Although T lymphocytes are considered essential for the control of EBV infection, it remains uncertain how this control occurs. We previously reported unexpected killing of EBV-transformed B-lymphoblastoid cells (LCLs) that did not express BHRF1 by CD4+ T cells specific for BHRF1, an EBV lytic cycle protein. Using LCLs transformed with an EBV mutant, in which the BHRF1 gene was deleted, we showed that killing of latently infected cells through the recognition of a protein produced during the lytic cycle is due to transfer of BHRF1 from lytically infected to latently infected cells, which occurs in culture. Accordingly, LCLs efficiently presented exogenous BHRF1 protein. Furthermore, we present evidence for persistence of captured BHRF1 Ag for several days. Due to this long-term persistence, repeated loading of suboptimal amounts of BHRF1 led to accumulation of BHRF1 Ags in LCLs and, ultimately, to their optimal recognition by BHRF1-specific CD4+ T cells. These results unveil an MHC class II-dependent pathway that could be important for the control of EBV latent infection through recognition of lytic cycle Ags.
Epstein-Barr virus is a human γ-herpesvirus that latently infects >95% of the population (1). The virus infects its host asymptomatically at a very young age and thereafter resides in the B cell population in a latent form (2). Although chronic EBV infection is free of complications in most individuals, EBV is, however, associated with lymphoid and epithelial cell tumors due to its cell transformation capacity (3). EBV is also an opportunistic agent often associated with malignancies occurring in immunosuppressed patients. This is the case for about half of B cell lymphomas developing in HIV-infected patients with a depleted CD4+ T cell pool (4). Besides, post-transplant lymphoproliferative diseases that occur sometimes upon immunosuppression are nearly always EBV associated (3).
After primary infection, EBV is carried for life as a latent infection of the circulating memory B cell pool, with low level reactivation from latency into virus productive (lytic) infection at oropharyngeal sites. This virus encodes eight latent cycle proteins, the nuclear Ags EBV-encoded nuclear Ag 1 (EBNA1),3 -2, -3A, -3B, -3C, and -LP and the latent membrane proteins 1 and 2, all of which being expressed in EBV-transformed B-lymphoblastoid cell lines (LCL). In contrast to the latent infection, during which only a limited number of proteins are expressed, activation of the EBV lytic cycle leads to the expression of as many as 80 virus-specific RNA species (1). Based on their time of appearance after induction, these transcripts are named immediate-early, early, or late. An intact immune system is capable of containing EBV infection and preventing transformation of infected B cells. In fact, the high frequency of EBV-associated lymphoproliferative diseases or lymphomas in immunocompromised individuals strongly supports a role for anti-EBV T cells in containing EBV infection (5). The CD8+ T cell response, which has been extensively studied, has been shown to be preferentially directed toward the early lytic proteins BZLF1 and BMLF1 (6, 7, 8, 9, 10, 11) and, to a lesser extent, toward the latent nuclear Ags EBNA3A, -3B, and -3C (see Ref.12 for a review).
EBV-specific CD4+ T cell responses have been explored only recently. These studies are of particular interest considering that EBV infects mainly B cells, in which the HLA class II pathway of Ag presentation is active. Upon recognition of EBV-infected B cells, EBV-specific CD4+ T cells might play a dual role not only in sustaining proliferation and functional maturation of CD8+ T cells, but also in directly contributing to elimination of EBV-infected B cells. The EBV-specific CD4+ T cell responses described to date are mainly directed against the EBNA proteins expressed during latency (13, 14, 15, 16, 17, 18, 19, 20).
Very few CD4+ T cell responses to EBV proteins produced during the lytic cycle have been characterized (21, 22, 23, 24, 25). We recently described a CD4+ T cell response directed against an epitope from the early EBV lytic protein BHRF1, presented in the HLA-DR*0401 context (25). We generated a large number of BHRF1-specific CD4+ T cell clones from several DR*0401+ individuals and showed that in all instances these T cell clones displayed strong killing of DR*0401-matched LCLs. In this study we explored the mechanism that could explain why the percentage of killed LCL cells far exceeded the small percentage of B cells entering the lytic cycle.
Materials and Methods
CD4+ T cell lines and T cell clones
CD4+ T cell lines and CD4+ T cell clones were derived from PBL from healthy virus carriers or from synovial fluid from patients suffering from rheumatoid arthritis, as previously described (25). BHRF1-specific DR*0401-restricted CD4+ T cells were sorted out from DR*0401+ T cell lines using magnetic beads coated with recombinant BHRF1/DR*0401 complexes. A large number of BHRF1-specific T cell clones were derived from these sorted cell lines (25) and were maintained in RPMI 1640 supplemented with 10% pooled human serum, 1 mM l-glutamine, and 150 IU/ml rIL-2. HLA class II typing was performed using standard molecular typing techniques at the Etablissement Français du Sang.
EBV-transformed B cell lines
DR*0401+ EBV-transformed LCLs were obtained by transforming PBL with B95.8 (wild-type (WT)-LCL) virus. Moreover, target LCLs were generated from DR*0401+ donors by in vitro infection of peripheral blood B cells with recombinant strains of EBV. An LCL transformed by a B95.8 virus mutant that had been rendered incapable of lytic cycle entry by disruption of the BZLF1 gene (BZLF1-knockout (BZLF1-KO) LCL) (26) and an LCL transformed with an EBV mutant in which the BHRF1 gene was deleted (BHRF1-KO LCL) (25, 27) were generated.
Dendritic cell (DC) preparation
PBMCs were isolated from a healthy DR*0401+ donor by Ficoll density gradient centrifugation. To minimize contamination of PBMC with platelets, the preparation was first centrifuged at 1000 rpm at room temperature. After removal of the top 20–25 ml, which contained most of the platelets, the tubes were centrifuged at 1500 rpm at room temperature. To generate DC, PBMC were depleted from T lymphocytes by adherence for 2 h at 37°C in Falcon culture flasks (BD Biosciences) at a density of 5 × 106 cells/ml in RPMI 1640 medium supplemented with 5% FCS. Nonadherent cells were discarded by five washes in PBS, and adherent cells were cultured in the presence of IL-4 (100 ng/ml) and GM-CSF (25 ng/ml) in complete RPMI 1640 medium. Cultures were fed on days 2 and 4 with IL-4 (100 ng/ml) and GM-CSF (25 ng/ml), and floating immature DC (iDC) were harvested on day 7. DC maturation was induced by addition of 10 ng/ml TNF-α and 1 mg/ml LPS on day 4, and mature DC (mDC) were used on day 7.
Peptides and recombinant BHRF1 protein
The 12-aa peptide, BHRF1122–133 PYYVVDLSVRGM, (hereafter, this epitope is identified by the underlined three letter code), was obtained from Genosys. Peptide stock solutions (20 mg/ml in DMSO) were diluted first to 2 mg/ml in acetic acid (1%) and then to the final concentration in RPMI 1640 culture medium.
Recombinant BHRF1 fusion protein, in which the C-terminal region was removed (residues 1–160), was produced in a bacterial expression system as previously described (28). The coding sequence of BHRF1 was amplified by PCR with primers encoding 5′ and 3′ restriction sites from a full-length BHRF1 cDNA (9) and cloned into pLM1 vector in-frame with a C-terminal His6 tag (DDDDLEHHHHHH). Constructs were verified by DNA sequencing. Expression of pLM1-BHRF1 in Escherichia coli strain BL21DE3star (Stratagene) was induced by addition of 1 mM isopropyl-β-d-thiogalactoside (Sigma-Aldrich). After 4 h, recombinant BHRF1 was purified from bacterial lysate by nickel-chelate affinity chromatography using HiTrap columns (Amersham Biosciences), and then dialyzed in 10 mM Tris and 50 mM NaCl.
CD4+ T cell clones were tested for killing of target LCLs at defined E:T cell ratios in a 4-h chromium release assay as described previously (25). The targets used were DR*0401+ WT-LCLs or DR*0401+ KO-LCLs, either unloaded or loaded for 30 min with the BHRF1122–133 peptide at 10 μM. Assays were performed in triplicate. The percent specific lysis was (experimental release − spontaneous release)/(maximal release − spontaneous release) × 100%. In some cytotoxicity assays, target cells were incubated at 37°C with various concentrations of BHRF1 protein in 1 ml of OptiMEM-1 serum-free medium (Invitrogen Life Technologies) for the specified time period or with the synthetic epitope peptide as a positive control. Preloaded target cells were then washed extensively and labeled with chromium for 1 h before being used in a standard chromium release assay.
To examine the Ca2+ dependency of BHRF1-specific, CTL-mediated cytotoxicity, EGTA (Sigma-Aldrich) was added at a final concentration ranging between 0.1 and 2 mM with 2 mM MgCl2. To evaluate the role of perforin in CD4+ CTL-mediated cytotoxicity, effector CTLs were pretreated for 2 h at 37°C with either concanamycin A (CMA; Sigma-Aldrich), an inhibitor of the perforin-based cytotoxic pathway, or brefeldin A (BFA; Sigma-Aldrich) at various concentrations before being incubated with the target cells. Treatment of the cells with EGTA, CMA, and BFA showed no toxic effect, as determined by the trypan blue exclusion test. To determine whether specific lysis was mediated by the Fas/Fas ligand (FasL) pathway, CTLs were preincubated for 1 h at 37°C with anti-FasL (BD Pharmingen) or anti-TRAIL mAb (BD Pharmingen) at concentrations ranging from 2.5 to 20 μg/ml before the priming cytotoxicity assay in the presence of these Abs.
Surface Ag staining was performed using mouse mAbs against CD4 (BD Biosciences) or HLA-B27 (mAb ME1 (29); provided by Dr. A. Toubert, Hôpital St. Louis, Paris, France). To detect intracytoplasmic perforin expression, CD4+ T cells were fixed in 4% paraformaldehyde and permeabilized using a 0.1% saponin solution in PBS for 30 min. The permeabilized cells were stained with the following labeled Abs (BD Pharmingen): perforin-PE or granzyme A-FITC. For granulysin staining, the cells were first incubated with monoclonal DH4 (30), a gift from Dr. A. Krensky (Stanford University, Stanford, CA) at 1 μg/ml for 30 min, then stained with FITC-labeled anti-mouse IgG. Control Abs were the respective isotype Abs conjugated with relevant fluoresceins (BD Biosciences). Cells were analyzed with a FACScan flow cytometer (BD Biosciences).
Transient transfections into COS-7 cells was performed by the DEAE-dextran chloroquine method using cDNAs encoding the EBV lytic protein BHRF1 or BRLF1, as described previously (9). Lysates were prepared from transfected COS cells with four cycles of freeze-thawing before being used for pulsing onto LCLs.
Sensitization of LCLs or DC with recombinant BHRF1 protein
BHRF1-KO LCLs or DC were incubated at 37°C with various concentrations of recombinant BHRF1 protein in serum-free medium for the specified time period or with PYY peptide at 10 μM, as a positive control. Preloaded target cells were then washed extensively and labeled with chromium for 1 h before being used in a chromium release assay with BHRF1-specific effectors. The effects of various inhibitors, including MHC class II Ag-processing inhibitors and one protein synthesis inhibitor, were examined to define the processing pathway used by the DR*0401-restricted BHRF1 epitope. BHRF1-KO LCLs were pretreated for 4 h with either chloroquine (Sigma-Aldrich) or BFA (Sigma-Aldrich) before sensitization for 1 h with BHRF1 protein. Target cells were incubated in growth medium supplemented with either chloroquine (0.1–200 μM) or BFA (0.1–100 μg/ml) for 4 h at 37°C before sensitization. During the Ag pulse, chloroquine was maintained at a concentration reduced to 25% of the initial concentration, whereas BFA was maintained at the initial concentration. For inhibition of protein synthesis, BHRF1-KO LCLs were incubated for 9 h in culture medium containing 1–100 μM emetine (Sigma-Aldrich) before sensitization with BHRF1 protein. BHRF1-KO LCLs treated with the various inhibitors were then used in cytotoxicity assays. As a control, in all sensitization assays some of the target cells were loaded for 30 min with PYY peptide at 10 μM before adding BHRF1-specific CD4+ T cells.
In coculture experiments, donor and acceptor cell lines were seeded as a 1:1 ratio and maintained for up to 25 days without adding more donor cells. At different times, cells were harvested, stained with anti-HLA-B27 ME1 mAb, sorted in B27+ and B27− fractions, and then used as targets in cytotoxicity assays as described above.
Killing of latently infected B cells, not expressing the EBV lytic protein BHRF1, by BHRF1-specific CD4+ T cells
We previously generated BHRF1-specific, DR*0401-restricted, CD4+ T cell lines from several donors by immunomagnetic sorting of PBL using beads covered with DR*0401/BHRF1 peptide complexes and derived a large number of BHRF1-specific CD4+ T cell clones from these cell lines (25). The most salient feature of these CD4+ T cell clones was their high lytic activity against HLA-DR*0401-matched LCLs transformed with WT EBV virus (WT-LCLs). In the present study, we tried to determine why a majority of LCL cells were killed when a minority (<5%) of them were replicating the virus and were therefore expected to express BHRF1. BHRF1-specific clones did not kill DR*0401+ LCLs transformed by a mutant virus with a deletion in the BHRF1 early gene (BHRF1-KO LCLs) unless loaded with the BHRF1122–133 cognate PYY (Fig. 1), suggesting that cytotoxicity was linked to direct recognition of a BHRF1-derived epitope. Furthermore, BHRF1-specific clones did not kill DR*0401+ LCLs transformed by a mutant virus that had been rendered incapable of lytic entry by disruption of the BZLF1 gene (BZLF1-KO LCLs) (26) unless loaded with peptide PYY (Fig. 1), excluding the possibility that latently infected B cells express BHRF1.
Because lysis occurred within a relatively short period of time (4-h 51Cr release assays), we studied the possible implication of perforin in this killing process. We found that perforin, granzyme, and granulysin were expressed by BHRF1-specific CD4+ T cell clones and BHRF1-specific CD4+ T cell lines (Fig. 2,A). Cytotoxicity assays were conducted in the presence of EGTA, CMA, or BFA or in the presence of blocking mAb specific for FasL or TRAIL. EGTA has been shown to efficiently inhibit perforin/granzyme-mediated cytotoxicity by chelating extracellular free calcium, which is required for exocytosis of cytolytic granules and pore formation by perforin (31). CMA acidifies intracellular vacuolar granules and is thought to inhibit perforin/granzyme-mediated cytotoxicity by increasing degradation of the content of exocytic granules (32), whereas BFA selectively inhibits FasL and other TNF-α family molecule-mediated cytotoxicity by inhibiting surface up-regulation of glycopeptide molecules (33). As shown in Fig. 2 B, the cytotoxicity of the BHRF1-specific CD4+ T cell clones against LCLs was almost completely inhibited by EGTA and CMA, but not by BFA or anti-FasL or anti-TRAIL mAb Taken together, these data indicate that granule exocytosis is the main pathway of cytotoxicity used by BHRF1-specific CD4+ effector T cells.
BHRF1 transfer from lytically to latently infected LCLs in in vitro culture
The consistent killing of DR*0401+ LCLs by BHRF1/DR*0401-specific CD4+ T cells highlights a paradox: if, as widely believed, BHRF1 protein is expressed in very few cells among LCLs (<5%), how could 40–80% of LCLs be killed by BHRF1-specific CD4+ T cells? The unexpectedly high sensitivity of LCLs to BHRF1-specific CD4+ T cells suggested that cross-presentation of BHRF1 could occur in vitro: BHRF1 protein could be released from a few cells undergoing the lytic cycle, captured by neighboring latent LCL cells, and finally processed as an exogenous Ag. To test this possibility we mixed WT LCLs expressing BHRF1, but lacking the DR*0401-restricting allele (donor LCLs), and equal numbers of BHRF1-KO LCLs lacking the cognate Ag, but expressing the appropriate HLA-DR*0401 allele (recipient LCLs). Recipient (HLA-B27−) and donor (HLA-B27+) LCLs could be distinguished by staining with the HLA-B27-specific mAb ME1. As expected, neither donor (DR*0401−) nor recipient (BHRF1−) LCLs were killed by BHRF1/DR*0401-specific CD4+ T cells (data not shown). At different coculture time points (8, 15, or 21 days), cells were stained with mAb ME1 and sorted on the basis of HLA-B27 expression (Fig. 3A). The data shown in Fig. 3 B are representative of the three experiments performed. Although HLA-B27− BHRF1-KO-LCLs were not killed after 8 or 15 days of coculture, they were efficiently killed after a 21-day coculture. In contrast, B27+, DR*0401− cells were not killed at any time.
Sensitization of LCLs or DC with BHRF1 protein or PYY peptide
The above data suggested that transfer of BHRF1 protein could occur in culture from lytically infected to latently infected LCLs. The capacity of LCLs to present exogenously added BHRF1 protein was more specifically assessed using BHRF1-KO LCLs. In the first set of experiments, COS-7 cells were transiently transfected for 2 days with a plasmid encoding either BHRF1 or BRLF1, two proteins produced during the lytic cycle of EBV. Freeze-thaw cell lysates from COS cell transfectants were then pulsed for 24 h onto BHRF1-KO LCLs, and sensitized BHRF1-KO LCLs were used as targets in cytotoxic assays. Sensitization of BHRF1-KO LCLs with lysates from COS cells transfected with a cDNA encoding BHRF1 resulted in lysis of BHRF1-KO LCLs by BHRF1-specific CD4+ T cell clones, whereas lysates from BRLF1-transfected COS cells did not sensitize BHRF1-KO LCLs (Fig. 4).
In the second series of experiments, BHRF1-KO LCLs were sensitized with purified BHRF1 protein at 4 μM for 24 h at 37°C in serum-free medium, then washed extensively, and used as targets in a cytoxicity assay. In parallel, KO-LCLs were sensitized with the PYY peptide at 10 μM or with DMSO solvent as a control. Incubation of BHRF1-KO LCLs with BHRF1 protein clearly sensitized these cells to killing by BHRF1-specific CD4+ T cell effectors (Fig. 5,A). Incubation of BHRF1-KO LCLs with BHRF1 protein for 5 min was sufficient to sensitize LCLs (data not shown). To determine the minimum time required for uptake plus processing of BHRF1, BHRF1-KO LCLs were loaded with BHRF1 protein at 10 μM for 30 min, then washed and incubated for different times before being fixed with paraformaldehyde and used to stimulate TNF-α production by BHRF1-specific CD4+ T cells. This kinetic study indicated efficient target cell sensitization after a 90-min incubation with BHRF1 protein (data not shown). To exclude the possibility that the cytotoxicity of BHRF1-specific CD4+ T cells toward BHRF1-KO LCLs sensitized with BHRF1 protein was due to BHRF1 peptides present in the protein stock, BHRF1-KO LCLs were fixed for 30 min in 2% paraformaldehyde, then incubated for 2 h with either purified BHRF1 protein or PYY peptide at 10 μM. These target cells were then tested for their ability to trigger TNF-α release by BHRF1-specific CD4+ T cells. Paraformaldehyde-fixed BHRF1-KO LCLs fed with the BHRF1 protein were unable to stimulate BHRF1-specific CD4+ T cells, whereas they induced strong stimulation of CD4+ T cells when sensitized with the PYY peptide (Fig. 5 B).
Finally, iDC and mDC were compared with LCLs for their capacity to present exogenous BHRF1 to BHRF1-specific CD4+ T cell clones. DR*0401+ iDC, mDC, and BHRF1-KO LCLs were sensitized with various concentrations of BHRF1 protein or PYY peptide for 30 min, then used as targets for BHRF1-specific CD4+ T cells. As shown in Fig. 5,C, loading and presentation of BHRF1 protein by iDC was ∼10-fold more efficient than that by LCLs. By contrast, mDC were ∼100-fold less efficient than LCLs. These differences might be accounted by different levels of pinocytic activity of iDC, mDC, or LCLs or to qualitative differences in term of the expression of receptors specifically involved in BHRF1 uptake. The recombinant BHRF1 protein appeared more efficient at sensitizing DC than the corresponding concentration of PYY peptide (Fig. 5 C, right and left panels). This could be due to the fact that the BHRF1122–144 peptide used throughout this study was not the optimal one.
Exogenously loaded BHRF1 protein follows the classical MHC class II pathway of presentation
We next addressed whether HLA-DR*0401-restricted presentation of BHRF1 epitope was dependent on the conventional MHC class II pathway, which requires peptide binding to MHC class II molecules in a low pH vacuolar compartment. BHRF1-KO LCLs were incubated with the lysosomotropic agent, chloroquine. This drug has been shown to block the acidification of late endosomes and lysosomes, thereby inhibiting endosomal processing by resident cathepsins. BHRF1-KO LCLs were treated with various concentrations of chloroquine before loading with the BHRF1 protein and were then used as targets for BHRF1-specific CD4+ T cells in a cytoxicity assay. As shown in Fig. 6, the presence of chloroquine at 100 μM during processing of exogenously added BHRF1 protein significantly reduced the level of LCL lysis by BHRF1-specific clones, indicating that processing and presentation of exogenous BHRF1 are dependent on endosome/lysosome acidification.
To test whether newly synthesized proteins were required for processing and presentation of BHRF1 Ags, the BHRF1-KO LCLs were treated for 9 h with emetine, which blocks protein synthesis, including de novo synthesis of MHC class II molecules, before sensitization with BHRF1 protein. Emetine is known to abrogate presentation of epitopes using the classical pathway, with no effect on the presentation of epitopes using the alternative pathway (34). Presentation of BHRF1 was completely blocked by emetine treatment and thus was entirely dependent on protein synthesis. BFA at 1 μg/ml, an inhibitor of anterograde movement from the endoplasmic reticulum to the Golgi complex that blocks presentation of epitopes by newly synthesized class II proteins without perturbing peptide display by recycling class II molecules, also blocked exogenous BHRF1 presentation by HLA-DR*0401 (Fig. 6). Importantly, none of the above inhibitors had any toxic effect in the range of concentrations tested and, in particular, did not affect recognition of LCLs loaded with the PYY peptide (Fig. 6). These data indicate that presentation of exogenous BHRF1 depended on epitope binding to newly synthesized MHC class II αβ heterodimers en route to the cell surface.
Long-term persistence of BHRF1 Ag after sensitization of BHRF1-KO LCLs with BHRF1 protein
The results of coculture experiments indicated that transfer of BHRF1 in in vitro culture from BHRF1+ to BHRF1− LCLs required at least 2 wk to induce recognition of BHRF1− LCLs by BHRF1-specific CD4+ T cells. This suggested either that productive lytic cycle rarely occurs in vitro or that BHRF1 was released at suboptimal levels and progressively accumulated in latently infected LCL. To test the latter hypothesis, we studied the persistence of BHRF1 Ag after exogenous loading of BHRF1 protein. We sensitized BHRF1-KO LCLs with a unique pulse of BHRF1 protein at 1 μM and assessed recognition of these cells by BHRF1-specific CD4+ T cells 24 h after sensitization and then every 7 days. A significant percentage of sensitized BHRF1-KO LCLs that were shown to have an average doubling time of 3.5 days were still killed by BHRF1-specific T cells 2 wk after a single pulse of BHRF1, indicating that BHRF1 Ags persist for a long time in LCLs (Fig. 7,A, left panel). Similarly, BHRF1-KO LCLs sensitized with one pulse of PYY peptide at 10 μM resulted in long-term sensitization of LCLs, because they were efficiently killed by BHRF1-specific CD4+ T cells at least 8 days after loading (Fig. 7,A, right panel). Furthermore, we tested the effect of repeated daily loading of suboptimal amounts of BHRF1 protein (30 nM) on BHRF1-KO LCLs. One to 15 daily pulses were performed. The data shown in Fig. 7 B indicated that three daily pulses with BHRF1 protein at 30 nM led to optimal recognition of these target cells by BHRF1-specific CD4+ T cells compared with a unique pulse with BHRF1 protein at 500 nM. Taken together, these data suggest that in LCLs maintained in in vitro culture, some cells occasionally enter the lytic cycle, releasing BHRF1 Ags at suboptimal levels, and are then captured and accumulated by latently infected B cells.
Although there is no doubt that EBV-specific T cell responses play a crucial role in immunosurveillance of EBV, the contribution of CD4+ T cells to EBV-specific immunity is still poorly documented. We previously described the killing of latently infected LCLs by CD4+ T cell clones recognizing the EBV lytic protein BHRF1. In this study we provide evidence for cross-presentation of BHRF1 resulting from intercellular transfer of BHRF1 Ag from lytically infected to latently infected LCL cells in culture, thus explaining why latently infected B cells that do not produce EBV lytic proteins were nevertheless efficiently killed by BHRF1-specific T cells.
The consistent killing of DR*0401+ LCLs by BHRF1/DR*0401-specific CD4+ T cells was unexpected, because only a small fraction of LCL cells (<5%) enter the lytic cycle. Expression of BHRF1 during latency has been suggested in the past, because differentially spliced BHRF1-containing mRNAs were detected by Northern blot analysis in latently infected LCLs (35) and in EBV-associated B cell lymphomas (36). However, Kieff et al. (37) reported that neither BHRF1 mRNA nor BHRF1 protein could be detected in latently infected cells. Staining of various LCL cell lines with a BHRF1-specific mAb was shown to be restricted to a few cells, corresponding to those undergoing a lytic cycle (38), and BHRF1 could not be detected in post-transplantation lymphoproliferative disorders (39). Moreover, the recent demonstration that the EBV genome from a latent stage III EBV cell line that expresses all latent genes contains a micro-RNA cluster located within the mRNA of the BHRF1 gene indicates RNA silencing of BHRF1 during latency (40). In the present study we ruled out the possibility that BHRF1 could be expressed at low levels by latently infected B cells by showing that LCLs transformed with a B95.8 mutant virus, incapable of lytic cycle entry after disruption of the BZLF1 gene, were not killed by BHRF1-specific CD4+ T cells.
Although the ability of LCLs to cross-present Ags from cocultured cells has remained controversial (16), these cells can efficiently present soluble Ags on MHC class II (13, 20). Accordingly, our data indicate that BHRF1-KO LCLs can perform efficient uptake of BHRF1 protein and subsequent processing of the DR*0401-restricted BHRF1 epitope, although less efficiently than iDC. This is in agreement with previous data showing that DC are more potent APC than B cells (41). Presentation of the BHRF1 epitope followed the classical MHC class II-restricted pathway of presentation of exogenous Ags (34, 42). Sensitization with the BHRF1 protein, followed by processing, occurred in <1.5 h, and presentation of the epitope was then stable for at least 2 wk. Sensitization of BHRF1-KO LCLs with one pulse of the BHRF1 peptide was also stable for at least 8 days. This suggests that long-term sensitization with the BHRF1 protein is due to the formation of stable MHC class II/BHRF1 peptide complexes. In this respect, binding to class II molecules has been shown to protect the peptide from further degradation during intracellular transit and processing (43), although epitope trimming continues after peptide binding to MHC proteins (44). Coculture experiments indicated that transfer of BHRF1 Ags from BHRF1− to BHRF1+ LCLs required more than a 2-wk coculture to obtain optimal recognition of BHRF1-sensitized LCLs. Most probably, the amount of BHRF1 protein released in the medium within a short period of time (e.g., in few days) is too low to sensitize LCLs for recognition by BHRF1-specific T cells. However, progressive uptake and accumulation of BHRF1 in latently infected LCLs, associated with long-term persistence of BHRF1 Ags may ultimately allow their efficient and persistent recognition by BHRF1-specific T cells. The persistence of peptide/class II MHC complexes has been previously reported (45), and recent data suggest that the kinetic stability of MHC class II/peptide complexes is a key parameter that dictates immunodominance (46). Taken together, our data suggest a mechanism by which persistence of Ag derived from an exogenous protein allows progressive target cell sensitization to recognition by MHC class II-restricted T lymphocytes. Such a mechanism might be particularly relevant in situations where low doses of proteins are chronically released, e.g., in some infectious or oncological processes.
The potent killing of unpulsed EBV-latently infected B cells by BHRF1-specific CD4+ T cell clones described in this study is in sharp contrast with the apparently weak killing potential of CD4+ T cell clones specific to latent EBV Ags. CD4+ T cell recall responses to EBNA1 have been generated by in vitro stimulation of PBL with DC, either infected with vaccinia vector-encoding EBV proteins or loaded with EBV peptide or EBV protein (19, 20, 21). In particular, CD4+ T cell clones specific for the TSL epitope in EBNA1 have been obtained by several groups. Although these clones killed autologous LCL loaded with the cognate Ag in protein or peptide loading assays, their ability to recognize unloaded LCLs has continued to be debated (13, 16, 17, 20, 47). Similarly, CD4+ T cell clones specific to the EBV latent proteins EBNA2, EBNA3A, and EBNA3C exhibited little, if any, killing of autologous LCLs, although they did kill autologous LCLs when EBV Ag expression was enhanced after pulsing with synthetic EBV peptide epitopes (15, 20, 25, 47). We show in this study that all BHRF1-specific CD4+ T cell clones tested killed latently infected LCLs via the perforin/granzyme pathway. This is in accordance with the previous description of an HLA-DR2-restricted, BHRF1-specific CD4+ T cell clone that is able to kill autologous LCLs (23).
Whether cross-presentation of BHRF1 occurs in vivo is not an easy question to answer. If so, it may explain the high frequency of BHRF1/DR*0401-specific CD4+ T cells that we have detected in all EBV-seropositive DR*0401+ donors tested to date (25), suggesting that these cells were efficiently reactivated in vivo. Proteins produced during the lytic cycle, particularly cytoplasmic proteins such as BHRF1, could be released during the lytic cycle and captured by APC. Cross-presentation of BHRF1 could be of particular importance due to the strong inhibition of MHC class II Ag presentation by EBV lytic cycle proteins. It was recently demonstrated that EBV, like other viruses, escape detection by CD4+ T cells by inhibiting induction of the expression of MHC class II genes. In fact, the EBV immediate-early protein BZLF1 was shown to block IFN-γ signal transduction and expression of the MHC CIITA CIITA (48). This viral inhibition is designed to prevent the presentation of endogenous viral Ags, but probably does not affect the presentation of exogenous viral Ags in professional APCs (49). Thus, our data suggest that CD4+ T cells specific to EBV lytic cycle Ags could be important for the control of EBV latency.
We thank Dr. J. Yates (Department of Genetics, Roswell Park Cancer Institute, Buffalo, NY) for kindly providing BHRF1-KO EBV mutant, Dr. E. Olejniczak (Abbott Laboratories, Abbott Park, IL) for helpful suggestions for the production of BHRF1 protein, and Drs. A. Rickinson and G. Taylor (Institute of Cancer Studies, University of Birmingham, Birmingham, U.K.) for helpful discussions.
The authors have no financial conflict of interest.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
This work was supported by grants from the Ligue Nationale Contre le Cancer, Centre Hospitalier Universitaire de Nantes, European Community (Grant QLK2-CT-2001-01205), and institutional grants from Institut National de la Santé et de la Recherche Médicale.
Abbreviations used in this paper: EBNA, EBV-encoded nuclear Ag; BFA, brefeldin A; CMA, concanamycin A; DC, dendritic cell; FasL, Fas ligand; iDC, immature DC; KO, knockout; LCL, EBV-transformed B-lymphoblastoid cell line; mDC, mature DC; WT, wild type.