vaccine (80). It thus appears that the (mostly) negative results obtained with TM peptides are unbiased and reflect an impairment of processing, presentation, and/or TCR recognition that is not taken into account by T cell epitope prediction algorithms (19). Such a conclusion is corroborated by the fact that in an analysis of a large set of viral glycoproteins from different virus families (Fig. 6), peptides of the TM domain were among the most frequently predicted, but when they were analyzed experimentally, they failed to be associated with a similar dominance (40, 59–67). There was a single exception, the N-terminal TM domain of HBsAg, which is described as being immunodominant in humans (50, 55–58). At least in the context of full-length HBsAg, however, this sequence element has been described to lie outside the membrane (81, 82) and therefore may have characteristics that differ from conventional TM domains. Since the IEDB T cell epitope prediction algorithms predict MHC-II–peptide binding affinities only (37, 38) (with an excellent match to experimentally determined affinities) (83), the discrepancies observed with peptides of the TM domains are likely to be associated with a suboptimal generation of such peptides through the antigen-processing machinery. This could be due to their shielding in lipid membranes or other structural features that affect proteolytic processing. Alternatively, some peptides could be more homologous to native human proteins than other regions of the viral proteins, and T cells specific for such sequences may have been deleted from the circulating repertoire. Since the IEDB algorithms consider only MHC-II–peptide binding, they do not account for holes in the CD4⁺ T cell repertoire due to central/peripheral tolerance. Additional research is thus required to identify influences of protein structure on immunodominance patterns of CD4⁺ T cell responses that may ultimately allow further improvements in CD4⁺ T cell epitope predictions.

TABLE 4.

Hydrophobicity of peptides from the TBE E TM domain and C peptides with similar hydrophobicity levels

Protein and peptide position (aa)	Peptide sequence	% Hydrophobic amino acids within the peptide	Vaccinated responders (%)a
E
441–455	LGGAFNSIFGGVGFL	53	0.0
445–459	FNSIFGGVGFLPKLL	60	8.8
449–463	FGGVGFLPKLLLGVA	67	14.7
453–467	GFLPKLLLGVALAWL	80	11.8
457–471	KLLLGVALAWLGLNM	73	0.0
461–475	GVALAWLGLNMRNPT	60	2.9
465–479	AWLGLNMRNPTMSMS	53	0.0
469–483	LNMRNPTMSMSFLLA	60	5.9
473–487	NPTMSMSFLLAGGLV	60	8.8
477–491	SMSFLLAGGLVLAMT	67	8.8
481–495	LLAGGLVLAMTLGVG	67	0.0
C
29–43	PRVQMPNGLVLMRMM	67	9.7
33–47	MPNGLVLMRMMGILW	73	12.9
37–51	LVLMRMMGILWHAVA	80	48.4
41–55	RMMGILWHAVAGTAR	60	64.5

^aPercentage of vaccinated responders recognizing the peptide.

TABLE 5.

IL-2 ELISpot results of CD4⁺ T cell response to TBE virus E transmembrane domain peptides using minipool and single-peptide analysis

Subject group and no.	CD4+ T cell response (no. of spots/106 cells) to:
	Minipoola	Single peptideb (aa)
		445–459	449–463	453–467	461–475	465–479	469–483	473–487	477–491	481–495
Vaccinated
209	73	21			21				36
211	89								59
212	84		54	54
214	33						25
222	83		38	33				23	28
227	29		26					36
229	68							22
230	269		143	118			33
231	45	22	27
242	51	25		45
Infected
114	26	41			31			56	36	51
118	59								28
132	38				59	34	24

^aMinipool containing peptides spanning aa 441 to 495.

^bResults in the ELISpot assay were negative for the peptides at aa 441 to 455 and aa 457 to 471.