Human CD8+ T Cell Responses to EBV EBNA1: HLA Class I Presentation of the (Gly-Ala)–Containing Protein Requires Exogenous Processing

Immunity, Vol. 7, 791–802, December, 1997, Copyright 1997 by Cell Press

Human CD81 T Cell Responses to EBV EBNA1: HLA Class I Presentation of the (Gly-Ala)–Containing Protein Requires Exogenous Processing Neil Blake,* Steven Lee,* Irina Redchenko,* Wendy Thomas,* Neil Steven,* Alison Leese,* Patty Steigerwald-Mullen,† Michael G. Kurilla,† Lori Frappier,‡ and Alan Rickinson*§ * CRC Institute for Cancer Studies University of Birmingham Birmingham B15 2TA United Kingdom † Department of Pathology and Microbiology University of Virginia Charlottesville, Virginia 22901 ‡ Department of Medical Genetics and Microbiology University of Toronto Toronto, Ontario M5S 1A8 Canada

Summary Epstein-Barr virus (EBV)–induced cytotoxic T lymphocyte (CTL) responses have been detected against many EBV antigens but not the nuclear antigen EBNA1; this has been attributed to the presence of a glycinealanine repeat (GAr) domain in the protein. Here we describe the isolation of human CD81 CTL clones recognizing EBNA1-specific peptides in the context of HLA-B35.01 and HLA-A2.03. Using these clones, we show that full-length EBNA1 is not presented when expressed endogenously in target cells, whereas the GAr-deleted form is presented efficiently. However, when supplied as an exogenous antigen, the fulllength protein can be presented on HLA class I molecules by a TAP-independent pathway; this may explain how EBNA1-specific CTLs are primed in vivo.

Introduction Virus-specific CD81 cytotoxic T lymphocytes (CTLs) recognize short viral peptides presented on the cell surface as a complex with major histocompatibility (MHC) class I molecules. Viral proteins synthesized in the cytosol are degraded, primarily by the proteasome system, and the resultant peptides are translocated into the endoplasmic reticulum (ER) by the transporter associated with antigen processing (TAP). Binding of high-affinity peptides to nascent MHC class I molecules in the ER generates stable MHC class I–peptide complexes which then move via the ER–Golgi network to the cell surface for presentation to the CD81 T cell repertoire (York and Rock, 1996). Various mechanisms have been identified whereby viruses might evade such CTL recognition; some of the more interesting recent examples have come from human herpesviruses of the a and b subfamilies. Herpes § To whom correspondence should be addressed (e-mail: williamsd

@cancer.bham.ac.uk).

simplex virus encodes an immediate-early protein, ICP47, and human cytomegalovirus (HCMV) a late protein, US6, which block peptide transport by interacting with the cytosolic and ER aspects, respectively, of the TAP complex (Fru¨ h et al., 1995; Hill et al., 1995; Ahn et al., 1997; Hengel et al., 1997). In addition, HCMV encodes other ER-resident proteins, one of which (immediate-early US3) acts to retain MHC class I molecules at this site (Ahn et al., 1996; Jones et al., 1996), while others (US11 and US2) subsequently transport the retained molecules back to the cytosol (Wiertz et al., 1996a, 1996b). For both of these viruses, the effect of the processing blockade is to divert the human CTL response toward epitopes derived from virion structural proteins, that is, antigens delivered into the cytosol at the initiation of infection, when the antigen processing pathway is still intact (Riddell et al., 1991; Tigges et al., 1992). With HCMV, such diversion is exaggerated by the ability of the virion matrix protein pp65 to further protect the immediate-early antigen IE1 from processing by posttranslational modification (Gilbert et al., 1996). By contrast, immediate-early and early proteins encoded during lytic replication of Epstein-Barr virus (EBV), the prototype human g herpesvirus, appear to be highly immunogenic to the CTL response (Steven et al., 1997), implying that this agent may have evolved different strategies of immune evasion operating at different times during its life cycle. In this context, EBV has a unique capacity to colonize the host through the virus-driven clonal expansion of latently infected B cells. This latent growth-transforming infection is also strongly immunogenic and elicits CD81 CTL responses directed against a range of latent cycle proteins, most particularly the EBV nuclear antigens EBNAs 3A, 3B, and 3C but also less frequently against EBNA2, EBNA-LP, and the latent membrane proteins LMP1 and LMP2 (Khanna et al., 1992; Murray et al., 1992; Rickinson and Moss, 1997). To date, however, no such responses have been detected against EBNA1, the virus genome maintenance protein and the only viral protein expressed in all in vitro forms of EBV latency and in all virus-associated malignancies (Rickinson and Kieff, 1996). The work of Levitskaya et al. (1995) has suggested that the presence of a glycine-alanine repeat (GAr) domain of more than 200 amino acids within the EBNA1 primary sequence prevents the protein from being presented to the CD81 CTL repertoire. Thus, insertion of an immunodominant epitope from EBNA3B (also known as EBNA4) into the full-length EBNA1 sequence, and insertion of the GAr domain into full-length EBNA3B, in both cases markedly reduced epitope-specific CTL recognition of these chimeric proteins when expressed from recombinant vaccinia vectors. Prompted by these important findings, we began a series of experiments to determine the extent to which this GAr-mediated silencing of EBNA1 as a CTL target antigen is absolute in natural EBV infections. Here we report the isolation of EBNA1 epitope–specific CTLs from three virus-immune individuals; confirm that the GAr domain does block presentation of these epitopes

Immunity 792

Figure 1. Screening of EBV-Specific CTL Clones from Donors RT and IM52 (A) Schematic representation of the fulllength EBNA1 protein and of the glycine– alanine–deleted form E1DGA. The GAr domain (striped box) and nuclear localization signal (nls, shaded boxes) are highlighted. Amino acid coordinates refer to the EBNA1 sequence in the reference EBV strain, B95.8. (B) Autologous target cells, infected overnight with a recombinant vaccinia virus expressing the indicated EBV latent protein, were labeled with (51Cr)O4 and used in standard chromium release assays with CTL clones derived from in vitro LCL stimulation of peripheral blood mononuclear cells from donors RT and IM52. Specific clones are illustrated to indicate the range of CTL responses generated. Results are expressed as percentage specific lysis seen in assays conducted at effector:target ratios of between 2:1 and 5:1.

from an endogenously expressed EBNA1 protein; and identify an alternative pathway whereby the GAr-containing protein can be presented to the CD81 CTL system. Results Isolation of EBNA1-Specific CTL Clones and Identification of Target Epitopes We first constructed a recombinant vaccinia virus that could express a GAr-deleted form of EBNA1 (E1DGA; Figure 1A) from the vaccinia P7.5 promoter. Constructs of this kind have never been obtained with full-length

EBNA1, since in the presence of the GAr domain, the only viable recombinants carry the EBNA1 gene in reverse orientation; vaccinia-mediated expression of the full-length protein could be achieved under the control of a T7 promoter, but this required coinfection with a second vaccinia expressing T7 RNA polymerase (Murray et al., 1992). The availability of the vacc-E1DGA recombinant therefore allowed us to screen for EBNA1-specific CTL reactivities without the requirement for a helper virus and without the potential interference in antigen processing mediated by the GAr domain. In a first series of experiments, memory CTLs were reactivated from healthy EBV-seropositive donors by

CD81 T Cell Responses to EBNA1 793

Figure 2. An E1DGA-Specific CTL Clone from Donor RT Is Restricted through HLA-B35.01 and Recognizes the Minimal Epitope Peptide HPVGEADYFEY (EBNA1 407–417) (A) CTL clone RT c28.90 (recognized as E1DGA-specific in screening assays of the kind illustrated in Figure 1) was tested in a chromium release assay against vacc-E1DGA–infected LCL targets matched at particular HLA class I loci, as indicated. As a control, the LCL targets were infected with a vaccinia vector expressing EBNA3B (vE3B). Note that the range of targets also included the TAP-negative T2:B35.01 cell line either infected with vacc-E1DGA or vE3B as described above, or preexposed to the EBNA1 407–417 epitope peptide at 2 3 1028 M or to DMSO alone as a control. (B) RT c28.90 was tested against autologous LCL cells preincubated with the indicated molar concentration of synthetic peptides from within the EBNA1 region 406–417. The minimal epitope was defined as HPVGEADYFEY (EBNA1 407–417). All results are expressed as percentage specific lysis seen in a standard chromium release assay at an effector:target ratio of 5:1.

stimulating peripheral blood lymphocytes with irradiated cells of the autologous EBV-transformed B lymphoblastoid cell line (LCL), followed by limiting dilution cloning. Individual CD81 T cell clones were then tested for evidence of EBV antigen–specific cytotoxicity by screening on autologous target cells expressing E1DGA or one of the other virus latent proteins from recombinant vaccinia vectors. Significantly, one of the first six healthy donors tested, donor RT (HLA-A2.01, -A24.02, -B27.05, -B35.01, -C2.02, -C4), gave evidence of an E1DGA-specific response. Figure 1B (left column) illustrates the four types of clonal reactivity detectable in the CTL memory of this donor. These included previously identified (Rickinson and Moss, 1997) reactivities to a B35.01-restricted epitope in EBNA3A (c58), a

B27.05-restricted epitope in EBNA 3C (c63), and an A2.01-restricted epitope in LMP2 (c10) plus novel clones, of which c16 is an example, specifically recognizing E1DGA-expressing targets. These E1DGA-specific clones were in a minority, but, overall, such reactivities have been generated from donor RT on three independent occasions by in vitro stimulation with the autologous B95.8 virus–transformed LCL. In a parallel series of experiments, T cell clones were established from the blood of patients with infectious mononucleosis (IM) undergoing primary EBV infection. Again one of six patients studied, IM52 (HLA-A11, -A30, -B13, -B35.01, -C6) yielded clones that recognized E1DGA-expressing targets; representative results are shown for one such clone, c66, in Figure 1B (right column) along with other clones from the same individual recognizing an A11-restricted epitope in EBNA3B (c52), a C6-restricted epitope in EBNA3C (c45), and a rare, as yet poorly characterized, epitope in EBNA-LP (c27) (Steven et al., 1996; N. S., unpublished data). The E1DGA-reactive clones from donor RT and IM52 were then tested in further assays designed to identify their HLA restriction element and viral epitope peptide. In experiments of the kind illustrated in Figure 2A, representative clones from donor RT were assayed on the autologous and on HLA-matched allogeneic LCL targets with or without vaccinia-mediated E1DGA expression. There was no lysis of the EBV-transformed LCLs per se (data not shown) and no lysis after their infection with an irrelevant vaccinia (vacc-EBNA3B); however, vaccE1DGA infection induced specific killing both of the autologous and of B35.01-matched LCLs (Figure 2A), indicating CTL restriction through the HLA-B35.01 allele. To identify the cognate viral epitope, E1DGA-specific clones such as RT c28.90 were then screened against synthetic peptides (15-mers overlapping 10) spanning the entire unique sequence of EBNA1. Using a visual assay of T cell–T cell killing for the initial screen, two overlapping peptides representing amino acids 403–417 and 408–422 of the B95.8 strain EBNA1 sequence appeared to contain the antigenic epitope. Smaller peptides from within this region were then titrated for their ability to sensitize autologous LCL targets to CTL recognition in standard chromium release assays. As shown by the representative results in Figure 2B, this identified the 11-mer peptide EBNA1 407–417 (HPVGEADYFEY) as the minimal epitope sequence. This accords well with the B35.01 consensus motif identified by sequencing eluted peptides, having a proline at position 2 and a tyrosine at the C-terminus (Rammensee et al., 1995). Precisely similar results were obtained using the E1DGA-reactive clones from IM52, indicating that these were also B35.01 restricted and specific for the EBNA1 407–417 epitope (data not shown). Identification of HLAB35.01 as the restriction element allowed us to carry out an additional experiment using as target cells T2:B35.01, an HLA-B35.01 transfectant of the TAP-negative T2 cell line (Salter et al., 1985; Takiguchi et al., 1994); this formally demonstrated that the processing and presentation of endogenously expressed E1DGA was TAP dependent. Thus, RT c28.90 effectors recognized T2:B35.01 cells when preloaded with the EBNA1

Immunity 794

vacc-E1DGA and only if the cells expressed the HLAA2.03 allele (Figure 3A). Screening the panel of EBNA1 peptides in a visual T cell–T cell killing assay first showed that the epitope lay within the amino acid region 568– 586, and subsequent titration assays with smaller peptides from the region identified the 9-mer sequence VLKDAIKDL (EBNA1 574–582) as the minimal epitope (Figure 3B).

Figure 3. E1DGA-Specific CTL Clones from Donor NPC4 Are Restricted through HLA-A2.03 and Recognize the Minimal Epitope Peptide VLKDAIKDL (EBNA1 574–582) (A) CTL clone NPC4 c11 (recognized as E1DGA-specific in screening assays of the kind illustrated in Figure 1) was tested in a chromium release assay against vacc-E1DGA–infected LCL targets matched at particular HLA class I loci as indicated. As a control, the same LCL targets were infected with a vaccinia vector expressing EBNA3C (vE3C). (B) A second E1DGA-specific clone, NPC c16, was tested against autologous LCL target cells preincubated with the indicated molar concentration of synthetic peptides within the EBNA1 region 573– 582 or with an equivalent dilution of DMSO as the ‘no peptide’ control. The minimal epitope was defined as VLKDAIKDL (EBNA1 574–582). All results are expressed as percentage specific lysis seen in a standard chromium release assay at an effector:target ratio of 5:1.

407–417 peptide but not when infected with vaccE1DGA (Figure 2A). In subsequent work, we screened the peripheral blood and/or tumor-infiltrating lymphocytes of three patients with EBV genome–positive nasopharyngeal carcinoma (NPC) for evidence of virus-specific CTL reactivities. Analyzing the T cell clones induced by autologous LCL stimulation in vitro identified one patient, NPC4 (HLAA2.03, -A2.07, -B38, -B46.01, -C1, -C7.02), in whom the tumor-infiltrating lymphocyte population contained an E1DGA-reactive component. Figure 3 presents the relevant data from two such clones, c11 and c16. These again recognized LCL targets only after infection with

Effect of the GAr Domain on the CTL Recognition of Endogenously Expressed Antigen Using the above CTL clones, we were able to test the effect of the GAr domain upon the processing of natural CD81 T cell epitopes within the EBNA1 primary sequence. As illustrated in Figures 4A and 4B, the first experiments of this kind used target LCLs either infected with a control vaccinia (vacc-TK) or infected with vaccE1DGA or coinfected with vacc-E1 (expressing fulllength EBNA1) plus the vacc-T7 helper virus. All of the B35.01-restricted, EBNA1 407-417–specific clones tested from donor RT and IM52 gave strong lysis of vaccE1DGA–infected targets, approaching that seen with target cells preexposed to the synthetic epitope peptide. In addition, these effectors consistently recognized vaccE1/vacc-T7–coinfected targets at levels that, though lower, were always significantly above the background seen for targets infected with the control vaccinia; representative results from one such clone, RT c28, are shown in Figure 4A. Similar results were also consistently observed with the A2.03-restricted CTL clones from NPC4 recognizing the EBNA1 574-582 epitope, although absolute levels of lysis were always lower with these effectors; representative results from two such clones, c11 and c16, are shown in Figure 4B. The problems inherent in using vaccinia vectors to express full-length EBNA1 were highlighted, however, when we checked the EBNA1 species being synthesized in vacc-E1/vacc-T7–coinfected cells. Figure 4C shows an immunoblot of protein extracts from vacc-E1DGA– infected and from vacc-E1/vacc-T7–coinfected cells, probed with a monoclonal antibody (MAb) 1H4–1 recognizing a motif in the unique C-terminal domain of EBNA1. This confirms that vacc-E1DGA expresses a single EBNA1 species whose size, approximately 52 kDa, corresponds to that expected for the GAr-deleted polypeptide. However, the vacc-E1 recombinant expresses a heterogeneous pool of EBNA1 species ranging from full length (approximately 80 kDa) down to a size equivalent to that of E1DGA itself. In light of these findings, we elected to study the CTL recognition of full-length versus GAr-deleted EBNA1 proteins using a different viral vector system in which the presence or absence of a GAr domain did not influence the generation of standard recombinants. Thus replication-deficient recombinant adenoviruses RAdE1 and RAdE1DGA were constructed expressing the relevant EBNA1 sequences from the CMV immediate early promoter. As illustrated in Figure 5C, immunoblotting of infected cell extracts with the MAb 1H4–1 confirmed that RAdE1DGA expressed the 52 kDa GAr-deleted product while RAdE1 (unlike the corresponding vaccinia recombinant) expressed a single polypeptide whose position

CD81 T Cell Responses to EBNA1 795

on the gel was equivalent to that of the full-length EBNA1 protein found in LCL cells. These viruses were then used to sensitize target cells for cytotoxicity assays. Representative results are shown in Figures 5A and 5B from assays involving B35.01-positive LCL targets and EBNA1 epitope–specific effectors RT c28.28 and IM52 c66. Target cells expressing the GAr-deleted form of EBNA1 were lysed almost as strongly as cells preloaded with the epitope peptide. Now, however, target cells expressing full-length EBNA1 as a single species were never recognized above the background levels shown by the control RAd35-infected targets. This pattern of results was obtained on several independent occasions of testing with a number of different CTL clones. CTL Recognition of Exogenously Loaded EBNA1 If full-length endogenously expressed EBNA1 is not processed for presentation to CD81 T cells, there must be another route whereby such epitope-specific responses are elicited in vivo. The fact that exogenously delivered antigens can under certain circumstances be presented via the MHC class I pathway (Rock, 1996) led us to examine this as a possibility. The initial studies were carried out using fragments of EBNA1 expressed in bacteria as fusions with maltosebinding protein (MBP); the fusion protein E1CD contains an EBNA1 fragment (residues 323–450) that includes the B35.01 epitope 407-417, whereas fusion protein E1NX contains two linked EBNA1 fragments (residues 8–39 and 420–641) but lacks the B35.01 epitope. Neither fusion protein carries the GAr domain; indeed it was not possible to express larger GAr-containing fragments of EBNA1 in a bacterial expression system. In an initial experiment B35.01-positive LCL cells were incubated in serum-free medium for 12 hr at 378C in the presence of E1CD, E1NX, or MBP, and then washed extensively, labeled with chromium, and used as targets in a standard cytotoxicity assay; parallel targets from the same LCL were incubated with the synthetic epitope peptide or with dimethyl sulfoxide (DMSO) solvent as a control, and then likewise washed extensively before use in the assay. Overnight (12 hr) incubation with the epitopecontaining E1CD fusion protein clearly sensitized these cells to recognition by EBNA1-specific effectors (Figure

Figure 4. Partial Recognition of vacc-E1/vacc-T7 Coinfected Target Cells by E1DGA-Specific CTL Clones from Donors RT and NPC4 (A) CTL clone RT c28 was tested against a standard B35.01-matched target LCL that had been infected with vacc-E1DGA (vE1DGA), with vacc-TK (vTK2) as a control or coinfected with vacc-E1 (vE1) and vacc-T7 (vT7) viruses. Control targets included the same LCL preexposed to the EBNA1 407–417 epitope peptide at 2 3 1028 M or to DMSO alone. Results are expressed as percentage specific lysis

served at effector:target ratios of 5:1 (solid bars) and 2:1 (striped bars). (B) CTL clones NPC4 c16 (solid bars) and c11 (striped bars) were tested against an A2.03-matched target LCL infected with the same vaccinia recombinants as in (A). Control targets included the same LCL preexposed to the EBNA1 574–582 epitope peptide at 2 3 1028 M or to DMSO alone. Results are expressed as percentage specific lysis observed at an effector:target ratio of 5:1 for both clones. (C) Immunoblot of protein extracts from human fibroblasts either mock-infected or infected at 10 moi with the following recombinant vaccinia viruses: vacc-TK (vTK2), vacc-E1 plus vacc-T7 (vE1/vT7), vacc-E1DGA (vE1DGA), or vacc-E1 alone (vE1). Cells were harvested 20 hr after infection and extracts (105 cell equivalents/track) resolved on a 7.5% SDS-PAGE gel along with reference extracts (10 6 cell equivalents/track) from the EBV-positive B95.8 LCL and from the BJAB cell line as controls. The blot was probed with the EBNA1specific MAb 1H4–1 and developed using enhanced chemiluminescence (Amersham).

Immunity 796

Figure 5. Nonrecognition of RAdE1-Infected Target Cells by E1DGA-Specific CTL Clones from Donors RT and IM52 (A) CTL clone RT c28.28 and (B) CTL clone IM52 c66 were tested against a standard B35.01-matched target LCL that had been infected with recombinant adenoviruses expressing E1DGA (RAdE1DGA), EBNA1 (RAdE1), or the control RAd35. Control targets included the same LCL preexposed to the EBNA1 407–417 epitope peptide at 2 3 1028 M or to DMSO alone. Results are expressed as percentage specific lysis observed at effector:target ratios of 5:1 (solid bars) and 2:1 (striped bars).

6A), with optimal lysis seen at fusion protein concentrations of 1027 M and greater (Figure 6B). Control experiments were then carried out to ensure that this E1CDmediated sensitization reflected active processing of the exogenously added protein and not just the direct binding to B35.01 molecules of preexisting peptide fragments within the fusion protein preparation. Thus when the E1CD-containing medium was spun through a Centricon C10 column with a 10 kDa molecular weight cut-off, the flow-through (which would contain any small peptide contaminants) did not sensitize cells to lysis, whereas the retained protein fraction was as effective as the original preparation (data not shown). Furthermore, varying the time of incubation with the fusion protein before the assay showed that E1CDmediated sensitization required overnight exposure, whereas the synthetic epitope peptide was effective within 45 min (Figure 6C). The fusion protein was not required throughout the whole 12 hr period, however, since a short (2 hr) exposure followed by washing and overnight incubation in fresh medium also sensitized target cells to specific lysis (data not shown). Last, whereas peptide sensitization remained optimal when carried out at 48C, sensitization by E1CD required incubation at 378C (Figure 6C), suggesting that active uptake and processing of the fusion protein were involved. In the final series of assays, again using CTL clones reactive to the B35.01-restricted epitope EBNA1 407417, we sought to determine whether full-length EBNA1 (containing the GAr domain) could be handled as an exogenous antigen and presented on HLA class I molecules as above. For this it was necessary to use proteins expressed in the baculovirus system, namely full-length EBNA1 (b1-641) and two smaller GAr-negative EBNA1 fragments, b330-641 containing the B35.01 epitope and b451-641 lacking the epitope. These preparations were tested in parallel with the E1CD, E1NX, and MBP bacterially expressed proteins for their ability to sensitize a standard B35.01-positive LCL to CTL recognition by overnight incubation. Representative assays using two CTL clones, RT c83 and RT c84, are illustrated in Figure 7A. There was strong sensitization mediated by b330-641, almost to the levels shown by the E1CD fusion protein and by the synthetic epitope peptide itself. More important, the cells preincubated with the full-length b1-641 protein were also consistently recognized, showing levels of lysis lower than seen for b330-641–treated targets but significantly higher than the background values seen for control targets preincubated with b451–641. Sensitization with these baculovirus-expressed proteins required similar conditions to those already identified for E1CD (data not shown). This series of assays was then extended to include the TAP-negative T2:B35.01 cell line as the source of target cells. The pattern of results obtained,

(C) Immunoblot of protein extracts from A549 cells infected at 200 moi with RAdE1, RAdE1DGA, or the control RAd35. Cells were harvested 20 hr after infection and extracts (2 3 10 5 cell equivalents/ track) resolved by 7.5% SDS-PAGE along with B95.8 LCL and BJAB reference extracts (106 cell equivalents/track); the blot was then probed with MAb 1H4–1 as described for Figure 4C.

CD81 T Cell Responses to EBNA1 797

Figure 6. EBNA1 Epitope–Specific CTL Clones Recognize B35.01-Matched LCL Target Cells Exogenously Loaded with an EBNA1/MBP Fusion Protein (A) Target cells were preincubated in 1 ml AIM-V serum free medium for 12 hr at 378C with bacterially expressed MBP or with the MBP fusion proteins E1NX or E1CD at 2 3 1027 M and then washed extensively and used in a chromium release assay with CTL clone RT c28.58 specific for the EBNA1 407– 417 epitope. As controls, target cells were similarly preexposed for 12 hr to the EBNA1 peptide 407–417 at 2 3 10 27 M or to DMSO alone. (B) Target cells were preloaded for 12 hr at 378C with different concentrations of E1CD and then washed extensively and used in a chromium release assay with CTL clone RT c83 specific for the EBNA1 407–417 epitope. Peptide- and DMSO-treated control targets are as in (A). (C) Target cells were preloaded with E1CD at 2 3 10 27 M in 1 ml of AIM-V media for the indicated time period and then washed extensively before being used in a chromium release assay with CTL clone RT c83. As controls, target cells were preexposed for 45 min to the EBNA1 epitope peptide 407–417 at 2 3 10 28 M or to DMSO alone. (D) Target cells were preincubated for 12 hr with MBP or E1CD at 2 3 10 27 M in 1 ml AIM-V media either at 378C or at 48C, before extensive washing and use in a chromium release assay with CTL clone RT c83. As controls, target cells were preexposed for 12 hr at 48C to the EBNA1 epitope peptide 407–417 at 2 3 10 28 M or to DMSO alone. Results throughout are expressed as percentage specific lysis seen at an effector:target ratio of 5:1.

illustrated in Figure 7B, was very similar to that seen with a standard TAP-positive LCL. Thus, preexposure of T2:B35.01 cells to E1CD or to b330-641 efficiently sensitized the cells to the EBNA1 epitope–specific CTL clones RT c83 and RT c28.90. Moreover, the full-length protein b1-641 again mediated significant sensitization, producing levels of lysis clearly above control values. Discussion Earlier work from several laboratories has shown that EBV-induced CTL responses are detectable against most of the eight virus-coded latent cycle proteins but apparently not against EBNA1 (Gavioli et al., 1992; Khanna et al., 1992; Murray et al., 1992; Tamaki et al., 1995). A possible explanation came from Levitskaya et al. (1995), who, working with chimaeric antigens expressed from vaccinia vectors, found that molecules containing the GAr domain were to a large extent protected from presentation via the HLA class I pathway. In the present report we show (1) that CD8 1 HLA class I–restricted CTL responses to EBNA1 epitopes are in fact detectable in humans; (2) that such CTLs recognize only target cells endogenously expressing the GArdeleted and not the full-length form of EBNA1; and (3) that, when delivered as an exogenous antigen, the fulllength protein can be processed and presented on HLA class I molecules via an alternative TAP-independent pathway. When screened for EBNA1-specific reactivities using vacc-E1DGA–infected targets, 3 of 15 individuals tested

gave CD81 CTL clones that recognized a specific combination of EBNA1 epitope peptide and HLA class I–restricting allele (Figures 1–3). Thus, donors RT and IM52 both yielded CTLs recognizing the 11-mer epitope EBNA1 407–417 in the context of B35.01, and NPC4 yielded CTLs recognizing the 9-mer epitope EBNA1 574–582 in the context of A2.03. In the case of the B35.01-restricted CTLs, their inability to recognize vacc-E1DGA–infected cells of the T2:B35.01 line formally proved that the endogenously expressed E1DGA protein was being processed and presented via the conventional TAP-dependent pathway. The circumstantial evidence strongly supports the contention that these are authentic EBVinduced responses and not rare examples of cross-reactivity, that is, CTLs induced by some other agent but fortuitously recognizing an epitope in the EBNA1 sequence. First, it would be remarkable if such crossreactivity were to explain two independent examples of EBNA1 epitope–specific lysis when such a mechanism has not been substantiated for the response to any other EBV target antigen; we have, for instance, never detected cross-reactive recognition of any EBV epitope in autologous LCL-stimulated T cell preparations from EBV-seronegative individuals (Rickinson and Moss, 1997). Furthermore, recognition of the above EBNA1 epitopes was detected down to the 10212 and 10210 M peptide concentrations respectively in titration assays, values that lie within the range of endpoints observed for immunodominant peptide epitopes in other EBV latent cycle proteins (Burrows et al., 1992a; Gavioli et al., 1993; Lee et al., 1995, 1997). Finally, and most significantly,

Immunity 798

Figure 7. Exogenously Loaded EBNA1 Is Processed in a TAP-Independent Manner (A) B35.01-matched LCL target cells were preincubated with either bacterially expressed proteins (MBP, E1NX, or E1CD) or baculovirus-expressed proteins (b451-641, b1-641, or b330-641) at 2 3 10 27 in 1 ml of AIM-V serum for 12 hr at 378C and then washed extensively and used in chromium release assays with CTL clones RT c83 (top) and RT c84 (bottom) specific for the EBNA1 407–417 epitope. (B) TAP-negative T2:B35.01 target cells were preincubated with either bacterially expressed proteins (MBP, E1NX, or E1CD) or baculovirus expressed proteins (b451-641, b1-641, or b330-641), washed, and used in chromium release assays exactly as described for Figure 7A, with CTL clones RT c83 (top) and RT c28.90 (bottom) specific for the EBNA1 407–417 epitope. Results are expressed as percentage specific lysis seen at an effector:target ratio of 5:1. Note that each assay included as control targets the same target line preexposed as described above to the EBNA1 407–417 epitope at 2 3 1028 M or to DMSO alone. nt, not tested.

one of the sources of the CTL clones, IM52, was a patient in the acute phase of primary infection, when the virusinduced CTL response is at its peak (Steven et al., 1996); another was the lymphocytic infiltrate of an EBV-positive malignancy, NPC, in which every tumor cell expresses EBNA1 in the absence of the other EBNAs (Rickinson and Kieff, 1996). Using these EBNA1 epitope–specific CTLs to study processing of the native protein, we observed significant lysis of vacc-E1/vacc-T7 coinfected target cells expressing full-length EBNA1, although at levels much lower than seen for vacc-E1DGA–infected targets (Figure 4). This result was consistent with the original findings on CTL recognition of a vaccinia-expressed reference antigen with or without GAr insertion (Levitskaya et al., 1995) and suggested that the GAr domain can indeed inhibit processing but not completely. However, the experiments are complicated by the finding that, even using plaque-purified stocks of the vacc-E1 recombinant, EBNA1 is expressed in the vaccinia system as a smear of protein species ranging from full-length (80 kDa) down to the size of the GAr-deleted product (52 kDa) (Figure 4C). This smearing is highly unusual and contrasts with the marked stability of the native EBVcoded EBNA1 in vaccinia-infected LCL cells (data not shown). We believe that such results are not a reflection of proteolytic breakdown from the full-length protein, but rather of heterogeneity in the size of the EBNA1 coding sequence within the vacc-E1 virus preparation itself. This heterogeneity is not due to plaque variation, since the same pattern was observed with independent virus isolates even after attempts at further plaque purification (M. G. K., unpublished data), but probably arises through homologous recombination of the vaccinia virus

genome during viral replication (Ball, 1987), resulting in varying lengths of the GAr domain at the nucleic acid level. We could not discount the possibility, therefore, that the low-level lysis of vacc-E1–infected targets stemmed from the processing of vector-encoded EBNA1 molecules with deletions in the GAr domain. Unambiguous results were obtained only upon switching to the recombinant adenoviral system (Grunhaus and Horwitz, 1992), where both full-length and GAr-deleted EBNA1 could be expressed from viral vectors of equivalent construction (Figure 5). These assays now proved that the pesence of the GAr domain in native EBNA1 can indeed offer the endogenously expressed protein efficient protection from the TAP-dependent processing pathway. Such findings nevertheless highlighted a paradox. If the native virus-coded EBNA1 protein is protected from processing, how are EBNA1 epitope–specific responses elicited in vivo? There is no evidence for the existence of GAr-negative strains of EBV in the human population (Hennessy and Kieff, 1983; Falk et al., 1995), and in fact we were able to isolate the resident EBV from two of our three CTL donors, RT and IM52, and in both cases confirm that it encoded a GAr-positive EBNA1 protein of standard size (data not shown). A possible solution was suggested to us by recent work showing that exogenous antigens can be processed for presentation on MHC class I molecules, especially if supplied in a particulate form that encourages their uptake by phagocytosis or macropinocytosis (Kovacsovics-Bankowski et al., 1993; Pfeifer et al., 1993; Ulmer et al., 1994; Norbury et al., 1995; Rock, 1996). We therefore asked whether LCL cells, which are often used as surrogate antigen-presenting cells in in vitro assays, could process exogenously added EBNA1 and present the derived epitopes

CD81 T Cell Responses to EBNA1 799

on HLA class I molecules. The results clearly showed that such processing could occur with exogenously supplied fragments of EBNA1 produced either as a MBP fusion protein in bacteria or from a baculovirus vector (Figures 6 and 7). Numerous controls indicated that presentation required active processing of the exogenous protein by cells and was not attributable to peptide contaminants preexisting in the preparation. More important, a purified preparation of full-length EBNA1 produced in the baculovirus system (b1–641) could also be processed and presented on HLA class I molecules, albeit somewhat less efficiently than the corresponding b330-641 EBNA1 fragment (Figure 7A). The ability to handle full-length EBNA1 clearly distinguishes this pathway of exogenous antigen processing from the conventional route taken by endogenously expressed proteins. A second important distinction between the two pathways became apparent when the experiments were extended to include the T2:B35.01 line as a source of antigen-presenting cells (Figure 7B). This showed that presentation of exogenously loaded EBNA1, whether full-length protein or a fragment, occurred via a TAP-independent pathway. The nature of this pathway remains to be determined, but it is worth noting that other reports have also described the TAPindependent processing of exogenous antigen by LCL cells (Liu et al., 1995; Schirmbeck et al., 1995). Recent work by Levitskaya et al. (submitted) indicates that GArmediated protection from the TAP-dependent processing pathway is mediated at the level of proteasome breakdown, and it will be important to determine whether the TAP-independent pathway identified here involves the proteasome at any stage. We propose that CD81 T cell responses against EBNA1 epitopes are induced in vivo through uptake of the exogenous protein by professional antigen-presenting cells. In this context, there must be a continued supply of EBNA1 protein released in vivo since, of all of the virus latent proteins, EBNA1 is the only one against which antibodies are consistently maintained in healthy virus carriers (Henle et al., 1987), and recent work has also identified EBNA1-specific CD41 T cells in virusimmune individuals (Khanna et al., 1995, 1997). However, we emphasize that, while CD81 CTLs are induced against EBNA1 epitopes, such effectors will not be capable of recognizing naturally infected cells expressing the full-length GAr-positive protein. If, as other work suggests (Tierney et al., 1994; Chen et al., 1995), EBV persists by adopting forms of latency in which EBNA1 is present in the absence of the other EBNAs, protection of the full-length protein from endogenous processing would offer such cells a means of avoiding detection in the immunocompetent host. Experimental Procedures Cell Lines, Antibodies, and Peptides LCLs were established and maintained as described (Steven et al., 1996); the T2:B35.01 line, established by stable transfection of T2 cells (Salter et al., 1985) with an HLA B35.01 cDNA (Takiguchi et al., 1994), was kindly provided by M. Takiguchi (University of Tokyo, Japan). The 293 human kidney (ATCC CRL 1573) and A549 human lung carcinoma (ATCC CCL 185) cell lines were passaged in Dulbecco’s modified Eagle’s medium supplemented with 10% (v/v) fetal

calf serum, 100 IU/ml penicillin, 100 mg/ml streptomycin, and 2 mM Primary human fibroblasts were established from small skin biopsies and passaged in RPMI 1640 medium supplemented as described above. The EBNA1-specific rat MAb 1H4–1 (Gra¨sser et al., 1994), which recognizes a motif in the unique C-terminal domain (F. Gra¨sser, unpublished data), was a kind gift from F. Gra¨ sser (University of Homburg-Saar, Germany). Peptides were synthesized by standard fluorenyl-methoxycarbonyl chemistry (Alta Bioscience, University of Birmingham, United Kingdom) and dissolved in DMSO, and their concentrations determined by biuret assay. L-glutamine.

CTL Cloning EBV-specific CTL clones were derived from healthy EBV-seropositive donors and from patients with IM by autologous LCL stimulation and limiting dilution cloning, using published procedures (Steven et al., 1996; Lee et al., 1997). Samples from patients with NPC were kindly provided by P. Johnson (Department of Clinical Oncology, Prince of Wales Hospital, Shatin, Hong Kong). T cell clones were isolated from peripheral blood lymphocytes as described above for healthy donors, and from tumor infiltrating lymphocytes as follows. The biopsy was finely minced with scalpel blades, and the cells released into suspension were cultured in 96-well round-bottom plates in growth medium containing 30% MLA 144 supernatant and recombinant interleukin-2 (100 IU/ml). After 7 days, cells were refed with this medium and stimulated with autologous LCL (irradiated at 4000 rad; responder-to-stimulator ratio 40:1). Cells were refed again on days 14 and 21 and then cloned by limiting dilution as described above. CTL Assays EBV-specific CTL clones were identified, using a chromium release assay, by their ability to lyse targets expressing individual EBV genes from recombinant vaccinia vectors. Targets were incubated with 50–100 mC of (51Cr)O4 for 90 min, washed, and incubated with CTL at known effector-to-target ratios in a standard 5 hr chromium release assay. The percentage specific lysis was calculated as (release by CTL 2 spontaneous release) 3 100/(total released in 1% SDS 2 spontaneous release). For experiments using recombinant vaccinia viruses, target cells were infected with virus for 90 min at a multiplicity of infection (moi) of 10 and then incubated in growth medium for an additional 15 hr prior to labeling with chromium. For experiments with recombinant adenoviruses, target cells were infected at a moi of 200 for 90 min, followed by incubation in growth medium for 15 hr before labeling with chromium. To identify viral epitopes, EBNA1-specific clones were first tested in a visual T cell–T cell killing assay (Burrows et al., 1992b) using overlapping 14- and 15-mer synthetic peptides representing the entire unique sequence of the EBNA1 protein. Minimal epitopes were mapped in chromium release assays using smaller peptides from within the region of interest. Peptides were added to target cells at known dilutions, using equivalent dilutions of peptide-free DMSO as a control. In cytotoxicity assays involving exogenous loading of protein, unless otherwise stated target cells were incubated at 378C with 2 3 1027 M protein in 1 ml AIM-V serum-free medium (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 90 min before being used in a standard chromium release assay. Construction of Plasmids and Recombinant Viruses E1DGA was generated by polymerase chain reaction (PCR) deletional mutagenesis from the plasmid pBS:E1 (which contains EBV genomic fragment 107,823–110,280 spanning the EBNA1 coding sequence; for nucleotide coordinates see Baer et al., 1984). The 59 and 39 regions flanking the GAr domain were amplified separately using the following primers: E1A: 59-GAATCATGTCTGACGAGGGG CCAG-39 (59 base 107,945) and E1Br: 59-ACCTCCTGCTCCTGTTCC ACGGTG-39 (39 base 108,205), producing a 284bp 59 fragment, and E1C: 59-ACAGGAGCAGGAGGTGGAGGCCGG-39 (39 base 108,936) and E1Dr: 59-GGGGTCCAGTGCTTGGGCCTTCTC-39 (39 base 109,277), producing a 388bp 39 fragment. Both fragments were mixed and reamplified with primer E1A and E1Dr, and the resulting

Immunity 800

657 bp fragment was digested with BspE1 and PflMI and ligated into the similarly digested pBS:E1, generating the plasmid pBS:E1DGA (corresponding to a deletion of EBNA1 codons 92–323). VaccE1DGA was constructed as described previously for recombinant vaccinia viruses expressing EBNAs 2, 3A, 3B, 3C, and LP and LMPs 1 and 2 (Murray et al., 1992). Vacc-E1 expressing full-length EBNA1 and the control vaccinia vacc-TK have been described previously (Murray et al., 1992). RAds expressing EBNA1 and E1DGA were constructed using established protocols (McGrory et al., 1988). A genomic fragment of EBV B95.8 strain spanning the EBNA1 coding region was excised from the plasmid p292 (provided by Bill Sugden, University of Wisconsin, Madison) and blunt-end–cloned into the Xba1 site of plasmid pJEF4 (Floettmann et al., 1996). The EBNA1 gene was then removed as a BamH1 fragment and cloned into the adenovirus transfer plasmid pMC3, generating the plasmid pMCE1. pMC3 is derived from pMV35 (Wilkinson and Akrigg, 1992) by excision of the lacZ gene. The E1DGA coding sequence was removed from pBS:E1DGA as a BstY1/HincII fragment, inserted into the Sma1 site of p1813 (Kay and McPherson, 1987), then excised as a BamH1 fragment, and cloned into pMC3, generating the plasmid pMC-E1DGA. Plasmids pMC-E1 and pMC-E1DGA were transfected into 293 cells along with pJM17 (Microbix Biosystems), and recombinant adenoviruses expressing either full-length EBNA1 (RAdE1) or E1DGA (RAdE1DGA) were rescued. Recombinant adenoviruses were plaque purified, propagated, and titered on 293 cells. The recombinant adenovirus, RAd35, containing the lacZ gene, has been described previously (Wilkinson and Akrigg, 1992). Bacterially Expressed EBNA1 Fusion Constructs The pMAL:E1NX deletion mutant consisting of an in-frame fusion of MBP with two linked EBNA1 fragments (residues 8–39 and 420– 641) has been described previously (Khanna et al., 1997). The resulting 637–amino acid fusion protein has a predicted molecular weight of 68.3 kDa. The pMAL:E1CD deletion mutant was constructed by PCR amplification of the EBNA1 sequence corresponding to amino acid positions 323–450, using the primers E1C and E1Dr. The 388 bp PCR product was cloned into the Sma1 site of the pMal-c expression vector producing an in-frame fusion of MBP with EBNA1 residues 323–450 and having a predicted molecular weight of 54.7 kDa. Fusion proteins were expressed in Escherichia coli BL 21 strain after transformation with pMal:E1NX, pMal:E1CD, or the pMal control plasmid and induction with IPTG (isopropyl-b-Dthiogalactopyranoside). Expression of appropriately sized proteins was confirmed by Coomassie staining and immunoblotting of 10% SDS–PAGE (polyacrylamide gel electrophoresis) gels. For preparative purposes, 200 ml cultures of Rich media supplemented with ampicillin were inoculated with lawns of bacterial clones that were grown under ampicillin selection, and recombinant protein expression was induced for 3 hr at 378C by the addition of 0.3 mM IPTG to cultures with an optical density at 550 nm (OD550) of between 0.5 and 1. Bacterial cell lysates were produced by incubation in 10 ml of lysis buffer (1 mg/ml lysozyme, 50 mM Tris, 100 mM NaCl, 10 mM EDTA, 1 mM EGTA, 1 mM PMSF [phenylmethylsulfonyl fluoride] [pH 8.1]) on ice for 15 min followed by a freeze–thaw cycle and brief sonication. The lysates were clarified by centrifugation at 38,000 3 g, 48C, for 30 min, and purified over columns containing a 3 ml bed volume of amylose resin, according to the manufacturer’s protocols with minor modification (New England Biolabs, Beverly, MA). Recombinant proteins were eluted with 10 ml of elution buffer (10 mM maltose, 150 mM NaCl, 20 mM HEPES [pH 7.4]) and collected in 1 ml fractions. Fractions containing protein, as determined by Bradford protein assay, were pooled and quantitated by Bradford assay and 10% SDS-PAGE analysis. Baculovirus Expressed EBNA1 Protein Construction and purification of baculovirus expressed EBNA1 fragments 330–641 (b330-641) and 451–641 (b451-641) have been described previously (Goldsmith et al., 1993). Baculovirus-expressed full-length EBNA1 (b1–641) was generated by excision of the EBNA1 coding sequence from plasmid p291 (provided by Bill Sugden, University of Wisconsin) and blunt-end–cloned into the BamH1 site of the baculovirus transfer plasmid pVL941. pVL941 containing fulllength EBNA1 was kindly supplied by Carl Schildkraut (Albert Einstein School of Medicine, New York, NY). A recombinant baculovirus

was generated using the BaculoGold system (Pharmingen, San Diego, CA) and EBNA1 protein expressed and purified as described previously (Goldsmith et al., 1993). All preparations were checked by Coomassie staining of SDS-PAGE gels and shown to contain a single protein species of the expected size with no detectable breakdown products. Acknowledgments This work was funded by the Cancer Research Campaign and by the Medical Research Council, United Kingdom. S. L. is supported by a Medical Research Council Career Development Award. Received August 12, 1997; revised October 20, 1997. References Ahn, K., Angulo, A, Ghazal, P., Peterson, P.A., Yang, Y., and Fru¨h, K. (1996). Human cytomegalovirus inhibits antigen presentation by a sequential multistep process. Proc. Natl. Acad. Sci. USA 93, 10990–10995. Ahn, K., Gruhler, A., Galocha, B., Jones, T.R., Wiertz, E.J.H.J., Ploegh, H.L., Peterson, P.A., Yang, Y., and Fru¨h, K. (1997). The ER-luminal domain of the HCMV glycoprotein US6 inhibits peptide translocation by TAP. Immunity 6, 613–621. Baer, R., Bankier, A.T., Biggin, M.D., Deininger, P.L., Farrell, P.J., Gibson, T.G., Hatfull, G., Hudson, G.S., Satchwell, S.C., Se´guin, C., et al. (1984). DNA sequence and expression of the B95-8 EpsteinBarr virus genome. Nature 310, 207–211. Ball, L.A. (1987). High-frequency homologous recombination in vaccinia virus DNA. J. Virol. 61, 1788–1795. Burrows, S.R., Rodda, S.J., Suhrbier, A., Geysen, H.M., and Moss, D.J. (1992a). The specificity of recognition of a cytotoxic T lymphocyte epitope. Eur. J. Immunol. 22, 191–195. Burrows, S.R., Suhrbier, A., Khanna, R., and Moss, D.J. (1992b). Rapid visual assay of cytotoxic T-cell specificity utilizing synthetic peptide induced T-cell-T-cell killing. Immunology 76, 174–175. Chen, F., Zou, J.Z., di Renzo, L., Winberg, G., Hu, L.F., Klein, E., Klein, G., and Ernberg, I. (1995). A subpopulation of latently EBVinfected normal B cells resemble Burkitt lymphoma (BL) cell in expressing EBNA1 but not EBNA2 or LMP1. J. Virol. 69, 3752–3758. Falk, K., Gratama, J.W., Rowe, M., Zou, J.Z., Khanim, F., Young, L.S., Oosterveer, M.A.P., and Ernberg, I. (1995). The role of repetitive DNA sequences in the size variation of Epstein-Barr virus (EBV) nuclear antigens and the identification of different EBV isolates using RFLP and PCR analysis. J. Gen. Virol. 76, 779–790. Floettmann, J.E., Ward, K., Rickinson, A.B., and Rowe, M. (1996). Cytostatic effect of Epstein-Barr virus latent membrane protein1 analyzed using tetracycline-regulated expression in B cell lines. Virology 223, 29–40. Fru¨h, K., Ahn, A., Djaballah, H., Sempe´, P., van Endert, P.M., Tampe´ , R., Peterson, P.A., and Yang, Y. (1995). A viral inhibitor of peptide transporters for antigen presentation. Nature 375, 415–418. Gavioli, R., de Campos-Lima, P.O., Kurilla, M.G., Kieff, E., Klein, G., and Masucci, M.G. (1992). Recognition of the Epstein-Barr virusencoded nuclear antigen EBNA4 and EBNA6 by HLA-A11-restricted cytotoxic T lymphocytes: implications for down-regulation of HLAA11 in Burkitt lymphoma. Proc. Natl. Acad. Sci. USA 89, 5862–5866. Gavioli, R., Kurilla, M.G., de Campos-Lima, P.O., Wallace, L.E., Dolcetti, R., Murray, R.J., Rickinson, A.B., and Masucci, M.G. (1993). Multiple HLA A11-restricted cytotoxic T-lymphocyte epitopes of different immunogenicities in the Epstein-Barr virus-encoded nuclear antigen 4. J. Virol. 67, 1572–1578. Gilbert, M.J., Riddell, S.R., Plachter, B., and Greenberg, P.D. (1996). Cytomegalovirus selectively blocks antigen processing and presentation of its immediate-early gene product. Nature 383, 720–722. Goldsmith, K., Bendell, L., and Frappier, L. (1993). Identification of EBNA1 amino acid sequences required for the interaction of the functional elements of the Epstein-Barr virus latent origin of DNA replication. J. Virol. 67, 3418–3426.

CD81 T Cell Responses to EBNA1 801

Gra¨sser, F.A., Murray, P.G., Kremmer, E., Klein, K., Remberger, K., Feiden, W., Reynolds, G., Niedobitek, G., Young, L.S., and MuellerLantzsch, N. (1994). Monoclonal antibodies directed against the Epstein-Barr virus-encoded nuclear antigen 1 (EBNA1): immunohistologic detection of EBNA1 in the mailignant cells of Hodgkin’s disease. Blood 84, 3792–3798. Grunhaus, A., and Horwitz, M.S. (1992). Adenoviruses as cloning vectors. Semin. Virol. 3, 237–252. Hengel, H., Koopmann, J-O., Flohr, T., Muranyi, W., Goulmy, E., Ha¨mmerling, G.J., Koszinowski, U.H., and Momburg, F. (1997). A viral ER-resident glycoprotein inactivates the MHC-encoded peptide transporter. Immunity 6, 623–632. Henle, W., Henle, G., Anderson, J., Ernberg, I., Klein, G., Horwitz, C. A., Marklund, G., Rymo, L., Wellinder, C., and Straus, S.E. (1987). Antibody responses to Epstein-Barr virus-determined nuclear antigen (EBNA)-1 and EBNA 2 in acute and chronic Epstein-Barr virus infection. Proc. Natl. Acad. Sci. USA 84, 570–574. Hennessy, K., and Kieff, E. (1983). One of two Epstein-Barr virus nuclear antigens contains a glycine-alanine copolymer domain. Proc. Natl. Acad. Sci. USA 80, 5665–5669. Hill, A., Jugovic, P., York, I., Russ, G., Bennink, J., Yewdell, J., Ploegh, H., and Johnson, D. (1995). Herpes simplex virus turns off the TAP to evade host immunity. Nature 375, 411–415.

Murray, R.J., Kurilla., M.G., Brooks, J.M., Thomas, W.A., Rowe, M., Kieff, E., and Rickinson, A.B. (1992). Identification of target antigens for the human cytotoxic T cell response to Epstein-Barr virus (EBV); implications for the immune control of EBV-positive malignancies. J. Exp. Med. 176, 157–168. Norbury, C.C., Hewlett, L.J., Prescott, A., Shastri, N., and Watts, C. (1995). Class I MHC presentation of exogenous soluble antigen via macropinocytosis in bone marrow macrophages. Immunity 3, 783–791. Pfeifer, J.D., Wick, M.J., Roberts, R.L., Findlay, K., Normark, S.J., and Harding, C.V. (1993). Phagocytic processing of bacterial antigens for class I MHC presentation to T cells. Nature 361, 359–362. Rammensee, H-G., Fride, T., and Stenvanovic, S. (1995). MHC ligands and peptide motifs: first listing. Immunogenetics 41, 178–228. Rickinson, A.B., and Kieff, E. (1996). Epstein-Barr Virus. In Fields Virology, B.N. Fields, D.M. Knipe, P.M. Howley, R.M. Chanock, J.L. Melnick, T.P. Monath, B. Roizman, and S.E. Straus, eds. (Philadelphia: Lippincott-Raven Publishers), pp. 2397–2446. Rickinson, A.B., and Moss, D.J. (1997). Human cytotoxic T lymphocyte responses to Epstein-Barr virus infection. Annu. Rev. Immunol. 15, 405–431. Riddell, S.R., Rabin, M, Geballe, A.P., Britt, W.J., and Greenberg, P.D. (1991). Class I MHC-restricted cytotoxic T lymphocyte recognition of cells infected with human cytomegalovius does not reduire endogenous viral gene expression. J. Immunol. 146, 2795–2804.

Jones, T.R., Wiertz, E.J.H.J., Sun, L., Fish, K.N., Nelson, J.A., and Ploegh, H.L. (1996). Human cytomegalovirus US3 impairs transport and maturation of major histocompatibility complex class I heavy chains. Proc. Natl. Acad. Sci. USA 93, 11327–11333.

Rock, K.L. (1996). A new foreign policy: MHC class I molecules monitor the outside world. Immunol. Today 17, 131–137.

Kay, R., and McPherson, J. (1987). Hybrid pUC vectors for the addition of new restriction enzyme sites to the ends of DNA fragments. Nucleic Acids Res. 15, 2778.

Salter, R.D., Howell, D.N., and Cresswell, P. (1985). Genes regulation HLA class-I antigen expression in T-B lymphoblast hybrids. Immunogenetics 21, 235–246.

Khanna, R., Burrows, S.R., Kurilla., M.G., Jacob, C.A., Misko, I.S., Sculley, T.B., Kieff, E., and Moss, D.J. (1992). Localization of EpsteinBarr virus cytotoxic T cell epitopes using recombinany vaccinia: implications for vaccine development. J. Exp. Med. 176, 169–176.

Schirmbeck, R., Melber, K., and Reimann, J. (1995). Hepatitis B virus small surface antigen particles are processed in a novel endosomal pathway for major histocompatibility complex class I-restricted epitope presentation. Eur. J. Immunol. 25, 1063–1070.

Khanna, R., Burrows, S.R., Steigerwald-Mullen, P.M., Thomson, S.A., Kurilla., M.G., and Moss, D.J. (1995). Isolation of cytotoxic T lymphocytes from healthy seropositive individuals specific for peptide epitopes from Epstein-Barr virus nuclear antigen 1: implications for viral persistence and tumor surveillance. Virology 214, 633–637.

Steven, N.M., Leese, A.M., Annels, N.E., Lee, S.P., and Rickinson, A.B. (1996). Epitope focusing in the primary cytotoxic T cell response to Epstein-Barr virus and its relationship to T cell memory. J. Exp. Med. 184, 1801–1813.

Khanna, R., Burrows, S.R., Steigerwald, P.M., Moss, D.J., Kurilla, M.G., and Cooper, L. (1997). Targeting Epstein-Barr virus (EBV) nuclear antigen 1 (EBNA1) through the class II pathway restores immune recognition by EBNA1-specific CD41 cytotoxic T lymphocytes: An evidence for HLA DM-independant processing. Int. Immunol. 9, 1537–1543. Kovacsovics-Bankowski, M., Clark, K., Benacerraf, B., and Rock, K.L. (1993). Efficient major histocompatibility complex class I presentation of exogenous antigen upon phagocytosis by macrophages. Proc. Natl. Acad. Sci. USA 90, 4924–4964. Lee, S.P., Morgan, S., Skinner, J., Thomas, W.A., Jones, S.R., Sutton, J., Khanna, R., Whittle, H.C., and Rickinson, A.B. (1995). EpsteinBarr virus isolates with the major HLA B35.01-restricted cytotoxic T-lymphocyte epitope are prevalent in a highly B35.01-positive African population. Eur. J. Immunol. 25, 102–110.

Steven, N.M., Annels, N.E., Kumar, A., Leese, A.M., Kurilla, M.G., and Rickinson, A.B. (1997). Immediate early and early lytic cycle proteins are frequent targets of the Epstein-Barr virus-induced cytotoxic T cell response. J. Exp. Med. 185, 1605–1617. Takiguchi, M., Kawaguchi, G., Sekimata, M., Hiraiwa, M., Kariyone, A., and Takamiya (1994). The role of the conserved residue in the pocket A and the polymorhic residue in pocket E of HLA-B35.01 in presentation of human minor histocompatibility peptides to T cells. Int. Immunol. 6, 1345–1352. Tamaki, H., Beaulieu, B.L., Somasundaran, M., and Sullivan, J.L. (1995). Major histocompatibility complex class-I restricted cytotoxic T lymphocyte response to Epstein-Barr virus in children. J. Infect. Dis. 172, 739–746. Tierney, R.J., Steven, N., Young, L.S., and Rickinson, A.B. (1994). Epstein-Barr virus latency in blood mononuclear cells: analysis of viral gene transcription during primary infection and in the carrier state. J. Virol. 68, 7374–7385.

Lee, S.P., Tierney, R.J., Thomas, W.A., Brooks, J.M., and Rickinson, A.B. (1997). Conserved CTL epitopes within EBV latent membrane protein 2: a potential target for CTL-based tumor therapy. J. Immunol. 158, 3325–3334.

Tigges, M.A., Koelle, D., Hartog, K., Sekulovich, R.E., Corey, L., and Burke, R.L. (1992). Human CD81 herpes simplex virus-specific cytotoxic T-lymphocyte clones recognize diverse virion protein antigens. J. Virol. 66, 1622–1634.

Levitskaya, J., Coram, M., Levitsky, V., Imreh, S., Steigerwald-Mullen, P.M., Klein, G., Kurilla., M.G., and Masucci, M. (1995). Inhibition of antigen processing by the internal repeat region of the EpsteinBarr virus nuclear antigen-1. Nature 375, 685–688.

Ulmer, J.B., Donnelly, J.J., and Liu, M.A. (1994). Presentation of an exogenous antigen by major histocompatibility complex class I molecules. Eur. J. Immunol. 24, 1590–1596.

Liu, T., Zhou, A., Overall, C., Lederer, E., Ljunggren, H-G., and Jondal, M. (1995). Heat-inactivated Sendai virus can enter multiple MHC class I processing pathways and generate cytotoxic T lymphocyte responses in vivo. J. Immunol. 154, 3147–3155. McGrory, W.J., Bautista, D.S., and Graham, F.G. (1988). A simple technique for the rescue of early region I mutations into infectious human adenovirus type 5. Virology 163, 614–617.

Wiertz, E.J.H.J., Jones, T.R., Sun, L., Bogyo, M., Geuze, H.J., and Ploegh, H.L. (1996a). The human cytomegalovirus US11 gene product dislocates MHC class I heavy chains from the endoplasmic reticulum to the cytosol. Cell 84, 769–779. Wiertz, E.J.H.J., Tortorella, D., Bogyo, M., Yu, J., Mothes, W., Jones, T. R., Rapoport, T.A., and Ploegh, H.L. (1996b). Sec61-mediated transfer of a membrane protein from the endoplasmic reticulum to the proteasome for destruction. Nature 384, 432–438.

Immunity 802

Wilkinson, G.W., and Akrigg, A. (1992). Constitutive and enhanced expression from the CMV major IE promoter in a defective adenovirus. Nucleic Acids Res. 20, 2233–2239. York, I.A., and Rock, K.L. (1996). Antigen processing and presentation by the class I major histocompatibility complex. Annu. Rev. Immunol. 14, 369–396. Note Added in Proof The data cited as Levitskaya et al., submitted, have now been published: Levitskaya, J., Sharipo, A., Leonchiks, A., Ciechanover, A., and Masucci, M.G. (1997). Inhibition of ubiquitin/proteasomedependent protein degradation by the Gly-Ala repeat domain of the Epstein-Barr virus nuclear antigen 1. Proc. Natl. Acad. Sci. USA 94, 12616–12621.