Immediate-Early Transactivator Rta of Epstein-Barr Virus (EBV) Shows Multiple Epitopes Recognized by EBV-Specific Cytotoxic T Lymphocytes

Abstract

We analyzed the immediate-early transactivator Rta of Epstein-Barr virus (EBV) for its role as a target for specific cytotoxic T lymphocytes (CTL). Panels of overlapping peptides covering the entire amino acid sequence of Rta were synthesized and used to induce and analyze specific CTL responses in EBV-positive donors. Using peptide-pulsed target cells, we found nine different CTL epitopes that are distributed over the entire protein sequence. One epitope restricted by HLA-A24 could be mapped to the decameric sequence DYCNVLNKEF between amino acid positions 28 and 37 of the Rta protein. A second epitope could be assigned to the same region of Rta (residues 25 to 39) and was shown to be restricted by HLA-B18. Another, minimal epitope could be mapped to the nonameric sequence ATIGTAMYK between amino acid positions 134 and 142; this peptide was restricted by HLA-A11. Another four epitopes were proven to be restricted by HLA-A2, -A3, -B61, and -Cw4 and were located between Rta residues 225 and 239, 145 and 159, 529 and 543, and 393 and 407, respectively. For two other epitopes, only the location within the Rta protein is known so far (residues 121 to 135 and 441 to 455); their exact HLA restriction patterns have not yet been identified. Using target cells infected with recombinant vaccinia virus containing the gene for Rta, we showed that six of eight Rta-specific CTL lines recognized the corresponding peptides also after endogenous processing. These data suggest that Rta comprises an important target for EBV-specific cellular cytotoxicity. Together with recent findings of other immediate-early and early proteins also acting as CTL targets, they reveal the role of proteins of the lytic cycle in the immune recognition of EBV-infected cells.

Epstein-Barr virus (EBV) is a ubiquitous human gamma herpesvirus with a wide dissemination in all human populations, with prevalences of more than 90%. The first contact with EBV inevitably results in a lifelong latent infection, with the virus persisting in circulating B lymphocytes (17, 25). Subsequent to primary infection, which may cause infectious mononucleosis or, more often, remain clinically silent, the pathogenic consequences differ considerably between immunocompetent and immunocompromised virus carriers, demonstrating the major impact of the immune system on the control of the EBV infection. While immunocompetent virus carriers in general show no risk of further EBV-associated diseases after primary infection, patients with immunodeficiencies may develop plasma viremia, lymphoproliferative disease, or malignant lymphoma (13, 47). The frequency of lymphoma may also be enhanced by increased plasma viremia (39).

Several components of the immune system contribute to the highly efficient control of virus replication and proliferation of immortalized EBV-infected cells in healthy individuals. NK cells as well as antibody-dependent cellular cytotoxicity mechanisms directed against the viral glycoprotein gp350/220 seem to play a role (24, 43); neutralizing antibodies prevent endogenous reinfection through virus particles released by epithelial cells or lymphocytes. The best-characterized and probably the most important components of these mechanisms, however, are HLA-restricted specific cytotoxic T lymphocytes (CTL) directed against viral gene products of the latent state. These include EBV nuclear antigens 1 to 6 (EBNA1 to EBNA6) and latent membrane proteins 1 and 2 (LMP1 and LMP2). More than 30 distinct CTL epitopes in proteins expressed during latency have been identified so far (5–9, 11, 16, 18, 19, 21, 22, 26, 27, 33). Infected cells expressing these proteins can be eliminated by specific CTL, and uncontrolled proliferation of immortalized cells can be prevented by a multicomponent CTL response. Resting B lymphocytes, however, express solely EBNA1, which cannot be recognized by CTL (23, 28). Resting B lymphocytes are able to switch directly into the lytic cycle without synthesizing further proteins of the latent state (40, 44): this way they escape from being killed by CTL directed against EBNA2 to EBNA6, LMP1, and LMP2. An uncontrolled further progression of the lytic cycle would lead to the production and release of progeny virions and result in endogenous reinfection.

However, considering the rare detection of virus released from the B-cell reservoir as well as the extremely low pathogenicity of latent EBV infection in immunocompetent carriers, it must be assumed that additional control mechanisms which prevent viral replication in peripheral B cells do exist. There is now convincing evidence for immune surveillance mechanisms directed against proteins of the lytic cycle (37). After several authors suggested a putative role of lytic cycle antigens as target structures for EBV-specific CTL (22, 36, 41), our group identified the immediate-early transactivator Zta of EBV as a target for specific CTL (4). These findings were recently confirmed by Steven et al. (42) who, in addition, showed that specific CTL responses are also directed against the immediate-early antigen Rta, encoded by the open reading frame BRLF1, and the early antigens encoded by BMLF1, BMRF1, and BALF2.

In this study, we wanted to analyze the potential role of the second immediate-early protein, Rta, as a target for EBV-specific CTL in more detail. Like Zta, Rta plays an important role during the switch from latency to the lytic cycle. In some cell types, such as epithelial cells, Rta alone is able to disrupt latency; in other cell types, a combination of Rta and Zta induces maximal activation of early viral promoters (3, 48). We identified nine different CTL epitopes distributed over the entire Rta protein sequence. In all nine, the CTL responses were restricted by class I major histocompatibility complex (MHC) molecules; for seven peptide epitopes, the restricting MHC alleles could be determined.

MATERIALS AND METHODS

Cells and viruses.

Lymphoid cells were cultivated in RPMI 1640 growth medium (Gibco BRL, Eggenstein, Germany) containing 10% heat-inactivated fetal calf serum, 2 mM glutamine, and 100 mg of kanamycin or gentamicin per ml. For the determination of the restricting MHC molecules, lymphoblastoid cell lines (LCL) 9007, 9009, 9011, 9022, 9028, 9037, 9049, 9050, 9060, 9074, and 9103 from the 10th International Histocompatibility Workshop panel and with known HLA types were used as allogeneic targets (12). EBV-transformed LCL 002, Koch, 145, 14279, and 15760 were isolated from EBV- and human immunodeficiency virus-positive patients. LCL 101 to 129 were established by culturing peripheral blood lymphocytes (PBL) from healthy EBV-positive donors 101 to 129, respectively, with supernatants from marmoset line B95.8 in the presence of 1 μg of cyclosporine (Sigma, Deisenhofen, Germany) per ml. Phytohemagglutinin P (PHA; Sigma)-activated blasts were established from PBL from donors 101 to 129. They were generated by incubation of PBL for 72 to 150 h in the presence of 5 μg of PHA per ml. HLA types from donor cells were determined by standard serological methods. Recombinant vaccinia viruses R-Vac and Z-Vac expressing Rta and Zta, respectively, under the control of the vaccinia virus 7.5-kDa early/late promoter were described previously (3).

Synthetic peptides.

We used a complete set of 75 peptides encompassing the entire amino acid sequence of the 605-residue-long Rta protein derived from the B95.8 strain of EBV. A total of 74 peptides comprised 15 amino acids, and 1 (the last one) comprised 13 amino acids; all of them overlapped by 7 amino acids. In addition, 8- to 14-amino-acid-long variants of immunogenic peptides Rta_25–39 and Rta_129–143 and the control peptide RIGPGRAFVTIG from the HIV envelope protein (HIV-env) were used. All peptides were synthesized on a model 9050 synthesizer (Milligen, Eschborn, Germany) by use of fluorenylmethyloxycarbonyl chemistry as described elsewhere (31) and purified by high-pressure liquid chromatography (Pharmacia, Uppsala, Sweden).

CTL lines.

EBV-specific CTL lines derived from healthy EBV-positive adults were established as described previously (4). In brief, PBL were purified by density gradient centrifugation with Ficoll-Histopaque (Sigma) and cultivated in T-cell medium (RPMI 1640 growth medium with 10% heat-inactivated human serum [Sigma] and 2 mM glutamine, 1% nonessential amino acids, 2 mM sodium pyruvate, and 10 μg of gentamicin or kanamycin [all from Gibco BRL]). Once a week, T cells were stimulated with peptide-pulsed, irradiated, PHA-activated blasts for the generation of EBV-specific CTL lines at a stimulator/responder ratio of 10:1 and supplemented with 20 U of recombinant human interleukin-2 (Boehringer GmbH, Mannheim, Germany) per ml. For pulsing of stimulator cells, peptides were used at concentrations of 10⁻⁶ M in 5 μl of RPMI 1640. When peptide pools were used for stimulation, the indicated concentrations are for each peptide. After 3 to 4 weeks of stimulation, CTL lines were tested for cytotoxicity.

Cytotoxicity assays.

Cytotoxicity assays were performed as described previously (4). Briefly, 10⁶ autologous or allogeneic target cells were labelled for 1 to 2 h with 0.15 mCi of Na⁵¹CrO₄ (DuPont, Bad Homburg, Germany) and loaded with synthetic peptide for 2 h at a concentration of 4 × 10⁻⁷ M in a final volume of 25 μl (in experiments with peptide pools, concentrations are for each peptide). Then, the target cells were coincubated with different numbers of effector cells suspended in 100 μl of T-cell medium for 4 h in V-bottom 96-well plates. Plates were then centrifuged, and supernatants were harvested for a standard chromium release assay. When vaccinia virus-infected target cells were used, they were infected 12 h prior to labelling at a multiplicity of infection of 10. For every target cell preparation, the expression of Rta protein was controlled by Western blotting with Rta-specific monoclonal antibodies (Viva Diagnostika, Hürth, Germany). Only when the target cells showed clearly visible Rta bands were results analyzed. In the inhibition experiments, monoclonal antibodies W6-32 (Dako Diagnostika, Hamburg, Germany), specific for MHC class I-peptide complexes, and L243, specific for MHC class II molecules (1), were added to the target cells 1 h prior to coincubation with the effector cells (1 μg of monoclonal antibody per ml).

RESULTS

Identification of eight peptides derived from Rta as targets for EBV-specific CTL.

A set of 75 overlapping synthetic peptides representing the entire amino acid sequence of Rta was used to induce and enrich specific CTL. The peptides were divided into seven pools of 10 or 15 peptides each, comprising peptides 1 to 10 (pool 1), 11 to 20 (pool 2), 21 to 30 (pool 3), 31 to 40 (pool 4), 41 to 50 (pool 5), 51 to 60 (pool 6), and 61 to 75 (pool 7). PBL from 16 healthy donors were stimulated weekly with autologous peptide-labelled stimulator cells and supplemented with recombinant human interleukin-2. After three rounds of stimulation, they were tested for cytotoxicity in a standard chromium release assay. As targets, autologous PHA-activated blasts labelled with the corresponding peptide pool were used. Effector/target (E/T) ratios were 30:1 and 6:1. A total of 22 PBL lines from 13 donors showed specific CTL activities; values of specific lysis ranged from 15 to 90%. In order to confine the target structures to single peptides, we tested the 22 Rta pool-specific CTL lines against autologous PHA-activated blasts labelled with individual Rta peptides of the relevant pool. For 11 CTL lines, we were able to define altogether eight single peptides that were specifically recognized (Table 1). For the other 11 CTL lines, recognizing peptides of different pools, no single structure from among the 75 examined synthetic peptides could be unambigously identified as a definite target.

TABLE 1.

Peptides of Rta specifically recognized by CTL lines from seven donors

Peptide	Amino acid positiona	Amino acid sequence	Recognizing CTL line	HLA types of the corresponding donor	% Specific lysisb
Rta-4	25–39	LVSDYCNVLNKEFTA	A-1	A1, A11, B18, B51(5), Cw3, Cw7	27 ± 6
			B-1	A3, A24, B8, B61, Cw2, Cw7	21 ± 2
			C-1	A24, A32, B51, B61	25 ± 4
Rta-16	121–135	FFIQAPSNRVMIPAT	G-1	A1, A24(9), B8, B27, Cw1, Cw3	20 ± 9
Rta-17	129–143	RVMIPATIGTAMYKL	A-2	A1, A11, B18, B51(5), Cw3, Cw7	67 ± 10
Rta-19	145–159	KHSRVRAYTYSKVLG	D-1	A1, A3, B7, B8, Cw4, Cw7	17 ± 3
Rta-29	225–139	RALIKTLPRASYSSH	F-1	A2, B44, Cw4	15 ± 5
Rta-50	393–407	ERPIFPHPSKPTFLP	E-1	A2, A29, B35, B51, Cw4	56 ± 17
Rta-56	441–455	EVCQPKRIRPFHPPG	B-2	A3, A24, B8, B61, Cw2, Cw7	44 ± 12
Rta-67	529–543	QKEEAAICGQMDLSH	B-3	A3, A24, B8, B61, Cw2, Cw7	52 ± 9
			C-2	A24, A32, B51, B61	48 ± 6

^aAmino acid positions of the Rta protein sequence (GenBank accession no. A43043 or A03771). 

^bObtained by subtracting the lysis of target cells labelled with the control peptide HIV-env from the lysis of target cells labelled with Rta peptides. Values above 10% were considered positive. The results of at least three different experiments are presented as means ± standard errors of the means. The E/T ratio in all experiments was 20:1. 

Identification of the HLA elements restricting the peptide-specific CTL responses.

In the following experiments, CTL lines obtained from donors A to G and directed against the peptides indicated in Table 1 were analyzed for HLA restriction. As shown in inhibition experiments with monoclonal antibodies directed against MHC class I (W6-32) and class II (L243) antigens (Fig. 1), the observed responses of these CTL lines were, in all cases, restricted by class I MHC antigens. For all CTL lines examined, specific lysis decreased by more than 70% after preincubation of autologous target cells with W6-32, whereas preincubation with L243 did not change specific lysis values significantly.

FIG. 1.

Inhibition of reactivity of Rta-specific CTL lines from various donors by preincubation of target cells with monoclonal antibodies against MHC class I (W6-32) (lighter bars) and class II (L243) (darker bars) molecules. Autologous target cells were labelled with peptides for 60 min and then incubated with antibody for another 60 min. The antibody concentration was 0.2 ng/μl for CTL lines A-1, A-2, and F-1; in all other cases, 0.03 ng/μl was used. Lysis in the absence of antibodies was as follows: A-1, 15%; A-2, 32%; B-1, 16%; B-2, 35%; B-3, 40%; C-1, 24%; C-2, 46%; D-1, 18%; E-1, 28%; F-1, 12%; and G-1 25%.

For further identification of the restricting MHC class I molecules, panels of allogeneic target cells (PHA-activated blasts and LCL) that shared one or two of their MHC class I alleles with a particular effector CTL line were used. Target cells were labelled with the relevant peptide and tested for specific lysis by the CTL lines in a 4-h standard chromium release assay.

Peptide Rta-4 was recognized by three CTL lines, A-1, B-1, and C-1 (Table 1). As HLA typing did not reveal among the three donors (A, B, and C) a common HLA allele that could be responsible for the presentation of this peptide, we assumed that its amino acid sequence, LVSDYCNVLNKEFTA, contained at least two differentially restricted CTL epitopes. Indeed, Rta-4-specific CTL derived from donors B and C exclusively recognized target cells expressing the HLA-A24 molecule (Table 2), whereas CTL derived from donor A lysed only HLA-B18-positive target cells (Table 2).

TABLE 2.

MHC class I restriction of CTL epitopes identified within the Rta protein sequence

CTL line	Peptidea	Amino acid position	% Specific lysisb in the presence of the following allogeneic MHC class I molecule:
			A1	A2	A3	A11	A24	A29	A32	B7	B8	B18	B27	B35	B44	B51	B61	Cw2	Cw3	Cw4	Cw7
A-1	Rta-4	25–39	0 ± 1			2 ± 1						30 ± 15				2 ± 1			0 ± 1		0 ± 1
A-2	Rta-17	129–135	0 ± 1			65 ± 6						4 ± 2				2 ± 1			0 ± 2		0 ± 2
B-1	Rta-4	25–39			1 ± 2		25 ± 6				0 ± 2						5 ± 2	5 ± 2			1 ± 1
B-3	Rta-67	529–543			0 ± 0		0 ± 1				1 ± 1						31 ± 7	1 ± 2			1 ± 1
C-1	Rta-4	25–39					19 ± 4		2 ± 1							0 ± 1	5 ± 2
C-2	Rta-67	529–543					0 ± 0		0 ± 0							0 ± 0	40 ± 10
D-1	Rta-19	145–159	1 ± 1		20 ± 4					0 ± 1	0 ± 1									2 ± 3	1 ± 1
E-1	Rta-50	393–407		0 ± 0				0 ± 0						5 ± 6						44 ± 11
F-1	Rta-29	225–139		18 ± 6											0 ± 1					0 ± 1

^aTarget peptide used to generate specific CTL lines. 

^bObtained by subtracting the lysis of target cells pulsed with the control peptide HIV-env from the lysis of target cells pulsed with the relevant target peptides. The results are given as means ± standard errors of the means. The E/T ratio in all experiments was 20:1. Restricting elements are shown in bold. 

Two CTL lines (from donors B and C) were both reactive against peptide Rta-67 (Table 1). They both recognized target cells from the other donor; therefore, analysis of MHC class I restriction of this epitope focused on HLA-A24 and HLA-B61. Further experiments with allogeneic targets showed that Rta-67-specific CTL recognized the peptide only in the context of HLA-B61 (Table 2).

Each of the peptides Rta-16, Rta-17, Rta-19, Rta-29, Rta-50, and Rta-56 induced a CTL response in only 1 of the 16 donors tested (Table 1). By using allogeneic target cells sharing one or two of their MHC alleles with an effector CTL line, we were able to define the restricting HLA elements for four of the peptides. Peptides Rta-17, Rta-19, Rta-29, and Rta-50 are presented in the context of the HLA-A11, HLA-A3, HLA-A2, and HLA-Cw4 molecules, respectively (Table 2). For peptides Rta-16 and Rta-56, the restricting MHC class I molecule could not be identified. Table 2 summarizes the HLA restriction results for all peptide epitopes identified within the Rta protein sequence.

Determination of the minimal lengths of peptides Rta-4 and Rta-17 bound to HLA-A24 and HLA-A11, respectively.

CTL lines obtained from donors B and C and stimulated with full-length peptide Rta-4 were tested against HLA-A24-positive target cells labelled with shortened versions of this peptide. Altogether, 14 peptides 8 to 14 amino acids long were analyzed (Table 3). Experiments with both CTL lines suggested that the decameric minimal epitope DYCNVLNKEF was located between amino acid positions 28 and 37 of the Rta protein sequence. Removal of the N-terminal aspartic acid (D) as well as the C-terminal phenylalanine (F) led to a drastic decrease in specific lysis of the labelled target cells. Data were obtained in four different experiments; Table 3 shows representative results. To localize the minimal amino acid sequence of another CTL epitope, CTL line A-1 stimulated with the 15-mer Rta_129–143 was tested with shortened versions of this peptide in a standard chromium release assay. The minimal amino acid sequence seemed to be ATIGTAMYK, located between amino acid positions 134 and 142, since removal of the N-terminal alanine (A) as well as the C-terminal lysine (K) resulted in significant decreases in specific lysis rates (Table 4).

TABLE 3.

Fine mapping of the CTL epitope between amino acid positions 25 and 39 of Rta recognized in the context of HLA-A24

Peptide	Amino acid sequence	% Specific lysisa
Rta25–39	LVSDYCNVLNKEFTA	15
Rta26–39	VSDYCNVLNKEFTA	14
Rta27–39	SDYCNVLNKEFTA	16
Rta28–39	DYCNVLNKEFTA	16
Rta29–39	YCNVLNKEFTA	7
Rta30–39	CNVLNKEFTA	4
Rta31–39	NVLNKEFTA	2
Rta32–39	VLNKEFTA	0
Rta25–38	LVSDYCNVLNKEFT	16
Rta25–37	LVSDYCNVLNKEF	12
Rta25–36	LVSDYCNVLNKE	1
Rta25–35	LVSDYCNVLNK	2
Rta25–34	LVSDYCNVLN	0
Rta25–33	LVSDYCNVL	0
Rta25–33	LVSDYCNV	0

^aObtained by subtracting the lysis of target cells pulsed with the control peptide HIV-env from the lysis of target cells pulsed with peptide Rta_25–39. The E/T ratio was 10:1. 

TABLE 4.

Fine mapping of the CTL epitope between amino acid positions 129 and 143 of Rta recognized in the context of HLA-A11

Peptide	Amino acid sequence	% Specific lysisa
Rta129–143	RVMIPATIGTAMYKL	16
Rta130–143	VMIPATIGTAMYKL	18
Rta131–143	MIPATIGTAMYKL	18
Rta132–143	IPATIGTAMYKL	23
Rta133–143	PATIGTAMYKL	27
Rta134–143	ATIGTAMYKL	37
Rta135–143	TIGTAMYKL	6
Rta129–142	RVMIPATIGTAMYK	12
Rta129–141	RVMIPATIGTAMY	4
Rta129–140	RVMIPATIGTAM	3
Rta129–139	RVMIPATIGT	2

^aObtained by subtracting the lysis of target cells pulsed with the control peptide HIV-env from the lysis of target cells pulsed with peptide Rta_129–143. The E/T ratio was 15:1. 

Comparison of the specific recognition of peptide-labelled target cells and target cells presenting naturally processed peptides.

Peptide-specific CTL were also tested against autologous target cells infected with recombinant vaccinia virus containing the Rta gene (Fig. 2). Six of eight CTL lines recognized the vaccinia virus-infected target cells, thus demonstrating that the same or similar peptides were presented by class I antigens after processing of the whole protein in vivo. Two cell lines, E-1 and B-3, however, did not lyse the vaccinia virus-infected cells. For cell line B-3, directed against peptide Rta-67, we assumed that this effect was due to donor-specific differences in peptide processing or presentation, as the Rta-67-specific cell line C-2 was able to recognize peptide-labelled as well as vaccinia virus-infected autologous target cells. On the other hand, C-2 did not recognize vaccinia virus-infected cells from donor B, although it lysed these cells readily after they were labelled with the corresponding peptide (data not shown).

FIG. 2.

Comparison of specific lysis rates for autologous peptide-labelled target cells (lighter bars) and autologous vaccinia virus-infected target cells (darker bars). Specific lysis of vaccinia virus-infected target cells was obtained by subtracting the lysis of target cells infected with the control recombinant vaccinia virus expressing Zta (Z-Vac) from the lysis of target cells infected with recombinant vaccinia virus expressing Rta (R-Vac), whereas specific lysis of peptide-labelled target cells was obtained by subtracting the lysis of target cells pulsed with the control peptide HIV-env from the lysis of target cells pulsed with the relevant peptide. Values above 10% specific lysis were considered positive. The E/T ratio was 30:1.

DISCUSSION

Strong evidence has been accumulated that B lymphocytes latently infected by EBV are controlled mainly by means of specific CTL (5–9, 16, 18, 19, 21, 22, 26, 27, 33). However, only little information exists about immune surveillance mechanisms directed against EBV-infected B lymphocytes entering the lytic cycle and producing infectious virus. We used synthetic peptides from EBV-encoded lytic cycle proteins to label autologous lymphocytes, which then could serve as stimulator and target cells. With this methodology, we showed that an EBV-encoded protein expressed during the lytic cycle, the immediate-early transactivator Zta, could indeed serve as a target for specific CTL (4, 42).

In this article, we performed a detailed analysis of the specific CTL response directed against the immediate-early antigen Rta encoded by BRLF1. We used a set of 75 overlapping peptides to screen several EBV-positive donors for Rta-specific memory CTL. We found eight peptides distributed over the entire sequence of the molecule and specifically recognized by in vitro-stimulated CTL. Restricting HLA alleles were HLA-A2, -A3, -A11, -A24, -B18, -B61, and -Cw4; thus, as most of these alleles are rather frequent in the Caucasian population (HLA-A2 and -A3 are among the most frequently found) (20), the majority of Caucasians should be able to mount a cellular immune response against this protein. Indeed, 27 of our 29 donors showed at least one of these alleles. Compared to the findings for other proteins that have been identified so far as target structures for CTL against EBV and other viruses, our findings represent a high accumulation of differentially restricted CTL epitopes within the sequence of just one protein.

Interestingly, one of the epitopes found on Rta was restricted by an HLA-C allele (HLA-Cw4). This finding is in accord with our earlier observation of one of the two restriction elements for Zta epitopes being HLA-Cw6 and is also in accord with the findings of Steven et al. (42), who detected 3 HLA-C-restricted epitopes among 11 epitopes derived from lytic cycle antigens. In contrast, for CTL specifically recognizing antigens of latency, the use of HLA-C as a restriction element appears to be rare (42). This finding reflects the fact that HLA-C antigens are expressed on the surface of latently infected LCL in considerably smaller amounts than HLA-A or HLA-B antigens (49), most probably due to rapid degradation of their mRNAs (30) or inefficient assembly with β₂-microglobulin (34). Thus, HLA-C alleles may, at least in special situations, play a role in the presentation of foreign epitopes to specific CTL (9, 46), in the same way as their HLA-A and HLA-B counterparts, in addition to serving as NK cell inhibitors (10) and monitors of the antigen processing machinery (15).

CTL epitopes recognized in vivo are generated by endogenous processing of newly synthesized proteins. To prove that our Rta epitopes were indeed operative in vivo, we stimulated CTL lines with peptide-labelled cells and tested them against autologous target cells infected with recombinant vaccinia virus containing the Rta gene. Eight CTL lines recognizing the seven different epitopes were examined in this way; six lines showed reactivity against the vaccinia virus-infected cells, whereas two (B-3 and E-1, specific for peptides Rta-67 and Rta-50, respectively) obviously were not able to kill vaccinia virus-infected cells. Inappropriate processing of vaccinia virus antigens has been described (2, 45); in such cases, CTL reactivities against a certain protein or epitope can be demonstrated by use of target cells transiently transfected with an expression vector for the protein in question or by application of other, e.g., adenovirus-based, expression systems (32, 42).

Since we found two differentially restricted CTL epitopes on peptide Rta-4, we wanted to localize the precise minimal epitopes within the 15-mer. The decameric amino acid sequence DYCNVLNKEF, located between residues 28 and 37 of Rta, was identified as the minimal epitope for HLA-A24-restricted Rta-4-specific CTL. The sequence characteristics of this peptide correspond well to the recently published sequence motif for HLA-A24 binding peptides (x-Y-N/E/L/M/P/G-D/P-V/I-F-N/Q-E/K-F/L/I) (29). Essential for binding to HLA-A24 are a tyrosine anchor at position 2 and a leucine, isoleucine, or phenylalanine anchor at the C terminus. Furthermore, positions 5 and 6 are preferentially hydrophobic. All of these requirements are fulfilled by our peptide, but in contrast to the published nonameric motif suggesting lysine (K) or glutamic acid (E) for position 8, our Rta peptide contains both of these amino acids. Variations of this kind have been observed and do not seem to necessarily affect the binding capacity of the peptide (14). Characterization of a second minimal epitope showed that the nonameric amino acid sequence ATIGTAMYK at positions 134 to 142 of the Rta protein correlates with the predicted motif for HLA-A11 binding peptides (x-V/I/F/Y-M/L/F/Y/I/A-x-x-x-I/L/Y//V/F-x-K) (38), since our motif contains isoleucine (I) as the anchor amino acid at position 3 and lysine (K) at the C terminus as well as alanine (A) at position 1 and tyrosine (T) at position 2, as predicted by ligand pool sequencing data.

Most investigations of the immunological control of EBV have dealt with cell-mediated immune responses to latent infection. These mechanisms, without doubt, represent a very important part of the immunity to EBV, as they should prevent the outgrowth of EBV-transformed cells and thus the development of lymphoproliferation and lymphoma. However, despite a large amount of data about immune surveillance mechanisms exerted by specific CTL recognizing latently infected cells, investigators are far from a complete understanding of immunity toward EBV infection. Our results highlight a new aspect of cell-mediated immunity against EBV, demonstrating that the cellular immune response against proteins of the lytic cycle is a frequent and possibly important phenomenon. This mechanism may represent a second line of defense against the virus by controlling its replication in vivo. A weakened immune response against lytic infection could explain the recurrence of virus replication and enhanced viral shedding seen in immunocompromised patients. Furthermore, it is tempting to assume that selective defects in the immune response against lytic cycle proteins may play a role in cases of chronic active EBV infection in which active virus replication is suspected (35).