Preservation of a critical epitope core region is associated with the high degree of flaviviral cross-reactivity exhibited by a dengue-specific CD4⁺ T cell clone

Abstract

Dengue is a member of the Flaviviridae, a large group of related viruses some of which co-circulate in certain regions (e.g. dengue and Yellow fever in South America). Immune responses cross-reactive between different dengue serotypes are important in the pathogenesis of dengue disease but it is not known whether previous infection with one flavivirus might affect the clinical course of subsequent infections with other members of the family. CD4⁺ T cells have been shown to be important in the production of cytokines in response to dengue infection and can demonstrate significant epitope cross-reactivity. Here, we describe the generation and characterisation of CD4⁺ T cell clones from a patient experiencing acute dengue infection. These clones were DRB1 *15⁺ and recognised epitope variants not only within other dengue viruses but certain other flaviviruses. This cross-reactivity was dependent upon the presence of a five-amino acid core region, consistent with structural observations of class II MHC binding to TCR demonstrating that only a subset of residues within an epitope bound to a class II molecule are “read out” by the TCR. This capacity of CD4⁺ T cell clones to recognise a given epitope despite considerable variation between viruses may be of pathological significance, particularly in regions where related viruses co-circulate.

Introduction

It was in the 1970s that epidemiological studies first demonstrated that most cases of clinically severe dengue disease were seen in those experiencing secondary infection with one of dengue’s four serotypes [1]. It is now known that anamnestic immune responses cannot only fail to protect but are capable of altering or even enhancing the clinical manifestations of an infection [2, 3]. That such responses are seen in the context of secondary infection with a closely related virus (such as might occur in an individual undergoing acute secondary dengue infection with a viral serotype different from that causing the primary) is clear [4]. Less apparent is whether previous exposure to a less closely related virus is of the same clinical significance in human disease as it is in experimental animal work [5]. Dengue is a member of a large group of related viruses, the Flaviviridae. Members include West Nile virus, Japanese encephalitis virus and yellow fever virus itself [6]. Certain of these viruses may circulate in the same geographical region and it is not known whether previous infection with one flavivirus (e.g. yellow fever) might affect the presentation of subsequent infection with another member of the family (e.g. dengue). Experimental observations suggest that it might be the case: dengue virus infection can be enhanced in vitro by anti-sera to other flaviviruses [7]. Moreover, a study examining immune responses to an experimental live attenuated dengue-vaccine candidate found that the recipients demonstrating immunity to yellow fever developed viremia earlier than the ones with no yellow fever immunity [8].

The association of heterologous secondary dengue infection with severe disease has been attributed to the enhancing effect of antibodies primarily specific for the first infection. It is likely that other immune components also play a part and an increasing body of evidence supports a role for cell-mediated immunity. Dengue and broader flaviviral cross-reactive responses are seen in both the humoral and cellular arms of the immune system [9, 10] following dengue infection. CD4⁺ and CD8⁺ T cell responses have been described [11-16] and may be beneficial or deleterious, contributing to pathogenesis through original antigenic-sin like mechanisms or the production of inflammatory cytokines [11, 12, 17]. CD4⁺ T cells are important sources of such cytokines and are necessary for the in vitro production of cytokines in response to dengue-specific stimulation by peripheral blood mononuclear cells (PBMC) taken from experimental dengue vaccine recipients [18].

T cells can demonstrate significant cross-reactivity. A single clone may recognise a large number of different peptide-MHC (pMHC) complexes and many different Tcell clones can respond to a single pMHC complex [19]. CD4⁺ T cells show greater tolerance of changes in an epitope than do CD8⁺ T cells and a single TCR may recognize class II epitopes with no sequence or physical homology at all [20]. This is thought to be a consequence of the positional flexibility the “open-groove” a class II molecule allows its long epitopes. Such “degeneracy” increases the repertoire of epitopes an individual’s limited array of TCR might be capable of recognising [21]. More than this, it implies that CD4⁺ memory T cells formed following a primary infection are more likely to respond to subsequent infection by a related pathogen. Such heterologous immunity could profoundly influence protective and potentially pathogenic immune responses to flaviviral infection in those regions where two or more members co-circulate.

Here, we describe the generation of CD4⁺ T cell clones showing a very high degree of cross-reactivity across different flaviviruses despite considerable differences in the relevant peptide sequence. Recognition appeared to be dependent upon the preservation of a critical motif of just five amino acids. Changes outside the motif were tolerated. These observations are consistent with the high degree of degeneracy exhibited by CD4⁺ T cells and suggest that there is a high degree of CD4⁺ T cell cross-reactivity between flaviviruses, a phenomenon that may be beneficial or deleterious and could be of great significance in those regions where two or more flaviviruses co-circulate.

Results and discussion

Generation and characterisation of cross-reactive dengue-specific cytolytic CD4⁺ T cell clones

CD4⁺ T cell clones recognising a DRB1*15-restricted epitope within the non-structural protein 3 (NS3) of dengue virus were generated by limiting dilution from PBMC taken from an HLA DRB1*15⁺ patient 2 weeks after an episode of severe (dengue haemorrhagic fever grade III) secondary dengue 4 infection. This epitope differed slightly between the four dengue virus serotypes (Fig. 1A) and all were recognised by the clones. Restriction was confirmed in a cultured ELISPOT assay using B-cell lines that had only DRB1*15 in common: the use of such lines allowed peptide recognition, lines lacking DRB1*15 did not (data not shown). All clones showed some degree of cytolytic activity in standard chromium-release assays and the clone showing the highest level, 4E3, was selected for further study. It demonstrated broad cross-reactivity across all four variants of the epitope and cytolytic activity was maintained at E:T ratios as low as 1.25:1 (Fig. 1B and C). Chromium-release assays carried out in the presence and absence of inhibitors of different killing mechanisms demonstrated that killing was predominantly mediated by perforin (Fig. 1D). The clone conformed to a Th1 cytokine production phenotype, producing high levels of TNF-α and IFN-γ in response to peptide stimulation (data not shown). Intracellular cytokine staining for IFN-γ combined with surface staining for CD107 (a marker of lymphocyte degranulation and considered an indicator of cytolytic activity [22]) demonstrated that the cells producing IFN-γ in response to stimulation were also degranulating (Fig. 1E). Inhibition of perforin-mediated killing by concanamycin did not affect the increased level of CD107 expression produced by antigen stimulation (data not shown). This is not surprising — concanamycin acts by increasing perforin degradation within granules rather than preventing degranulation itself.

Figure 1.

(A) Variations in the sequence of the NS3 epitope from the dengue 1 videx sequence. Dashes indicate idential residues. (B and C) Clone 4E3 shows broad cross-reactivity across all four variants of the NS3 epitope. Chromium-release assays for clone 4E3 using DRB1*15 B cells pulsed with each of the four variants of the dengue NS3 epitope for 1 h at 37°C. Lysis in the absence of peptide was less than 2%. (B) At different E:T ratios with target B cells being pulsed with peptide at 25 μM. (C) At different peptide concentrations (micM = μM) with an E:T ratio of 20:1. (D) Cytolytic activity is abolished by the addition of concanamycin A. Chromium-release assays were performed at an E:T ratio of 10:1 with targets being pulsed with 25 μM of the dengue 1 variant of the NS3 epitope. Perforin activity was inhibited by incubating the clones with concanamycin A (concentration 1 μM) for 1 h before the addition of B cells. Fas-dependent killing was blocked by incubating B cell with antibody ZB4 (5 μL in 200 μL) for 1 h before the addition of T cells. Results shown are representative of three independent experiments. (E) Clone 4E3 cells increased CD107a/b surface expression following antigen stimulation. Matched B cells (100 000; unpulsed or previously pulsed with 20 μM of dengue 1 peptide) were incubated with 100 000 cells of clone 4E3 in the presence of anti-CD107a and anti-CD107b FITC. After 1 h, monensin was added and after 5 h, the cells were washed, permeabilized and stained with IFN-γ APC and CD4 PerCP. Analysis was conducted after gating on lymphocytes using a Dako Cyan-ADP. Ninety percent of cells fell below threshold for IFN-γ and CD107 staining before stimulation. After stimulation, 93% of cells fell above.

We then tested the ability of clone 4E3 to recognise truncated peptides. Deletions from the C-terminal of the peptide representing DEN-1 produced modest falls in lysis, whereas deletions from the N-terminal produced significant falls (Fig. 2A). To our surprise, both 5R and 5L precipitated similar levels of lytic activity despite overlapping by just five amino acids (PIRYQ). Once titrated in a cultured ELISPOT assay the difference between the two was revealed: responses dropped off quickly at low levels of 5R stimulation, indicating perhaps the importance of C-terminal residues in the longer peptides (Fig. 2B). Recognition of peptide 5L was better than the preceding 4L, presumably because the methionine in 4L is a less favourable terminal amino acid than the proline in 5L. Once this proline was removed, recognition halved. No truncated peptide was as efficiently recognised as the 15mer.

Figure 2.

(A) The variation in lysis efficacy with truncated versions of the NS3 epitope. Clones were incubated for 4 h with chromium-labelled B cells pulsed with the truncated versions of the dengue 1 epitope at an E:T ratio of 10:1. Unstimulated cells produced 1.1% background lysis. (B) 5L shows a more sustained response in peptide titrations than 5R. Clone 4E3 was stimulated with B cells pulsed with a titration of peptide in a cultured ELISPOT. The response to dengue full-length epitope was sustained throughout. The response to 5L dropped off at low concentrations but not as strikingly as 5R. (C) Single alanine substitutions within the epitope’s PIRYQ motif abolishes recognition. Clone 4E3 was stimulated with B-cells pulsed with either 1micM or 5micM of synthetic peptide in a cultured ELISPOT at an E:T ratio of 1:50. The dengue 4 variant of the epitope was used as the index (D4WT). Alanine substitutions outside the critical PIRYQ motif produced modest falls in response, whereas substitutions within it eliminated response. (micM = μM).

It was possible that the five amino acids which peptide 5R and 5L had in common (PIRYQ) represented a critical motif necessary for peptide recognition. Using the D4 variant of the epitope as the index, we synthesised a series of peptides with a single alanine substitution in each position along the epitope (Fig. 2C). The peptides in which this substitution fell within the PIRYQ motif were not recognised by clone 4E3, whereas the peptides in which it fell outside this region were recognised, confirming that the preservation of this motif is critical for recognition.

These assays raise interesting questions. The best lytic response is seen with the 15mer, yet significant activity is preserved using certain peptides truncated from either end. It is interesting that peptides 5R and 5L, both of which demonstrate a relatively well-preserved ability to precipitate lysis have only five amino acids in common, PIRYQ. How could this be sufficient for TCR recognition when class II epitopes are “supposed” to be at least 13 amino acids long [23]? Although the 15mer is optimal, significantly shorter peptides can contain a critical, necessary portion of the epitope. This has been observed in clones recognizing HIV epitopes [24]. The PIRYQ portion may represent a critical motif necessary for binding. However, the regions flanking the “peptide core” are required for optimal recognition. This phenomenon is thought to be mediated primarily by the creation of a more stable interaction with the class II molecule rather than directly playing a part in binding to the TCR and may explain the step-wise fall in lytic activity with progressively truncated peptides [25].

The significance of the cytolytic effector function demonstrated by these clones is not clear. Cytotoxic CD4⁺ T cells have been described in vivo in association with viral infections such as CMV [26] and HIV [27] and ex vivo phenotypic analysis suggests they represent antigen-experienced cells (memory) at an advanced stage of cellular differentiation [28]. Unlike memory CD8⁺ Tcells, perforin expression is dependent upon cell activation in memory CD4⁺ Tcells [29]. As such, it is likely that cytotoxic CD4⁺ Tcells are of most importance in chronic disease (e.g. autoimmune) and persistent viral infection with immune activation a key driving force behind their differentiation. It is not known whether such cytotoxic activity is of significance in an acute viral infection.

Clone 4E3 demonstrates broad flaviviral cross-reactivity

Clone 4E3 was tested in chromium-release assays against targets pulsed with variant peptides representing the corresponding region of the NS3 epitope within other flaviviruses (Fig. 1A). Despite considerable sequence variations it was able to lyse effectively B cells pulsed with most variants (Fig. 3A). Activity against B cells pulsed with the West Nile virus (WNV) variant (the most similar) was as good as that against dengue. Activity was reduced against those with the greatest variation from the dengue sequence. At high peptide concentrations the lytic activity against Japanese encephalitis (JE) and West Nile appeared similar. Titrations of peptides representing the heterologous viruses confirmed that WNV continued to precipitate an effective cytolytic response at low concentration whereas JE did not (Fig. 3B). This is consistent with their respective divergence from the dengue sequence, WNV differing by three amino acids, JE by five.

Figure 3.

Clone 4E3 shows broad cross-reactivity across flavivirus variants of the dengue NS3 epitope (TBE = tick-borne encephalitis; YF = yellow fever; D1 = dengue 1 epitope; no peptide = B cells with no peptide pulse). (A) Clones were incubated for 4 h with peptide-pulsed chromium-labelled B cells at an E:T ratio of 10:1 in a standard CTL lysis assay. (B) Peptide titration in a cultured ELISPOT assay demonstrates the rapid fall in response with JE variant stimulation despite similar levels of activity at high concentrations (micM = μM). There was no response in the absence of peptide.

The high degree of flaviviral cross-reactivity exhibited by this clone may be explained by the preservation of the PIRYQ common core. The West Nile virus variant of the dengue epitope was recognized at least as well as dengue itself despite three amino acid differences and the PIRYQ motif remains intact. The Japanese encephalitis variant was much less well recognized, perhaps as a consequence of a substitution within this motif. The other variants tested had still more differences within this region and were poorly recognized. The extent of cross-reactivity may be of significance. There is evidence that CD4⁺ T cell memory responses mounted in response to secondary dengue infections may play a part in the pathogenesis of severe disease [15, 16, 18]. The fact that dengue T cell clones are cross-reactive with other flaviviruses raises once again the question as to whether prior infection with another flavivirus might produce an anamnesiatic immune response to a primary dengue infection. Such responses might be protective or contribute to pathogenesis, particularly if the effector responses described above are present in vivo.

Clones recognising elements within the DEN3 variant of the epitope described here have been previously described [30]. Zeng et al.[30] reported the generation of a CD4⁺DEN3-specific clone from PBMC taken from the recipient of an experimental DEN-3 vaccine, which recognised the minimal epitope IRYQTTATK (NS3_241-249). Unlike the clones described above it failed to lyse cells pulsed with DEN1, 2, 4, yellow fever or West Nile virus variants. Such extensive cross-reactivity to this epitope has not been previously described. The DEN-1 version of the minimal epitope described by Zeng was poorly recognised by clones 4E3 - lysis dropped to around a third of maximal and it is worth noting that this minimal epitope stops short of containing the entire PIRYQ sequence. They defined their minimal epitope as IRYQTTATK on the basis that lysis dropped to5% with truncated peptide RYATTATK. Clone 4E3 produced around 23% lysis of B cells loaded with PIRYQTTAVK, dropping to 11% with the loss of the proline (IRYQTTAVK). It is therefore possible that these two minimal peptides represent two distinct epitopes, which although presented by the same HLAmolecule are recognized by the T cell in distinct manners.

Concluding remarks

We have described how the presence of a critical five-amino acid-core region within an epitope preserves the ability of CD4⁺ T cell clones derived from patients with dengue to recognise not only epitopes within other dengue viruses but within other flaviviruses. This is despite significant sequence variation outside the critical region. Substitutions within this region, however, are associated with a loss of recognition.

The TCR of a CD4⁺ T cell recognises peptide antigen bound to the groove set in the outer face of a MHC II molecule [31]. Structural studies of the interaction of the TCR with peptide-MHC class II demonstrate that although class II epitopes are longer than class I epitopes [32] the TCR interaction remains restricted to a nine-amino acid portion of the whole [33]. The remainder of a MHC class II bound peptide (the flanking region) is situated outside of this core. The importance of the flanking region appears to lie in its ability to stabilize the interaction of the peptide its MHC molecules [25]. CD4⁺ T cell clones are therefore capable of sustained recognition of given epitope despite considerable variation, particularly if this variation is outside the critical core region involved in TCR binding. This capacity may be of pathological significance in viral infections.

It could be hypothesized that the promiscuous nature of CD4⁺ T cells facilitates the rapid expansion of cross-reactive populations from memory early in secondary dengue infection, or perhaps even primary dengue infection in an individual who had experienced infection with a flavivirus demonstrating sufficient homology. In most cases, these populations would be beneficial, providing help to CD8⁺ T cells and other immune system components in addition to possible cytotoxic effector activity. However, in some settings (for example, a flanking region mutation that increases epitope affinity for MHC) overproduction of vasoactive cytokines by these populations might contribute to the fluid leak syndrome that is the key feature of severe dengue. This has implications for the development of a safe, effective dengue-virus vaccine. The association of severe disease with secondary infection has led to legitimate fears of immunization-mediated disease enhancement [34]. The work described here also raises the possibility that previous infection with other flaviviruses may be of importance - a question that could only be answered by extensive epidemiological surveys. It will be important to assess the cytokine production, dengue and flaviviral cross-reactivity of CD4⁺ T cells elicited by putative dengue-vaccine candidates in order to elicit protective immunity against all virus serotypes whilst avoiding iatrogenic immunopathology.

Materials and methods

Patient samples and HLA typing

The blood sample from which the clones described here were generated was obtained from a cohort of patients attending the Hospital for Tropical Diseases in Ho Chi Minh City, Viet Nam. The infecting virus was identified using RT-PCR as previously described [35] and the serological status of the patient established by measuring IgM and IgG on paired samples [36]. The patient’s HLA type was determined using amplification refractory mutation system PCR (ARMS-PCR) with sequence specific primers, as previously described [37], and found to be - Class I: A0203, A24, B15, B54, BW6, CW7 and Class II: DRB1*15(01/06/09/12/13/16), DRB1*04, DRB3, DRB4, DQB1*05, DQB1*04. The study protocol was approved by the Scientific and Ethics committee at The Hospital for Tropical Diseases and the Oxford Tropical Research Ethics committee.

Production of peptide epitopes

Peptide epitopes were synthesised by standard, solid-phase 9-fluorenylmethoxy carbonyl chemistry, based on the consensus sequence for the dengue NS3 of each dengue serotype (1–4). Purity was >70%.

IFN-γ ELISPOT assays with T cell lines and clones

Clones and lines were tested for specificity in a 24-h cultured IFN-γ ELISPOT assay as previously described [38]. HLA-matched B cell lines pulsed with peptide at the desired concentration for 1 hour at 37°C were used as the targets at an E:T ratio of 1:10. Unpulsed B cells were used as a negative control.

Establishment of CD4⁺ T cell lines and clones

PBMC were stimulated with a pool of all four variants of the specific NS3 dengue epitope peptide—each peptide being present at 10 μM concentration. IL-2 was added on day 3, and the specificity of the line tested using lysis assay (described below) or cultured IFN-γ ELISPOT on days 10 and 20. Clones were established from CD4⁺ T cell lines by limiting dilution [39] (0.3 cells/well) after selection of antigen-specific cells with a commercial magnetic bead-based IFN-γ capture kit according to the manufacturer’s protocol (Miltenyi Biotec, Germany).

CD4⁺ T cell lysis assays

CD4⁺ T cell cytolytic activity was assessed using standard ⁵¹Chromium-release assays [39]. Briefly, HLA class II matched B cell lines were labelled with ⁵¹Chromium for 1 h and then pulsed with the appropriate peptide for 1 h after washing. Cells were cocultured with CD4⁺ Tcell clones at appropriate effector to target (E:T) ratios in 96-well plates at 37°C for 4 h. The supernatant (30 μL) was harvested from each well and transferred into 150 μL of Optiphase Supermix in a 96-well reading plate (Wallac, Finland). Radioactivity was counted using a Beta-plate counter. Specific lysis was calculated from the formula:

% lysis = (experimental counts - media control)/(detergent control- media control) × 100%.

Cytokine beads assay

In parallel with the lysis assays, we performed a duplicate set of experiments using the scheme described above but without ⁵¹Cr labelling of the B cell lines. Supernatant from the CD4⁺ Tcell clone and target cell co-culture was harvested after overnight incubation, and cytokines were measured by Luminex cytokine bead array analysis according to the manufacturer’s instructions (Bio-Rad Laboratories, USA).

Abbreviations

JE

Japanese encephalitis

NS3

non-structural protein 3

WNV

West Nile virus