An HLA-A2-Restricted T-Cell Epitope Mapped to the BNLF2a Immune Evasion Protein of Epstein-Barr Virus That Inhibits TAP

Abstract

The early lytic cycle protein of Epstein-Barr virus (EBV), BNLF2a, has recently been shown to play a critical role in immune evasion by inhibiting the peptide transporter associated with antigen processing (TAP), thereby blocking antigen-specific CD8⁺ T-cell recognition of many lytic cycle antigens. Surprisingly, we now show that a peptide (⁵⁰VLFGLLCLL⁵⁸) from the hydrophobic C-terminal region of this small (60-amino-acid) EBV protein is efficiently presented by the common class I allele HLA-A2 for recognition by cytotoxic T lymphocytes. The mechanism for this unexpected finding was revealed by experiments showing that this epitope is processed and presented independently of TAP.

CD8⁺ T lymphocytes control viral infection by recognizing short peptides derived from the intracellular breakdown of viral proteins and presented on the infected-cell surface by major histocompatibility complex class I molecules (15). This T-cell subset plays a critical role in controlling infection with Epstein-Barr virus (EBV), a gamma-1 herpesvirus which preferentially infects B lymphocytes (11). Different sets of proteins expressed during the lytic cycle of EBV replication induce quantitatively different cellular immune responses, with CD8⁺ T-cell responses consistently skewed toward epitopes in immediate early proteins and a subset of early proteins, with only occasional responses to novel epitopes in late proteins (16). Consistent with this observation, as EBV-infected cells move through the lytic cycle, their susceptibility to EBV-specific CD8⁺ T-cell recognition falls dramatically, concomitant with a reduction in transporter associated with antigen processing (TAP) function and surface human histocompatibility leukocyte antigen (HLA) class I expression (16, 17). This hierarchy of immunodominance is observed across EBV-infected individuals expressing different HLA alleles.

Importantly, a major factor governing the immunogenicity of EBV proteins has recently been revealed. The small EBV lytic cycle early protein BNLF2a was shown previously to block HLA class I antigen presentation by inactivating the TAP1/TAP2 peptide transporter (10). The BNLF2a protein mediates its effects by interacting with the TAP complex and inhibiting both its peptide- and ATP-binding functions. BNLF2a transcripts initially appear in the early phase of the lytic cycle, peaking 8 to 12 h after lytic cycle induction in B-cell lines (23). The majority of EBV lytic cycle transcripts are expressed in the late phase, after BNLF2a expression, at a time when the protein's effects are likely to be well-established. This pattern explains why the EBV late proteins are poorly processed and presented by lytically infected cells.

Defining a CD8⁺ T-cell epitope within the TAP inhibitor EBV protein BNLF2a.

Following reports that BNLF2a plays a critical role in controlling the EBV-specific T-cell response, we set out to investigate the possibility that this protein includes sequences capable of stimulating T cells. Nine peptides of 20 amino acids each, overlapping by 15 residues and covering the full BNLF2a protein sequence, were pooled (with each peptide at 10 μg/ml) and used to stimulate peripheral blood mononuclear cells (PBMCs) from five EBV-seropositive healthy individuals. After in vitro expansion for 2 weeks with recombinant interleukin-2, the resultant T-cell lines were screened for reactivity to each of the nine peptides by using gamma interferon enzyme-linked immunospot (ELISPOT) assays (3). Only one donor (donor B7; HLA-A2 HLA-A3 HLA-B7 HLA-B37 HLA-Cw6 HLA-Cw7) showed T-cell reactivity toward a BNLF2a peptide, with a clear response toward a peptide corresponding to BNLF2a residues 41 to 60 (the 41-60 peptide) observed (Fig. 1A).

FIG. 1.

Defining a CD8⁺ T-cell epitope within the TAP inhibitor EBV protein BNLF2a. (A) A T-cell line generated from the EBV-seropositive (sero⁺) healthy donor B7 by in vitro stimulation of PBMCs with a pool of nine overlapping peptides from BNLF2a was screened for reactivity to each of the nine peptides by a gamma interferon ELISPOT assay. a5 to h8 indicate ELISPOT plate well coordinates. (B and C) This T-cell line (B) and PBMCs (C) from donor B7 were also tested by flow cytometry for intracellular gamma interferon (IFN-gamma) expression, following incubation with the BNLF2a 41-60 peptide. Numbers in the boxes represent the percentages of gamma interferon-producing CD4⁺ or CD8⁺ lymphocytes. A phycoerythrin-conjugated anti-gamma interferon antibody, a peridinin chlorophyll-conjugated anti-CD8 antibody, and a fluorescein-conjugated anti-CD4 antibody were used.

To determine if the responding T cells expressed CD8 or CD4 and to measure the strength of this response, the T-cell culture and PBMCs from healthy donor B7 were tested by flow cytometry for intracellular gamma interferon expression (21) following incubation with the BNLF2a 41-60 peptide. As shown in Fig. 1B, the T cells that responded to this BNLF2a peptide were drawn from the CD8⁺ subset, indicating that this epitope is presented in association with HLA class I. Although these BNLF2a-specific T cells could be detected ex vivo using peptide stimulation in ELISPOT assays with PBMCs from donor B7 (data not shown), the response was relatively weak and difficult to observe using flow cytometry (Fig. 1C).

Characterizing the HLA restriction and minimal length of the BNLF2a epitope.

To determine the HLA class I allele involved in presenting the antigenic determinant within this region of BNLF2a, the T-cell line from donor B7 was tested by flow cytometry for intracellular gamma interferon expression (21) following exposure to a variety of normal lymphoblastoid cell lines (LCLs; 1 to 9) that had been pretreated with the BNLF2a 41-60 peptide (10 μg/ml) and washed extensively (Fig. 2A). Significant T-cell stimulation was observed only with LCLs that shared HLA-A2 with donor B7, indicating that this HLA molecule presents the epitope. To confirm this conclusion, the mutant HLA class I-negative 721.221 cell line (20) that had been transfected to express only HLA-A2 (7) and pretreated with the BNLF2a 41-60 peptide was also shown to stimulate the T cells. In contrast, untransfected 721.221 cells that had also been pretreated with the peptide failed to stimulate the T cells (Fig. 2A).

FIG. 2.

Characterizing the HLA restriction and minimal length of the BNLF2a epitope. (A) A T-cell line from donor B7, generated by in vitro stimulation with the BNLF2a 41-60 peptide, was tested by flow cytometry for intracellular gamma interferon (IFN-gamma) expression following exposure to a variety of normal LCLs that had been pretreated with the BNLF2a 41-60 peptide and washed extensively. These LCLs shared up to three HLA alleles with donor B7, as indicated in the figure. The mutant HLA class I-negative 721.221 cell lines that had been transfected to express only HLA-A2 or left untransfected were also pretreated with the BNLF2a 41-60 peptide and used to stimulate the T cells. Data are expressed as the percentage of CD8⁺ cells producing gamma interferon in response to stimulation with each LCL. −, none. (B) Truncated versions of the 41-60 peptide were tested in duplicate over a wide range of concentrations by a chromium release assay for recognition by the T-cell line from donor B7. The peptide-treated target cells were a phytohemagglutinin (PHA)-stimulated T-cell line derived from an HLA-A2⁺ individual. The effector/target cell ratio was 5:1.

To map this T-cell epitope more precisely within the BNLF2a sequence, truncated versions of the 41-60 peptide were tested over a wide range of concentrations for recognition by the T-cell line from donor B7. This chromium release assay (7) utilized a phytohemagglutinin-stimulated T-cell line (4), derived from an HLA-A2⁺ individual, as the peptide-treated target cells. As shown in Fig. 2B, the nonamer sequence ⁵⁰VLFGLLCLL⁵⁸ was recognized at much lower concentrations than the longer peptides, suggesting that this sequence is the naturally processed T-cell epitope. This conclusion is supported by the observation that the ⁴¹LRLLLVVLCVLFGLLCL⁵⁷ peptide, which excludes the Leu⁵⁸ residue, failed to be recognized even at the highest concentrations (Fig. 2B). Furthermore, the ⁵⁰VLFGLLCLL⁵⁸ nonamer conforms to the well-established HLA-A2 peptide-binding motif (5).

Endogenous processing of the BNLF2a T-cell epitope is independent of TAP.

To determine if this BNLF2a epitope is endogenously processed and presented by EBV-infected cells, an enzyme-linked immunosorbent assay was performed to detect gamma interferon production by the B7 T-cell line after exposure to various stimulator LCLs. This assay relies on the T cells' secreting gamma interferon upon the recognition of cognate antigen expressed by the small number of lymphoblastoid cells spontaneously reactivating lytic replication (16). The LCLs were from an HLA-A2⁺ individual and were transformed with either wild-type EBV or a recombinant strain of EBV in which the BNLF2a gene was deleted. As shown in Fig. 3A, stimulation with the normal HLA-A2⁺ LCL resulted in significant release of gamma interferon from the T cells while the corresponding LCL infected with the BNLF2a knockout EBV strain failed to significantly stimulate the T cells unless they were pretreated with the synthetic peptide.

FIG. 3.

Endogenous processing of the BNLF2a T-cell epitope is independent of TAP. (A) Endogenous processing and presentation of the ⁵⁰VLFGLLCLL⁵⁸ epitope in EBV-infected cells. Five thousand cells of the ⁵⁰VLFGLLCLL⁵⁸-specific T-cell line B7 were incubated for 16 h in the presence of 50,000 HLA-A2⁺ lymphoblastoid cells that had been transformed with either wild-type B95.8 EBV or a recombinant EBV in which the BNLF2a gene had been deleted by homologous recombination (BNLF2a-KO EBV) or in the presence of VLFGLLCLL synthetic peptide (Pep BNLF2a 50-58)-sensitized BNLF2a knockout EBV-infected LCLs as a control. Recognition of the BNLF2a epitope by the T-cell line was assessed by measuring gamma interferon (IFN-gamma) secretion. (B and C) Endogenous processing and presentation of the ⁵⁰VLFGLLCLL⁵⁸ epitope from recombinant vaccinia virus. The T-cell line from donor B7 was used as an effector in a cytotoxicity assay against a normal TAP⁺ LCL from an HLA-A2⁺ individual (B) or the TAP-negative cell line T2 (C). Target cells were infected with Vacc-BNLF2a or its parent TK⁻ control virus (Vacc-TK−) at a multiplicity of infection of 10:1 for 1 h at 37°C. After overnight infection, cells were washed, incubated with ⁵¹Cr for 90 min, and used as targets in a standard 5-h ⁵¹Cr release assay. Uninfected target cells exposed to the VLFGLLCLL peptide at a final concentration of 10 μg/ml or to an equivalent concentration of dimethyl sulfoxide solvent (untreated) during ⁵¹Cr labeling served as positive and negative controls. (D and E) Endogenous processing and presentation of the TAP-dependent ⁴⁰⁷HPVGEADYFEY⁴¹⁷ epitope from recombinant vaccinia virus. A T-cell clone specific for the EBNA1 epitope comprising residues 407 to 417 (Pep EBNA1 407-417) (21) was used as an effector in a cytotoxicity assay against a normal TAP⁺ LCL from an HLA-B35⁺ individual (D) or T2 cells that had been stably transfected with HLA-B35 (E). Target cells were infected with a recombinant vaccinia virus encoding the EBNA1 protein without the large glycine-alanine repeat domain (Vacc-EBNA1dGA) or its parent TK⁻ control virus, as described above. Uninfected target cells exposed to the HPVGEADYFEY peptide at a final concentration of 10 μg/ml or to an equivalent concentration of dimethyl sulfoxide solvent were also included.

To further examine endogenous processing of this epitope, a series of chromium release cytotoxicity assays were performed that made use of a recombinant vaccinia virus carrying the coding sequence for BNLF2a (Vacc-BNLF2a), generated as described previously (16). Only a small percentage of cells within LCLs express the lytic EBV antigens (2), and therefore, the BNLF2a-specific B7 T-cell line failed to lyse a normal LCL from an HLA-A2⁺ individual unless the LCL was first infected with Vacc-BNLF2a (Fig. 3B). Interestingly, the mutant LCL-T-lymphoblastoid hybrid cell line 174 × CEM.T2 (referred to hereinafter as T2), which carries the gene for HLA-A2 but fails to express TAP (19), was very well recognized by these T cells following either infection with Vacc-BNLF2a or pretreatment with the ⁵⁰VLFGLLCLL⁵⁸ BNLF2a peptide (Fig. 3C). These data indicate that this BNLF2a epitope is processed independently of TAP, thereby providing an explanation for how a CD8⁺ T-cell epitope is sourced from this EBV protein that inhibits TAP transport. As a TAP-dependent control, a T-cell clone specific for the ⁴⁰⁷HPVGEADYFEY⁴¹⁷ epitope from the EBNA1 EBV protein, which is presented in association with HLA-B35 (21), was used in similar assays (Fig. 3D and E). These T cells failed to recognize a T2 cell line stably transfected with HLA-B35 and infected with a vaccinia virus construct encoding the EBNA1 protein sequence without the glycine-alanine repeat domain (21).

Finally, we measured the strength of the T-cell response to this epitope in PBMCs from several HLA-A2⁺ individuals by using a multimer of the HLA-A2-⁵⁰VLFGLLCLL⁵⁸ complex (ProImmune, Oxford, United Kingdom). Significant multimer staining of cells from three of four healthy EBV-seropositive individuals was observed (Fig. 4A to D), with up to 1.1% of CD8⁺ cells from the strongest responder showing specificity for this epitope. As expected, PBMCs from an HLA-A2⁺ EBV-seronegative individual showed no significant staining with the multimer (Fig. 4E). PBMCs from two HLA-A2⁺ patients with acute infectious mononucleosis were also included (Fig. 4F and G), and the cell sample from one showed a significant population (0.6% of CD8⁺ cells) of T cells specific for the BNLF2a epitope. PBMCs from the healthy individuals HD1, HD3, and HD4 were also incubated with the BNLF2a peptide (5 μg/ml) and cultured in the presence of interleukin-2 for 7 days before the repetition of the multimer staining (Fig. 4H to J). Large increases in the numbers of BNLF2a-specific T cells in the cultures of samples from HD1 and HD3 were observed.

FIG. 4.

Screening for BNLF2a-specific T cells with an HLA-A2-VLFGLLCLL multimer. (A to G) PBMCs from four EBV-seropositive healthy donors (A to D), one EBV-seronegative (−ive) healthy donor (E), and two patients with acute infectious mononucleosis (F and G) were included in the screening. (H to J) PBMCs from the healthy individuals HD1, HD3, and HD4 were incubated with the BNLF2a peptide, cultured in the presence of interleukin-2 for 7 days, and then screened for reactivity. Cells were stained with an allophycocyanin-conjugated HLA-A2-VLFGLLCLL pentamer per the instructions of the manufacturer (ProImmune, Oxford, United Kingdom). Cells were then washed and incubated with saturating amounts of anti-human CD8 antibodies before being washed and analyzed on an LSRII flow cytometer. Analysis was conducted using FlowJo software (TreeStar). Numbers in the top right corners represent the percentages of CD8⁺ lymphocytes that stained with the pentamer. Designations above the graphs identify the donors.

Earlier reports have described TAP-independent processing of epitopes from the EBV latent membrane protein 2 (13). Interestingly, TAP independence was shown to correlate with the hydrophobicity of these epitope sequences, although the pathway of access into the endoplasmic reticulum remains to be determined. It is notable that the ⁵⁰VLFGLLCLL⁵⁸ epitope is also very hydrophobic, being drawn from the hydrophobic C terminus of the BNLF2a protein.

TAP-specific inhibitors such as the BNLF2a protein of EBV have also been identified previously in the betaherpesvirus human cytomegalovirus (1, 8, 14) and in simplex viruses (6, 9) and varicelloviruses (12), both members of the alphaherpesvirus subfamily. These proteins use a variety of mechanisms to inhibit TAP function, and their existence clearly testifies to the strength of the evolutionary pressure exerted by CD8⁺ T cells on many herpesviruses. This report, however, highlights the limitations of this immune evasion strategy and explains why most herpesviruses target multiple points of the antigen-processing pathway for inhibition (18, 22). Although only a few lytic EBV proteins appear to be targeted by CD8⁺ T cells, the virus has certainly been unsuccessful in completely evading this arm of the immune system during lytic replication. Indeed, the immediate early antigens, which are expressed before BNLF2a, stimulate huge numbers of CD8⁺ T cells, perhaps highlighting the limited role that BNLF2a plays in immune evasion and the outcome of infection. It will now be important to determine if the few epitopes that have been mapped to the late lytic EBV antigens are also processed independently of TAP.